Bloom reported a strong Q3 with revenue beating estimates by more than $90 million, strong margin expansion including GAAP operating margin shifting to positive territory. However, the reason the stock surged after hours was not found in the earnings report, rather the stock was up as much as 20% after hours from management commentary on FY2026 during the earnings call.
Bloom Energy has seen incredibly strong price action this year of nearly 400% YTD and is nearly 600% off the April lows. Therefore, it would take something unexpected to get the stock to soar after an earnings report – yet management delivered exactly that by stating: “As we have previously announced, we are doubling our capacity to 2 gigawatts by December 2026, which will support about 4x our 2025 revenue. That expansion is all systems go. Bloom's capacity will not be a bottleneck for our customers.”
Although management stated “about 4X our 2025 revenue,” a conservative approach would be to assume the revenue will be recognized across both 2026 and 2027. Analyst estimates are for $1.9B in 2025 and a mere uptick to 16.6% growth next year for revenue of $2.2B. However, helping the bull case further that last night’s comments offer alpha is that Bloom detailing further their primary benefit, which is quick time to power. Therefore, I imagine some of the “4X from 2025 revenue” from doubling the capacity will occur in 2026 and some will occur in 2027. I’m not choosy on which exact quarter given there is a wide disconnect in analyst estimates as 2027 estimates are for $3.5B, or less than 2X 2025 revenue.
The 2GW represents 100% growth in capacity as AEP had contracted for 1GW going into this year. My interpretation is that Bloom Energy must be able to charge more for its power given how rapid their solid oxide fuel cells are deployed with expectations of 90 days, yet they actually delivered in an astonishing 55 days for Oracle. In the past, management had hinted they were doubling GWs but did not correlate it to a quadrupling of their revenue. From Q2 earnings call: “Now our robust product has robust demand. We will double our factory capacity from 1 gigawatt a year now to 2 gigawatts a year by the end of next year. Our mission has never felt more urgent, and we are ready.”
Even with this blockbuster statement of “about 4x our 2025 revenue,” one has to wonder if 2GW is the baseline for 2026 capacity. The company counts one massive energy partner Brookfield, two hyperscalers and one neocloud as customers (ORCL, AWS via AEP and CRWV) plus they hinted of a fourth large customer on the call today via a gas company partnership. Secondly, management explained why their product is in high demand and will likely remain that way for some time. We cover this and more below!
Please note, we are shifting our post-earnings formatting to have the contextual information including Q&A commentary first and financials second.
“About 4X our FY2025” – The Comment that Caused the Stock to Surge
I want to double click on the comment that 2GW will be “about 4X our FY2025 revenue.” Surprisingly, the first few analysts on the call breezed over the comment before an Evercore analyst asked for clarity. I’m quoting this in full given the importance of the comment in the opening remarks with the CEO confirming the analyst’s understanding and stating, “we will never be the constraint in our customers growing their data center“ and that “all systems are go.” My translation is that Bloom can increase its capacity faster than many other energy options and now that “lighthouse” customers such as Oracle have taken note, that we will see what Bloom is truly capable of in the coming quarters. As Bloom’s CEO stated, commercial demand is “accelerating.”
Here is a further breakdown on the 4X comment:
“Nicholas Amicucci
Evercore ISI Institutional Equities, Research Division
I just wanted to build upon on kind of the doubling of capacity by the end of 2026 and kind of the commentary that would support 4x the fiscal '25 revenue.
How should we think about kind of the utilization on that capacity as we kind of enter into — again, as we enter into 2027 and we have that — the 2 gigawatts kind of up and running. I mean because if we're exploring opportunities to go beyond that 2 gigawatts, it seems like 4x full year '25 revenue, that seems like a big number that we could get there relatively quickly. So I just wanted to parse that out a little bit.
K. Sridhar
Co-Founder, CEO & Chairman
Yes. So here is a simple way to think about it, right? We didn't get to where we are today to deliver what I just explained, this purpose-built factory based on just meeting a market demand as we see it right now, we just prepared ourselves. What is the beauty of Bloom being able to expand its capacity and offer what we do? Is the return on investment like invested capital? So we are fiscally very disciplined, and we only make decisions based on that added cost and its absorption, will it have a great rate of return.
So we have a very disciplined process on this. And on top of that, we have a very clear understanding right now given time to power shortages and the importance of this as a nation-state issue for AI.
We are committing to strive and work as hard as we need to and stay ahead such that we will never be the constraint to our customer on growing their data center. That's what we are positioned for. And we will increase capacity. We will increase it in whatever steps necessary as we see fit. But as you saw, this 2-gigawatt capacity, all systems go based on that.
Would we use it for peak capacity? When we use it, will we use it for steady capacity? All that, you'll hear from us as we talk about our backlog and other things next year. But we are now using our OpEx wisely to invest in capability and talent to think about how do we expand beyond 2 gigawatts. That's all I can say right now. Thanks for that question.”
Regarding how fast Bloom Energy can build the additional GW, management was confident they can do so faster than anyone else: “Today, we are able to provide our power faster than most of the others who have supply chain constraints. We can expand our capacities a lot faster than anybody else.”
Why Bloom Energy Remains in High Demand Amidst a Crowded Energy Industry
We’ve covered Bloom’s products extensively, yet it doesn’t hurt to refresh our understanding of what makes the company stand out given the market dynamics around how data centers are scrambling to secure power is shifting nearly daily.
Our primary message has been “time to power” for Bloom, which management emphasized stating: “We are going to strive to make sure we are able to provide power for our customers before they are ready for it. We will not be the bottleneck. And we designed our factories; we built it with that in mind.”
Looking beyond speed, management also focused on price-to-performance, especially when they were asked how Bloom plans to compete with small-scale gas turbines with management stating a clear benefit to their solid oxide fuel cells (SOFC) is that hyperscalers can put out a lot more tokens with their fuel cells, stating: “With the same amount of gas that's available, same amount of space that's available, we can produce a lot more tokens for the hyperscaler than any other technology can today, end-to-end. And so the value for a hyperscaler is not about the cost of power. It's about that cost of the entire value chain across the board. So price-performance ratio, we can compete with anybody.” This was reiterated in the opening comments with management stating their fuel cells produce “10x more power in the same footprint than they did 10 years ago.”
It was also discussed that mechanical combustion solutions require a lot of batteries, whereas Bloom does not require batteries, implying that gas turbines are a band-aid solution compared to SOFCs, which can provide offer steady output without the grid or batteries.
Updates on Bloom’s Deals with Brookfield, Oracle, Hints of New Hyperscaler Customer
Earlier this month, Bloom shares surged more than 26% to $110 on the backs of a partnership with Brookfield, which will see the asset management firm invest up to $5 billion in Bloom’s fuel cell tech to be deployed at data centers worldwide. While timelines are rather unclear for deployment of the fuel cells, the two state that the partnership does include an AI inference focused site in Europe that will be announced before year-end. To put in perspective the potential size of the partnership, this would represent nearly 3x of Bloom’s estimated revenue for fiscal 2025.
Under the partnership, Bloom will become Brookfield’s “preferred on-site provider for Brookfield's trillion-dollar infrastructure portfolio of AI factories, data center operators, corporate facilities and factories.” What makes this partnership important is not only the fact that Brookfield has invested $50 billion towards AI and “is tripling the size of its AI strategy over the next 3 years,” but that Brookfield is willing to finance for Bloom.
CEO KR Sridhar explained that “if there are Bloom-sourced deals that require financing, so we can offer a customer a PPA, they are willing to step in and be the financier for that. It’s an inaugural investment.” This would remove an important layer on the equation for growth for Bloom as it would help them accelerate deployments without them or customers bearing the capital or financing risks.
Management also hinted of another hyperscaler customer in the works, but declined to provide any further details: “We signed our first deal with a major gas provider who will convert its gas to electricity with Bloom fuel cells and sell that on-site power to a third hyperscaler. The hyperscaler will announce details of this installation when it is ready.”
Bloom also struck a deal with Oracle back in July to deploy its fuel cells for onsite power at select Oracle Cloud Infrastructure (OCI) data centers over the next 90 days. While terms of the deal such as size were not disclosed, Bloom completed shipments in just 55 days, highlighting its ability to deliver power to data centers rapidly.
Bloom had signed a partnership with CoreWeave in July 2024, with the first fuel cells expected to be commissioned in Q3 2025 for a data center in Illinois. Bloom briefly updated on this, saying that its fuel cells are generating power at the facility. However, it is rumored that the data center is just 14MW, essentially making it a pilot/validation deployment rather than a full-scale commercial deployment.
Nvidia’s Rubin is Coming; Bloom Energy is Ready
There was discussion around how Bloom Energy’s solutions could become more attractive with the Rubin generation of GPUs with an analyst asserting DC/DC power would be more efficient than DC/AC power.
Our team has covered quite closely the power requirements for Rubin Kyber racks, which could draw 600kw or 5X that of the NVL72 systems that are shipping now. You can read more here on this topic (an important read if you are invested in AI energy stocks).
What was discussed on the earnings call is the power requirements will put immense pressure on voltage, stating: “the laws of physics dictate that you have to go to an 800-volt DC architecture if you want AI chips that have more power density, which is the only way you can improve upon AI in the next generation. This is not an if, this is not a nice-to-have. This is a must-have.”
Bloom is asserting they are prepared for the 800-volt DC architecture (whereas most energy solutions are not such as 50MW turbines) as they are adaptable when moving beyond the 48-volt DC architectures of today.
“That is the 48-volt DC because the small wire through which a small amount of water comes into the straw, that water was sufficient to satiate the thirst. That was when CPU racks were 13 kilowatts. We have put a lot of Band-Aids on it to make sure Blackwell chips that come somewhere near the 130 kilowatts can handle it through the straw […] We saw this coming one day. We didn't know what day in 2000 when we initially created architecture. We built an architecture where we can feed these straws appropriately right at that 800 volts, and we decided every unit we have shipped for the last 15 years has that.”
Bloom Q3 Revenue Beat of 21%
Bloom smashed analysts' revenue estimates by 21.3%. The company reported record revenue of $519.05 million, versus estimates of $428.07 million. Revenue grew by a solid 57.1% YoY and 29.4% sequential growth, accelerating 37.6 percentage points from the previous quarter’s YoY growth of 19.5%.
The company’s fourth consecutive record revenue was driven by the strong demand for its fuel cell technology, driving AI data centers. We have discussed the fuel cells opportunity as a key catalyst in our article here. The company’s fuel cells are very efficient and are currently producing 10x power within the same footprint than produced previously a decade ago.
The management highlighted three major tailwinds that are positioning the company to become a global standard for on-site power generation; a market expected to reach a trillion dollars. First, AI buildouts have increasingly made on-site power generation a core necessity. Secondly, since AI is a national priority, government policy on on-site power generation is now more liberal. Third, the company’s fuel cells are highly efficient and are witnessing double-digit YoY cost reductions.
Revenue growth decelerates in Q4 due to tough comps, as last year’s Q4 revenue grew by 60.4%. Analysts expect Q4 revenue to grow 6.4% YoY to $608.7 million. Revenue growth will accelerate to 20% in Q1 2026 and to 23.6% growth in Q2 2026.
Looking ahead, analysts expect 2026 revenue to grow 24% YoY and accelerate to 62% growth in 2027. Management sounded optimistic on the future growth as the company’s co-founder and CEO, K. Sridhar said in the Q3 earnings call, “This seminal year for Bloom positions us for an even stronger 2026 and beyond with higher growth and more profitability”.
The analysts' estimates would trend higher after management's comments during the recent earnings call that doubling capacity to 2 gigawatts would support 4x the company’s 2025 revenue. Using a conservative approach, we believe revenue may be recognized over the next two years.
Key Segments
Products, installation, and service revenue growth showed acceleration from the previous quarter. While Electricity segment declined sequentially.
Products revenue grew by 64% YoY to $384.3 million, accelerating from the 31% growth in Q2.
Installation revenue growth spiked 105% YoY to $65.78 million, accelerating from a (13%) decline in Q2.
Service revenue grew by 16% YoY to $58.6 million, accelerating from the 4% growth reported in the previous quarter.
Electricity revenue was down (25%) YoY to $10.35 million, decelerating from a decline of (10%) in Q2.
Margins Continue to Expand
Bloom’s margins are improving, primarily driven by operational efficiency, product cost improvements, and operating leverage. Bloom is fundamentally transforming into a stronger company, as its GAAP operating margins were previously deep in the red, in double digits.
Q3 gross profits grew by 92.7% YoY to $151.68 million or a gross margin of 29.2%, up 5.4 percentage points YoY and 2.5 percentage points sequentially. Similarly, adjusted gross margins showed strong YoY and sequential improvement, primarily driven by product cost improvements and manufacturing efficiencies.
Operating margins improved 4.4 percentage points YoY and 2.4 percentage points sequentially to 1.5%, primarily driven by strong operational efficiencies. Adjusted operating profits grew by 470% YoY to $46.2 million or an adjusted operating margin of 8.9% compared to 2.5% in the same period last year and 7.1% in the previous quarter.
Net margin was (4.4%) compared to (4.5%) in the same period last year and (10.6%) in the previous quarter. Adjusted net income was $35.45 million compared to ($1.5 million) in the same period last year. Adjusted net margin improved 7.2 percentage points YoY and 1.3 percentage points sequentially to 6.8%.
Adjusted EPS beat of 47%
GAAP EPS came at ($0.10) in Q3 compared to ($0.06) in the same period last year. GAAP EPS was negatively impacted by a one-time loss related to unconsolidated affiliates of ($19.6 million) or a ($0.08) per share. The company reported adjusted EPS of $0.15, beating estimates by 47%, and was up from ($0.01) in the same period last year and $0.10 in the previous quarter. Bloom reported strong profits growth driven by operational efficiency, product cost improvements, and operating leverage.
Analysts expect adjusted EPS of $0.31 in Q4 and $0.04 in Q1. Looking forward, adjusted EPS is expected to grow strongly by 84.7% YoY to $0.93 in 2026 and 122.4% to $2.07 in 2027.
Cash Flow and Balance Sheet
The company reported positive operating cash flows and free cash flows in the recent quarter after negative cash flows in the first two quarters of the year.
Q3 operating cash flows were $19.67 million or 3.8% of revenue compared to ($69.5M) or (21%) of revenue in the same period last year. Operating cash flow improvement was primarily driven by higher profits and working capital improvements.
Strong operating cash flows also led to higher free cash flows. Q3 free cash flow was $7.4 million or 1.4% of revenue compared to ($83.8 million) or (25.4%) in the same period last year.
While management has not provided concrete 2025 guidance, it noted on the earnings call that “we expect fiscal 2025 to be better than our previously stated annual guidance on our financial metrics”. It suggests that the company’s cash flows and free cash flows would be strong in Q4, based on management's guidance during Q2 results that cash flows would be at the same level as in 2024. To give a perspective, the company reported operating cash flow of $92 million in 2024 and free cash flow of $33.2 million. Year to date, the company reported operating cash flow of ($304 million) and free cash flow of ($338 million), which means operating cash flow will be about $396 million and free cash flow will be about $371 million, respectively, in Q4.
Cash was $595.1 million and debt of $1.13 billion at the end of Q3 2025. While debt remained unchanged, cash improved by $20.3 million from the previous quarter.
Conclusion:
We are in the era of “what you see is what you get” – meaning, those offering strong earnings reports right now are setting up for a strong runway as future generations of GPUs will only be more power hungry. There is far less speculation than there was at the start of the year when we first covered Bloom in terms of which energy solutions can answer the demands of the AI data center buildout. The test for investors will be figuring out how to hold-on while this market unfolds in the coming years.
We hope to help with all of the above from being early to the products and solutions driving forward this massive market, to carefully examining the financials for confirmation the company is delivering, and providing the technicals to help stay the course while also not getting too emotional during the highs and lows.
Our earnings season is off to a strong start, we have dozens of reports to cover for you alongside real-time trade alerts that do what few can offer – analyze the complex AI market yet also execute.
Regarding the flawless execution, I want to thank the team on this one: Knox, Damien and Royston. It’s been a pleasure to see the pieces come together, and we hope there are many more like it to come.
I/O Fund Equity Analysts Damien Robbins and Royston Roche contributed to this article.
Please note: The I/O Fund conducts research and draws conclusions for the Fund’s positions. We then share that information with our readers. This is not a guarantee of a stock’s performance. Please consult your personal financial advisor before buying any stock in the companies mentioned in this analysis.
Last quarter, I began providing a comprehensive ranking of the top AI stocks in our universe. Our portfolio benefited from this exercise, and we hope yours did too.
Perhaps most beneficial is taking a moment on a quarterly basis to be as objective as possible. Information in the form of earnings reports, product announcements, news headlines, and even social media exuberance can cloud an investor’s decisions. It is not only the volume of information, but the speed at which information comes daily to where it can be hard to discern the true winners from those catching a fleeting headline.
The report below is designed to be objective in a no holds barred approach. Because we manage our own money full-time, we are laser-focused on ways to improve on a quarterly basis. Similar to the Q3 2025 report, the analysis below tops out at over 16,000 words, totals 43 pages and took three weeks to write. Even if we’ve held a stock for years, we update the investment thesis and re-examine the fundamentals. There is no stock that is immune to being cut from our portfolio if it cannot prove it’s one of the best AI stocks at this moment. Vice versa, to be added for the first time, a stock must be able to prove it can hold its own among already-strong choices.
After rigorous examination, the list below summarizes the strongest AI stocks of Q4 2025 and the top 3 Thematic Trends that I believe will help drive the AI market to new heights.
AI Networking: The Exponential Investment Opportunity
Each quarter, the I/O Fund scans hundreds of tech companies and their quarterly earnings reports for anomalies. To be precise, we track 188 data points on each portfolio company and top-ranking stocks we don't own totaling about 40 stocks a quarter for a total of over 7,000 data points. This quarter, one metric stands out as truly exceptional: Nvidia’s networking growth of 46% QoQ and 98% YoY to $7.25 billion. In the previous quarter, Nvidia’s networking grew 64% QoQ after declining (3%) in the previous quarter. Management stated the networking to compute attach rate is 75% in the fiscal Q4 earnings call.
Here is why this data point is unique – if Nvidia’s networking segment were a standalone segment, it would place #3 in the world on AI revenue with Azure reporting about $36 billion per year in AI revenue (at $75 billion total with a statement that half is AI-related) versus Nvidia’s networking now on a $28 billion run rate. Broadcom would be fourth at $6.2 billion AI revenue guided for next quarter, or about a $25B run rate.
There’s an argument to be made that Google Cloud or AWS would be in the running, however, they do not breakout AI revenue yet. An investor can also reasonably assume if it was at Azure’s level or higher, they would share their first place standing among the Big 3.
Yet, if any of those leading AI companies grew their AI segment 46% QoQ, their stock would go wild. The metric was missed because it was buried in Nvidia’s Compute weakness/China noise, yet this number is truly the bellwether as we move into Q3.
It also gives me a nostalgic pause as I was early to cover the importance of the Mellanox acquisition in both 2019 and 2020 stating: “Mellanox’s Ethernet switch systems are the most used internal system in the top 500 in a recent report released at ISC High Performance, with 247 systems, and InfiniBand is the second most-used, with 140 systems. However, InfiniBand, a computer-networking communications standard, connects the most high-powered computers where the presence of Ethernet is nearly non-existent […] This is a strategic acquisition for Nvidia and Mellanox to become the strongest combination for artificial-intelligence and machine-learning computations.” You will be hard-pressed to find an equity analyst covering the AI market that closely back in 2019-2020 at the height of Covid.
Turning our attention to 2025 and beyond, we want to look at what stocks are downstream from the number one growth market at scale.at scale. As stated, it’s not only the largest QoQ growth across AI, but rather it’s that Nvidia was able to grow its networking segment 46% QoQ while operating as the world’s third-largest AI segment.
My readthrough is that something important is ramping, and it will be back-half weighted. You’ve known for many quarters that the “something” is the new Blackwell GB200 system. What is newer information is that Blackwell Ultra GB300 is also ramping nearly simultaneously.
Per Nvidia, the strong performance in Networking was driven by “growth of NVLink compute fabric for GB200 and GB300 systems, the ramp of XDR InfiniBand products, and adoption of Ethernet for AI solutions.”
Double-Clicking on the Network Requirements of the GB200s and GB300s
Nvidia’s networking segment is a proxy for the AI networking allocation we hold in our portfolio. The quarterly numbers that Nvidia provide relate to their proprietary InfiniBand networking and the NVSwitches that help to route NVlink connections for GPU-to-GPU communication.
However, as the 75% attach rate above describes, there is a roughly 25% opportunity for other networking vendors to participate. Additionally, there are times that Nvidia will outsource components or cabling rather than build every single component in-house. In some instances, sourcing raw earth materials such as indium phosphide (InP) would prove quite challenging. Therefore, the information below may relate to Nvidia’ proprietary networking, yet it has important readthroughs to carefully selected suppliers.
Generally speaking, networking cables, ASICs and components required for GPU interconnects will increase 5X to 9X as we move from the HGX/DGX systems to the NVL72 systems. Given the GPU count will increase 9X from 8 GPUs to 72 GPUs, it makes sense that networking will increase similarly.
This is inferred by a few key points in terms of how the architecture is shifting from the 8-GPU HGX/DGX systems in the Hopper generation. These older systems that shipped in 2023-2024 used four NVSwitch ASICs to connect eight TensorCore H100 GPUs.
Networking is at the heart of the Blackwell architecture as the increased bandwidth is instrumental in driving the higher performance. The NVLink domain moves from supporting eight GPUs to 72 GPUs with a speed of 1.8 TB/s. Nvidia’s 5th generation NVLink interconnects will deliver a higher aggregate bandwidth of 9X to 18X compared to Ethernet and InfiniBand in previous generations. By increasing the bandwidth, the NVL72 systems will pool together compute and memory for up to 4X faster training and 30X faster inference.
This is accomplished with 18 NVSwitch ASICs, up from four in the HGX/DGX systems. That’s a 4.5X increase in NVSwitch ASICs.
If you look below at the specs between the larger NVL systems and the two HGX SKUs, you can see it’s not only compute performance that increases 10X (petaFLOPs) but also aggregate memory bandwidth and speeds.
The main takeaway is that scale-up architectures are as much about networking as they are about compute. The goal is to add more GPUs that can effectively distribute training and inference across a large cluster, which includes sharing data and memory, synchronizing, and exchanging model parameters.
Therefore, the interconnect or “connectivity layer” must scale accordingly to prevent GPUs from idling and waiting for data across the increased number of GPUs that are operating in parallel and at scale. High-speed communication is central to Blackwell and future generations of GPUs because the goal of a GPU-cluster is to act as a single processor. As AI systems grow GPUs by 9X, but also clusters grow from 100s of thousands to millions, there will be significantly more components required such as retimers, switch fabrics, silicon photonics, transceivers, cables and controllers.
However, despite Nvidia and Broadcom dominating the AI networking space, there remains immense opportunity for smaller, lesser-known suppliers.
Key opportunities mentioned above include PCIe components, as this is needed to connect the GPU to CPUs, memory and storage. Specialized copper cables are needed to connect AI servers to networking switches, and for linking many servers together. Retimers help to extend data integrity beyond short distances. The jump from 400G to 800G to eventually 1.6T on data rates will require an additional upgrade to nearly every component in the network layer, such as switch ASICs, optical transceiver modules, faster digital signal processors (DSPs) and retimers/redrivers. When it comes to products like EML lasers, the indium phosphide (InP) is hard to source. There is expected to be a new market for co-package optics (CPOs) with the new Rubin architecture – to name just a few of the AI networking opportunities currently in play and also squarely in front of us.
AI Energy: How Much and How Fast?
AI energy is all the rage today, yet it was rarely spoken about when we first covered the trend in the article AI Power Consumption Becoming Mission Critical. That article would later grow into lengthier premium thematic coverage, which would spark winning positions such as Bloom Energy, Oklo, NuScale and our first Bitcoin Miner/AI Data Center position Core Scientific in early 2025 (the first Bitcoin miner to retrofit for AI DCs). When we say we work hard to be early to new trends, we mean exactly that.
While networking can get bogged down in jargon and specs, and suppliers can be dropped quite quickly in the ever-shifting landscape, energy presents an entirely different investing landscape.
The problem that AI energy stocks seek to solve is simple to conceptualize compared to the intricacies of networking. Once a power solution is confirmed, the chances it is dropped from a qualified supplier list in the same manner as a networking stock is less likely. Think of energy to AI as water to humans – a daily essential that is less about the competitive landscape and more about the sheer necessity for survival.
Hyperscalers are spending hundreds of billions of dollars annually on AI data center capex, from physical data center space, GPUs and servers, hardware and networking. With these substantial sums flowing towards GPUs that are now being refreshed on an annual cadence, the impetus for hyperscalers, neoclouds and other cloud providers turns to how quickly these GPUs can secure power and be deployed.
There are three reasons the race is incredibly fierce to power these new systems:
The first reason is the stock market, as the current capex numbers are significant, and investors will want to see a return on this investment. If a company like Microsoft buys tens of billions of Nvidia’s Blackwell GPUs, the longer the massive investment in GPUs waits for power, the more delayed that revenue and profits become.
Secondly, for competitive reasons and to keep up with Nvidia’s product road map, the next generation of GPUs will arrive every 1-2 years and Big Tech will want to maximize returns before the next generation comes online. Competitors who can energize the newest generation of GPUs faster will have a critical head start over those that are waiting for power. But there’s a catch, every generation of GPUs requires more power – for example, there will be a 5X increase between the power requirements of the GB200 NVL72 and the Vera Rubin NVL576 “Kyber” rack over a two-year design timeframe. These figures do not include networking, interconnects, cooling and other hardware, which will further boost power draw per rack.
Third, the AI race is not merely a battle between companies like Google, Amazon and Microsoft. Rather, it is a battle among global powers. While the news has latched onto China-fears such as DeepSeek, tariffs or rare earth materials, the true challenge lies in the fact that China has significantly more power than the United States. In a recent Fortune article, energy experts stated China’s reserve margin has never dipped below 80% to 100% nationwide, meaning it’s at 2X the capacity the country needs. Meanwhile, the United States is at a 15% reserve margin.
These three reasons are simple in concept, yet the lack of power having vast consequences cannot be overstated if you combine the sheer size of investments being made in AI alongside fierce, heightened competition.
The push to 600kW racks over the next few years means this is not a transient problem, rather it is one that the industry will continue to face, meaning continuous new construction may be needed to handle surging power demand. For example, Vantage’s upcoming 1.4 GW campus in Texas for Oracle is designed for ultra-high density racks up to 250kW, yet this will not be enough power to able to host NVL576 racks in just two to three years’ time. Additionally, a former Microsoft Azure AI executive reportedly said he estimated that the “requirements in terms of power for the data center would probably at least double every three years and maybe exponentially so over a period of time,” further reinforcing this.
AI Accelerators:
AI accelerators such as GPUs and custom silicon need no introduction – raw compute and compute performance has driven the AI market up to this point. The analysis would be remiss to not acknowledge the trend that is so powerful, it is displacing the FAANGs of the last decade with Nvidia firmly the world’s most valuable company now and Broadcom within striking distance of passing up companies like Meta.
As the market weighs the so-called AI bubble, there are many disparate facts thrown at investors: dot-com fears, China/tariff concerns, stock pullbacks when there are minor announcements, and things like circular investments from OpenAI.
Forgive me if I sound repetitive, but what truly matters is Big Tech capex. This is the single, most important number as it far outweighs the importance of earnings reports, fiscal year guidance, Nvidia’s networking growth or their product roadmap, if AMD has a new deal from OpenAI, Oracle’s insane RPO, Broadcom’s networking chips and custom silicon announcements – all of the above is being single-handedly driven by Big Tech’s large capital expenditure (capex) budgets.
Let me throw a few stats at you.
Analyst estimates cannot keep up with the capital expenditures being spent on AI infrastructure. This time last year, the expectations were for $250 billion in Big Tech capex. Morgan Stanley later forecast $300 billion in Big Tech capex for 2025. That number stands at $365 billion for 2025 with one quarter left to go.
It’s easy to tune out the words “big tech capex” at this point but zoom out for a minute and consider that Big Tech’s TTM capex was $24B at the start of 2015, or up 15X over ten years.
In terms of the opportunity looking forward, McKinsey is predicting 3.5X growth in gigawatts for AI data centers between 2025-2030. The costs associated with AI data centers range from $3 trillion to $8 trillion, or about $5 trillion at the midpoint. This correlates to about 3X growth if we assume the current run rate is $1.8 trillion at the current capex of $365 billion.
On a more near-term basis, Goldman Sachs sees hyperscaler capex increasing sharply through 2027 – capex is projected to be $1.15 trillion from 2025 through 2027, more than double the $477 billion spent from 2022 through 2024.
Going back to the first point, analysts thus far have missed the mark in their estimates. Every quarter, sell side analysts rush to update their models. We are penciling in that 3x is a baseline to work with over a 5-year time frame.
To be objective, there are analysts calling for a stock market crash based on the risks around the consumer and a GDP that is propped up by capex spending.
Stifel stated in August:
“While the capex boom around AI temporarily supports GDP and asset prices, Stifel forecasts this bump will fade as corporate tech spending plateaus. Such a build-out, after all, occurs only once, while consumer spending power is entering a lull that could expose markets to abrupt correction.”
There is weight to what Stifel is describing, which is why tariffs remained a risk on our last Top 15 report and remain a risk for this report, as well. You can read more here about how the consumer is fairly weak under the hood, and how capex spending is creating a false impression that GDP is stronger than it is.
Where I disagree with Stifel is the idea that “such a build-out, after all, occurs only once” This is an antiquated opinion where perhaps the parallels between AI and electricity have gone too far, or perhaps it’s leftover from the cloud era where there were digestion periods every few years.
Portions of the AI buildout will be occurring every 1-2 years with new generations of AI systems. Therefore, AI is less of a static end point that is readily achieved – and rather it is an evolving architecture that enables ambitions to expand each year. Although cloud was also architecture-driven, it reached its end goal rather quickly in terms of driving down costs and improving productivity, allowing companies to quickly scale, and providing pay-as-you-go compute and services rather than significant up-front costs from on-premise servers. The end goal for AI is far more ambitious, as it could take a decade or more before Big Tech accomplishes commercially viable AGI (general artificial intelligence).
When listening to commentary on earnings calls, in sharp contrast to what macro analysts are saying, what we hear is that Big Tech management teams are nearly in a panic to add more capacity. This one is from AWS’ Andy Jassey: “The faster we grow, the more CapEx we end up spending because we have to procure data center and hardware and chips and networking gear ahead of when we're able to monetize it. We don't procure it unless we see significant signals of demand.”
Although the tone on capex can temporarily shift from time to time, the risk that capex will dry up from a one-time buildout is low. In many ways, the greatest risk to AI isn’t within the AI economy itself, but in broader macro conditions. We are seeing many macroeconomists attempt to forecast an AI downturn, yet having followed tech cycles for years, I’ve learned that while the macro economy waxes and wanes, technology consistently resurfaces as the best way to participate in the market — and this is especially true for AI.
For years, there’s been a debate on whether Big Tech’s AI spending will translate to revenue and profits. Meanwhile, during those years, the I/O Fund has been laser focused on where that AI capital is actually being allocated. Rather than thinking of our approach as the picks and shovels for those chasing a gold rush, we think of it as an “AI stack” strategy—investing in the lesser-known layers and components that are driving forward an ecosystem capable of massive GDP.
With that, I present my current Top 15 AI Stocks List.
#1: Nvidia is Simultaneously Shipping Two GPU Systems
When looking at YTD performance, Nvidia has returned 35% compared to SMH at 42%. Admittedly, it’s not great to underperform your industry. Peers within AI have outperformed Nvidia YTD by a wide margin, such as AMD and MU outperforming 2-3X as they’re up 93% and 135% on a YTD basis.
I was recently asked on Fox if AI chips are due for consolidation. My reply was that any dips should be bought. Although I talk a lot about Nvidia and AI, my process for covering this stock has not changed since the first day I spoke of the company. My process is that we stick close to the GPU releases rather than the earnings reports as we are dealing with the world’s best design company (fact). Therefore, revenue is a step-function of a new design being released. That may seem overly simplistic, but I can assure you, it’s worked quite well for this stock.
While AI hardware players must contend with competitive forces and supply chain qualifications that can change their positioning very quickly, that is not the case with the King of Parallel Processing. Rather, Nvidia’s near-monopoly has been built carefully as each GPU release has virtually zero competitive pressure.
Generally speaking, Blackwell and Blackwell Ultra are shipping simultaneously. There are nuances to this statement, such as the GB200 NVL72 systems began shipping in volume in Q2 2025 and GB300 NVL72 began shipping this quarter and will ship in volume next quarter in Q4 2025.
I want to keep this really simple as the enormity of what Nvidia is shipping would be easy to miss. The market won’t be caught off guard like it was two years ago – we all know who the best slugger in the game is. However, there are two home runs lining up (dare I say, a grand slam) as Blackwell and Blackwell Ultra ship within 6 months of each other. In this analogy of a baseball game, Nvidia has been in the dugout for a few quarters now with the Blackwell delay. I am still in the front row of the stadium awaiting the number one slugger to return to the plate.
Overall Revenue Growth:
Nvidia reported $46.74 billion in revenue in Q2, slightly ahead of estimates for $46.13 billion. This corresponded to growth of 55.6% YoY, decelerating more than 13 points from 69.2% in Q1; on a QoQ basis, revenue increased just 6.1%, slowing from 12% in Q1.
However, its Nvidia’s guide for this upcoming quarter that solidifies the stock as our top pick in AI chips for now. The guide for $54 billion in revenue corresponds to 53.8% YoY growth and a rebound to 15.5% QoQ growth.
As you’ll see below, Broadcom is expected to report higher QoQ growth yet it’s the scale at which Nvidia is putting up these numbers that separates the company from its peers.
AI Segment Growth:
Nvidia’s data center revenue increased 56% YoY and 5% QoQ to $41.1 billion, a marginal miss versus estimates for ~$41.2 billion in the quarter. This is also the smallest sequential increase since Hopper’s breakout quarter at just ~$2 billion.
While compute declined (1%) due to timing of shipments and loss of China revenue, networking was up a whopping 46% QoQ and 78% YoY to $7.25 billion. As stated above under the AI Networking thematic section, this spells good things for what is in the pipeline.
Earnings:
Nvidia reported just a 3.9% adjusted EPS beat in Q2, reporting $1.05 in earnings versus estimates for $1.01. This corresponds to growth of 54.4% YoY, rebounding substantially from Q1’s H20-affected 32.8% growth.
Adjusted EPS growth is expected to remain strong through the rest of the fiscal year, at 47.2% and 53.2% in Q3 and Q4.
Margins:
GAAP operating margin was 60.8% in Q2, improving nearly 12 points QoQ and coming in 1.7 points ahead of guidance for 59.1%. Adjusted operating margin was 64.5%, also up nearly 12 points QoQ and 1.4 points ahead of guidance for 63.1%.
Cash:
Free cash flow was $13.45 billion, down from $26.14 billion in Q1. FCF margin was 28.8%, again down more than 30 points QoQ and more than 16 points lower YoY.
Typically, Nvidia has very strong cash flows and this is not a concern, rather is a transient quarter for cash. Operating cash flow margin and free cash flow margin have been in the 60% range in recent quarters.
Valuation:
There is not a major takeaway based on valuation.
Forward PS Ratio:
Nvidia trades at 20 forward PS and is a strong buy in the 10 forward PS range yet is a strong sell in the 30 forward PS range. We are in the middle of the clear buy/sell indicators for sales valuation.
Forward PE Ratio:
Nvidia trades at 40 forward PE ratio and is a strong buy in the 20 forward PE range yet is a strong sell in the 50 forward PE ratio.
Notable Risks:
Nvidia has far fewer risks than other stocks in this report. China tariffs can affect peers in the supply chain especially since there are hundreds of components in each AI system.
#2: Broadcom: Well-Deserved Second Place Contender
Financials: 8/10
Thematic: 10/10
Valuation: 3/10
The networking opportunity that Broadcom is positioned to capture has been evolving every quarter to where Hock Tan himself has not been able to correctly anticipate its size. Here is what Tan stated on a recent earnings call:
“In fact, the increased density in scale up is 5 to 10x more than in scale out. And that's the part that kind of pleasantly surprised us and which is why this past quarter, Q2, the AI networking portion continues at about 40% from what we reported a quarter ago for Q1. And at that time, I said I expect it to drop. It hasn't.”
This quote illustrates a few things – the strength of the networking market is surprising even to Broadcom, it helps to quantify scale-up versus scale-out in terms of networking components, and it shows that (likely) this is not priced in yet as it’s a relatively new inflection. In fiscal Q2, networking growth was 170% YoY although it’s not expressly broken down in earnings reports.
Regarding custom silicon, the chances ASICs can keep pace with Nvidia on sheer compute is low (to nearly impossible). When it comes to raw compute density combined with the software ecosystem that Nvidia offers, Big Tech’s side projects may never catch up. The rapid product road map that Nvidia offers deepens the moat the company has firmly established with universal CUDA. Meanwhile, custom silicon programs can take years to fully develop and move into production.
So then why are Big Tech companies turning to Broadcom for less flexible (yet highly optimized) AI chips with product road maps that are considerably longer than Nvidia’s? Decades ago Jeff Bezos stated “your margin is my opportunity,” referring to the fact that Amazon’s value proposition was to offer goods at a lower cost to consumers. Broadcom is similar, in that Nvidia’s margin is their opportunity to offer AI accelerators at a lower cost. The same can be true for AMD’s value proposition, as well, especially as we enter the inference stage of the AI market. Nvidia’s gross margin of 72% (and up to 78% about a year ago) is attractive to investors who seek a quality stock, yet the margin is also communicating gluttonous pricing power.
According to a recent article by VentureBeat, industry conversations and analysis suggest that “Google may be obtaining its AI compute power at roughly 20% of the cost incurred by those purchasing high-end Nvidia GPUs. While the exact numbers are internal, the implication is a 4x-6x cost efficiency advantage per unit of compute for Google at the hardware level.”
Where it becomes attractive to drive down costs is the inference market. Hundreds of millions of users interact daily with AI assistants, causing inference to become the focal point for providers such as OpenAI and Google. Meeting these levels of growing demand, without significant response delays or downtime, requires more and more accelerators, networking and interconnect products.
Broadcom’s edge goes beyond the fact that custom accelerators are often multiples cheaper than Nvidia’s GPUs for inference tasks – it's that custom silicon is increasingly performant with each generation. By optimizing algorithms (software), Big Tech can drive higher performance from large language models (LLMs) while continuing to use Nvidia’s compute power excellence for training (and also some inference tasks).
Lastly, the VMWare acquisition has been particularly fun to watch as it’s one of the best execution M&A moments recently. A few quarters back, Tan stated VMWare was the “star of the show” as it’s been reporting accelerating bookings and backlog. Here is why it’s done well post-acquisition: “This allows customers to deploy their AI models on-prem. And wherever they do business without having to compromise on privacy and data — in control of their data. And we are seeing this capability drive strong demand for VCF, from enterprises seeking to run their growing AI workloads on-prem.”
Overall Revenue Growth:
Q3’25 revenue was $15.95 billion, beating estimates for $15.82 billion, and reflecting top line growth of 22.0% YoY and 6.3% QoQ. Looking ahead, management provided Q4’25 guidance of $17.4 billion of revenue, implying 24% YoY growth and a slight uptick to 9% QoQ growth.
AI Segment Growth:
Semiconductor Solutions accelerated nine points to 26% YoY growth due to a rebound in AI accelerators (+63% YoY). Within this, AI Semiconductor revenue surged 63% YoY to $5.2B, showing re-acceleration after a slower Q2 (+46% YoY). AI now represents 57% of Semiconductor revenue and 32% of total company revenue.
Management guided Q4 AI revenue to $6.2 billion, which would represent ~19% sequential growth and eleven consecutive quarters of YoY growth.
Earnings:
Non-GAAP EPS growth of 38% outpaced revenue growth of 22%. EBITDA margin was 67%.
Margins:
GAAP Gross Margin of 67.1%, down 90 bps QoQ from 68.0% in Q2’25, essentially flat from 66.8% in Q3’24.
GAAP operating margin of 36.9%, down 190 bps QoQ from 38.8% in Q2’25, but up significantly from 30.3% in Q3’24.
Non-GAAP operating margin of 65.5%, slightly up QoQ from 65.3% in Q2’25 and up 180 bps from 63.7% in Q3’24.
Cash:
Free Cash Flow of $7.0B represents a free cash flow margin of 44.0%, compared to 42.7% in Q2’25 and 35.6% in Q3’24.
Valuation:
Broadcom’s valuation is in unchartered territory.
Broadcom trades at a 52 forward PE ratio and has traded as low as 13 forward PE two years ago and as low as 21 forward PE in the April rout.
Broadcom trades at a 26 forward PS ratio and has traded as low as 10 forward PS in April and was at 6.5 two years ago.
Notable Risks:
Valuation is the predominant risk as Broadcom has never traded at these levels in its multi-decade listing history.
The company secured a long-term deal with OpenAI to supply 6GW of GPUs with the first GW to be delivered in H2 2026. We saw a flurry of sell-side activity with one analyst raising their price target from $185 to $310 stating the Open AI deal could generate $80 billion in chip revenue for AMD over the next few years.
As stated above under the Broadcom section, this is not a matter of AMD offering the best end-to-end performance. That will remain Nvidia for the foreseeable future. Rather, this is about driving down costs for Big Tech (and Open AI) while focusing less on raw compute power for training and more on memory and throughput for inference.
If we zoom out (as I like to do), you may recall the MI400s are expected to be the moment that AMD tightens the competition with Nvidia. The MI400 series will be the start of rack-scale systems for AMD, starting with Helios, which will connect up to 72 GPUs similar to Nvidia’s NVL72 systems.
According to AMD, Helios will “deliver up to a 10x generational performance increase for the most advanced Frontier models, and we believe it will be the highest-performance AI system in the world when it launches.” The last part is doubtful, yet the effort to close the gap with Nvidia will likely go a long way when coupled with lower pricing.
AMD stated in their last earnings report they have an ambitious goal of reaching tens of billions in MI400 sales. Investors should take note that management is specifically calling out the MI400 for this, arriving in H2 2026. The readthrough is that OpenAI is an early validator that the MI400s have serious chops, and where OpenAI goes, the rest of the AI market tends to follow.
Overall Revenue Growth:
AMD reported a slight beat on the top line at $7.685B in revenue compared to estimates of $7.43B. This represents growth of 31.6% compared to growth of 27.4% expected.
AI Segment Growth:
Last quarter, management had stated, “we expect data center segment to decrease due to the exclusion of MI308 revenue.” Therefore, it was not a surprise when data center was down (11.8%) QoQ yet was up 14% YoY for revenue of $3.24B.
Earnings:
EPS was in line with expectations at $0.48 yet was down (30%) from $0.69 in the year ago quarter. The company is expected to rebound quickly with EPS of $1.15 next quarter.
Margins:
Operating margin of (2%) for operating profits of ($134M) also included the $800M in inventory changes. The adjusted operating margin of 12% was guided correctly and was in line with expectations.
Cash:
AMD’s cash flow margins sustained well at 20% operating cash flow margin compared to 13% last quarter and 10% OCF margin last year. Free cash flow margin of 15% also expanded from a year ago at 8% margin and up from 10% FCF margin last quarter.
Valuation:
Forward PS Ratio:
AMD is trading at 13 current PS with the stock failing to hold 14 current PS two times in the past (precedes a larger selloff). The forward PS ratio of 11.7 is at the stock’s peak forward PS ratio of 12.
Forward PE Ratio:
AMD is trading at 61 forward PE ratio, the highest the stock has traded since the AI boom began in early 2023.
Notable Risks:
As the contender to the world’s most valuable company, AMD has execution risk. The company’s lead in Data Center CPUs are often at risk due to companies like Nvidia wanting to cut down costs on the instructions side of AI systems.
The valuation remains the most notable risk.
#4: Micron Quietly up 120% YTD
Financials: 8/10
Thematic: 10/10
Valuation: 7/10
Micron deserves a second look as the company is no longer tied to consumer device cycles. Instead, high bandwidth memory (HBM) had led to higher margins and multi-year supplier agreements, resulting in a leveraged approach to participate in the AI infrastructure buildout.
As pointed out in our free analysis last week, HBM is seeing a 3.5X increase in per-GPU capacity across the last three years and AI systems are commanding an increase of 34X as the number of GPUs rises and is further compounded by each GPU system requiring more HBM per package.
The B200 features 180GB of HBM3e content, more than double the H100 and a 28% increase versus the H200. In an 8-GPU server configuration, the B200 boasted 1.44TB of HBM content.
The B300 boasts 288GB of HBM3e content, a 60% increase versus the B200 and over 3.5x more than the H100. In an 8-server configuration, the B300 has 2.3TB of HBM content. This chip is beginning to ship now in Q3-Q4 2025.
Putting in context Nvidia’s rack-scale solutions, the GB200 and GB300 NVL72, shows just how rapidly HBM content is increasing. The GB200 supports up to 13.4TB of HBM content, while the GB300 supports up to 21.7TB of HBM, nearly 34X higher than the 640GB of HBM content in the 8-GPU DGX H100 servers.
The line in the sand for AI hardware companies is the margins and Micron has performed beautifully in that regard. I have to admit, when I saw their last earnings report at end of September, I had to look twice to make sure it was really Micron.
Overall Revenue Growth:
Micron reported record FQ4 revenue of $11.32 billion. Revenue growth accelerated 9.4 percentage points sequentially to 46% YoY, and on a sequential basis, growth was 21.7% QoQ, a solid 6.2-point acceleration.
AI Segment Growth:
FQ4 DRAM revenue grew by 69% YoY and 27% QoQ to $8.98 billion, a second consecutive quarter of strong sequential growth.
Earnings:
In Q4, Micron reported adjusted EPS of $3.03, up 157% YoY and beating estimates by 5.9%.
For Q1, Micron guided for adjusted EPS to be $3.75, +/- $0.15, more than 23% ahead of consensus and corresponding to YoY growth of 110%. Earnings growth is expected to reaccelerate to 155% in Q2 but then decelerate to 126% in Q3 (but still growing handsomely).
Margins:
Micron’s margin turnaround story has been impressive, with gross margin up more than 55 points over the last two years and operating margin up more than 66 points.
Adjusted gross margin in Q4 was 45.7%, up 6.7 points QoQ and 9.2 points YoY, aided by strong growth in CMBU which carried a 59% gross margin in the quarter, DRAM pricing, favorable product mix, and cost controls. For Q1, adjusted gross margin was guided to be 51.5% at midpoint, a 5.8 points sequential expansion and up by a solid 12 points YoY.
FQ4 adjusted operating margin was 35%, up 8.2 points QoQ and 12.5 points YoY, driven by operating leverage.
Cash:
FQ4 adjusted free cash flow grew by 149% YoY to $803 million or 7.1% of revenue, an improvement of 2.9 percentage points YoY. Management expects adjusted free cash flow to strengthen in FQ1 and to be significantly higher in FY2026.
Valuation:
Micron trades at an attractive valuation of 4.2 forward PS. The stock has traded as low as 2 forward PS and as high as 7 forward PS.
Micron trades at 12 forward PE ratio. The stock has traded as low as 7 yet as high as 124 due to the lumpy bottom line from the previous cyclical low in 2023 timeframe.
This goes back to the debate on if MU is a cyclical stock that deserves a lower valuation or is it emerging as a major, secular AI player. Should it be the latter, there is quite a bit of room in the valuation.
Notable Risks:
If Micron announces set pricing like they did in 2024, the stock could plateau. There are fierce competitors in the space, such as SK Hynix and Samsung. If pricing proves cyclical, the current valuation will not hold. If the pricing proves more of a secular trend, there is ample room in valuation – how the market will view the stock 2026-2027 is not clear although with technicals, some of the risk can be mitigated.
TSM is typically off cycle from AI semiconductors, meaning, we will see a boom in the high-performance computing segment about 5-6 quarters before we see a boom in the AI chip market.
It would be easy to assume TSM is a quarter or two ahead of shipment times when, in fact, it’s more like three quarters when factoring in HBM and CoWoS packaging. From there, it takes an additional two quarters for system integration and assembly from companies like Dell, Foxconn, and Supermicro before the servers are shipped.
Perhaps second to capex, TSM can be used as a strong proxy for the health of the AI market as the common denominator across AI chips. The HPC segment is communicating that we have a few quarters of strong growth in the pipeline. From there, we will need to monitor how long the QoQ decline in HPC lasts as a couple of quarters is typical; anything longer would be a concern to monitor further.
On the topic of timing, the chart below helps to illustrate that even though AI is a secular trend, due to shipping cycles, TSM and even MU can still see cyclical results. When a new generation is ramping, TSM will naturally see the results first as will Micron before the systems are shipped. If we draw a similar parallel to the last QoQ decline in TSM’s shipping cycle, we can see that putting our money in the AI Chips companies makes more sense right now.
Pictured Above: When TSM’s HPC segment declined for three quarters in the past, the stock underperformed compared to a stock like Nvidia, which was reaping the benefits of the new generation finally shipping in volume (Hopper).
Overall Revenue Growth:
Q3 revenue grew by 40.8% YoY and 10.1% QoQ to $33.10 billion, beating the guidance midpoint by 2.2%. TSMC boosted the full year revenue guidance by 5 percentage points for the second consecutive quarter to mid-30% on continued strong AI demand. This is up from the 30% growth provided in Q2.
For Q4, TSMC guided revenue of $32.2 billion to $33.4 billion. At midpoint of $32.8 billion, this represents a YoY growth of 22% and down (0.9%) sequentially.
AI Segment Growth:
TSMC stated that HPC revenue was flat QoQ in NT$ in Q3, though there was more pronounced increase on a US$ basis due to FX. TSMC’s revenue is recognized in US$, so every 1% appreciation of the NT$ adversely impacts NT$ reported revenue by ~1%. Management sounded very optimistic during the Q3 earnings call about long-term AI growth opportunities.
Earnings:
The company’s Q3 EPS grew by 50.5% YoY to $2.92, beating estimates by an impressive 12.3% with the strongest beat in the last two years. Analysts expect Q4 EPS to grow 26.8% YoY to $2.84 in Q4 and grow 25.5% in Q1. Looking forward, they expect EPS to grow 19.8% YoY to $12.34 in 2026 and 24.6% YoY to $15.48 in 2027.
Margins:
Margins continue to expand due to cost controls, higher capacity utilization rates, economies of scale, and better price negotiation with customers and suppliers.
The company’s gross margins improved 170 basis points YoY and 90 basis points sequentially to 59.5%. Cost improvements, better capacity utilization, and better price negotiation with customers and suppliers primarily drove the strong margins.
Q3 operating profits grew by 50% YoY to $16.74 billion, with an operating margin of 50.6%, an improvement of 310 basis points YoY and 100 basis points sequentially, primarily driven by higher gross profits and operating leverage.
Cash:
Despite higher capex, cash was also strong with operating cash flow at 43.9% compared to 51.6% in the same period last year. Q3 free cash flows were down (16.1%) YoY to $4.8 billion or 14.6% of revenue compared to 24.4% in the same period last year. The free cash flows were down due to higher capex which grew by 51.6% YoY to $9.7 billion to support strong further growth.
Valuation:
Forward PS Ratio:
TSM is trading at 13 forward PS ratio compared to 6 at the April low and 6 at the start of the year. Regarding the current PS ratio of 17 – I cannot find a higher valuation going back 10 years.
Forward PE Ratio:
TSM is trading at 31 forward PE ratio, the highest it’s been going back to early 2023 during the AI boom. The current PE ratio of 35 is the highest its been going back to ten years except on rare exception briefly at the 2021 top.
Notable Risks:
Similar to Broadcom, the predominant risk is valuation.
AI Networking Stocks:
My number one ranked trend is AI networking. As discussed in the AI Chips section, stocks such as Nvidia and Broadcom are also the networking leaders. However, the below stocks are networking pureplays, with many on a tear since the April lows.
As you have likely noticed recently, the networking ecosystem is nuanced as recent announcements from Oracle/AMD and Nvidia/Intel have caused some networking stocks to plunge. To provide the bigger picture, I am revisiting the thesis and rankings for our networking stocks.
Tied for #1: Astera Labs: Increased ASPs from Scorpio
Financials: 10/10
Thematic: 10/10
Valuation: 4/10
When Astera is asked why they stand apart within a crowded networking market, management responds that the drive for low latency PCIe is the primary contributor to the beat/raise across both the Aries and Scorpio products.
Last March, Astera announced further collaboration with Nvidia by offering NVLink solutions for PCIe/CXL within servers (scale up): “Most recently, Astera Labs demonstrated the industry’s first end-to-end PCIe 6 interoperability with Scorpio P-Series Fabric Switches, Aries 6 Retimers and a NVIDIA Blackwell GPU at NVIDIA GTC 2025. Scorpio P-Series Fabric Switches have also been integrated with the NVIDIA MGX platform for PCIe 6-ready modular designs.”
Launched only this year, Scorpio now exceeds 10% of total revenue “making it the fastest-ramping product line in Astera Labs’ history.” Keep an eye specifically on Scorpio as a GPU-to-GPU scale-out and scale-up product line.
On the custom silicon side, it’s widely understood that Astera supplies Amazon as there were disclosures around Amazon having a warrant to buy shares in exchange for guaranteeing $650 million in orders in the SEC filing. As of mid-2025, GuruFocus confirmed Amazon still holds shares in Astera Labs.
Astera’s agile ability to compete head-on with Broadcom goes beyond only Amazon and Nvidia (though certainly, those two are enough). Management stated they had design wins with ten customers including merchant GPUs and ASICs, plus a line of sight to further sales growth from their many product lines in H2 2025 and 2026, and the upcoming UALink consortium in 2027.
It takes some time for a company like Astera to become a qualified supplier. I would need at a minimum an earnings report to state otherwise or a QoQ decline of some kind to be convinced this has changed. Instead, what I’m seeing is quite the opposite.
Revenue:
Astera Labs reported revenue of $191.9 million, beating consensus of $172.5 million for growth of 150% YoY and 20% QoQ. About eight months ago in November, analyst consensus for the June quarter was for 85% growth — thus the company has nearly doubled these expectations in less than a year.
AI Segment:
Same as revenue growth (pureplay) – up 150% YoY and 20% QoQ.
Earnings:
Adjusted EPS of $0.44 beat by 36%. Consensus is $0.39 for 69% growth. GAAP EPS was $0.29
Margins:
Astera delivered in that regard with a GAAP operating margin of 20.7% compared to 7.9% expected. This operating margin is a major win for ALAB investors as the company is now comfortably GAAP profitable despite stock-based compensation being around 20% of revenue.
Cash:
Astera’s cash from operations increased significantly with an operating cash flow margin of 70.5%, up from 38.7% last year. This totaled operating cash flow of $135.4M with $1.07B in cash on the balance sheet and no debt.
Valuation:
Astera’s valuation of 35 forward PS is in the mid-range as the company has seen as low as 11 forward PS and as high as 60 forward PS.
The stock’s forward PE ratio of 103 is steep, yet the company is recently GAAP profitable, thus it’s hard to go by this ratio.
Typically we use technicals in a situation where a hypergrowth stock is surging on revenue and has strong bottom line results, causing the valuation to send mixed signals. On one hand, stocks like this should be highly valued. On the other hand, how rich of a valuation are buyers willing to pay? Rather than get too stuck in the weeds, technicals can help us participate in the upside while protecting the downside.
Notable Risks:
Every stock has risks yet Astera less so than others on my Top 15 list. I can see a scenario where the market softens on AI valuations (temporarily) and a scenario where there are strong earnings and valuations march onward.
Tied for #1 Credo: Active Electric Cables (AECs) for Miles
Financials: 10/10
Thematic: 10/10
Valuation: 4/10
The saying “I can see for miles” implies visibility into the future. The saying makes me think of Credo as there is over two miles of copper cabling for each NVL72 system. We can literally see Credo’s AECs for miles, and figuratively, there is more visibility than usual for this particular stock given the inflection of 274% YoY and 31% QoQ growth last quarter helps to confirm its positioning in Nvidia’s Blackwell systems.
Credo’s new 800G HiWire ZeroFlap AECs are designed to reach 7 meters with full host-to-switch connectivity, and are especially designed for liquid cooled servers. The over 7 meter distance helps to enable large AI clusters sized into hundreds of thousands of GPUs.
Credo competes with 800G OSFPs AOCs, yet these are particularly troublesome due to physical constraints that cause the connectors to break. There is also link lapse with AOCs, which are “momentary disruptions in network links.” Credo’s AECs aim to solve these issues, and the results speak for themselves.
For distances between two meters and seven meters (or about six to 24 feet), active electric cables (AECs) are also seeing heightened demand as servers scale up to eight GPUs to now 36 GPU to 72 GPU per rack-scale AI system.
In a nutshell, this is why Credo is reporting surging growth in a highly competitive market: “Reliability and power efficiency [leads] to choosing AECs over optical solutions as they are up to 1,000x more reliable and consume half the power. AECs virtually eliminate link fabs, which are intermittent losses of connection, boosting cluster reliability and productivity while reducing power consumption.”
Regarding “consume half the power” … Credo’s proprietary serializer/deserialzer (Ser/Des) technology, active electric cables and digital signal processing (DSPs) give the company a significant competitive advantage as it enables power-efficient connectivity that is reasonably priced.
Revenue:
The company reported growth of 274% YoY and 31% QoQ for revenue of $223.1 million. This beat estimates on the top line by 17% with management raising full-year revenue growth outlook by 35 points, from 85% YoY to 120% YoY.
AI Segment:
Product Revenue came in at $217.1 million, up 279% YoY and 31% QoQ.
Earnings:
The bottom line also shined with adjusted EPS beating estimates by 44.4%. This represents growth of 1,200% YoY from a thin $0.04 in the prior-year quarter. Triple-digit growth of 425% on the bottom line is expected to follow although flat QoQ.
Margins:
GAAP Operating Margin was 27.2%, up from 19.9% in the last quarter and up from (24.2%) in prior-year quarter.
Adjusted Operating Margin was 43.1%, up from 36.8% last quarter and up from 3.7% in prior-year quarter.
This is the standout – massive operating leverage as opex grew only ~11% QoQ vs. the 31% pick up in revenue.
Cash:
FCF Margin of 23.8%, down from 31.9% last quarter but up more than 45 points from (21.9%) in the prior-year quarter. Debt free.
Valuation:
The forward PS ratio is 24 and on the higher range of where Credo trades with the upper region being 33 earlier this year yet has traded as low as 8 at the April low.
Credo trades at a 65 forward PE Ratio yet similar to Astera Labs, this is hard to put much weight into as the company is newly profitable.
Notable Risks:
Every stock has risks yet Credo less so than others on my Top 15 list. I can see a scenario where the market softens on AI valuations (temporarily) and a scenario where there are strong earnings and valuations march onward.
#3: Small Cap Networking Stock with Strong QoQ 400G Growth and Incoming 800G/1.6T Growth
This past quarter, the I/O Fund was on the hunt for networking stocks that reported an inflection. Given Nvidia is reporting 46% QoQ growth at scale on their networking segment, we figured there would be some breadcrumbs to follow in the supply chain as to companies that are beneficiaries of the incoming AI networking boom.
The stock we identified for our Discovery members reported 40% QoQ growth last quarter and is forecasting 17% QoQ growth in the upcoming quarter – some of the highest we’ve seen in the supply chain landscape. Management discussed the ability to grow capacity 8.5X this year before doubling this by mid-2026.
Financials: 6/10 (not a pureplay)
Thematic: 7/10
Valuation: 7/10
Lumentum has been on our radar for more than one year, as the company supplies components for datacom transceivers and optical interconnects with tech that has caught the attention of heavyweight Nvidia. We’ve been closely monitoring Lumentum and waiting patiently for their EML lasers for 200G to ship, enabling 800G and 1.6T bandwidths.
As discussed in the past, optical interconnects help data centers accelerate data throughput between data centers and inside the data center between servers or racks, while reducing latency and power consumption. AI is driving cloud demand higher from the hyperscalers, leading to more data being created and processed, thus helping drive a need for these interconnects to meet demand for high-speed, low power data transmission in data centers.
Specifically, Q4’s report provided confirmation of the EML laser ramp, as EMLs achieved an all-time high for shipments with revenue more than doubling versus its June 2024 baseline. Management also cited a “substantial” 200G EML order to be fulfilled in the December quarter, although offered little additional clarity on the size of the order.
Revenue:
Lumentum reported Q4 revenue of $480.7 million, beating analyst estimates by a modest 2.29%. This was a notable uptick on the top-line, with growth of 13.1% QoQ, accelerating 7 points, and 55.9% YoY, accelerating 42 points.
AI Segment:
Cloud & Networking Q4 revenue came in at $424.1 million, representing 66.5% YoY growth. Additionally, the segment’s QoQ growth of 16.1% accelerated sharply from 7.6% QoQ in Q3, coming in much stronger than Coherent, where growth decelerated from 10% QoQ to 5%.
Earnings:
Lumentum reported adjusted EPS of $0.88 in Q4, beating estimates of $0.81 and improving against the $0.57 reported in Q3 and $0.06 reported in the year ago quarter.
Q4’s adjusted net income margin of 13.1% reflects continued operational improvements and the fourth consecutive quarter of sequential improvement (from 9.6% in Q3, 7.5% in Q2, and 3.6% in Q1).
Margins:
Q4’s GAAP operating loss of ($8.4 million) represents a (1.7%) operating margin, compared to (8.9%) in Q3 FY25 and (43.3%) in Q4 FY24.
Non-GAAP operating margin of 15.0% expanded nicely compared to prior quarter of 10.8% and prior year quarter of (5.1%).
Cash:
For FY25, operating cash flow was up ~5x to $126.4 million, for a 7.7% margin, improving from a 1.8% margin in FY24.
Valuation:
Lumentum’s Forward PS ratio is above what it typically trades at 5 forward PS. It’s previous top this year was 4 forward PS and it failed to hold that valuation twice.
The company’s forward PE Ratio is 33 and is within a reasonable range as the company has traded as high as 50-60 and as low as 18 over the past 12 months.
Notable Risks:
Lumentum has many competitors and its supplier agreements are not as clear to our firm as Credo and Astera Labs. However, the fundamentals are communicating that Lumentum’s management commentary is reliable, which is that their EML lasers are in high demand.
AI Software and AI Data Layer
As it stands today, there are a handful of AI software companies and AI data layer companies that have set themselves apart in fundamental strength. As the introduction pointed out, the noise in stock investing is often quite loud and the rate at which stock tickers are exuberantly discussed is detached from reality. We do a fairly comprehensive scan and find the list of software stocks that offer concrete proof of participating in the AI trend is far fewer than one would originally imagine.
I’ve stated below that AI hardware companies will also be some of the best AI software companies (incoming tangent here):
To illustrate this, Broadcom is preparing to displace a few of the FAANGs at a $1.6T valuation compared to Meta at $1.8T valuation and Amazon at $2.3T valuation. This is on the cusp of Broadcom stating their serviceable addressable market will be $60B to $90B in revenue by fiscal 2027, which is in two years from today as their fiscal year ends in October (roughly 200% at the midpoint given the current $25B run rate).
One has to wonder, why do the AI software juggernauts not discuss their revenue as openly as the AI hardware companies, especially given software tends to be recurring and easier to model? On the other end of the spectrum, we see up to 100X forward sales valuations in this category.
Yet, if we look at Nvidia AI systems as an example, it’s easy to see that software is what will accelerate AI over the next few years. There are four layers to Nvidia’s full-stack accelerated computing: hardware, system software, platform software, and applications. When you consider Blackwell, for example, there are transformer engine libraries, integrated with CUDA and cuDNN that determine when to use FP8 versus FP4 during training or inference to maximize speed yet also maintain accuracy. Software determines which precision to use for every layer and every tensor, which helps to deliver a 2X to 4X improvement in training and inference. There are additional revenue segments, such as automotive and robotics, where Nvidia will be able to license its software and gradually add this high-margin revenue to its already profound stock market performance.
If you look below, you’ll also notice that I’ve put Oracle in the software/data layer category for two quarters now. Although it's Oracle Cloud Infrastructure (OCI) driving growth of 55% and expected to accelerate to 77%+ growth and also strong RPO of 359%, it’s the software and data layer that separates Oracle from other cloud infrastructure providers. For example, software-defined remote direct memory access (RDMA) helps to lower latency and increase bandwidth by bypassing the CPU during training and inference tasks. Oracle says it “consistently charges less than Amazon Web Services (AWS) for the equivalent compute capacity.” But we have to look closely at why that is.
Vectorized data is another area where Oracle sets itself apart using the data layer as it allows both structured and unstructured data to be understood by AI models, which helps to increase the use of private data alongside public data.
As we approach the AI software/AI data list now and into the future, the following should be kept in mind:
Flexibility in what defines an AI software winner (wait until you see my #2 below)
It can’t rank until we see real, tangible evidence of a stock participating in the AI trend. Why pre-emptively invest in a company that is not reporting strong AI revenue when there are so many choices on the market? (Looking at you Tesla Optimus fans).
I believe the stocks listed below fit this criteria.
#1: Reddit is an AI Data Layer Frontrunner
Financials: 10/10
Thematic: 7/10
Valuation: 4/10
If you had asked me a few years back to describe Reddit’s value proposition, descriptions like “front page of the internet,” “world’s largest forum,” “best place to query the crowd” may have come to mind. Yet, there is something far more valuable that Reddit provides in the AI era, which is a continuous supply of human-generated conversations. To some extent, Reddit is transforming into a human data farm for AI models as it provides continuous, conversational data. What was once a forum is now a wealth of opinions and loads of sentiment that AI models desperately need to produce more natural and sentient-sounding responses.
Therefore, Reddit ranks higher than you might originally guess going into an AI software discussion. The reason for the high-ranking is that Reddit’s forum-based, real-world discussions are at the top of the list of data sets that can help advance AI models. Therefore, this is less about forum users and more about the licensing of data (and what Reddit gets in exchange).
Google clearly agrees as the company is licensing data from Reddit, and what’s interesting about this partnership is that it’s easy to see Google lacks highly contextual, human sentiment type data that social platforms provide. Meta, for example, has something similar to Reddit – whereas Google does not have this social aspect from search. In exchange, Google is boosting Reddit in search results.
Although the world’s leading forum site has only 416 million weekly active users compared to Facebook’s 2 billion, it ranks fifth behind Facebook as the most visited site in the United States. In addition, due to a few changes in how Google surfaces content with AI overviews, Reddit is now the second most visible site in the United States – ranking above Facebook for example – and the top line results show the company is reaping the rewards of being in the search giant’s good favor.
Over the last two years, Reddit has seen an explosion in SEO visibility on Google, with data from Sistrix placing growth from July 2023 to April 2024 at a whopping 1,328%. This moved Reddit from 85th most visible site to the 7th most visible.
Now, as of October 2025, Reddit has moved to the 3rd most visible site in the US, per Sistrix, behind Wikipedia and YouTube, and ahead of popular sites such as Facebook in 7th and Amazon in 4th. This major improvement in SEO ranking may be a potential contributor to Reddit’s accelerating growth over the past five quarters – yet as stated, the surge in growth is out of Reddit’s control and relies on Google SEO placement, which could change at anytime.
In terms of user engagement, Reddit notched 3.8 billion visits as of September 2025, according to data from Similarweb, compared to 11.4 billion for Facebook, 6.5 billion for Instagram, and 4.3 billion for X. Users visited an average of 4.77 pages per visit with an average visit duration of nearly 6 minutes, compared to 12.18 pages per visit and an average duration of 10 minutes for Facebook. Similarweb places Reddit as the fifth-most visited site in the US, behind Facebook in fourth place.
This rise in search ranking has created a fundamentals profile that is hard to ignore.
Revenue:
Reddit delivered a rather impressive Q2 on July 31st with revenue beating estimates by more than 17%.
Q2 marked Reddit’s fastest growth since the start of 2022, and a significant improvement over the past two years from just 12% growth at the start of 2023. What’s even more impressive is that Reddit delivered this 77.7% growth on top of a rather difficult 53.6% comp, yet this may shape up to be the peak growth quarter for the year as comps get tougher.
AI Segment:
Behind the substantial revenue beat in Q2 was 84% growth in advertising revenue to $465 million. This marked a sharp 23 point acceleration from 61% growth in Q1. Sequentially, advertising revenue grew almost 30%, with growth of more $106 million QoQ, outpacing Q4 2024’s $79 million sequential increase.
There was 50% YoY growth in active advertisers as it continued to acquire new advertisers. Additionally, performance ads and brand ads both increased more than 80% YoY, reflecting strong engagement from advertisers on the platform.
This past quarter marked the highest sequential growth in ARPU in more than three years at 25%, outpacing even Q4 24’s 18% growth. Management believes global ARPU is “still low on an absolute basis and remains an opportunity” for long-term improvement – for example, Meta’s global ARPU is around 3x of Reddit’s at $13.65 as of Q2, and though Meta hasn’t updated regional metrics since the end of 2023, it’s possible that US ARPU is 10x that of Reddit’s.
Earnings:
Over the past month, consensus EPS estimates through Q4 2026, or the next six quarters, have been revised 23% to 66% higher; over the past six months, estimates have moved 20% to 57% higher, as margins strengthen. For example, Q2 2026 has seen its estimate move from $0.42 to $0.70 over the past month, and Q3 2026 from $0.55 to $0.83. This now projects three consecutive quarters of triple-digit YoY growth followed by three consecutive quarters of >50% growth.
Margins:
Gross margin was 90.8% in Q2, up 1.3 points YoY and marginally higher QoQ.
Operating margin was 13.6%, up nearly 25 points YoY and 12.6 points QoQ. Notably, this also exceeded Q4 2024’s operating margin of 12.4%.
Cash:
Operating cash flow was $111.3 million in Q2, up 292% YoY. OCF margin was 22.3%, down 10 points QoQ but up more than 12 points YoY. Free cash flow and operating cash flow are correlated nearly 1:1. Free cash flow was $110.8 million in Q2 for a 22.2% margin.
Valuation:
Reddit has some room at 19 forward PS but not a ton of room before it will test its previous top at 24. After about a 25% move from here, Reddit will be testing a level it has not held two times in the past year. The stock has traded as low as 8.3.
The bottom line valuation is harder to identify a trend as the company was not GAAP profitable when it went public, although is firmly GAAP profitable now. The stock trades at 52 forward PE yet has traded as high as 70 forward PE and as low as 25 in recent quarters.
Risks:
Reddit’s primary risk is the surge in traffic relies on a third-party relationship with Google that could be terminated at anytime. It may not be terminated given the emphasis on contextual data for models, yet the recent success hinges on this data licensing deal.
#2: CoreWeave: Beating the Big 3 on Utilization Rates
CoreWeave is perhaps the first stock in the history of the market to have short sellers before it was a public listing. While it’s risky and novel to collaterize GPUs, we want to remain open-minded as CoreWeave’s software-defined infrastructure is one of the most advanced in the industry for AI workloads.
Please note, CoreWeave carries outsized risk as there is high debt leverage, negative free cash flow and is not profitable. The stock’s thematic ranking is quite high, yet we would only approach this stock using technicals for risk management.
The company’s software stack is specifically designed to maximize GPU utilization, elasticity, and cost efficiency to the point of booking out GPUs like it’s a leading hyperscaler. Remarkably, it’s the first company in almost twenty years to meaningfully disrupt the dominance of the Big Three in cloud infrastructure.
By focusing only on GPUs and software optimizations, CoreWeave offers bare metal servers at a cost that is up to 20% to 50% cheaper than hyperscalers. Its value proposition is best summarized in its utilization rates for GPUs. CoreWeave has published its MFU rate of 35% to 45%, stating it is 20% higher than competitors, which means other AI data centers have MFU rates more in the 30% range. Due to FLOPs performing an astronomical number of calculations, small percentages translate to an important advantage.
The company is able to scale quickly with new GPUs due to the Mission Control automation layer that provides automated deployments of systems like the GB300 NVL72s. The company stated: “Mission Control continues to be the cornerstone of CoreWeave's ability to scale at breakneck speed, building a fully automated and rigorous process for cluster life cycle management with unmatched visibility for our customers.”
CoreWeave also offers a Virtual Private Cloud for a private network space. By combining an isolated virtual private cloud with Nvidia’s Quantum InfiniBand, customers get ultra-low latency with enhanced security.
The company's Kubernetes Service is an AI-optimized Kubernetes environment for scheduling AI workloads and scaling up/down for the right mix of CPU, GPU, memory and storage (known as elasticity). SUNK, known as Slurm on Kubernetes, combines container orchestration with a job scheduler to manage large batch jobs. AI labs use this service to combine scheduling for high performance computing with a cloud-native environment.
Local Object Transport Accelerator (LOTA) for AI object storage is another feature that is optimized for AI workloads by focusing on performance and cost efficiency. The company recently added archive tier object storage, which allows data to move between hot and cold storage based on access patterns, which optimizes costs. In the recent earnings call, the company stated they are seeing customers “shifting petabytes of their core storage to CoreWeave in the form of multiyear contracts.”
CoreWeave recently completed its acquisition of Weights & Biases in May to add observability, which means engineers can quickly diagnose a failure or inefficiency in the software layer and infrastructure layer. For example, if a model is training slowly, the observability platform will help an AI engineer identify and resolve this quickly.
More recently, CoreWeave integrated W&B for a joint launch of its Inference-as-a-service feature, which allows developers to use APIs to tap into AI models from OpenAI, Meta, DeepSeek, etcetera. Inference is key for CoreWeave to fully monetize its investments in capex-heavy infrastructure. For example, these popular LLMs combined with chain of reasoning inference, which means generating step-by-step reasoning, will become compute-intensive especially at scale. This will lead to CoreWeave monetizing every chatbot response, API calls and applications to easily payback their initial investments plus some (in time).
The paragraph above is quite important and worth summarizing – essentially, CoreWeave’s path to monetization will become clearer as the inference market takes off in the coming quarters/years. What you see is not what you get, rather what you see could surge should CoreWeave execute well – and find financing.
This stock comes with outsized risk. Please note the cash/debt section below. Despite being in the Pro tier’s Top 15 due to a strong thematic, this is one we will hold only with close risk management.
Revenue:
CoreWeave reached a new milestone of over $1.0 billion in revenue in Q2 2025, growing 206.7% YoY to $1.21 billion. On a sequential basis, the Q2 revenue grew by 23.6%. The company beat analyst consensus estimates by 12.2%, driven by strong demand for the company’s AI cloud infrastructure services.
Looking forward, revenue is expected to grow by 174% YoY to $5.26 billion in the year 2025 and 129.6% YoY to $12.08 billion in 2026 and 48.3% growth in 2027. Most importantly, management has increased the full-year revenue guidance for the second quarter in a row due to the strong customer demand. Management increased guidance by $250 million at the midpoint to a new range of $5.15 billion to $5.35 billion for the year 2025.
AI Segment:
The company’s backlog was $30.1 billion at the end of Q2, up 86% YoY driven by the company’s strategic deal with OpenAI in March 2025 and the signing of subsequent expansion deals with the company.
The company derived 77% of 2024 revenue from its two largest customers, i.e., Microsoft and Nvidia. While in the recent quarter, Microsoft accounted for 71% of the total revenue. Goldman Sachs estimates that Microsoft’s share is expected to drop to 38% in 2026, followed by OpenAI at 21%, Nvidia at 6%, and the remaining 35% to be attributed to other customers.
Earnings:
The company reported GAAP loss per share of (-$0.60) in Q2 compared to the analyst consensus estimate of (-$0.49), missing estimates by –21.7% due to the higher operating expenses, particularly the technology and infrastructure expenses.
Analysts expect GAAP loss per share of (-$2.67) for this year, followed by (-$0.90) for 2026. They expect a positive GAAP EPS of $1.59 in 2027.
Margins:
Q2 gross margin was 74%, up 200 basis points YoY and up 100 basis points QoQ. The company is investing heavily in data center and server infrastructure to meet very strong AI demand from its customers. The operating margin was 2%, compared to 20% in the same period last year and (3%) in Q1. Operating expenses increased by 276% YoY, driven by high technology and infrastructure expenses. The management tried to explain in the Q2 earnings call that expenses are front-loaded and have a short-term impact on the margins.
Cash:
CoreWeave’s business model is based on aggressive capacity expansion, currently fueled primarily by debt. As a result, cash is rather thin and gets spent quickly, and free cash flow is widely negative.
CoreWeave reported $1.15 billion in cash and equivalents (excluding $0.56 billion in restricted cash and equivalents), though CoreWeave updated in an 8-K related to its now upsized $1.75 billion raise that total cash will be closer to $5 billion.
Operating cash flow was ($251.3 million) for a (21%) margin in Q2, widening from ($117.8 million) in the year ago quarter. Free cash flow was ($2.7 billion) for a (223%) margin, widening slightly from ($2.36 billion) in the year ago quarter.
Debt was reported at $11.05 billion in Q2, with $3.62 billion being current. Current debt is likely closer to $12 billion now, with a majority (~$6.7 billion) tied to its two existing delayed draw term loan facilities; if the new DDTL is drawn upon, debt could rise to $14.6 billion. On the other hand, the company is trying to reduce its cost of capital by raising cash through secured debt, using its highly valuable GPUs as collateral, which is positive.
As discussed above, capex for the second half of the year is expected to be >$15 billion, with cash on hand only covering one-third of that at maximum. This will place the emphasis on finding alternative funding to finance this spending.
Valuation:
CoreWeave’s valuation of 12.4 forward PS is at the higher range for where the stock has traded this year. The stock traded as high as 15 forward PS and as low as 3.4 during April.
The company is not profitable for a bottom line valuation and is the only stock on our list that is not profitable.
Risks:
Cash/Debt is a very high risk with debt at $12 billion and growing with only $1.15 billion in cash and negative free cash flow.
#3: Oracle Inflecting with 8% QoQ Growth, RPO of 359% Soars and The Future is Bright
Financials: 7/10 (other segments weigh on AI segment)
Thematic: 7/10 (not a pureplay)
Valuation: 2/10
As stated in the intro, even though bare metal servers are a large part of the story for both CoreWeave and Oracle, it’s also the software-defined optimizations that set these cloud infrastructure players apart from the Big 3. For Oracle, it’s also RDMA, which is a networking fabric that increases performance and density, plus the data layer (of course) as Oracle is able to leverage its deep roots in relational databases. Inference will need a mix of private data sets to augment public data sets to fine-tune reasoning tasks in an effective way for enterprises to use the models internally, and Oracle is positioned to capture this opportunity.
Oracle offers the widest range of bare metal GPU instances among major cloud providers, and scalability at any size up to 65,536 Hopper GPU clusters and 131,072 B200 GPU clusters, which are expected to come online in 2025. Oracle also offers very flexible VM instances, letting customers pay for only the capacity they need as they need it for any size workload, rather than offering fixed instance sizes.
With less overhead and fewer CPU cycles, RDMA helps Oracle offer its AI clusters at a lower cost: Oracle says it “consistently charges less than Amazon Web Services (AWS) for the equivalent compute capacity.”
Oracle says that it can offer less than 10 microseconds of latency between nodes, improving efficiency. In the most recent earnings call, Oracle emphasized how cheap they are compared to the Big 3, stating: “We have gotten the entire Oracle Cloud, the whole thing, every feature, every function of the Oracle Cloud down to something we can put into a handful of racks, 3 racks, we call it Butterfly that costs $6 million. So we can give you a private version of the Oracle Cloud with every feature, every security feature, every function, everything we do for $6 million. I think the cost for the other hyperscalers is more than 100x that.”
Oracle’s AI vector capabilities also stand out given Oracle’s database roots, offering native AI vector search capabilities with seamless integration to leading AI models from OpenAI, xAI, Meta, Cohere and more. AI vector search lets enterprises search both structured and unstructured data in a variety of manners, enabling intelligent, relevant and accurate AI responses utilizing their data.
The announcement of Oracle’s AI database is particularly interesting in terms of ways the stock can extend its run. As explained in the earnings call, the combination of vectorizing data to where it can be understood by AI models with the ability to connect private databases to AI reasoning models will result in enterprises unlocking higher value from AI.
Oracle is teasing a more beefed-up AI database, which management stated will officially launch at Oracle World Cloud, describing a combination of private enterprise data, large reasoning models and automated agents: “Who's offering that to customers? We'll be the first when we deliver it and demonstrate it at AI World next month.”
Oracle has already made major headway with AI embedded databases with 23ai, which converts vector data into contextual information. By connecting a database to Chat-GPT, there is more reasoning layered into the results.
The inference market will define by size and quality of data for reasoning purposes, and Oracle sits on arguably the world’s largest enterprise data sets. Although we have grown used to compute driving the AI training market, there will be an important shift toward the data layer driving the inference market.
With Oracle embedding the AI database, inference will happen inside the database where the data resides. This is distinct from pulling data out of the database into the large language model, which is inefficient. Oracle’s move to embed the database supports a sustained, upward trajectory in the stock price.
Revenue:
Oracle delivered Q1 revenue of $14.9 billion, growing 12% YoY but slipping 6% sequentially, coming in just shy of the Street’s $15.0 billion estimate.
AI Segment:
Remaining performance obligations (RPO) grew 359% YoY with cloud RPO growing “nearly 500%” on top of 83% growth last year. This compares to RPO growth of 41% YoY last quarter and cloud RPO growth of 83% last year.
Oracle Cloud Infrastructure (OCI) was forecast to “grow 77% to $18 billion this fiscal year and then increase to $32 billion, $73 billion, $114 billion and $144 billion over the following 4 years.”
You can think of this as an acceleration from roughly 50% growth on IaaS in recent quarters to up to 128% growth in future years, specifically from the $32B to $73B in the medium-term of two years out.
OCI (IaaS) revenue grew 55% YoY to $3.3 billion, faster than hyperscaler peers.
This led to multi-cloud database revenue with Amazon, Google, and Microsoft surging 1,529% YoY.
Earnings:
GAAP EPS of $1.01, down (15%) QoQ from $1.19 in Q4’25 and flat YoY vs. $1.03 in Q1’FY25. This figure was also lower than the analyst estimates of $1.04.
Non-GAAP EPS of $1.37, up 6% YoY from $1.39 in Q1’25 but down (14%) QoQ from $1.70 in Q4’25.
Margins:
Q1 figures represent a 29% operating margin, down from 32% in the prior quarter and 30% in the prior year quarter.
Cash:
(39%) FCF and 55% OCF. Highly leveraged cash to debt ratio
Valuation:
Oracle’s valuation is at the top range of where the stock has traded this year at 12 forward PS. The stock briefly touched 13 before selling off, and otherwise, has not traded higher than 12 this year (or for decades really, but is transforming into an AI stock)
The forward PE Ratio of 42 is the highest it’s ever traded this year or in previous years.
Risks:
Highly leveraged cash to debt ratio is the predominant risk.
#4: AppLovin Has the Best Operating Margins Sector-Wide
Financials: 10/10
Thematic: 7/10
Valuation: 3/10
APP has many aspects to focus on for the bull story, yet the margins are truly one-of-a-kind. It is hard to take the title of “most impressive margins” in a software category as AppLovin is up against the most operationally efficient, cash loaded companies worldwide.
Regarding its segments, management has repeatedly stated that gaming ads alone can sustain growth of 20% to 30% YoY. Therefore, the catalyst for the next few years is securing additional supply, such as e-commerce, as well as opening up the AXON ad platform to more advertisers.
The AXON ads manager recently became self-service, which means it can scale at levels not previously seen by offering self-service interface for AppLovin’s 1 billion reach. As of now, AppLovin is limited in the number of advertisers it can manually onboard. According to the opening remarks: “With the rollout going smoothly, we are ready to widen access. On October 1, 2025, we plan to open the AXON ads manager on a referral basis, perfectly timed for the holiday season. Feedback from these partners will guide our global public launch in the first half of 2026. To date, web advertising campaigns have been limited to the United States. On October 1, we plan to open our platform to most major international markets.”
Management also hinted these improvements will lead to “a lot of upside in the numbers we’re able to report.”
Here is the full quote:
“We expect that will increase the advertiser count quite quickly and also allow us to go through live examples of advertisers coming in self-service all the way to scale on our product. Assuming all that goes well, then we talked about opening up the platform entirely to the world in first half of next year. We think as advertiser count grows on our business, especially in categories outside of gaming, you're going to see a lot of upside in the numbers that we're able to report”.
The last earnings report was stellar with all fundamental boxes ticked, a team that has proven to execute, and incoming catalysts that are quite well-timed to where we maybe have two quarters (or less) to wait for an inflection.
And check out those margins!
Revenue:
AppLovin reported revenue of $1.26 billion compared to consensus of $1.28 billion according to some sources yet others reflect the consensus we had of $1.22 billion, thereby it’s debatable if the top line beat. Overall revenue last quarter was $1.48B versus this quarter at $1.26B. As we covered in the past, this is due to AppLovin divesting its mobile gaming “Apps” business, with the sale completed on June 30th. Therefore, if you adjust for this sale, revenue for the ads business in Q1 was $1.15B for QoQ growth of 8.7%.
AI Segment:
Same as revenue (AXON ad engine is powered by AI).
Earnings:
On the bottom line, the company had a large beat with EPS of $2.39 compared to $1.99 EPS expected, representing growth of 169%. This was a 45 point beat on growth rate for the bottom line. Adjusted EBITDA doubled to $1.02 billion, up from $943 million last quarter. This represents an adjusted EBITDA margin of 81%.
Margins:
Operating margin of 76% expanded from 44.7% last quarter and more than doubled from the year ago quarter at 36.2%. Wow!
Cash:
As management alluded to on the earnings call, the company “prints cash” with a 61.3% operating cash flow margin and a 61% free cash flow margin.
Valuation:
The company is trading at a PS ratio of 40.3 and a forward PS ratio of 37.6. The company’s forward PS ratio peaked at 46.7 on September 30 and is currently trading about 20% below its peak. While the forward PS ratio above 30 is considered high, the market is giving the company a premium valuation due to a remarkable turnaround in margins.
Another key catalyst is that the company completed the divestment of its low margin mobile gaming business in Q2 2025. The company’s adjusted EBITDA margin has swiftly moved higher from 68% in Q1 to 80.9% in the recent quarter. It would also make more sense to look at the EV/EBITDA ratio for AppLovin. It currently trades at an EV/EBITDA ratio of 63.4 and a forward EV/EBITDA ratio of 49.1. The company has traded at a peak EV/EBITDA ratio of 110 during June 2021 and around 88 during the tech market bubble in November 2021. With profitability improving post-divestiture, there could be further room for valuation expansion.
Risks:
APP is the subject of short reports and has a business model the SEC, short sellers and others find suspicious. However, we think the management team, AI-powered ads business model and strong market presence is worth a shot especially when using technicals alongside any entries or exits.
#5: Cloudflare is Locked and Loaded for AI Inference Market
The Workers Platform, known as Act 3, is positioned to take advantage of the massive inference trend. The I/O Fund team recently dug up a stat that inference is expected to account for 60% to 70% of AI workloads by 2030. In particular, Cloudflare emphasizes their position is what will help the company win this market: “The fact that we sit in front of so much of the web and that more than half of our dynamic traffic is already between APIs means that we are strategically positioned to deliver the agentic web of the future.”
Revenue:
Cloudflare reported its largest revenue beat in the last six quarters at 2.1% above consensus, with Q2 revenue up 27.8% YoY to $512.3 million. This also marked a slight 1.3 point acceleration on the top-line from 26.5% growth in Q1.
For Q3, Cloudflare guided for revenue of $543.5 to $544.5 million, ahead of estimates at the time for $538.9 million. This corresponds to a slight deceleration to the mid-to-high 26% YoY growth range, where Cloudflare is expected to remain through Q4. This provides no clear indication yet that the company is able to drive a sustained revenue acceleration aided by AI.
Current RPO accounted for 66% of total RPO, or ~$1.30 billion, increasing 33% YoY in Q2, a four point acceleration from 29% growth in Q1. This is also a notable uplift from 26% growth in the year ago quarter.
AI Segment:
There is no official AI segment yet.
Earnings:
Cloudflare topped estimates in Q2 driven by the revenue beat and stronger adjusted margins, and boosted its FY25 adjusted EPS outlook as a result.
GAAP EPS was ($0.15), missing estimates for ($0.08) as GAAP margins drifted lower.
Adjusted EPS was $0.21, beating estimates for $0.18, fueled the outperformance in adjusted operating margin in the quarter.
Margins:
GAAP gross margin was 74.9% in Q2, down nearly 3 points YoY and 1 point QoQ. Adjusted gross margin was 76.3%, down 2.7 points YoY and 0.8 points QoQ.
GAAP operating margin was (13.1%), down 4.4 points YoY and 2 points QoQ. Adjusted operating margin was 14.1%, approximately flat YoY and up 2.4 points QoQ; this was also ahead of guidance for 12.6%.
For Q3, Cloudflare guided for adjusted operating income of $75-76 million, pointing to adjusted operating margin of 13.9%, down nearly 1 point YoY and moderating slightly QoQ.
Cash:
Operating cash flow was $99.8 million for a 19% margin, flat YoY but down from a 30% margin in Q1.
Free cash flow was $33.3 million for a 6% margin, down 4 points YoY and 5 points QoQ.
Network capex was 11% of revenue in Q2, down from 17% of revenue in Q2. Cloudflare stuck to its guidance for network capex to be 12-13% of revenue for the year, suggesting slight moderation in 2H.
Valuation:
Cloudflare is officially trading at its highest forward since the tech bubble popped in late 2021-early 2022. The current PS of 40 and forward PS of 36 does not offer much support in terms of the stock holding well at these levels.
Cloudflare is trading at a wild forward PE ratio of 255, which is far above any forward PE ratio over the past year (the previous highest forward earnings ratio was 205 where it sharply reversed twice).
Risks:
Valuation is the predominant risk coupled with little evidence of real AI revenue right now (more of a future winner that we want a placeholder for).
AI Energy
#1: Right Place, Right Time for Bloom Energy
Bloom Energy provides on-site 24/7 power generation using their proprietary solid oxide fuel cells (SOFCs). The SOFCs are stacked up by the hundreds to thousands in Bloom Energy Servers (BES), which enable the conversion of fuels like natural gas, biogas and hydrogen to electricity.
Bloom Energy is securing data center deals due to fast deployment of about three months. Here is what management described as the competitive advantages regarding time to power for fuel cells: “A big shift in our business today is time to power. We are providing solutions to meet the urgent needs of our customers who cannot fulfill their power needs from the grid. In these cases, we rapidly book, build, ship, install and power sites for our customers in a matter of months, a much faster timeline than a grid connection. Such rapid drill activities will necessarily come with timeline variances, both pull-ins and delays, and will affect our quarterly revenue line. You are seeing this in our Q3 numbers.”
Although we expect Bloom to be a very volatile stock, the fact is that very few alternative energy companies can move as quickly as BE in what our firm has dubbed an energy crisis in getting power to data centers.
As the CEO stated on the call, to wait 5-7 years is “untenable.” To compare, Bloom will power Oracle with on-site power solutions in as soon as 90 days. Additional key customers for BE include American Electric Power (AEP), Quanta and Equinix. Notably, Amazon and Cologix are customers of Bloom through AEP in Ohio.
Perhaps the most important statement on the earnings call was when the CEO stated: “We expect new orders from other AI hardware ecosystem players soon, complementing demand we see from our more traditional commercial and industrial customers.”
Revenue:
Bloom reported a nearly 6% beat to estimates in Q2, reporting $401.2 million in revenue versus estimates for $378.9 million. Revenue grew 19.5% YoY, slowing from 38.6% growth in Q1.
AI Segment:
Product revenue increased 31.1% YoY to $296.6 million, slowing from 38% growth in Q1. As the CEO stated on the call, to wait 5-7 years is “untenable.” To compare, Bloom will power Oracle with on-site power solutions in as soon as 90 days.
Earnings:
Adjusted EPS of $0.10 beat estimates for $0.02, and represented a notable $0.16 improvement YoY.
GAAP EPS was ($0.18), missing estimates for ($0.10) as GAAP net margin declined sequentially.
Margins:
GAAP operating margin is approaching positive territory at (0.9%) in Q2, up nearly 5 points QoQ and 6 points YoY.
Adjusted operating margin was 7.1%, up more than 3 points QoQ and 8 points YoY. Adjusted EBITDA was $41.2 million
Cash:
Free cash flow was ($220.4 million) in Q2 for a (54.9%) margin. For the first half, FCF was ($345.3 million), just over a 1% improvement YoY.
Unrestricted cash and equivalents totaled $574.8 million, down from $794.8 million in Q1. This raises the risk that Bloom will turn to financing methods as Bloom likely awaits cash flows meaningfully improving in Q4.
Valuation:
If you want to see what AI can do for a valuation, look no further than BE. The stock used to trade in the range of 1.5 forward PS for many years and is now trading at a 14.5 forward PS. I can’t offer much in terms of where the valuation trends as it’s far beyond anything BE has traded at historically.
However, the forward PE ratio offers a bit more data as the forward PE of 219 is an area the stock has traded at consistently over the past year. With that said, the stock has traded as low as 33 forward PE, as well.
Risks:
The valuation is a risk yet we are less concerned with what BE does as an individual stock and would see any selloff being more of a broader catalyst.
#2: GEV: All Roads Lead to GEV
GE Vernova is the world’s largest gas turbine supplier at 25% ahead of Schneider at 24%. Even still, GEV nearly tripled its gas turbine equipment in the second quarter – a statement that has us sitting up in our seats. Per the earnings call: “Power orders grew 44%, led by Gas Power equipment nearly tripling year-over-year.”
Also, consider that we have been covering Bitcoin miners and other energy sources that can quickly help hyperscalers secure powered shells in the 1GW to 3GW range – yet GEV has 62 GW in backlog for gas equipment contracts, already surpassing expectations of reaching 60 GW by the end of this year. In other words, the chances that GEV is not a significant player in supplying energy to data centers for many years to come is nil.
In a bid to supply options quickly to alleviate bottlenecks, GEV is also shipping aeroderivative gas turbine packages and doing extensive R&D on a small modular reactor (SMR) design. As detailed below, how exactly GEV evolves to solve the crucial bottleneck around AI power consumption is not set in stone, rather the company is experimenting rapidly with how to leverage their deep experience in natural gas, electrification and renewables like wind to meet global demand.
This year, the company is expected to report $37 billion in revenue with strong earnings growth of 45.3%. The stock is not a hypergrowth profile, rather, it is a quality, defensible position that could do well during any periods of doubt in the broader AI trend.
The defensibility is particularly attractive when you consider that gas turbines is the crux of the issue for expanding gas power plants. According to Bloomberg’s calculations, more than “$400 billion worth of gas-fired power plants through the end of the decade are in jeopardy of delay or cancellation because of the lack of capacity to meet future turbine orders.” The same article points toward GE Vernova filling orders as far out as 2030.
Revenue:
The company’s Q3 revenue grew by 12% to $9.97 billion, beating estimates by 8.8%. Organically, revenue grew by 10% YoY to $9.83 billion, driven by strong electrification and power. Analysts expect revenue to decline (2.3%) YoY in Q4 but rise 3% QoQ, before rebounding to 8.7% in Q1 2026.
On the back of strong demand for power and equipment, management reiterated full-year revenue guidance and expects 2025 organic revenue to come at the high end of the guidance of $36 billion to $37 billion. The power segment organic revenue guide was maintained at 6-7%, and electrification segment was raised to ~25%, up from 20% previously in Q2. On a side note, the wind segment is expected to be down high-single digits, lowered from down mid-single digits due to the more challenging market conditions.
Revenue growth is set to accelerate over the next three years. Analysts expect a 6.6% increase in 2025, bringing the total revenue to $37.2 billion. Momentum is projected to build further, with revenue climbing to $41.0 billion in 2026, up 10.2% and to $46.7 billion in 2027, up 13.9% YoY.
AI Segment:
In Q3, GEV signed just over 12GW of new gas equipment contracts with ~1GW going directly to orders and ~12GW going into what’s called a slot reservation agreement. During the quarter, the company also converted 7GW of SRAs into orders and shipped 4GW of equipment.
Management had previously stated they would exit the year with 60GW “at better margins with significant momentum into ‘26.” Here is the breakdown from that comment:
33GW are in the backlog, up 4GW sequentially.
Slot Reservation Agreements (SRA) grew from 25GW to 29GW.
Total backlog including SRAs is 62GW, up from 55GW last quarter.
There was a discussion in Q2’s call that this represents about 3 years of backlog: “And we've talked about the fact that we'll get to at least 60 gigawatts by the end of the year. So that's directionally 3 years of backlog.” There was mention of eventually seeing 80GW to 100GW in the backlog but no date or other details were discussed, other than that’s the goal over time.
Earnings:
Q3 GAAP EPS came at $1.64, missing estimates by a (18.6%) but up from ($0.35) in the year-ago quarter. Despite the miss in Q3, earnings growth is projected to be strong over the next few quarters, with GAAP EPS up 87.3% to $3.24 in Q4 and accelerating to 128.6% to $2.08 in Q1 2026.
Analysts continue to expect strong EPS growth in the coming years. For 2025, analysts expect GAAP EPS to grow 40.9% YoY to $7.86, and 62.8% and 44.8% YoY in the subsequent two years, reaching $18.53 in 2027.
Margins:
Q3 adjusted EBITDA grew by 234% YoY to $811 million, driven by strong growth in the electrification and power segments, and a strong rebound in wind. Adjusted EBITDA margin improved 540 basis points YoY to 8.1% driven by profitable volume, better pricing, and productivity gains. The company is witnessing an annual EBITDA margin expansion, increasing from 2.4% in 2023 to 5.8% in 2024, and management has further guided expansion in the range of 8-9% for 2025.
Q3 operating margin was 3.7%, down slightly from 4.2% in Q2 but up from (4.0%) in the year ago quarter.
Cash:
Most importantly, management has maintained its FY free cash flow guidance from Q2, where it was raised from a range of $2.0 billion to $2.5 billion to a new range of $3.0 billion to $3.5 billion. This was primarily driven by a higher profit outlook and increased down payments due to rising orders. Through Q3, the company has generated $1.9 billion in free cash flow, implying a strong Q4 to finish the year inside its guided range.
The company generated free cash flows of $732 million in Q3 compared to $968 million in the same quarter last year. On a sequential basis, this was a sharp increase from $194 million in Q2.
Valuation:
Similar to other names in this group, GEV’s valuation has reset higher as the company demonstrates clear AI product–market fit. The stock is trading at 4.2 forward PS despite trading as low as 1 earlier this year.
The forward PE ratio offers room in the valuation. The stock is trading at 83 forward PE ratio yet has traded as high as 130 in the past.
Risks:
Of the companies featured here, GEV offers less risk as the sheer GWs it can provide are desperately needed for AI data center buildouts. However, it’s not a hypergrowth stock like the others. It’s included here to say – we are eyeing the stock as one that could outpace the legacy FAANGs for example, as it’s a leader within one of our largest and most timely thematic trends.
#3 Bitcoin Miners
Bitcoin Miners offer an exceptional risk/reward as these companies are pivoting from unprofitable Bitcoin mining operations to the high margin business of supplying powered shells for AI data centers. According to CoreWeave, powered shells are the biggest constraint in the AI build out, marking a critical shift to where Nvidia’s GPUs are no longer the primary constraint. With existing power, cooling, and network infrastructure, miners can deploy AI-ready capacity faster than new construction—offering Big Tech a critical shortcut in the race to scale.
Our Discovery Tier highlights a Bitcoin miner that is expected to report 325% YoY growth in the upcoming quarter with a positive operating margin. Knox sees significant upside and is ranking this Bitcoin miner as his #1 among all Bitcoin Miners.
The best way to train for a marathon is to run a marathon. You can consider this report a training exercise while the real test is portfolio returns. Hopefully after reading this 43-page report, a few things are evident – we have a strong grasp of where the AI market plans to go next, that our process is nimble enough to capture winners in niche micro-trends, and we are capable of offering a level of conviction rare among AI investors. Consider that Nvidia is not one of our top performers this year (so far) – yet it’s been one of our strongest years to date.
With one quarter left, we look forward to making it our best of 2025.
Want to know what the I/O Fund is eyeing next for a new entry? Our Discovery tier surfaces new ideas in an effort to provide a significant edge to AI investors. Last week, we published our Top 10 New Ideas List for our Discovery members that pinpoints the stocks we are most likely to add to our portfolio next. Discovery was first to discover Bloom Energy, Core Scientific and Oklo to our portfolio, and these new additions became some of our biggest wins (thus far) in 2025. Current Pro and Advanced Members: To subscribe to Discovery with 30% off, click here to email us or email premium@io-fund.com and mention code DISCOVERY30.click here to email us or email premium@io-fund.com and mention code DISCOVERY30.
Advanced Members – stay tuned! This week, you will be receiving technical setups from Knox in his Quarterly Positions Report offering a complete picture of how we plan to enter or layer into the stocks listed above. These setups will also be covered in Knox’s upcoming webinar this Thursday at 4:30 PM Eastern. If you’re a Pro Member wanting information on Advanced or Discovery tiers, please email us at premium@io-fund.compremium@io-fund.com
Damien Robbins and Royston Roche, Equity Analysts at I/O Fund contributed to this analysis.
Please note: The I/O Fund conducts research and draws conclusions for the Fund’s positions. We then share that information with our readers. This is not a guarantee of a stock’s performance. Please consult your personal financial advisor before buying any stock in the companies mentioned in this analysis.
Taiwan Semiconductor (NYSE: TSM) recently announced fiscal Q3 earnings, stating its longer-term AI revenue outlook is stronger than anticipated. The company reported record Q3 revenue of $33.1 billion, which was up by a solid 40.8% YoY and 10.1% sequentially.
Over the last few months, the stock has surged as it’s up 50% since early June. At the time, we covered the stock in the article “Taiwan Semiconductor Stock: AI Growth Amid Geopolitical Risk”, we provided a scenario that could take the stock toward the $300s: “If we see the next drop take the form of a 3-wave drop that holds over $146, then we could be setting up for a rally toward the $300s.” At time of writing, the stock was at $203 and touched $304 on October 15th.
For Taiwan Semiconductor (TSMC), the outlook is two-fold. On one hand, TSMC is deepening its moat with advanced nodes, such as N2 and A16. The company already powers tens of trillions in market cap on the stock market when you consider Apple, Nvidia, Broadcom, Amazon, AMD and Google are customers of TSMC. Essentially, all mega cap stocks have an AI strategy spanning merchant GPUs and custom silicon, and of course, software – yet the common denominator to these strategies is they all funnel into TSMC.
Yet, on the other hand, many pundits have been asking – are we in an AI bubble? While AI leaders, related suppliers and strong R&D teams march onward with exceptional earnings reports, ongoing deal announcements and provide strong commentary that AI demand greatly outstrips supply – investors are left wondering what the true value is for the outsized growth. They say a picture is worth a thousand words and TSMC’s valuation chart certainly fits that description:
Although no risk management plan will be entirely error-free, our firm has an outstanding 5-year record of handling the tech sector’s inevitable volatility. Whether it’s the trade plan we provided for free on TSM to our stock newsletter subscribers or the weekly trade alerts and webinars we offer behind our paywall on dozens of lesser-known AI stocks – we always have a plan as we seek to maximize the upside, while protecting to the downside.
Below, we discuss the key items to track whether TSMC’s moat is intact – even as Intel increasingly becomes a United States darling with $11B in funding – plus not-to-miss highlights from the most recent earnings report. Last, we won’t leave you hanging with a note on how we managed our position after the strong run-up in June.
TSMC Reports Record Q3 Revenue Growth of 40.8% on Surging AI Chip Demand
TSMC revenue beat management guidance by 2.2% primarily due to robust AI-related demand.
Riding strong AI chip demand, TSMC once again boosts its full-year revenue growth guidance by 5 percentage points.
TSMC reported a record Q3 revenue of $33.1 billion, up 40.8% YoY and 10.1% sequentially. The growth was primarily driven by strong demand for its leading process technologies (including 3nm, 5nm, and 7nm process nodes) due to the continuing demand for AI and high-performance computing chips.
prompt
TSMC guided Q4 revenue of $32.2 billion to $33.4 billion, representing YoY growth of 22% and down (-0.9%) sequentially. Although management emphasizes strong AI-related demand, the guide shows a break in strong QoQ HPC revenue growth (more on this below).
TSMC’s Q3 revenue rose 40.8% year-over-year, driven by strong AI and high-performance chip demand.
TSMC boosted full year revenue guidance by 5 percentage points for the second consecutive quarter to mid-30% on continued strong AI demand. This is up from the 30% growth provided in Q2.
Management sounded very optimistic during the Q3 earnings call about long-term AI growth opportunities. As discussed below, AI momentum is stronger than expected. The growing adoption of AI consumer models is driving strong demand for AI chips. Similarly, enterprise AI demand is strong, and the rising sovereign AI opportunity will become another strong growth driver.
TSMC: AI Accelerator Growth to Hit Mid-40% CAGR by 2029
Management commentary suggested the chipmaker’s AI outlook through 2029 may be revised higher than the previously-guided mid-40% CAGR, though TSMC declined to provide an updated figure, opting to hold off until early 2026:
“AI demand actually continues to be very strong, [it's] stronger than we thought 3 months ago, okay? So in today's situation, we have talked to customers and then we talk to customers' customer.
So the CAGR we previously we announced is about mid-40s. It's [still] a little bit better than that. We will update you probably in beginning of next year so we have a more clear picture. Today, the [numbers] are insane.”
TSMC Reports Softer HPC Growth After Historic Q2 Gains
TSMC reported a softer QoQ increase in HPC revenue in Q3 after the chipmaker posted its largest sequential growth in Q2 at nearly $3 billion. Q3 saw sequential growth of just over ~$800 million in USD terms, or ~5% QoQ, versus over 20% QoQ growth seen in Q2. However, Q3’s USD-visible growth continued to be inflated by forex, as HPC revenue in NTD was approximately flat QoQ at ~NT$564 billion, versus over 13% growth in Q2.
However, upon looking closer, HPC also saw its percentage of total revenue decline three points from Q2 to Q3, from 60% of total revenue to 57%. Meanwhile, smartphone revenue came in quite strong, gaining three points in total revenue to 30% share as revenue surged nearly 23% QoQ to $9.9 billion.
The softness in HPC revenue may stem from the upcoming platform shift to Nvidia’s Rubin architecture, which relies on a customized 3nm node, N3P, whereas Blackwell relies on TSMC’s N4P process, a customized 5nm node. As you can see in the chart below, TSMC reported a decline in HPC QoQ and YoY revenue at the time that Nvidia’s previous generation Hopper began to ship around Q1 2023. The decline persisted for a few quarters before TSMC resumed very strong QoQ growth in early 2024.
HPC revenue grew 57% YoY and 5% QoQ in Q3 2025, easing from record sequential gains in Q2 as demand pauses ahead of Nvidia’s Rubin ramp.
Nvidia’s upcoming Rubin generation is slated to be released in the second half of 2026 as six variations entered trial production in August. According to CFO Collette Kress, "The chips of the Rubin platform are in the fab. The Vera CPU, Rubin GPU, CX9 Super NIC, NVLink 144 scale-up switch, Spectrum X scale-out and scale-across switch, and the silicon photonics processor [for co-packages optics]. Rubin remains on schedule for volume production next year.”
According to Tom’s Hardware, Nvidia and its partners have “successfully created the photomasks and put Nvidia's Rubin GPU, Vera CPU, and various scale-up and scale-out switching ASICs to production.” The article states that Nvidia is now awaiting them to verify performance, power and other criteria.
Although TSMC serves many high-profile customers in its HPC segment, such as AMD, Broadcom and Intel, it goes without saying that Nvidia is the heavyweight driving the larger swings in this segment as Nvidia has 10X AI revenue in the $200B range compared to second-place Broadcom in the $20B-ish range. As Tom’s Hardware estimates, it could take 9-12 months or longer for Rubin to enter production.
Meanwhile, TSMC will likely report a lull in the HPC segment between the Blackwell and Rubin generations as Nvidia represents the lion’s share of this segment.
TSMC’s Q3 EPS Surges 50.5% Year-over-Year, Beating Estimates by Over 12%
As much as TSMC is a top-line story, it is equally a bottom-line story. The company’s EPS grew by 50.5% YoY to $2.92, beating estimates by an impressive 12.3% with the strongest beat in the last two years.
Analysts expect Q4 EPS to grow 26.8% YoY to $2.84 in Q4 and grow 25.5% in Q1. Looking forward, they expect EPS to grow 19.8% YoY to 12.34 in 2026 and 24.6% YoY to $15.48 in 2027. Due to TSMC’s economies of scale, the margin dilution from overseas fab expansion is expected to be 1%-2% in 2025, lower than their initial estimate of 2% to 3%.
TSMC’s Q3 EPS rose 50.5% year-over-year to $2.92, marking its strongest beat in two years and exceeding analyst estimates by 12.3%, driven by cost efficiency and economies of scale.
TSMC Stock Trades at Very High Valuation
Despite TSMC’s strong bottom line growth, the stock trades at one of the highest bottom line valuations since the AI boom began. As visible below, even the strong fiscal Q3 EPS beat of 12.3 points and strong forward EPS guide is not making a dent in the stretched valuation.
The AI trade has been particularly hard to nail on valuations. For example, we see some stocks such as Palantir trading as high as 100 forward PS, there are long-standing stocks like Broadcom trading higher than it ever has in its history and it seems there are new deals being announced every day followed by 20%+ pops.
Regarding TSM, it’s up 290% since early 2023 while the Nasdaq is up 127% and market favorites like Apple are up 98%. There is no denying that TSMC has been as strong performer as it’s the common denominator across the AI accelerator supply chain.
Below we look at the following to determine how to approach the combination of a strong AI stock with a high valuation:
The top reason that TSMC has seen strong price action and what must happen for the strong price action to continue
How to weigh the United States providing more funding to Intel (to date) compared to TSMC
A new trade setup for TSM now that our original trade plan has played out as described last June
TSMC Strengthens Pricing Power as GPUs Transition to Advanced 3nm and 2nm Nodes
TSMC’s impressive price performance reflects many factors, but at its core, it comes down to the company’s unmatched pricing power in the AI ecosystem. As we had discussed in our previous newsletter on TSMC, Taiwan Semiconductor Stock: AI Growth Amid Geopolitical Risk, the chipmaker is the common denominator to essentially all mega cap stocks’ AI strategies, and the company is deepening its moat with more advanced nodes, such as N2 and A16.
As chips progress to these more advanced nodes, such as Nvidia’s Rubin moving to 3nm and AMD now building CPUs on 2nm, TSMC stands to benefit from increased pricing power. This is because wafers at these new and upcoming nodes carry significant premiums versus the prior node, accounting for substantial increases in performance or power efficiency.
Pricing for TSMC’s 3nm node was originally pegged by analysts at ~$18,000 per wafer at the start of 2025, before rising just over 10% to $20,000 by September, per reports. This would give the 2nm node a ~50% premium with its reported pricing at $30,000 per wafer.
Now, reports claim that supply chain sources suggest 3nm pricing is closer to $25,000 to $27,000 per wafer, a 50% uplift from the start of the year. This is important for TSMC as this process underpins Nvidia’s Rubin, which is expected to see shipments of ~5.7 million units next year per JPMorgan. Assuming this pricing stands, this could provide a meaningful tailwind to TSMC’s AI revenue as Rubin ramps, though it would shrink the premium gap with 2nm to just 12-20%.
TSMC Strong Margins Support Strong Pricing Power
Margins continue to expand due to cost controls, higher capacity utilization rates, economies of scale, and better price negotiation with customers and suppliers.
Operating margin crosses 50% in Q3.
Net income grew by 50.3% YoY.
TSMC’s ability to generate exceptionally strong profits showcases that the company is one of the best-managed companies in the world. Despite the rising inflation, tariff concerns, technological advancement, trade wars, overseas fab expansion, and geopolitical tensions, TSMC has overcome these challenges by continuing to generate superior profits.
The company’s gross margins improved 170 basis points YoY and 90 basis points sequentially to 59.5%. Cost improvements, better capacity utilization, and better price negotiation with customers and suppliers primarily drove the strong margins.
At the same time, it was partially offset by dilution from overseas fab and unfavorable foreign exchange rates. Management expects gross margin to improve 50 basis points further sequentially and 100 basis points YoY in Q4 to 60% with a boost from favorable foreign exchange rates next quarter.
Operating profits grew by 50% YoY to $16.74 billion, with an operating margin of 50.6%, an improvement of 310 basis points YoY and 100 basis points sequentially, primarily driven by higher gross profits and operating leverage. TSMC beat its operating margin guidance by a solid 410 basis points. Management Q4 guide is 50%.
TSMC’s operating margin climbed to 50.6% in Q3 2024, up 310 basis points year-over-year, reflecting strong cost control, operating leverage, and continued AI-driven demand.
Despite higher capex, cash was also strong with operating cash flow at 43.9% compared to 51.6% in the same period last year. Q3 free cash flows were down (-16.1%) YoY to $4.8 billion or 14.6% of revenue compared to 24.4% in the same period last year. The free cash flows were down due to higher capex which grew by 51.6% YoY to $9.7 billion to support strong further growth.
Risks: China, Margins following Onshoring and Intel
As we look into the future, there are puts and takes to onshoring. Last Friday, TSM and Nvidia announced that the first US-made Blackwell wafer was produced at TSM’s Phoenix fab, marking an important milestone in building domestic supply chains for advanced chip production. However, this will mark a new test for TSMC’s pricing power as AMD CEO Lisa Su has stated that US-made chips will be between 5% to 20% more expensive than Taiwan-made chips, with rumors that US chips could be as much as 30% more expensive. While bringing this level of capacity to Arizona could provide a strong tailwind for revenue over the next handful of years, TSMC will have to prove it can maintain strong margins should US-made chips continue to command a premium versus Taiwanese chips.
Although Intel has largely been written off, given it has nowhere near the IP as TSMC on advanced nodes, it’s important to note that Intel is seeing outsized support from the United States with $11.1 billion in funding compared to TSMC’s $11.6 billion despite the company being the market leader in the manufacturing of most advanced chips (so far).
Broadly understood yet important to note, geopolitical tensions related to China could negatively affect the stock. The company has been reducing this risk by setting up fabs outside of Taiwan. However, it will dilute margins, since the cost of setting up fabs and operating fabs in the US is higher than in Taiwan.
TSMC’s Technicals Overview:
In our last report in June, we stated that “If we see the next drop take the form of a 3-wave drop that holds over $146, then we could be setting up for a rally toward the $300s.”
This is exactly what happened, as we only saw a dip that tested $224, in what was obviously 3-wave move. TSM then pressed toward $309, and starting to turn lower.
Considering that we saw TSM pushing toward our target zone on decelerating volume and momentum, it was apparent that we were in some type of 5th wave move, which is the final swing in a trend. There are two scenarios that make the most sense based solely on the price action.
Blue – We hit the lower bounds of a topping zone between $309 – $340. If TSM can hold $264, we can see another swing into the upper regions of this topping zone in the coming weeks – months. However, if this push to new highs is accompanied with a continued pattern of lower volume and momentum, then the odds will increase that a bigger correction will likely follow.
Green – We have completed wave 5 of 3, which would put us in a 4th wave correction. We should break below $264, which will invalidate the blue count above. This larger 4th wave should see $243 – $227, then turn toward the $378 – $420 region to complete the large 5 wave pattern that started on the April low. We cannot break below $227, if this plays out.
Conclusion:
The I/O Fund often weighs stocks through an either/or lens. Today’s question: own TSMC at a valuation it hasn’t proven it can sustain—or Nvidia, as it moves into volume shipments of its Blackwell and Blackwell Ultra AI systems?
We’ve recently chosen to stay on the sidelines with TSMC, but not on AI. We remain heavily invested in Nvidia and several suppliers we believe have been qualified to supply the AI systems ramping in volume as we speak.
Join us this Thursday for a one-hour webinar, where we’ll outline our buy and sell strategies on under-the-radar AI stocks and discuss how we’re positioning in a market where some valuations look stretched while others still have room to run. Learn more here
Damien Robbins and Royston Roche, Equity Analysts at I/O Fund contributed to this analysis
Please note: The I/O Fund conducts research and draws conclusions for the Fund’s positions. We then share that information with our readers. This is not a guarantee of a stock’s performance. Please consult your personal financial advisor before buying any stock in the companies mentioned in this analysis.
Galaxy’s Q3 was the best quarter in company history, per management, with the company seeing record crypto trading volumes, facilitating the sale of 80,000 Bitcoin worth more than $9 billion for a major client. Initial Data Center revenue was recorded in Q3 of just $2.7 million, though management explained that Data Center would be immaterial to results until the first half of 2026 when the first phase at Helios is delivered to CoreWeave.
Management had some interesting comments about being tied to CoreWeave when it comes to future financing and the economic opportunities around diversifying to possible hyperscale customers, as well as why they will not pursue a GPUaaS model like IREN. Management also teased a “big update” in Q2 next year for the newly-launched consumer-facing platform GalaxyOne.
Adjusted Gross Profit Surges in Q3
Galaxy reported revenue of $29.2 billion in Q3, though as we had explained in our initial analysis, Galaxy Bridges Crypto and Asset Management with AI Data Centers, this reflects gross principal trading (Digital Asset sales) that are subsequently offset by transaction expenses.
As such, adjusted gross profit is a better metric for investors to track, as it strips out the inflated appearance of the top line from Digital Asset sales, which were $28.8 billion in Q3 with transaction fees of $28.3 billion. The reported revenue figure does provide clues into the health of Galaxy’s crypto business, as Digital Asset sales surged nearly 231% QoQ, including the 80K Bitcoin trade, but even backing that out, this still would represent north of 100% QoQ growth.
Galaxy reported record adjusted gross profit in Q3, driven by the sharp uptick in Digital Assets and an improvement in adjusted gross margin in the segment, as well as strong performance in Treasury & Corporate. Data Center did contribute ~$2.7 million in adjusted gross profit in the quarter, though it will not become a more material driver until 2026 when revenue ramps as the first phase of Helios is delivered to CoreWeave.
Adjusted gross profit was $728.4 million this quarter, up more than 143% QoQ and 976% YoY. Adjusted gross margin was 2.5%, down from 3.3% in Q2 due to the much higher revenue base ($29.2 billion versus $9.1 billion).
Brief Updates on Digital Assets, Data Center
Galaxy provided some insights into its Digital Assets business, which saw its strongest monthly performance in July, bolstered by that 80K Bitcoin sale. Management said that this momentum carried through the remainder of the quarter, with a meaningful 140% QoQ increase in trading volumes, growth in its client base and increased institutional engagement.
Assets under management and under stake also saw strong QoQ growth, reaching a combined $15.4 billion, with $8.8 billion AUM and $6.6 billion AUS, up from $8.9 billion at the end of Q2. Sequentially, this represented growth of ~$3 billion in AUM and ~$3.5 billion in AUS. Galaxy explained that this level of AUM/AUS would represent more than $40 million in annual recurring revenue from fees, as management expects a majority of the new growth in AUM to be long-lasting, high-fee paying assets.
Galaxy also launched its first consumer-facing platform GalaxyOne in the quarter, with four initial product offerings: Premium Yield, offering 8% yield on cash for accredited investors; GalaxyOne Cash, an FDIC-insured high-yield cash account; GalaxyOne Crypto, support trading and transfers of select crypto products, and GalaxyOne Brokerage, support commission-free equities trading. This boosts Galaxy’s competitiveness with Coinbase and Robinhood in the realm of consumer-facing equity and crypto platforms.
Data Center Progress & Funding
Galaxy’s delivery of 800MW of contracted power to CoreWeave at Helios remains on track and on time, with the first 133MW phase set to be delivered in the first half of 2026, the second 260MW in 2027 and the third and final 133MW in 2028. Galaxy owns a 345kV substation at the campus, capable of supporting 900MW of capacity, with an adjacent 345kV substation owned by Wind Energy Transmission of Texas (WETT) under development to help meet future power needs.
Galaxy noted that it is also in queue with a handful of other interested parties for more power in Texas, and hinted that 800MW will not be their total powered footprint, with an additional 2.7GW under study at Helios, though the future pipeline may remain limited beyond that.
In Q3, Galaxy secured a $1.4 billion, three-year project financing facility to fund the $1.7 billion build-out of the first phase. Management stated that they have been drawing on the facility pretty regularly and on a straight-line basis, and by the end of the quarter the company had drawn ~$430 million so far, taking its total debt outstanding to $1.15 billion. Additionally, Galaxy closed a $460 million PIPE deal for the data center side, with the $325 million in net proceeds going to further fund the expansion of Helios beyond the first phase.
The project financing draw provides a rough idea of how much capital Galaxy will need through 2028 to support the entire 800MW capacity for CoreWeave. Assuming further financing needs come at similar terms and construction at similar costs, Galaxy may need to finance $4.2 billion in total, with $2.8 billion for the second phase. Galaxy now has $1.9 billion in cash and stablecoins on its balance sheet, up ~$700 million sequentially, providing ample room for self-funding the equity to secure financing for the upcoming phases.
Net Income and EPS
Driven by the surge in adjusted gross profit, Galaxy reported a very strong $505.1 million in net income in Q3, up a whopping 1,546% QoQ. Net margin on the reported revenue figure was 1.7%, up from 0.3% in Q2 and up from (0.4%) in the year ago quarter.
This translated to $1.01 in GAAP EPS, or $1.12 in adjusted EPS, or ~5x higher than consensus estimates for $0.21 in the quarter. However, it is important to note that this figure should not be annualized as it is likely to be reliant on the health of the crypto industry and trading volumes, which can vary wildly from quarter to quarter.
Adjusted EBITDA was $629.4 million, nearly 3x the $211.4 million generated in Q2. Adjusted EBITDA margin was 2.2%, down slightly from 2.3% in the prior quarter.
Earnings Q&A:
More Context and Clarity on Helios Build-out
Management provided an important update on the construction progress for the first phase of Helios, confirming that it remains on schedule, with the first data hall to be powered on in December of this year before reaching full operational status in the first half of 2026.
CIO Chris Ferraro explained that 70% of Galaxy’s concrete and civil work is complete and that equipment deliveries and installation is underway. He added that Galaxy is “now placing chillers and putting together the piping system that will form the backbone of our advanced liquid cooling design, an essential component to support next-gen GPUs at industry-leading cabinet densities. Our e-houses, which contain the critical electrical infrastructure have started to ship from the integrators and medium-voltage switchgear and transformers are already being set on their pads.”
With this progress, Galaxy expects the building to be sealed from weather within the next few weeks, allowing work on the interior to proceed regardless of outdoor weather. The next major milestone on deck is powering on the first data hall in early December, then shifting to preparing the hall for service.
For the second and third phases, management provided a quick update, stating that they are “proactively securing long lead time items like backup diesel generators and medium-voltage switchgear early, locking in cost certainty and delivery time lines.” This is important as it will help de-risk the build-out from an infrastructural viewpoint and mitigate potential impacts from equipment delays.
Update on Power
Considering that Galaxy has already contracted out all 800MW of its approved power, management provided a deeper look into power delays and connection requests in ERCOT’s grid.
Ferraro explained that “ERCOT's interim process and the level of scrutiny applied to large loads requesting to interconnect to the system has led to delays in additional capacity approvals” across Texas, as the state’s grid has been overwhelmed over the last 12 months and ERCOT does not want to take on projects that would risk destabilizing the grid.
Piper Sandler’s Patrick Moley questioned about approval timelines, considering Galaxy has a substantial 2.7GW awaiting approval, to which management responded that they will not have an exact timeline for approval until it happens and that “predicting the date is probably a fool's game.” However, they believe that signs from ERCOT, WETT and their utility partner AEP approving and finalizing studies at a faster pace give them a higher degree of confidence in approval for capacity coming online in late 2028 to 2029.
Management had mentioned that WETT’s Pitchfork 345kV substation is expected to “deliver an additional 3 gigawatts of power capacity with 2 synchronous condensers adjacent to the Helios campus starting in 2028,” which would support the entirety of the campus’ remaining buildout with some buffer room. Galaxy stated that there are “increasing proactive reach outs to us from very large customers in addition to our current partner, CoreWeave, who all want to know when are we getting approval for how much and over what time period,” so getting approval for the entire 2.7GW of additional capacity will unlock a tremendous opportunity to meet future demand.
Looking beyond Helios, management implied that there may not be much more in the pipeline until prices come down: “Markets for some of these companies without contracts, without customers, the market is pricing in a tremendous amount of optimism. And so that feeds through to the price of projects. And so in the short run, I don't think you're going to see us reaching out and buying a whole lot more power at these prices.” Considering the capital intensity of simultaneous multi-GW build-outs, this is a smart move as it would prevent Galaxy from overstretching its cash with too many builds.
CoreWeave’s Credit Quality as Anchor Tenant
Another interesting discussion circled around to CoreWeave as Galaxy’s primary anchor tenant, the economics of being this closely linked to CoreWeave and the transition to a multi-tenant site at full capacity.
When questioned about the potentiality of refinancing the project financing facility to unlock some capital, CIO Chris Ferraro explained that their ability to refinance this debt and raise future debt at attractive rates is linked rather directly to CoreWeave’s credit profile:
“The views of CoreWeave's credit profile, which lenders are very focused on, in addition to Galaxy's credit profile are changing and getting better by the day on both fronts. And so the ultimate outcome is really going to be a function of where we and CoreWeave and the markets are then.” He added that if rates were in the high single-digits and if Galaxy and CoreWeave remain successful, financing rates may more lower.
However, CEO Mike Novogratz seemed to contradict this, stating that the market is in a period of “trying to understand what CoreWeave's credit quality is today and what it should be on the forward. And that's going to be a big, big determinant of their ability to get better lease rates, their ability to get financing, [and the] ability for us as the landlord to finance our projects. And so where that goes, how the market evolves is thinking on CoreWeave is something we're very focused on and something that's a little unknown today.”
This is quite an important quote, as Galaxy’s financials and ability to secure future financing at attractive rates is inextricably linked to CoreWeave’s financial health. As we discussed in our free newsletter on CoreWeave, the neocloud is facing a wall of capex in Q4 and may continue to turn to the debt market to fund its expansion, likely weakening its credit profile. This could then hamper Galaxy’s ability to find attractive financing for its upcoming projects.
Considering CoreWeave is Galaxy’s only AI tenant for the time being, it raised a crucial point about diversifying the customer base in the future to a multi-tenant structure as Helios expands beyond 800MW. Novogratz stated that there is a “real decision to be made as to whether on a net economic basis, whether a lower-yielding lease from a higher credit quality tenant, net balances out to a better economic equation for us as we think about broadening the portfolio.”
What he is implying here is that an investment-grade hyperscaler will be able to secure a lower-yield lease, carrying a lower annualized opportunity to Galaxy, but because the creditworthiness of the hyperscaler exceeds that of CoreWeaves, Galaxy’s associated costs and funding will be more attractive and translate to a better return on investment (long-term revenue minus funding costs and interest expenses on debt).
Shying Away from GPU-as-a-Service
Unlike IREN, Galaxy is not going to prioritize or even attempt to shift towards renting GPUs out as a service, opting instead to take deals like the one it signed with CoreWeave. Galaxy said this is because they do not have the same software + hardware advantage that CoreWeave and others may have where they have a strong value-add proposition on top of the bare metal, and it is not something they have invested in.
Management also chalked this decision up to a lack of understanding over the useful life of GPUs, preferring to invest instead in long-lived infrastructure where their expertise is more suited:
“We’re not confident in what useful life of GPUs are ultimately going to be, and the cycles of GPU efficiency are pretty nascent still. And so we like very much investing in long-lived infrastructure that we understand useful life of, and we don't quite yet understand what the useful life of GPUs are. And so the business model around return on capital on GPUs, particularly if you haven't added real expertise in real value add, I think is a really challenging thing to decide to do. So we're not thinking about it.”
GalaxyOne: Expect “Big Update” in Q2 2026
Considering how recent the launch of GalaxyOne was, the impact on growth is likely to be minimal in the near-term as Galaxy works to build out the product suite. Management expects it to take a handful of quarters to get momentum on the platform, and tentatively expects to have a “big update” for investors in Q2 2026.
Galaxy teased that it has a “really ambitious” road map over the next six to 18 months to transform GalaxyOne into a “one-stop serves all wallet,” with the aim to add products and offerings that help high net worth consumers invest and store wealth, and reallocate across equities, bonds, digital assets and more. The platform also would open up an ability for cross-platform collaboration between trading, asset management and staking, and serve as a new outlet for Galaxy to continue growing its AUM/AUS and increase its fee-based ARR rather substantially.
Conclusion
There is a lot to like from Galaxy’s Q3 report, with adjusted gross profit surging from a strong crypto backdrop while data center updates show continued progress towards start of service for CoreWeave in 2026. Management remains focused on building out Helios and is encouraged by signs from utility partners and regulators about receiving approval for additional power to expand the campus later this decade.
Damien Robbins, Equity Analyst at I/O Fund contributed to this analysis.
Please note: The I/O Fund conducts research and draws conclusions for the Fund’s positions. We then share that information with our readers. This is not a guarantee of a stock’s performance. Please consult your personal financial advisor before buying any stock in the companies mentioned in this analysis.
Micron’s stock is up 120% YTD – or 3X more YTD than AI heavyweight Nvidia. Recently, the high-bandwidth memory content that Micron supplies has increased 3.5X between GPU generations, leading to a quiet memory boom across DRAM and NAND suppliers.
Memory is typically a cyclical industry that is low margin and lumpy, yet memory is seeing a newfound resurgence from AI that is strong enough to transform commoditized hardware into a secular trend as the AI economy is built out. AI servers use more DRAM and NAND than traditional servers, relying heavily on high-bandwidth memory (HBM) for training and inference.
In fiscal 2025, Micron’s HBM, high-capacity dual in-line memory modules (DIMMs) and low-power (LP) server DRAM revenue reached $10 billion, up more than fivefold from the prior year, while HBM alone reached $2 billion in revenue in Q4. Data center is already proving to be a strong driver of growth for Micron, accounting for 56% of sales in FY25, up from 35% in FY24. On a dollar basis, data center revenue surged 137% YoY to $20.75 billion.
Below, we look at how Micron has quietly outperformed some of the biggest players in AI YTD and if its ability to defy the odds can continue.
Micron Delivers 3.5× More Memory per GPU, Powering 34× Larger AI Models
HBM capacity per chip continues to rise with each new generation of GPU, as its ability to offer higher bandwidth, performance, and lower latency is crucial for increasingly powerful large language models.
For example, we’ve seen a ~3.5x increase in HBM content in short fashion on Nvidia’s GPUs within about three years’ time frame:
The H100 featured 80GB of HBM2e content per chip. This chip began shipping in Q4 2022 and ramped in early 2023.
The H200 featured 141GB of HBM3e content per chip, 1.76x higher than its predecessor, which helped drive 1.4x to 1.9x faster inference on leading AI models.
The B200 features 180GB of HBM3e content, more than double the H100 and a 28% increase versus the H200. In an 8-GPU server configuration, the B200 boasted 1.44TB of HBM content.
The B300 boasts 288GB of HBM3e content, a 60% increase versus the B200 and over 3.5x more than the H100. In an 8-server configuration, the B300 has 2.3TB of HBM content. This chip is beginning to ship now in Q3-Q4 2025.
Putting in context Nvidia’s rack-scale solutions, the GB200 and GB300 NVL72, shows just how rapidly HBM content is increasing. The GB200 supports up to 13.4TB of HBM content, while the GB300 supports up to 21.7TB of HBM, nearly 34X higher than the 640GB of HBM content in the 8-GPU DGX H100 servers.
AMD is also showing surging memory requirements, to the tune of 3.5X across two main generations:
The Instinct MI250 featured 128GB of HBM2e memory.
The MI350X featured 288GB of HBM3e memory, a 125% increase versus the MI250 and on par with Nvidia’s Blackwell Ultra.
The MI400 series is expected to feature 432GB of HBM4 memory, a 50% increase versus the MI350X and the Blackwell Ultra. In the Helios rack configuration slated for 2026, the MI400 will boast 31.1TB of HBM content, 1.5x more than the GB300 NVL72.
NVIDIA’s Grace CPU Boosts Demand for Micron’s LPDDR5X Memory
When thinking about Nvidia’s GPU platforms, it oftentimes is overlooked that the GB200, GB300 and the older GH200 generations are all paired with Nvidia’s Arm-based Grace CPU (hence the G-based nomenclature). The Grace CPU accelerates CPU-to-GPU connections with Nvidia’s NVLink C2C and helps boost performance and energy efficiency of AI workloads utilizing its Arm Neoverse V2 cores and LPDDR5X (low power double data rate 5) memory.
For example, Micron tested inference performance on Meta’s Llama 3-70B with LPDDR5X memory on the GH200 versus DDR5 on the H100, and found that LPDDR5X delivered 73% less energy consumption with 5x higher throughput and 80% better latency.
The timing and ramp of the GB200 and GB300 throughout the second half 2025 and into 2026 suggests the new LPDDR5X growth curve is at its strongest. To put in perspective, shipments of ~30K GB200/300 racks in 2025 would require more than one million LPDDR5X modules, with each NVL72 rack featuring 17.3TB, or 36 480G modules. If rack shipments double to ~60K in 2026, this would also double LPDDR5X module demand.
Driven by the Blackwell platform, Micron saw revenue from LPDDR5X for servers up 50% QoQ in fiscal Q4 to a record level, though the company did not disclose its exact revenue contribution.
Micron’s Expanding Role in AMD and NVIDIA’s HBM Supply Chain
The HBM market is quite competitive between Micron, Samsung and SK Hynix with Micron historically ranking third. However, Micron plays an increasingly important role in Nvidia’s supply chain, and to a lesser extent, AMD’s. Micron is expanding its presence within HBM, stating in Q4 that it has expanded its HBM customer base from four customers in Q3 to six.
Micron has a range of products designed into Nvidia’s leading platforms:
Micron’s HBM3e 8-high 24GB cubes are designed into the HGX B200 and GB200 NVL72 platforms.
Micron’s LPDDR5X supports Nvidia’s GB300 Superchip, with Micron stating in Q4 that “since NVIDIA's launch of LPDRAM in their GB-product family Micron has been the sole supplier of LPDRAM in the data center.”
While Samsung remains a key HBM supplier for AMD, Micron has collaborated with the Nvidia challenger on the Instinct MI350 GPU family as well as its EPYC CPUs. Micron’s HBM3e 12-high 36GB cubes support the MI350X series, while its 128GB DDR5 RDIMM modules support AMD’s 4th gen EPYC CPUs, providing “up to 22% improved energy efficiency and up to 16% lower latency over competitive 3DS through-silicon via (TSV) products.”
mid prompt
Moving through 2026, the industry is shifting to HBM4 products, where Micron believes it outperforms Samsung and SK in terms of performance and power efficiency. The chipmaker noted it is in active discussions with customers for HBM4 volumes and expects to sell out of capacity for 2026 over the next few months.
This role of supplying both Nvidia and AMD with core memory products and leading on performance and power on the upcoming HBM generation positions Micron well for growth in 2026 and 2027.
Micron Reports Strong 44% Revenue Growth Driven by AI Memory Demand
Surging AI data center demand drives record FQ4 revenue.
Strong AI-demand for high-performance memory is creating tight supply, which in turn is driving higher DRAM and NAND prices.
Data center market growth is complemented by improvement in other end markets.
Micron reported record FQ4 revenue of $11.32 billion. The primary drivers of last quarter’s record revenue were the company’s DRAM segment, specifically High Bandwidth Memory (HBM) products, which benefited from the rapid expansion of AI datacenters.
Revenue growth accelerated 9.4 percentage points sequentially to 46% YoY, and on a sequential basis, growth was 21.7% QoQ, a solid 6.2-point acceleration. Micron guided a fresh record in FQ1 at $12.5 billion at midpoint, pointing to 43.5% YoY growth and a 10.5% sequential growth. Analysts expect revenue growth to accelerate to 60.6% in FQ2.
Micron’s FQ4 revenue growth accelerated by 9.4 percentage points sequentially to 46%, marking continued momentum in AI-driven demand.
FQ4 DRAM revenue grew by 69% YoY and 27% QoQ to $8.98 billion, a second consecutive quarter of strong sequential growth. DRAM revenue accounted for 79% of total revenue. The growth was driven by bit shipments in the low-teens percent sequentially and prices also increased in the low double-digit percentage range.
FQ4 NAND revenue was down (-5%) YoY and up 5% sequentially to $2.25 billion. NAND bit shipments declined in the mid-single digits, and prices increased in the high single digits percentage sequentially due to a favorable mix and tight supply.
Micron benefits from other markets such as smartphones and automotive, represented by the mobile and client business unit (MCBU) up 5% YoY and up 16% sequentially to $3.76 billion.
For FY25, revenue rose 49% YoY to $37.38 billion, driven primarily by DRAM and HBM revenue, which rose more than 62% YoY to $28.58 billion. HBM reached an annualized run rate of $8 billion in FQ4, with HBM share expected to grow again in FQ1 and HBM4 capacity in discussions to soon be sold out for calendar 2026.
Micron has not provided a full-year guide for revenue, but current consensus estimates call for 43% growth to $53.5 billion in revenue.
Micron Leads the AI Memory Supercycle with Record Data Center Growth
The AI-driven demand for memory (especially HBM and high-performance DRAM) is still in the early stages of a multiyear growth cycle. The company’s CEO and Chairman, Sanjay Mehrotra, also mentioned in the recent earnings call, “Memory is very much at the heart of this AI revolution. This means a tremendous opportunity for memory and certainly a tremendous opportunity for HBM."
In fiscal 2025, Micron's data center reached a record 56% of company revenue, with growth primarily driven by DRAM products and aided by data center SSDs and NAND components. Overall, data center revenue increased 137% YoY to $20.75 billion.
Micron’s Cloud Memory Business Unit (CMBU), which consists of its HBM, high-capacity dual in-line memory modules (DIMMs), and low-power server DRAM solutions, saw revenue surge to …
Sign up for FREE below to unlock the following information on Micron’s 120% YTD stock performance:
Micron’s AI-related segment that surged over 200% last fiscal year and what management is saying about growth in the upcoming fiscal year
The key line item in the income statement for every AI-related hardware player, and if Micron has what it takes to see sustained, bullish price action
What Micron’s valuation is communicating and if the stock still has room to run
Micron’s Cloud Memory Business Unit (CMBU), which consists of its HBM, high-capacity dual in-line memory modules (DIMMs), and low-power server DRAM solutions, saw revenue surge to 257% YoY in fiscal 2025 to $13.5 billion, or YoY growth of nearly $10 billion. Micron’s other data center unit, its Core Data Center Business Unit (CDBU), consists primarily of data center SSDs and NAND components. This unit saw revenue growth of 45% YoY to $7.23 billion.
For Q4, CMBU revenue rose 214% YoY to $4.54 billion, while CDBU revenue declined (23%) YoY. CMBU revenue growth was driven by HBM and strong bit shipment growth, though Micron offered no commentary behind the decline for CDBU.
Management expects robust AI server demand, the shift to HBM4 and tight DRAM supply to remain tailwinds to growth and profitability moving through 2026: “we expect healthy demand supply environment in 2026 for overall DRAM, and that bodes well for profitability of DRAM, profitability of HBM and of course, profitability of non-HBM as well, which is experiencing tight supply.”
When looking toward industry-level estimates, Goldman Sachs projected the HBM market to reach $45 to $51 billion, up ~37% YoY at midpoint from Micron’s $35 billion 2025 TAM estimate. With Micron’s HBM share expected to match its DRAM share in the low-to-mid 20% level, this would project HBM revenues likely surpassing $10 billion next year. Over the long run, Micron expects HBM’s total addressable market (TAM) to hit $100 billion or higher by 2030, which assuming some continual share gains would potentially give the chipmaker a $25 billion-plus HBM business by the end of the decade, or more than 4X from 2025.
Micron Clears 50% Gross Margin as Profitability Surges
Solid margin expansion driven by favorable pricing, improved product mix, disciplined cost management, and operating leverage.
50% gross margin bar cleared.
The growing proportion of higher value products mix to further aid margin expansion.
Micron’s margin turnaround story has been impressive, with gross margin up more than 55 points over the last two years and operating margin up more than 66 points. Adjusted gross margin in Q4 was 45.7%, up 6.7 points QoQ and 9.2 points YoY, aided by strong growth in CMBU which carried a 59% gross margin in the quarter, DRAM pricing, favorable product mix, and cost controls. For Q1, adjusted gross margin was guided to be 51.5% at midpoint, a 5.8 points sequential expansion and up by a solid 12 points YoY.
Adjusted operating margin was 35%, up 8.2 points QoQ and 12.5 points YoY, driven by operating leverage. For Q1, Micron expects adjusted operating margin to be 40.8%, up by 5.8 points QoQ and by 13.3 points YoY, signaling strong operating leverage.
The strong margins are expected to continue in FY2026 due to continued increase in memory prices due to supply tightness, cost controls, and growing proportion of higher value product mix.
Micron’s adjusted gross margin guidance for FQ1 stands at 51.5%, up 12 percentage points year over year.
Gross margin is the key line item for every hardware player in the AI market to see sustained, bullish price movements as it helps to communicate if the stock is participating in a commoditized trend, or if there’s something more secular and unique to the hardware player’s product that can lead to charging higher prices. Of course, driving down costs is also good to see – but nothing beats being able to charge more for a hardware product in terms of communicating the strength of an AI hardware stock.
Micron has a lot it must prove in terms of its lumpy margins yet is off to a solid start by moving from the mid-30% range to expected >50% next quarter.
Adjusted EPS Soars 157% Amid Strong AI Memory Demand
Micron is expected to see earnings double this fiscal year as margins have swiftly recovered from late 2023 and early 2024. In Q4, Micron reported adjusted EPS of $3.03, up 157% YoY and beating estimates by 5.9%.
For Q1, Micron guided for adjusted EPS to be $3.75, +/- $0.15, more than 23% ahead of guidance and corresponding to YoY growth of 110%. Earnings growth is expected to reaccelerate to 155% in Q2 but then decelerate to 126% in Q3. For the full year, Micron is expected to see 100% YoY growth to $16.63 primarily driven by better pricing, cost improvement, operating leverage, and a growing proportion of higher value products.
Micron’s FQ4 adjusted EPS climbed 157% YoY to $3.03, marking a strong earnings rebound as AI-driven demand and pricing strength lifted profitability.
Micron’s Free Cash Flow Surges 10x with AI Memory Boom
Adjusted free cash flow is on track to improve significantly in FY2026, driven by margin expansion.
Strong revenue and profits are leading to higher cash flows. Micron has seen a significant turnaround in the adjusted free cash flow from (-$5.45 billion) in FY2023 to $368 million in FY2024 and up nearly 10x to $3.72 billion in the recently concluded FY2025.
FQ4 adjusted free cash flow grew by 149% YoY to $803 million or 7.1% of revenue, an improvement of 2.9 percentage points YoY. Management expects adjusted free cash flow to strengthen in FQ1 and to be significantly higher in FY2026.
Valuation
Despite its recent rally, Micron trades at reasonable multiples, well below its peak from 2024. On the top line, Micron trades at 4x forward revenue, 10% above its average 3.6x multiple and far below its peak of 6.8x from mid 2024.
On the bottom line, Micron trades at 11.6x forward earnings, though its 43.9x average is skewed higher by 2024’s >100x multiples when margins were razor-thin. Since the start of 2025, Micron has traded as high as 16x forward earnings and as low as 8x.
Micron’s growth to this point and beyond has been centered around HBM, both on the top and bottom lines. CMBU is the only unit that sees operating margins above the corporate total, at 48% versus 25%, 29% and 20% for its other segments, meaning that future operating margin expansion will be tied solely to growth from CMBU, and felt more the faster CMBU grows. This may mean that margin and earnings upside in 2026 may be limited come 2027.
Conclusion
Micron’s stock has quietly outperformed AI leaders YTD by a wide margin such as Nvidia, Broadcom, Oracle, Meta, Microsoft and more. This has been achieved by AI-related segments surging over 200% YoY, leading to a critical improvement to their margins and by also rebounding to become cash flow positive again.
Clearly, Micron is no longer tied to consumer device cycles. Instead, HBM had led to higher margins and multi-year supplier agreements, resulting in a leveraged approach to participating in the AI infrastructure buildout.
In terms of whether Micron can become more secular, this past year has proven that HBM memory is a component multiplier when compared to GPUs in the hardware stack, as HBM scales faster than GPUs on a per-system basis. Each generation of GPU, from Hopper to Blackwell to Rubin, requires more memory capacity and bandwidth per chip. Therefore, there is compounded effect, as the number of GPUs rises combined with each GPU system requiring more HBM per package.
Join our next webinar on Thursday at 4:30 p.m. ET as the I/O Fund discusses potential buy and sell plans and real time trade alerts with premium members. The I/O Fund has two entries up over 300% in 2025 plus others up over 80%. Our cumulative returns of 210% over a five-year period would place us as #2 if we were a hedge fund and #5 if we were an ETF. Learn more here
Please note: The I/O Fund conducts research and draws conclusions for the Fund’s positions. We then share that information with our readers. This is not a guarantee of a stock’s performance. Please consult your personal financial advisor before buying any stock in the companies mentioned in this analysis.
Applied Optoelectronics (AOI) (NASDAQ: AAOI) is a lesser-known optical component and transceiver supplier. The small-cap has recently caught our attention for its hyperscaler deals and its ambitious capacity growth targets, aiming to become one of the largest domestic manufacturers for 800G transceivers.
The VSCEL laser, EML chip, and optical transceiver supplier has provided some of the strongest commentary for future growth, with management outlining 8.5x capacity growth by year-end and substantial capacity growth in 2026. This builds upon a comment slipped into Q1’s call that AOI believes it can become the largest 800G/1.6T optics supplier to Amazon, after signing a deal in March that management says is worth more than $4 billion over ten years.
AOI also stood out from its 40% QoQ growth in data center revenue in Q2, though this came after a soft Q1 affected by inventory digestion at its largest hyperscale customer and seasonality. Interestingly, data center was not the core driver of growth in Q2, with cable TV taking the reins with 8x YoY growth in the quarter. Cable TV is expected to provide an additional lever of growth come 2026, with management tentatively outlining early visibility into $300M+ in demand.
The crux with AOI is that cash flows are much softer than more-established AI networking stocks, and GAAP margins are still negative. However, it’s hard to argue with the implications from the profound expansion in capacity and how optics growth can transform margins.
Plotting Out the Optical Opportunity
Here’s a quick recap from our first thematic coverage on optical interconnects from over a year ago and why optical components would become a necessity as AI workloads rise.
Copper had long been standard for data center interconnects, but it cannot support network speeds of 800 gigabits (800G) to 1.6 terabits (1.6T) over long distances due to substantial signal loss. This isn’t to say copper is dead – Nvidia’s GB200 NVL72 utilized copper over optics (with more than 2 miles of copper cabling in the rack) to reduce power consumption by 20 kw (the system still draws 120kw of power). According to a representative from Marvell’s Cloud Optics division, “optical is the only technology that can give you the bandwidth and reach needed to connect hundreds and thousands and tens of thousands of servers across the whole data center.”
Optical transceivers are crucial in enabling high-speed data transfer, by transmitting and receiving data from optical (light) signals to electrical signals. In data centers, optical interconnects and transceivers are becoming the de facto standard to handle AI workloads, since they can function at significantly higher speeds than copper (currently at 800G+ speeds and moving quickly to 1.6T), with longer range, higher data capacity, and lower latency with minimal signal loss. One drawback, however, is that due to the electronic complexity of optical products, costs are higher as well as power consumption versus copper.
To read the full analysis on optical interconnects, click here.
Major optics manufacturer Innolight reportedly raised its forecast for global 800G unit shipments by 50%, now seeing 15 million shipments globally for 2025, up from 10 million previously. Other reports out of China were more optimistic, with one firm projecting >16 million 800G shipments this year with 1.6T demand exceeding 5 million units, and another projecting demand for 17 million 800G and 4 million 1.6T units. FS had placed a rough estimate of 5 million shipments for 800G products in 2024, so these projections would represent >3X growth YoY if they come to fruition.
Nvidia’s math also aligns with the upper end of those demand scenarios for >800G transceivers for scale-out connectivity, with the GPU leader claiming a 1:6 GPU-to-transceiver ratio for 100K 4-GPU servers. Nvidia states that this scenario, with 400K total GPUs, would require 2.4 million transceivers.
Nvidia estimates that optical module requirements for scale-out networks may be at a ratio 1:6 per GPU. Source: Nvidia via The Next Platform
While a majority of Nvidia’s shipments are likely to focus on the 72-GPU rack-scale solutions rather than smaller servers, a rough estimate for 750,000 servers, or 3 million GPUs (less than half of Nvidia’s estimated 2025 unit shipments), would project 18 million in transceiver demand solely to meet Nvidia’s needs.
For intra-rack connections for the GB200 NVL72, reports suggest that “the ratio of GPUs to 1.6T optical transceivers is 1:2 in the dual-layer InfiniBand network and 1:3 in the three-layer InfiniBand network. Compared to the DGX H100, the NVIDIA GB200 NVL72 architecture doubles the port speed for servers and switches, significantly advancing the adoption of 1.6T transceivers and increasing the need for the 1.6T network.” Assuming roughly 30,000 NVL72 units in demand, this would project to between 4.32 million to 6.48 million 1.6T transceivers for intra-rack connections.
AOI Targeting 8.5X Capacity Growth for 800G/1.6T Products by Year-end
At OFC in April, AOI outlined one of the most impressive capacity expansion stats that we have seen: an 8.5x increase for 800G and 1.6T products by the end of the year, with management reaffirming in both Q1 and Q2 that they remain on track to reach said target. Capacity for these high-speed products is split between two facilities, one in Taipei, Taiwan and the other in Sugar Land, Texas. Overlaying potential 3x growth industry this year with 8.5x capacity growth for AOI implies the company is eyeing market share gains into year end persisting through 2026.
Below is how AOI’s management is charting out the growth in capacity, with consistent monthly expansion taking it to its 8.5x target:
Capacity growth figures are relative to a March 2025 baseline for Taiwan production and a September 2025 baseline for Texas production. Source: AOIAOI
Note that these figures are cumulative, meaning that AOI is essentially targeting 2.35x more capacity by June from its March baseline in Taiwan for its 800G 2xFR4/DR8 product, before scaling to 3.65x by August and so forth. This is also the case in Texas, where AOI is targeting the start of production in September (100%), before scaling 4x by December. Equipment that was ordered in Q1 was said to be arriving as of Q2, further supporting this expansion effort.
Putting this in perspective, the 8.5x growth is targeting approximately 100K monthly capacity globally of 800G/1.6T products by year-end, according to management. This suggests that capacity earlier in 2025 was roughly 12K per month. AOI expects ~40% of 800G production to be domestic at its Texas facility, which they believe will make them one of the largest domestic manufacturers. More importantly, AOI says they can accommodate this expansion under its current facility footprint, saving on capex.
Management Eyeing 2X Growth in 2026
If 8.5X growth was not enough, AOI is going further and aiming to double capacity again by mid-2026, stating in Q2’s call that they are expecting to be able to produce >200K 800G/1.6T products per month, with the majority produced in Texas. This corresponds to annual production of ~2.4 million 800G/1.6T products, up more than 16x from less than 150K annually prior to these expansion plans. Compared to market leaders Innolight and Eoptolink, where monthly capacity is estimated at 500K and 300K units respectively (for annual capacity of ~9.6M), AOI still lags but is quickly catching up. The company believes it holds an advantage from its vertically-integrated model with in-house laser and subassembly manufacturing, along with substantial factory automation that allow it to quickly ramp capacity to this degree.
Based on Nvidia’s calculations, AAOI alone may be able to support 400K GPU shipments at full capacity, or more than 5% of UBS’ estimate for Nvidia’s GPU shipments of 7.4 million next year. More importantly, none of this capacity so far, not even the 8.5X growth, has translated into revenue, with 800G contribution immaterial to revenue as of Q2 as products remain in the qualification stage.
While pricing is unknown, an analyst had questioned management about the ramp of 800G/1.6T products in Q1 and penciled in $0.75/gig: “But I guess rough math, yeah, $0.75 a gig, that would be well over $100 million in a quarter. Am I doing something wrong in that thinking?” The question was quickly shot down by management, yet assuming pricing at this degree as the market becomes more competitive through 2026 provides an interesting look into the potential growth ahead for AOI.
Assuming by Q3, AOI is operating at max capacity of 200,000 pieces per month, with 80% of those being 800G and 20% 1.6T. This would roughly project ~$96 million monthly revenue for 800G products and ~$48 million monthly revenue for 1.6T products, assuming capacity is sold out.
On a quarterly basis, this would be north of $430 million in quarterly revenue simply from >800G products, not including 400G or cable TV, already more than double current consensus estimates for $194 million in Q3 and $217 million in Q4.
More Clarity on 800G Ramp, 1.6T
Considering the combination of commentary for >16X capacity growth for 800G and 1.6T products in just one year and the fact that the opportunity is squarely ahead, tracking the timing and scale of this revenue ramp is critical. Based on commentary from Q2, 800G accounted for, at maximum, 1% of data center revenue, considering shipments were only for qualification purposes.
Management said that they “continue to believe that we will produce meaningful shipments of 800G products sometime in the second half of 2025, likely in late Q3 or Q4. The schedule is constrained by our ability to build and qualify production capacity. We believe that the demand for 800G is strong, and we expect that when our production is ready, we will see a fairly quick ramp in revenue for 800G.”
The timing of the ramp comes down to two factors: AOI has to have meaningful production capacity available in order to quickly shift to volume production, and also wait for its prospective customers to finish production qualification. October and November are expected to see larger jumps in production capacity, with Taiwan capacity rising from 5x to 8.3x, and Texas up to 3x from its baseline, suggesting the ramp is imminent yet pending the conclusion of qualification.
AOI believes that one of its major hyperscale customers is in the final stages for securing 800G qualification, having recently audited and approved AOI’s Taiwan factory for 800G production. This adds an additional layer of confidence that the revenue ramp for 800G is near, as management expects that this customer could become a >10% customer in Q3.
For 1.6T, AOI expects to kick off volume manufacturing around June to July 2026, noting that it is already working with several customers and expects to have a minimum of three tier 1 customers. All three are pre-existing customers, so this could include Microsoft, Amazon or possibly even Oracle.
Another reason that the ramp of 800G and 1.6T is important is tied to margins: management explained that for 1.6T, gross margins should be above 40%, while 800G margins should be close to 40%. As discussed further below in the Financials section, AOI’s current gross margin is hovering at 30%, meaning a strong ramp for these products will drive gross and likely operating and net margins higher – management has guided for 35% to 40% gross margins by the end of 2026 and these products will be crucial to reach that.
AOI shares surged in early March when the company announced that it signed a 10-year supply deal with an Amazon subsidiary, widely understood to be AWS, while offering the company warrants for 7.94 million shares.
This is not AOI’s first long-term agreement with a major hyperscaler for data center products – the company signed a five-year supply agreement with Microsoft in 2023 for >400G products through 2028, and expects revenue from this contract to increase in 2025 relative to 2024. Terms for the Microsoft deal were not disclosed.
AOI has disclosed additional details about the Amazon deal: 1.32 million shares vested upon the signing of the agreement, but the remaining 6.62 million shares may vest over the next 10 years “dependent on aggregate purchases by Amazon of $4 billion of our products over this time period.” CEO Thompson Li said in Q1 that he believes the deal could be “much more than $4 billion in the next 10 years.” This would correspond to annual revenue on average of at least $400 million solely from Amazon.
Q2 had quite an interesting snippet from management regarding the deal:
“Our belief, our expectation, is that we can grow to be, the largest supplier of 800G and faster, higher data rate optics for Amazon. Now, that's not guaranteed by any means, but I don't see any reason why we couldn't be there. And that would imply a market share; typically, they're going to have two or three suppliers, so that would be, maybe 30, maybe even up to 40%.”
This quote is very important as it indicates the Amazon deal could create a long-term, high-volume customer contributing hundreds of millions in annual revenue, it also signals product validation from an industry leader and could lead to future supply deals, assuming capacity supports it.
Significant Customer Concentration a Risk to Consider
AOI exhibits a higher degree of risk related to its significant customer concentration, with its top ten customers accounting for 98% of revenue in Q2, a slight increase from 94% in the year ago quarter.
AOI’s largest customer was in its CATV segment, contributing 54% of total revenue, while its second largest customer was in its data center segment and contributed 34% of total revenue. These two customers alone, likely Digicomm and Microsoft based on prior revenue trends, combined for 88% of revenue. As mentioned previously, management believes they could have a third >10% customer in Q3, a major hyperscaler for 400G and 800G products.
Oracle was previously a >10% customer as recently as 2024, contributing 12.4% of total revenue last year, up from 8.8% in 2023. Microsoft had contributed nearly 44% of revenue as AOI’s largest singular customer in 2024, while Digicomm accounted for 34%, suggesting the recent growth in CATV has swapped its position.
High China Revenue Concentration
AOI has elevated China revenue exposure, which raises risk considering that geopolitical tensions are flaring up again. In Q2, China contributed $62.4 million in revenue, up 403% YoY and accounting for nearly 61% of revenue, with this likely driven by the growth in cable TV. Taiwan revenue was $39.5 million, up 37% YoY to 38% of revenue.
On the other hand, AOI is aiming to reduce its China content in its transceivers to near zero as they scale production, with current China content at <10% for its 800G and 1.6T products. AOI says that a majority of its key components for these products, such as laser chips, are already manufactured in the US.
Financials
Revenue
AOI reported revenue of $102.95 million in Q2, up 138% YoY and 3% QoQ, representing a modest (2.7%) miss versus consensus estimates. This was a slight deceleration from 146% growth in the previous quarter. Cable TV was the primary growth driver with revenue up 862% YoY to $56 million, while data center revenue rose 30% YoY to $44.8 million.
For Q3, AOI guided for revenue to be between $115 million to $127 million, pointing to 86% YoY growth at midpoint, though on a sequential basis this is a sharp uptick to 17% QoQ growth. Management expects that there may be some initial contribution from 800G modules late in the quarter: “Looking ahead to Q3, we expect a sequential increase in our datacenter revenue, driven by continued growth in our 100G and 400G products with the possibility of layering some additional increased 800G revenue late in the quarter.”
For fiscal 2025, revenue is currently projected to be $467.3 million, up 87.4% YoY, before slowing to 55.9% growth in fiscal 2026 to $728.4 million. However, there is potential that the 800G and 1.6T capacity ramp meaningfully accelerates AOI’s run rate by the end of 2026.
Key Segments:
CATV (Cable Television)
As noted above, CATV revenue in Q2 was up 862% YoY but down (13%) QoQ from a record Q1, in line with management’s expectations as production was retooled for Motorola-style amplifiers. The YoY growth was driven by the ramp of its 1.8 GHz amplifiers.
AOI also finished testing and certified its Motorola- and GameMaker-style amplifiers for Charter Communications, who will deploy AOI’s 1.8 GHz amplifies and QuantumLink remote management software. For Q3, AOI expects CATV revenue to be record or near-record.
Data Center
For Q2, data center revenue was $44.8 million, up 30.4% YoY and 39.8% QoQ. The QoQ increase comes after a (27.6%) QoQ decline in Q1 from inventory digestion at one of AOI’s largest hyperscaler customers and seasonality. Data center revenue has been lumpy so far, though growth is expected to pick up in Q3 and into Q4 as 400G ramps and 800G begins contributing.
Revenue for 10/40G products rose 68% YoY and 26% QoQ to $4.0 million, or 10% of data center revenue
Revenue for 100G products rose 25% YoY and 25% QoQ to $31.4 million, or 70% of data center revenue.
Revenue for 200G/400G products rose 43% YoY and 180% QoQ to $8.9 million, or 20% of data center revenue. AOI also completed its first volume shipment for high-speed single-mode 400G transceivers to a recently re-engaged hyperscaler, with increased sequential demand from other hyperscalers arising in Q2.
Telecom
Telecom and other revenue was down (45%) YoY to $2.1 million. Management expects telecom revenue, which was $1.9 million, to fluctuate from quarter to quarter.
Margins
AOI is not the strongest on margins, with operating margin widening for a second consecutive quarter on increased expenditures related to customer qualification projects for 800G and 1.6T products.
AOI reported GAAP gross margin of 30.3%, up more than 8 points YoY but down 0.3 points QoQ. Adjusted gross margin was 30.4%, with management guiding this to be essentially flat next quarter at 30.3%. Management has a target of 35% to 40% by late 2026 to 2027, emphasizing that they will need a few quarters for higher-margin 800G and cost reduction efforts from shifting to larger wafers to be seen. In the near-term, the ramp of single-mode 400G transceivers, which carry higher ASPs and gross margins, may provide a tailwind for expansion.
GAAP operating margin was (15.5%), a notable improvement from over (60%) in the year ago quarter, but widening from (6.5%) in Q4 and (8.9%) in Q1. Adjusted operating margin as (10.5%), down from (4.8%) in Q1 but up from (37.6%) a year ago.
Net margin was (8.8%), improving from (9.2%) in Q1 and more than (60%) a year ago. Adjusted net margin was (8.6%), down from (0.9%) in Q1 but improving from (25.1%) a year ago. For Q3, management’s guide implies adjusted net margin improving to (3.3%) at midpoint.
EPS
AOI reported a ($0.16) loss per share in Q2, missing estimates for ($0.07), as net margin did not improve much sequentially. For Q3, AOI guided for ($0.03) to ($0.10), a decent sequential improvement, likely driven by the ramp in 400G and a potential rebound in operating margin
Looking ahead, AOI is expected to shift to GAAP profitability by Q4 and expand earnings through mid-2026, with current estimates pointing to $0.03 in Q4 and rising to $0.19 by Q2 ’26, driven by the ramp of 800G.
Cash
Cash flows are also quite weak, though this has been mostly from surging accounts receivables and capex to support capacity expansion, a solid signal for the upcoming 800G ramp in conjunction with management’s commentary.
Operating cash flow in Q2 was ($65.5 million), widening from ($50.9 million) in Q1 as inventories and accounts receivables surged, up $36 million and $40 million QoQ respectively. OCF margin was (63.6%), down from (4.6%) a year ago and (51%) in Q1.
Free cash flow was ($104.3 million), widening from ($87.2 million) in Q1. Capex for the quarter was $38.8 million as AOI is quickly purchasing equipment to meet its aggressive capacity expansion targets for year-end and mid-year 2026. Full-year capex guidance is $120 to $150 million in preparation for increased 400G, 800G and 1.6T production.
Inventories were $138.9 million, up more than $36 million from Q1, driven by raw material purchases for use in production over the next few months.
Accounts receivable were $211.5 million in Q2, up $40 million QoQ. Digicomm accounted for $171.6 million of these receivables, tied to upcoming cable TV growth: “We talked about the dynamics, because we wanted to get some of the cable TV products in particular into the country and ready to be staged, ready for customer acceptance, we have offered some extended payment terms to certain customers in that channel chain to be able to accommodate that additional amount of revenue so that it's here when the customers need it.”
Cash and equivalents totaled $87.2 million, while debt was $188.2 million. During Q2, AOI completed its ATM program and raised $98 million net of fees, which it will utilize for equipment and machinery for its capacity expansion in Taiwan and Texas.
Earnings Q&A:
Gross Margin Increase Driven by Wafer Sizing
A five to ten point expansion in gross margins from 30% to 35-40% in four to six quarters (end of 2026 to early 2027) is a tall task to achieve while greatly expanding production capacity, but management provided some insights into the different levers available to reach this target. First, a shift from 2-inch wafers to 3-inch wafers is expected to bring substantial cost savings, along with 800G products ramping and improvements in cable TV:
Q, Simon Leopold, Raymond James:
Where I was trying to go with the question was to try to get a better sense of one of the elements to help the gross margin move towards that long term of 40%. So what I was trying to tease out in this question was the degree that you're outsourcing today versus a change towards more vertical integration in the future as a lever for gross margin improvement. So maybe the question is off-base and maybe I'm going down the wrong path. More bluntly, what will help the gross margin improve?
A, CEO Thompson Lin:
“Yes. I think the key is wafer, okay? Right now, we are doing 2-inch wafer. But we're going to 3-inch wafers. The cost will reduce by, I don't know, 50%, 60%. Then we'll go to 4-inch wafer by end of next year. This is a major, much bigger cost savings than what you're talking about. I think right now, yes, we're only maybe using 30% to 40% of our lasers. We were using, I would say, 2/3 of AOI lasers, okay? It will depend on customer by customer [basis]. Some customers prefer all the AOI lasers. Some customers prefer 50-50, okay? So that's why it's different. But more importantly, the cost funnel of AOI lasers changed from 2-inch to 3-inch to 4-inch.”
Management added that they need a few quarters to see a bigger impact from 800G products, which will have gross margin near 40%, and increased software revenue mix in cable TV, such as the QuantumLink remote management software deployment with Charter Communications.
Cable TV Demand to Reach $300-350M
Though the data center opportunity is what we’re most interested in, AOI’s commentary on cable TV pointed to potential 40% growth in fiscal 2026. Management said they have a “pretty clear line of sight” into channel inventory and demand, which supports this confidence in reaching $300 to $350 million in cable TV revenue next year. For perspective, cable TV revenue is on a ~$240 to $260 million run rate with $120 million in revenue in 1H:
CEO Thompson Li:
“So right now, next year, we are very comfortable, besides Charter, we should have more than 10 customers next year. And right now, based on the feedback from this customer, I think the real demand from these customers next year is, I would say, minimum $300 million to $350 million. … But the demand is pretty big, all right? Just next year, that's the number we see right now, $300 million to $350 million of real demand, all right, for this customer in, I would say, U.S., Canada, all right?”
CW Laser Production Increasing
AOI has been rather quiet about CW lasers for silicon photonics, with the focus primarily on the 800G transceivers, though management discussed increasing capacity for the high-power CW lasers to 2.5 million per month, or 30 million annually, by “sometime next year.”
We have previously discussed the importance of CW lasers for SiPho in our Lumentum analysis, Lumentum at Inflection Point with 20% QoQ Growth in AI-Related Segment, where Lumentum said it was “having challenges actually getting enough CW lasers for our own transceivers.” This will give AOI easily enough capacity to meet its planned transceiver capacity growth in a vertically-integrated manner:
“We are increasing our high-power CW laser for silicon photonics to maybe 2.5 million lasers per month by sometime next year. So right now, our in-house capacity is 100G EML. We should have 200G EML sometime soon, next year, for sure, the high-power laser for silicon photonics and VCSEL, the other new project, the 200G photo detector, okay? So this is all manufactured, 100%, in Houston.”
Conclusion
The primary risks with AOI stem from its margin profile as it is much weaker than more established networking players as well as the hypergrowth players we track. While there is an expectation for margins and EPS to improve through 2026 as 800G/1.6T products ramp, this will need to be proven over the coming few quarters. The weak cash flows also present a risk if there is a prolonged recovery, given the thinner cash position AOI has, while China concentration also poses a larger red flag given geopolitical tensions may be rising again.
However, it’s hard to argue with the strong growth expected across the optical transceiver industry emerging through 2026 from 800G and 1.6T products, combined with AOI’s 17X increase in capacity through mid-2026 for said products. Rough math suggests these products could generate more than double current quarterly revenue estimates when selling out at full capacity, while cable TV is expected to contribute nicely to growth in 2026 as well.
Damien Robbins, Equity Analyst at I/O Fund contributed to this analysis.
Every Thursday at 4:30 pm Eastern, the I/O Fund team holds a webinar for premium members to discuss how to navigate the broad market, as well as various stock and crypto entries and exits. Beth Kindig offers weekly deep dives including lesser-known cryptocurrencies and AI stocks, plus the team offers trade alerts. The I/O Fund team is one of the only audited portfolios available to individual investors. If you’d like to subscribe to the Advanced Market Signals plan, email us at premium@io-fund.compremium@io-fund.com.
Please note: The I/O Fund conducts research and draws conclusions for the Fund’s positions. We then share that information with our readers. This is not a guarantee of a stock’s performance. Please consult your personal financial advisor before buying any stock in the companies mentioned in this analysis.
Since launching the Discovery tier less than a year ago, The I/O Fund’s internal process for identifying new winners has greatly improved. Based on the research we produced from this tier, we added stocks like Bloom Energy, Core Scientific and Oklo to our portfolio, and these new additions became some of our biggest wins (thus far) in 2025.
As we continue to build out the Discovery library, we’d like to make it as easy as possible for our readers to follow along. We want to cast a wide net, explore and leave no stone unturned – therefore, we anticipate our coverage to include dozens of stocks over time. However, we also want to make sure we get down to brass tracks by providing you a clear takeaway by ranking what we’ve dug up every quarter. The ranking will also help clarify which ones we are eyeing an entry for and what setups Knox will cover in his Discovery Tier webinars. This will be called the I/O Fund’s Top 10 New Ideas List.
The ranking we provide is an estimate, which means all 10 stocks are of interest. Given the nature of momentum stocks, the ranking could shift quickly. Please check back to the Discovery Ranking list provided on the Dashboard for any changes in the interim.
Themes:
AI Networking:
Networking is at the heart of the new architecture that Nvidia is shipping now as the increased bandwidth is instrumental in driving higher performance. For example, the NVL72 systems will deliver 4X faster training and 30X faster inference compared to HGX systems. Notably, this is accomplished with many more GPUs from eight to 72 per system.
To support the new systems, the NVLink domain moves from supporting eight GPUs to 72 GPUs and 36 CPUs with a speed of 1.8 TB/s with 18 NVSwitch ASICs, up from four in the HGX/DGX systems due to increasing the number of GPUs but also due to doubling the per-GPU links. The 5th generation of NVLink supports up to 576 GPUs compared to the fourth generation of up to eight GPUs.
Scale-up refers to increasing the number of GPUs in an AI system. Proprietary NVLink remains the highest performance option for scale-up interconnects, although PCIe and scale-up Ethernet are also used. For the cabling, copper is used for intra-rack scale-up with up to 5,184 cables per system. Future generations of NVLink are likely to integrate optical I/O so that GPUs can communicate across racks without requiring costly retimers.
GB200 NVL72 with 72 GPUs and 36 CPUs has 18 NVSwitch chips and 72 InfiniBand NICs for scale-out networking and 36 Bluefield-3 Ethernet NICs for front-end networking. Compare this to the HGX systems with 8 GPUs and the DGX systems with 8 GPUs and 2 CPUs has 4 NVSwitch chips and 8 InfiniBand NICs.
This means the new architecture that Nvidia is shipping now results in 9X more GPUs, 4.5X more NVSwitches, 9X more InfiniBand NICs and 18X more front-end NICs. Each GPU requires its own InfiniBand link for scale-out whereas NVSwitch components grow faster than GPU count as each GPU must talk to every other GPU, therefore, it has more of an exponential growth.
Although NVLink is proprietary, it acts as a bellwether for the importance of AI networking and lesser-known suppliers. Generally speaking, what we can see from looking more closely at Nvidia’s networking fabric is that networking components are increasing 5X to 9X, and in some cases up to 18X.
Scale-out racks refer to connecting multiple racks across a cluster. InfiniBand switches and Ethernet switches are used for this purpose. As you can see below, Nvidia offers scalable units called SuperPODs that offer tens of thousands of GPUs. For a very large SuperPOD with 16,834 GPUs and 2048 nodes, there would be hundreds of InfiniBand switches required (or a hyperscaler can also use Ethernet switches) and extensive cabling is also required.
Also consider that as the networking and interconnects market matures, there will be new opportunities to participate, for example, co-package optics are expected to be introduced for the Rubin generation of GPUs in 2026-2027.
Regardless of the exact networking-to-compute ratio, as we scale up AI systems, the architecture becomes a networking and interconnect problem that must continually be solved for, and we want to be correctly positioned within the networking supply chain. Therefore, AI networking will remain a key focus for the I/O Fund’s Top 10 New Ideas in the coming quarters and years. You can expect our team to deliver additional names in this trend and to increase our allocation as needed.
AI Energy:
Data center power demand is expected to grow at an accelerated clip through the end of the decade and beyond, with more powerful GPUs and surging growth in inference two main drivers.
McKinsey projects data center capacity will rise ~2.5x to 219 GW by 2030, up from 82 GW in 2025, with AI contributing 70% of that demand. This corresponds to total capacity growth of 137 GW over the next five years, with 112 GW coming from AI.
Goldman Sachs estimated global data center power usage at 55 GW in early 2025, far below BCG’s 82 GW figure. However, GS projects power usage to reach 84 GW in 2027 and increase further to 122 GW by 2030, corresponding to total growth of just 67 GW. However, considering 2025 and 2026 capex spend could support >20 GW of new capacity, this forecast may understate the pace of capacity growth.
As that analysis pointed out, Nvidia’s Blackwell lineup is bringing a significant increase in power consumption, nearly double the H200’s 70 kW at 120 kW for the GB200 NVL72 and 140 kW for the upcoming GB300 racks.
Beyond Blackwell, Nvidia’s future design lineup shows continual increases in power consumption. Its Rubin generation is expected to boost thermal design power (TDP) by 50% over Blackwell at up to 180 kW per rack, with the upgraded Vera Rubin then doubling this to 360 kW per rack by 2027.
In its largest configuration, the Vera Rubin NVL576, dubbed the ‘Kyber’ rack, could draw as much as 600 kW (0.6 MW), or 5x that of the GB200 NVL72 in just a two-year design timeframe. These figures do not include networking, interconnects, cooling and other hardware, which will further boost power draw per rack.
This rapid increase in power consumption per GPU generation is critical, as existing infrastructure is simply unable to meet these escalating power demands. For example, Applied Digital pointed out that nearly 70% of current data centers “contain racks requiring between four and nine kW of power, and less than two percent of data centers have racks with greater than 50kW.”
For comparison, Super Micro’s GB200 NVL72 SuperCluster requires 132kW, while the upcoming Kyber rack could more than quadruple that to 600kW. Because Blackwell-based servers are now 15x to 30x the power density, cooling and power delivery strategies have to be redesigned, as liquid cooling now becomes a necessity.
This sharp rise in power density means current infrastructure may be unable to transition from 4-9kW racks to >130kW racks without incurring significant retrofitting costs, while building new infrastructure bypasses that hurdle and allows for optimization for high-powered racks.
For example, Vantage’s upcoming 1.4 GW campus in Texas for Oracle is designed for ultra-high density racks up to 250kW, yet this will not be enough power to able to host NVL576 racks in just two to three years’ time. Additionally, a former Microsoft Azure AI executive reportedly said he estimated that the “requirements in terms of power for the data center would probably at least double every three years and maybe exponentially so over a period of time,” further reinforcing this.
The push to 600kW racks over the next few years means this is not a transient problem, rather it is one that the industry will continue to face, meaning continuous new construction may be needed to handle surging power demand.
In the latest earnings call with CoreWeave, management agreed with the analysis we have presented to Pro Members, stating: “at the end of the day, right now, it's the powered shells that are the choke point that is causing the struggle to get enough infrastructure online for the demand signals that we are seeing […]”
Takeaway: When building our portfolio, we have to balance many things when it comes to AI energy stocks. Time to power is paramount as some energy infrastructure is 5-10 years out from commencing operations and generating revenue. Secondly, many energy stocks require significant cash to secure energy sources, get government approval and build the underlying infrastructure, which also includes taking into consideration regional differences, transmission capacity and distances from generation sites to metro areas.
AI Data Layer:
The initial years of AI development were compute intensive to where training created a compute hierarchy. We still see evidence of this hierarchy as companies with access to GPUs, networking and energy have an advantage, and the barrier to entry is high in both costs and by the limited supply preventing widespread access.
Nvidia has enjoyed a near-monopoly by being the parallel processing leader decades before AI took over as the primary market that demanded massive compute and data throughput for matrix multiplications and vector math. The lack of supply has afforded the world’s largest companies a head start in training and deploying models while startups and enterprises patiently wait for access. As more frontier LLMs are deployed by R&D labs and Big Tech, the emphasis moving forward will be on inference rather than only on training models.
The inference market is when enterprises and companies sitting on large private data sets will be able to increase the accuracy of open sourced models and licensed models. There will be an important shift to where companies that can offer domain-specific data in various industries, such as finance, healthcare, manufacturing, will do well by optimizing processes to generate more revenue and achieve better margins. It will start with the Fortune 500, the Global Fortune 2000 and well-funded startups. Meanwhile, investors should also not overlook the fact that R&D labs are growing closer to cracking the consumer market, as well, with apps such as OpenAI’s ChatGPT and Sora, or Perplexity’s search.
While training is benefiting those who sell the compute or own the infrastructure (and we will continue to own these stocks), there will also be a shift toward companies that manage the data pipes by sitting across the many database and software services that enterprises use. Think of all the ERP systems, CRMs, legacy databases, etc, where private data is stored. There will be an emphasis toward combining the data, keeping it private, yet utilizing it to increase the quality of inference.
Takeaway: Compute will continue to drive the scale for inference; data will drive the quality of inference. Therefore, a key focus for I/O Fund’s Top 10 New Ideas list over the next few quarters (and years ahead) will be AI data stocks that help private enterprises use their valuable data to feed data-hungry reasoning models. We will also, in tandem with AI networking and AI energy, be looking to build our portfolio with exposure to stocks that will participate in the AI inference market, which spans hardware, software, the data layer, and more.
The stocks below are new ideas and at time of publishing, the I/O Fund does not own these stocks although they are under strong consideration for the portfolio. To find out the stocks the I/O Fund owns, subscribe to our Pro tier for Research or our flagship tier Advanced with additional research, real-time trade alerts, allocations to stocks and weekly webinars.
CoreWeave: $20B+ in New Deals, Targeting Aggressive Power Growth
CoreWeave brands itself as the world’s first “AI hyperscaler” as they offer both infrastructure and a software platform for developing large language models and deploying them. Being dubbed an AI infrastructure player means CoreWeave must offer a compelling value proposition to attract business from arguably the largest competitors in the world – AWS, Microsoft Azure and Google Cloud.
One of their primary value propositions is offering bare metal servers, as the company does not need to offer shared GPUs like the hyperscalers. By stripping away the virtualization layer, raw performance goes up for R&D labs, who do not need to want to lower performance for workload flexibility. Although CoreWeave offers shared infrastructure in terms of storage and networking, the company’s key differentiation from the Big 3 is by offering dedicated bare-metal access. CoreWeave also offers proprietary software to help achieve higher total system performance and more favorable uptime relative to competitors.
CoreWeave has already reached a $5 billion run rate with 470MW (~20%) of its 2.2GW contracted power active and operational, leaving an additional ~1.73GW to be developed. The company is planning to have 900MW active by year-end, hence the need for high capex to support growth. At full-scale, the company may be able to support a $25 billion revenue run rate, aligning with FY29’s revenue estimate.
Overall Revenue Growth Overall Revenue Growth
CoreWeave reached a new milestone of over $1.0 billion in revenue in Q2 2025, growing 206.7% YoY and 23.6% QoQ to $1.21 billion, driven by strong demand for the company’s AI cloud infrastructure services. Revenue growth is expected to be strong in the coming quarters, driven by the robust demand due to training and inference workloads.
Management revenue guidance for Q3 is in the range of $1.26 billion to $1.30 billion, representing YoY growth of 119.2% and 5.5% QoQ at the midpoint. Revenue growth is expected to show a 20% acceleration QoQ in Q4 with revenue growing 139.5% YoY and a further 16 percent acceleration QoQ in Q1 2026, highlighting large deals signed in the recent quarters.
Looking forward, revenue is expected to grow by 174% YoY to $5.26 billion in the year 2025 and 129.6% YoY to $12.08 billion in 2026.
Key Metric Key Metric
CoreWeave reported backlog of $30.1 billion at the end of Q2, up 86% YoY driven by the company’s strategic deal with OpenAI in March 2025 and a subsequent expansion deal in May. CoreWeave expects 50% of this backlog, or ~$15 billion, to convert to revenue over the next 24 months, providing a strong degree of visibility into future growth.
However, CoreWeave’s backlog has now likely surpassed $50 billion, considering the company signed an additional expansion deal with OpenAI worth $6.5 billion and a large-scale deal with Meta worth $14.2 billion, both lasting through 2031.
Earnings Earnings
CoreWeave reported GAAP loss per share of (-$0.60) in Q2 compared to the consensus estimate of (-$0.49), with the miss stemming from higher operating expenses, particularly technology and infrastructure expenses.
CoreWeave is not expected to shift to GAAP profitability until FY27: analysts expect GAAP loss per share of (-$2.67) for this year, followed by (-$0.90) for 2026, before shifting to positive GAAP EPS of $1.59 in 2027.
Margins Margins
While gross margin has expanded over the last two quarters, operating margin remains thin from heightened expenditures and aggressive investments in data center capacity and servers.
Gross margin was 51.2%, up from 50.7% in Q1 and 43% in Q4. This is now slightly below the highest gross margin that CoreWeave has reported publicly at 53.7%.
Operating margin was just 1.6% in Q2, inflecting from a (2.8%) margin in Q1 but substantially lower than the 20% margin in the year ago quarter as tech and infrastructure expenses have surged more than 260% YoY.
Net margin was (24.0%) in Q2, marking a slight improvement from (32.1%) in Q1, with this pressured heavily by high interest payments on debt.
Cash Cash
CoreWeave’s business model is based on aggressive capacity expansion, currently fueled primarily by debt. As a result, cash is rather thin and gets spent quickly, and free cash flow is widely negative.
CoreWeave reported $1.15 billion in cash and equivalents (excluding $0.56 billion in restricted cash and equivalents), though CoreWeave updated in an 8-K related to its now upsized $1.75 billion raise that total cash will be closer to $5 billion.
Operating cash flow was ($251.3 million) for a (21%) margin in Q2, widening from ($117.8 million) in the year ago quarter. Free cash flow was ($2.7 billion) for a (223%) margin, widening slightly from ($2.36 billion) in the year ago quarter.
Debt was reported at $11.05 billion in Q2, with $3.62 billion being current. Current debt is likely closer to $12 billion now, with a majority (~$6.7 billion) tied to its two existing delayed draw term loan facilities. If its new DDTL is drawn upon, debt could rise to $14.6 billion. On the other hand, the company is trying to reduce its cost of capital by raising cash through secured debt, using its highly valuable GPUs as collateral.
Valuation Valuation
Given its limited history on the public markets, CoreWeave’s valuation comps are limited. The company is trading above its average forward PS multiple since IPO of 9.6x, currently valued at 12.8x. This is approaching its peak forward PS multiple so far of 16.8x.
Notable Risks Notable Risks
Based on company guidance, CoreWeave’s capex for the second half of the year is expected to be >$15 billion, with cash on hand only covering one-third of that at maximum. The majority of this capex, around >$12 billion, is expected to hit in Q4 due to timing of standing up new infrastructure, placing emphasis on finding funding to finance this spending. Debt is expensive for CoreWeave, with recent raises at >9% rates, making tapping the debt markets a more expensive endeavor – interest expenses were 22% of revenue in Q2.
Customer concentration also presents a risk, though this is now beginning to ease as other anchor customers Meta and OpenAI ramp. Microsoft had accounted for 72% of revenue in 1H, with the former two providing much needed diversification away from Microsoft.
IREN: Aggressively Building an AI GPU Cloud, Targeting $500M ARR by Q1
There are a few key things that separate IREN from other Bitcoin miners. The first is that IREN is still a Bitcoin miner whereas others have retrofitted their Bitcoin operations for the data center entirely or plan to very soon as the operations were not profitable. In contrast, IREN is able to turn a profit from Bitcoin mining and plans to use that cash to help fund its AI data center expansions.
Secondly, IREN can offer a hybrid mix of both colocation and AI cloud services, which is essentially bare metal servers without a hypervisor or virtual layer. This is attractive to hyperscalers as virtualization can lead to performance loss. For the AI cloud, IREN functions similar to CoreWeave as the GPUs are leased “as a service.” For colocation, IREN provides the facility, power and cooling for customers to deploy and manage their own hardware.
IREN is working to rapidly build out its GPU fleet considering its power pipeline totals 3GW, with the firm now owning more than 23,000 GPUs consisting of primarily Blackwells, more than doubling this month. IREN does have capacity to support >660K GPUs in total, though this will likely cost upwards of $60 billion in full.
Overall Revenue Growth Overall Revenue Growth
IREN delivered Q4 revenue of $187.3 million, up 229% from the $56.8 million earned in the year-ago period and up 65% sequentially from Q3’s $113.6 million. This kind of sequential growth is rarely seen outside of hypergrowth SaaS, let alone in a miner. The bulk of Q4 revenue came from Bitcoin at $180.3 million.
IREN reported FY25 revenue of $501.0 million, up 168% year-over-year from $187.2 million, underscoring one of the fastest growth rates in the sector. Looking ahead to 2026, IREN is estimated to report 113% growth to $1.09 billion, with growth each quarter expected to be >80% or higher.
AI Segment Revenue Growth AI Segment Revenue Growth
IREN’s AI cloud business is currently a small contributor to growth, with revenue of just $7 million in fiscal Q4. While this represented growth of 180% YoY and 94% QoQ, the segment accounted for less than 4% of total revenue in the quarter. For FY25, AI cloud revenue was $16.4 million, more than 5x higher than its FY24 revenue of $3.1 million.
In accordance with its doubled fleet at 23K GPUs, IREN unveiled a new AI Cloud annualized revenue (ARR) guidance, now targeting >$500 million in ARR by Q1 2026. This is more than double its guidance from August for $200 million to $250 million by year-end 2025, based on a fleet size of 10.9K GPUs. IREN said in early October that it has secured customer contracts for the 10.9K fleet representing $225 million in ARR, and that it remains on track to reach its Q1 target.
Earnings Earnings
IREN reported Q4 GAAP EPS of $0.66, though much of this (~$0.54) was attributed to a $147.7 million gain on financial instruments.
IREN’s EPS flipped positive in FY25, with diluted EPS of $0.39 versus a ($0.29) loss in FY24, marking its first full-year profit on a per-share basis, though GAAP EPS is heavily influenced by fair-value accounting marks. Looking forward, analysts expect GAAP profitability to more than triple through FY26 to $1.23.
Margins Margins
Operating margin shifted to positive in FY25, with Q4 seeing operating margin reach double-digit territory, a stark contrast to other miner peers.
Gross margin was 71.8% in Q4, reflecting benefits from scaling mining operations and disciplined power cost management. FY25 gross margin was 68.3%, up nearly 15 points YoY and signaling more room for expansion with Q4’s print.
Operating margin was 11.0% in Q4, down from 20.1% in the prior quarter. For FY25, operating margin was 3.5%, shifting positive from a (14.6%) margin in FY24.
Net margin was 94.4% in Q4, though stripping out the substantial gain on financial instruments would place net margin at 15.6%. For FY25, net margin was 17.3%, including 15.5 points contribution from gains on financial instruments.
Cash Cash
While operating cash flows were solid, free cash flow was largely negative at more than 2.5x revenue due to elevated capex.
Operating cash flow was $245.9 million in FY25, up from just $52.2 million in FY24. This equated to a strong 48.9% margin, up from 27.9% in the prior year. However, free cash flow widened from ($427.7 million) in FY24 to ($1.13 billion) in FY25, with IREN spending $574 million on PP&E and $799 million on GPUs and hardware.
IREN reported $564.6 million in cash against $962.8 million in convertible debt in Q4. IREN also priced an additional $875 million in convertible debt in early October, which will likely go towards additional GPU purchases.
Valuation Valuation
With its strong multi-month rally, IREN is now trading at peak forward PS multiples at 16.5x, far above its historical average 3.9x forward multiple.
On the bottom line, IREN is trading at the upper end of its forward PE range over the past year at 70x forward earnings. This is notably elevated from early August’s 17x forward PE multiple.
Notable Risks Notable Risks
IREN marks the first (and perhaps only) attempt across Bitcoin miners to double up as a neocloud, which may carry substantial capital risks to self-fund GPU fleet expansion. Analysts expect IREN to grow its GPU fleet by 5x by year-end 2026, though this firm’s estimate hinged on IREN taking on $6 billion in new debt related to GPU purchases. This would quickly consume cash and possibly could cost hundreds of millions quarterly in interest, depending on terms, which would be hard to cover with cash flows.
Additionally, IREN is targeting a rather aggressive ramp in its AI cloud business, which relies on internal projects for utilization, hourly rental rates, and on-time GPU delivery, all factors that may change quite quickly. Monthly AI cloud revenue has still yet to reach an inflection point as of August, suggesting little to no new GPU deliveries have occurred since June.
Palantir: AI Platform Drives Eight-Quarter Revenue Acceleration
The difference between Palantir and other AI-enabled database competitors is that Palantir is able to answer questions a model cannot answer. Traditional business intelligence companies require a complete data set whereas Palantir is able to tackle situations where there is not a complete data set. You can think of the competitive advantage as being actionable depth, which Palantir has described as “the reasoning that goes into decision-making, not just data.”
Palantir’s Artificial Intelligence Platform (AIP) integrates generative AI with operational data and workflows, and when combined with Palantir’s other platforms Foundry and Apollo, it provides an AI service mesh that can run hundreds of microservices, scale compute through its Rubix engine and orchestrate updates through Apollo.
Additionally, Palantir’s knowledge graph referred to as Ontology is a distinct advantage. The graph offers better context than a large language model would on its own – or as Palantir states, it’s “the reasoning that goes into decision-making.”
Overall Revenue Growth Overall Revenue Growth
Palantir cracked the $1 billion quarterly revenue milestone in Q2, with revenue of $1.003 billion in the quarter. Growth accelerated nine points to 48% YoY, and seven points to 14% QoQ. For Q3, Palantir guided for growth to accelerate slightly to 50%, what would mark its ninth straight quarter with accelerating revenue growth.
Management also boosted its full-year revenue growth forecast by ~$250 million, from $3.9 billion to $4.15 billion, corresponding to growth of 45% YoY. This marks a sharp acceleration from 29% growth in FY24.
AI Segment Revenue Growth AI Segment Revenue Growth
Palantir’s US Commercial segment is the primary vector for its AIP-driven growth, with the company seeing robust momentum from the segment. US Commercial revenue grew 93% YoY (a 22 point sequential acceleration) and 20% QoQ to $306 million.
Palantir also scored US commercial remaining deal value of $2.79 billion, up 145% YoY and 20% QoQ, while US commercial total contract value (TCV) rose 222% YoY to $843 million. US commercial customers rose 64% YoY to 485.
For the full year, Palantir boosted its guide to >85% YoY growth to $1.302 billion, compared to its Q1 guide for 68% growth to $1.178 billion.
Earnings Earnings
Palantir reported adjusted EPS of $0.16 in Q2, up 78% YoY, and GAAP EPS of $0.13, both coming in ahead of estimates. For Q3, Palantir is expected to report adjusted EPS growth of 67% YoY to $0.17.
For fiscal 2025, Palantir is estimated to see adjusted EPS rise 57% YoY to $0.64, before slowing to 32% growth to $0.85 in fiscal 2026.
Margins Margins
Palantir is a standout in terms of margins, as the company continues to drive operating margin expansion while accelerating revenue growth. This helps the company’s Rule of 40 metric, which stands at 94 as it combines adjusted EBITDA margin with revenue – or more than double the ideal 40 that many SaaS companies set out to achieve yet cannot due to a lack of GAAP margins.
Gross margin was 80.8% in Q2, down from 81% in the year ago quarter but up marginally from 80.4% in Q1.
Operating margin was 26.8%, expanding significantly from 15.5% in the year ago quarter and from 20% in Q1. Adjusted operating margin was 46.3%, up from 37.4% in the year ago quarter and 44.2% in Q1; Palantir guided for continued strength in Q3 at a 45.6% margin.
Net margin was 32.6%, up nearly 13 points from 19.8% in the year ago quarter and up from 24.2% in Q1.
Cash Cash
Palantir also stands out for its ridiculously strong cash flows, with margins above 50%, putting it in rare territory for a high-growth AI stock when combined with its operating and net margin profile.
Operating cash flow was $539.3 million in Q2 for a margin of 54%, well above its 20% margin from the year ago quarter but slightly below its peak OCF margin of 58% in Q3 2024.
Adjusted free cash flow was $568.8 million for a 57% margin, again up substantially from 22% a year ago but below its peak FCF margin of 63%. Palantir raised its adjusted FCF guidance for FY25 by $200 million, from $1.6-1.8 billion to now $1.8-2.0 billion, or a margin of 45.8% at midpoint.
Cash and equivalents totaled $6.0 billion, while debt was zero.
Valuation Valuation
On the topline, Palantir trades at 105.4x forward PS, far above its 5-year average multiple of 33.4x and well above the second most expensive software stock in Cloudflare at 35.7x forward PS. Palantir is in uncharted territory as it is at its peak multiple ever sustained for this metric.
On the bottom line, Palantir trades at 288x forward PE, again at its peak and far above its average of 103.4x. On an FCF basis, the company trades at 273.3x, also above its average multiple of 154.5x.
Notable Risks Notable Risks
The valuation with Palantir is a gamble as the company is attempting to set a new bar for AI software, with >100x forward sales multiples only achieved for short fashion in 2021 for a handful of prior market darlings, whose stocks have yet to return to those prices. Palantir’s elevated valuation may also present a risk if/when the company reaches peak revenue growth, as it cannot accelerate the topline forever.
Lumentum: Cloud & Networking Growth Accelerates to 16% QoQ
Optical interconnects are a trend the I/O Fund has been tracking for more than a year, as these interconnects help data centers accelerate data throughput between and inside the data center between servers or racks, while reducing latency and power consumption. Lumentum supplies components for datacom transceivers (including VSCELs, CW lasers for silicon photonics and EML-based lasers) and optical interconnects with tech that has caught the attention of heavyweight Nvidia.
While we have been closely monitoring Lumentum and waiting patiently for their EML lasers for 200G to ship, the company’s most recent report provided confirmation of the EML laser ramp, as EMLs achieved an all-time high for shipments with revenue more than doubling versus its June 2024 baseline. Management also cited a “substantial” 200G EML order to be fulfilled in the December quarter, although offered little additional clarity on the size of the order.
Lumentum also remains quite confident in its growth opportunities from EML lasers, optical circuit switches and co-packaged optics heading into 2026. The company provided a new $600 million quarterly revenue target it expects to reach by next June or earlier, up from $424 million in the current quarter.
Overall Revenue Growth Overall Revenue Growth
Lumentum reported Q4 revenue of $480.7 million, beating analyst estimates by a modest 2.29%. This was a notable uptick on the top-line, with growth of 13.1% QoQ, accelerating 7 points, and 55.9% YoY, accelerating 42 points, beginning to support the thesis that Lumentum is past the cyclical low it experienced in FY24.
Guidance for Q1 FY26 came in at $510 to $540 million, representing 55.8% growth YoY and a slight moderation to 6% QoQ growth. Management attributes the strong forward revenue guide to surging AI workloads and the “shift toward high-speed photonics for hyperscale cloud operators.”
FY25 revenue came in at $1.65 billion, with growth rebounding to 21% YoY from FY24’s (23%) YoY decline. FY26 is expected to see revenue growth accelerate nearly 20 points to almost 40% YoY, with current estimates at $2.29 billion.
AI Segment Revenue GrowthAI Segment Revenue Growth
Lumentum’s Cloud and Networking revenue came in at $424.1 million in Q4, representing 66.5% YoY growth. Additionally, the segment’s QoQ growth of 16.1% accelerated sharply from 7.6% QoQ in Q3.
Management cited a few factors behind the outperformance in Q4: strong hyperscaler demand driving more than 50% QoQ growth in cloud modules, all-time high in EML shipments, and strong transceiver demand. Of these, EML drove the results this current quarter: “In Components, we achieved an all-time high in EML shipments, nearly doubling the revenue compared to our June quarter 2024 baseline.”
For Q1, management expects Cloud & Networking to be up QoQ, and based on guidance, this would likely correspond to approximately 10% QoQ growth at midpoint. Management indicated there was potential for strong growth starting in the upcoming December quarter: “Recently, we received a substantial order for 200-gig lane speed EML chips, which we expect to fulfill in the December quarter. Overall, we expect 2026 to be a breakout year for laser chip sales of both 100-gig and 200-gig lane speeds.”
Earnings Earnings
Lumentum reported adjusted EPS of $0.88 in Q4, beating estimates of $0.81 and improving against the $0.57 reported in Q3 and $0.06 reported in the year ago quarter. Management guided for $0.95 to $1.10 in adjusted EPS in Q1, up nearly 470% YoY, while Q2 is expected to see 176% growth to $1.16.
Looking ahead, Lumentum is expected to see adjusted EPS rise at a 77% CAGR through fiscal 2027, rising from $2.06 in FY25 to $4.86 in FY26 to $6.48 in FY27.
Margins Margins
The re-acceleration in revenue drove solid in GAAP gross margin to 33.3%, higher than the 28.8% percentage seen in Q3 and double the 16.6% margin from Q4 FY24. Non-GAAP gross reiterates this story, as Q4’s 37.8% margin continued to expand versus the prior quarter comp of 35.2% and prior year comp of 27.8%.
Lumentum showed continued progress on operating margins as well in Q4 with a (1.7%) operating margin, compared to (8.9%) in Q3 FY25 and (43.3%) in Q4 FY24. Non-GAAP operating margin of 15.0% expanded nicely compared to prior quarter of 10.8% and prior year quarter of (5.1%). Most of the profit margin is driven by the Cloud and Networking segment, which boasted a margin of 23.6% in Q4, up more than 13 points YoY.
Q4’s adjusted net income margin of 13.1% reflects continued operational improvements and the fourth consecutive quarter of sequential improvement (from 9.6% in Q3, 7.5% in Q2, and 3.6% in Q1).
Cash Cash
Cash flows have been lumpy, though Q4 saw Lumentum report the highest operating cash flow margin in the past two years. Despite this, free cash flows are pressured as Lumentum reinvests to expand manufacturing capacity.
Operating cash flow increased 80% YoY to $64 million, representing a 13.3% margin. This expanded nearly 2 points from 11.5% in the year ago quarter and was a notable uptick from (0.4%) in Q3. For FY25, operating cash flow was up ~5x to $126.4 million, for a 7.7% margin, improving from a 1.8% margin in FY24.
Free cash flow declined (7%) YoY to $10.1 million, for a 2.1% margin, down from 3.5% in the year ago quarter. As a result of elevated capex, FY25 free cash flow was ($104.7 million), for a (6.3%) margin, improving only slightly from (8.3%) in FY24.
Cash and equivalents of $877.1 million in Q4, largely in line with the $866.7 million reported in Q3. Debt remained largely consistent with the prior few quarters at $2.57 billion.
Valuation Valuation
Lumentum is trading at peak multiples on a forward PS basis, at 5.2x, nearly twice its average 5-year multiple of 2.8x, where it was valued at as recently as June.
On the bottom line, Lumentum is trading below average multiples, at 35.3x forward PE versus its 5-year average of 40x and its 2025 peak at 50x in January. However, similar to the topline, this has expanded significantly since June, when Lumentum traded at 19x forward PE.
Notable Risks Notable Risks
Customer concentration is a risk present in Lumentum, as two customers currently represent 31% of total revenue. When asked about customer concentration for specific products, such as cloud modules and OCS, management stated that due to their being capacity constrained, it was unlikely they would take on new customers.
Trading at peak multiples on the top-line also presents a risk, as the company is expected to see quarterly revenue growth peak in fiscal Q1 at 56% and then decelerate to ~40% for the next two quarters, an unfavorable position to be in with an elevated valuation.
GE Vernova: Aiming for 60GW Backlog, Supplying AI Data Centers
GE Vernova is part of the spinoff that General Electric first announced in 2021 and later completed in 2024. The company broke up its three biggest segments into separate units: GE Healthcare was spun off first in 2023, GE Vernova for the Energy business was spun off in 2024 and began trading as GEV, and GE Aerospace was the business that remained with the existing stock ticker GE. At the time the Energy segment was split up, it was seeing $33 billion in revenue and was helping to generate 30% of the world’s electricity with 55,000 wind turbines and 7,000 gas turbines.
This year, the company is expected to report $37 billion in revenue with strong earnings growth of 45.3%. The stock is not a hypergrowth profile, but rather, it is a quality, defensible position that could do well during any periods of doubt in the broader AI trend.
The defensibility is particularly attractive when you consider that gas turbines is the crux of the issue for expanding gas power plants. According to Bloomberg’s calculations, more than “$400 billion worth of gas-fired power plants through the end of the decade are in jeopardy of delay or cancellation because of the lack of capacity to meet future turbine orders.” The same article points toward GE Vernova filling orders as far out as 2030.
GE Vernova is the world’s largest gas turbine supplier at 25% ahead of Schneider at 24%. Even still, GEV nearly tripled its gas turbine equipment orders this past quarter – a statement that has us sitting up in our seats. Per the earnings call: “Power orders grew 44%, led by Gas Power equipment nearly tripling year-over-year.” In a bid to supply options quickly to alleviate bottlenecks, GEV is also shipping aeroderivative gas turbine packages and doing extensive R&D on a small modular reactor (SMR) design.
Overall Revenue Growth Overall Revenue Growth
GE Vernova is on a path to revenue acceleration in 2026, as it is a major beneficiary of the spending surge of hyperscalers stemming from the increasing energy requirements from the global AI infrastructure build-out.
The company’s Q2 revenue grew by 11% to $9.11 billion, beating estimates by 3.6%. Organically, revenue grew by 12% YoY to $9.04 billion, primarily due to higher equipment and services revenue. Analysts expect revenue to grow by 2.7% YoY in the next two quarters and revenue growth to accelerate to 8.8% in Q1 2026.
On the back of strong demand for power and equipment, management has raised its full-year revenue guidance and expects 2025 organic revenue to come at the high end of the guidance of $36 billion to $37 billion. The power segment organic revenue guide is increased to 6-7%, up from the previous single-digit guide, and the electrification segment is expected to grow 20%, up from the previous mid-high teens percentage. On a side note, the wind segment is expected to be down mid-single digits due to the more challenging market conditions.
Revenue growth is set to accelerate over the next three years. Analysts expect a 6.4% increase in 2025, bringing the total revenue to $37.17 billion, a 5.1% growth in 2024. Momentum is projected to build further, with revenue climbing to $40.7 billion in 2026, up 9.5% and to $45.5 billion in 2027, up 11.9% YoY.
Key AI Metric Key AI Metric
In Q2, GEV signed 9GW of new gas equipment contracts with 2GW going directly to orders and 7GW going into what’s called a slot reservation. During the quarter, the company also converted 3GW into orders and shipped 5GW of equipment. GEV is also witnessing robust demand for its aeroderivative technology to support data centers, securing 27 aeroderivative units in the recent quarter compared to only one in the same period last year.
Overall backlog is now up to 55GW, ahead of the 50GW guided in April, including 29GW in backlog and 25GW in slot reservation agreements (SRA), up from 21GW. GEV is expecting the backlog will reach 60 GW by the end of this year. There was a discussion on the call that this represents 3 years of backlog: “And we've talked about the fact that we'll get to at least 60 gigawatts by the end of the year. So that's directionally 3 years of backlog.” In other words, the chances that GEV is not a significant player in supplying energy to data centers for many years to come is nil.
Earnings Earnings
Q2 GAAP EPS came at $1.86, beating estimates by a solid 14.3% driven by profitable volume, better pricing, and productivity gains. Analysts expect GAAP EPS of $1.85 for Q3 2025 compared to (-$0.35) in the same period last year. They expect GAAP EPS to grow 95.4% YoY to $3.38 in Q4 and 137.4% YoY to $2.16 in Q1 2026.
Analysts continue to expect strong EPS growth in the coming years. For the year 2025, analysts expect GAAP EPS to grow 45.3% YoY to $8.11, and 58.9% and 41% YoY in the subsequent two years, reaching $18.16 in 2027.
Margins Margins
GEV has rather thin margins, with operating margin in the mid-single digits, but management expects that higher turbine prices and strong demand will lead to improved margins as the backlog approached 60GW by year-end.
Gross margin was 21% in Q2, flat YoY but up 2 points from Q1. Operating margin was 5.6%, rebounding 4.7 points QoQ but down 2.4 points YoY. Net margin was 5.6%, up 2.3 points QoQ but down 10.2 points YoY.
Adjusted EBITDA margin was 8.5%, up more than 2 points YoY, with the improvement driven by volume and price, offsetting tariff impacts and investments. While GEV did raise its full-year adjusted EBITDA margin guide to 8-9%, management was clear to say that this included approximately one point of negative EBITDA margin related to tariffs.
Cash Cash
Management raised full-year free cash flow guidance from a range of $2.0 billion to $2.5 billion to a new range of $3.0 billion to $3.5 billion, primarily driven by a higher profit outlook and increased down payments due to rising orders. Year-to-date, GEV has generated $1.17 billion in free cash flow, implying a strong acceleration of free cash flow in the second half of the year to $2.1 billion.
Q2 free cash flow of $194 million represented a sharper decline versus $821 million in the same quarter last year, though GEV had benefited from a $300 million arbitration in the comparable quarter.
GEV has $7.9 billion in cash and equivalents and no debt.
Valuation Valuation
GEV is trading near peak multiples on the top line, but there is more room on the bottom line as multiples are not nearly as stretched.
The stock is trading at approximately 4.6x forward revenue, just below its peak of nearly 4.9x and above its average 2.6x multiple, though data is limited as the spinoff has only traded publicly for a year and a half.
On the bottom line, GEV trades at 83.3x forward earnings, slightly above its average 73.9x multiple, though shares have traded in as wide a range of 36x to 137x over the past year.
Notable Risks Notable Risks
Wind is providing a notable drag on growth and earnings, with GEV stating that they have lost approximately $300 million in the segment through the first half but expect to be closer to breakeven in the second half. Q2 adjusted EBITDA for Wind widened approximately ($50 million) YoY to ($165 million). Additionally, Q3 growth was guided to be down mid-teens YoY, but excluding a one-time settlement in the year ago quarter, growth would be up low single digits.
Applied Optoelectronics (AOI) is a lesser-known optical component and transceiver supplier, though the small-cap has recently caught our attention for its ambitious capacity growth targets, aiming to become one of the largest domestic manufacturers for 800G transceivers.
The VSCEL laser, EML chip, and optical transceiver supplier has provided some of the strongest commentary for future growth, with management outlining 8.5x capacity growth by year-end and another 2x capacity growth in 2026. This builds upon a comment slipped into Q1’s call that AOI believes it can become the largest 800G/1.6T optics supplier to Amazon, after signing a deal in March that management says is worth more than $4 billion over ten years.
AOI also stood out from its 40% QoQ growth in data center revenue in Q2, though this came after a soft Q1 affected by inventory digestion at its largest hyperscale customer and seasonality. Interestingly, data center was not the core driver of growth in Q2, with cable TV taking the reign with 8x YoY growth in the quarter. However, management is eyeing 800G transceivers to begin ramping as early as late Q3, supported by that substantial capacity growth come 2026.
The crux with AOI is that cash flows are much softer than more-established AI networking stocks, and GAAP margins are still negative, though it’s hard to argue with the implications from the profound expansion in capacity and how optics growth can transform margins.
Overall Revenue Growth Overall Revenue Growth
AOI reported revenue of $102.95 million in Q2, up 138% YoY and 3% QoQ, though this was a modest (2.7%) miss versus consensus estimates. This was a slight deceleration from 146% growth in the previous quarter. Cable TV was the primary growth driver with revenue up 862% YoY to $56 million, while data center revenue rose 30% YoY to $44.8 million. Telecom and other revenue was $2.1 million, down (45%) YoY.
For Q3, AOI guided for revenue to be between $115 million to $127 million, pointing to 86% YoY growth at midpoint, though on a sequential basis this is a sharp uptick to 17% QoQ growth. Management expects that there may be some initial contribution from 800G modules late in the quarter: “Looking ahead to Q3, we expect a sequential increase in our datacenter revenue, driven by continued growth in our 100G and 400G products with the possibility of layering some additional increased 800G revenue late in the quarter.”
AI Revenue Growth AI Revenue Growth
AOI’s data center revenue rose 40% QoQ and 30% YoY to $44.8 million, though the high sequential growth rate (the highest among AI networking stocks we track) was due to a soft comp in Q1, where revenue declined (29%) QoQ on inventory digestion at one of AOI’s largest hyperscaler customers and seasonality.
>800G revenue has not yet begun to appear as AOI has not yet begun shipments, with volume only going to qualification this quarter and accounted for 1% of data center revenue at maximum. Revenue from 200G/400G products accounted for 20% of revenue, or ~$9 million, while 100G accounted for 70% of revenue, or $31.4 million.
For Q3, AOI guided for a sequential increase in data center revenue, saying that “400G is picking up so strong in Q3, Q4.” There is also the possibility that some initial 800G revenue begins layering in late in Q3, but primarily in Q4 and into 2026 as capacity expands and customers move ahead with projects.
Earnings Earnings
AOI reported a ($0.16) loss per share in Q2, missing estimates for ($0.07), as margins did not improve much sequentially. For Q3, AOI guided for ($0.03) to ($0.10), a decent sequential improvement, likely driven by the ramp in 400G.
Looking ahead, AOI is expected to shift to GAAP profitability by Q4 and expand earnings through mid-2026, with current estimates pointing to $0.03 in Q4 and rising to $0.19 by Q2 ’26, driven by the ramp of 800G.
Margins Margins
AOI is not the strongest on margins, with operating margin widening for a second consecutive quarter on increased expenditures related to customer qualification projects for 800G and 1.6T products.
AOI reported GAAP gross margin of 30.3%, up more than 8 points YoY but down 0.3 points QoQ. Adjusted gross margin was 30.4%, with management guiding this to be essentially flat next quarter at 30.3%. Management has a medium-term target of 40%, emphasizing that they will need a few quarters for higher-margin 800G and cost reduction efforts from shifting to larger wafers to be seen.
GAAP operating margin was (15.5%), a notable improvement from over (60%) in the year ago quarter, but widening from (6.5%) in Q4 and (8.9%) in Q1. Adjusted operating margin as (10.5%), down from (4.8%) in Q1 but up from (37.6%) a year ago.
Net margin was (8.8%), improving from (9.2%) in Q1 and more than (60%) a year ago. Adjusted net margin was (8.6%), down from (0.9%) in Q1 but improving from (25.1%) a year ago. For Q3, management’s guide implies adjusted net margin improving to (3.3%) at midpoint.
Cash Cash
Cash flows are also quite weak, though this has been mostly from surging accounts receivables and capex to support capacity expansion, a solid signal for the upcoming 800G ramp in conjunction with management’s commentary.
Operating cash flow in Q2 was ($65.5 million), widening from ($50.9 million) in Q1 as inventories and accounts receivables surged, up $36 million and $40 million QoQ respectively. OCF margin was (63.6%), down from (4.6%) a year ago and (51%) in Q1.
Free cash flow was ($104.3 million), widening from ($87.2 million) in Q1. Capex for the quarter was $38.8 million as AOI is quickly purchasing equipment to meet its aggressive capacity expansion targets for year-end and mid-year 2026.
Cash totaled $87.2 million versus $188.2 million in debt, with AOI recently completing its ATM program in Q2, raising $98 million net cash that will help support its capacity expansion and R&D.
Valuation Valuation
AOI is trading a 4.2x forward revenue multiple, well below its 8x peak at the end of 2024 but also far above its low of 1.2x in April, emphasizing the volatility that stems from its small-cap profile with strong growth and weak cash.
On the bottom line, shares are trading at 37.5x estimated EPS of $0.85 in 2026, with this expected to be the company’s first year of GAAP profitability since 2017.
Notable Risks Notable Risks
The primary risks with AOI stem from its margin profile as it is much weaker than more established networking players as well as the hypergrowth players we track. While there is an expectation for margins and EPS to improve through 2026 as 800G/1.6T products ramp, this will need to be proven over the coming few quarters. The weak cash flows also present a risk if there is a prolonged recovery, given the thinner cash position AOI has.
Micron: HBM, LP Server DRAM Driving Strong Growth
Thematic: 9/10
Fundamentals: 6/10
Valuation: 7/10
Brief Overview: Brief Overview:
Micron is a primary beneficiary of rapidly increasing dollar content of high-bandwidth memory (HBM) chips with each new generation of GPUs. AI training and inference rely heavily on HBM for the massive memory bandwidth that complex models require. AI servers also use more DRAM and NAND than a traditional server. These are reasons that Micron’s cyclical fundamentals could become more secular as the AI economy is built out.
In fiscal 2025, Micron’s HBM, high-capacity dual in-line memory modules (DIMMs) and low-power (LP) server DRAM revenue reached $10 billion, up more than fivefold from the prior year, while HBM alone reached $2 billion in revenue in Q4.
Management expects robust AI server demand, the shift to HBM4 and tight DRAM supply to remain tailwinds to growth and profitability moving through 2026: “we expect healthy demand supply environment in 2026 for overall DRAM, and that bodes well for profitability of DRAM, profitability of HBM and of course, profitability of non-HBM as well, which is experiencing tight supply.”
What Micron will need to answer is if the cyclical nature of the memory market will smooth out as the dollar content of memory is rapidly increasing. Data center is already proving to be a strong driver of growth for Micron, accounting for 56% of sales in FY25, up from 35% in FY24. On a dollar basis, data center revenue surged 137% YoY to $20.75 billion.
Overall Revenue Growth Overall Revenue Growth
Micron reported record Q4 revenue of $11.32 billion, driven by DRAM products (and within that HBM) with revenue up 27% QoQ to $9 billion. Growth accelerated nearly 10 points sequentially to 46% YoY, and on a sequential basis, growth was 22% QoQ, a six point acceleration. Micron guided to a fresh record in Q1 at $12.5 billion at midpoint, pointing to 44% YoY growth and 9% QoQ.
For FY25, revenue rose 49% YoY to $37.38 billion, driven primarily by DRAM and HBM revenue, which rose more than 62% YoY to $28.58 billion. HBM reached an annualized run rate of $8 billion in Q4, with HBM share to grow again in Q1 and HBM4 capacity in discussions to be sold out for calendar 2026. Micron has not provided a full-year guide for revenue, but current consensus estimates call for 43% growth to $53.5 billion in revenue.
AI Revenue Growth AI Revenue Growth
In fiscal 2025, Micron's data center reached a record 56% of company revenue, with growth primarily driven by DRAM products and aided by data center SSDs and NAND components. Overall, data center revenue increased 137% YoY to $20.75 billion.
Micron’s Cloud Memory Business Unit (CMBU), which consists of its HBM, high-capacity dual in-line memory modules (DIMMs), and low-power server DRAM solutions, saw revenue surge 257% YoY in fiscal 2025 to $13.57 billion, or YoY growth of nearly $10 billion. Micron’s other data center unit, its Core Data Center Business Unit (CDBU), consists primarily of data center SSDs and NAND components. This unit saw revenue growth of 45% YoY to $7.23 billion.
For Q4, CMBU revenue rose 214% YoY to $4.54 billion, while CDBU revenue declined (23%) YoY. CMBU revenue growth was driven by HBM and strong bit shipment growth, though Micron offered no commentary behind the decline for CDBU.
Earnings Earnings
Micron is expected to see earnings double this fiscal year as margins have swiftly recovered from late 2023 and early 2024. In Q4, Micron reported adjusted EPS of $3.04, up 157% YoY and beating estimates by 6%.
For Q1, Micron guided for adjusted EPS to be $3.75, +/- $0.15, more than 23% ahead of guidance and corresponding to YoY growth of 110%. Earnings growth is expected to reaccelerate to 155% in Q2 but then decelerate to 126% in Q3. For the full year, Micron is expected to see 100% YoY growth to $16.63.
Margins Margins
Micron’s margin turnaround story has been impressive, with gross margin up more than 55 points over the last two years and operating margin up more than 66 points.
GAAP gross margin in Q4 was 44.7%, up 7 points QoQ and 9.4 points YoY, aided by strong growth in CMBU which carried a 59% gross margin in the quarter, DRAM pricing and favorable product mix. For Q1, gross margin was guided to be 50.5% at midpoint, a nearly 6 point sequential expansion and up more than 12 points YoY.
Operating margin was 32.3%, up 9 points QoQ and 12.7 points YoY, again aided by CMBU which carried a 48% margin in the fourth quarter. For Q1, Micron expects operating margin to be 38.6%, up more than 6 points QoQ and more than 13.5 points YoY, signaling some tailwinds from operating leverage from CMBU.
Net margin was 28.3% in Q4, up 8 points QoQ and nearly 17 points YoY. The trajectory within gross and operating margins suggests that there is a path for net margin to potentially expand to the mid-30% range in FY26.
Cash Cash
Operating cash flow was $5.73 billion for Q4, up 68% YoY and more than 24% QoQ. OCF margin was 50.6%, up 1 point from Q3 and up 6.7 points YoY. For the year, Micron generated operating cash flow of $17.53 billion, more than doubling from $8.51 billion in fiscal 2024, with OCF margin expanding 13 points to 46.9%.
Adjusted free cash flow was $801 million in Q4, shrinking from $1.95 billion in Q3 on surging capex. Adjusted FCF margin was 7.1%, up from 4.2% in the year ago quarter but down from 21% in Q3.
Micron reported total cash and equivalents of $11.9 billion and total debt of $14.6 billion.
Valuation Valuation
Despite its recent rally, Micron trades at reasonable multiples, well below its peak from 2024. On the top line, Micron trades at 4x forward revenue, 10% above its average 3.6x multiple and far below its peak of 6.8x from mid 2024.
On the bottom line, Micron trades at 11.6x forward earnings, though its 43.9x average is skewed higher by 2024’s >100x multiples when margins were razor-thin. Since the start of 2025, Micron has traded as high as 16x forward earnings and as low as 8x.
Notable Risks Notable Risks
Micron’s growth to this point and beyond has been centered around HBM, both on the top and bottom lines. CMBU is the only unit that sees operating margins above the corporate total, at 48% versus 25%, 29% and 20% for its other segments, meaning that future operating margin expansion will be tied solely to growth from CMBU, and felt more the faster CMBU grows. This may mean that margin and earnings upside in 2026 may be limited come 2027.
Talen: $18 Billion, 17-Year Nuclear Deal With Amazon
Talen is an independent power producer with its power assets primarily located in the PJM region, with more than 10GW of generation capacity with 2.2GW nuclear. With assets primarily located in Pennsylvania, Maryland and now Ohio, Talen has exposure to growing data center regions, having already locked in a long-term power agreement with Amazon to fuel data centers in Pennsylvania.
Talen’s deal with Amazon not only locked in substantial power generation for Amazon’s data center assets in a time the industry is facing a power crunch, but also locked in substantial revenue and free cash flow generation for Talen. The deal is expected to add a visible 50% uplift in cash flows as the deal ramps into full capacity by 2032, while providing a repeatable colocation model for Talen’s other assets to meet rising data center demand in the PJM region.
Talen is also expanding its power production portfolio with recent acquisitions of the Freedom Energy Center and Guernsey Power Station for ~$3.8 billion gross, expected to add 2.8 GW of combined-cycle natural gas generation capacity in the PJM region. The two plants are suitable for hyperscale data center supply, and are also expected to be immediately accretive to cash flows.
Overall Revenue Growth Overall Revenue Growth
Talen’s Q2 revenue rose 29% YoY and 62% QoQ to $630 million, positively impacted by a $176 million gain on derivatives (compared to a $76 million gain in the year ago quarter). Revenue from contracts with customers was $409 million, up 18% YoY but down (37%) QoQ. This shows the quarterly fluctuations that can stem from commodity contract hedging. For the first half of the year, Talen’s revenue was $1.02 billion, up 2% YoY, though revenue from customer contracts was $1.06 billion, up 40% YoY.
Looking ahead, Talen has some visibility into 2026 revenue from the PJM auction in July, where Talen cleared 6.7 GW of capacity translating to ~$805 million in capacity revenue for the 2026-27 planning year lasting June 2026 through May 2027. Talen expects 2026 capacity revenue of ~$747 million, up ~124% from $333 million projected in 2025. For additional perspective, capacity revenues were just $88 million in Q2, or ~22% of total contracted revenue with customers.
Key AI Metric Key AI Metric
Talen’s AI ties are to Amazon, having expanded and signed a 17-year power purchase agreement through 2042 with the hyperscaler to power its data center adjacent to the Susquehanna plant and potentially other facilities in the Pennsylvania region.
Talen signed the 1.92 GW agreement worth $18 billion in June, and expects to deliver the first 240MW by mid-2026, scaling to 360MW by mid-2027 and 480MW by mid-2028. Full volume is expected to be reached no later than 2032 with potential to accelerate this timeline, with price escalators through 2042.
Though Talen has not provided a view into how it believes revenue will ramp, at face value the deal is worth more than $1 billion in average annual revenue, though it will take up to seven years to reach full volume. Talen has provided a glimpse into how the deal will accrete to FCF in the ramp stage. The company expects $1.55 in FCF from Amazon in fiscal 2026, before rising to $4.00 to $5.75 by fiscal 2029 (with 840-1,200MW delivered), and to $7.00 to $8.25 by fiscal 2032 (1,600MW to full capacity).
Earnings Earnings
Talen reported GAAP EPS of $1.50 in Q2, rebounding from ($2.94) in Q4 as sharper losses on commodity contracts ate into revenue. This is also not comparable to the year ago quarter where Talen reported $7.60 in EPS, as this included an approximate $9.40/share benefit from the sale of its Cumulus data center to Amazon.
For fiscal 2025, Talen is expected to report $5.39 in GAAP EPS, signaling a strong second half of the year, with 2026 EPS projected to surge 285% to $20.74.
Margins Margins
Talen’s gross margin (operating revenues including derivatives minus energy expenses) was 60% in Q2, down from 63.3% in the year ago quarter. Operating margin was 10.5%, up from 5.5% in the year ago quarter and (27.2%) in Q1, driven by the adverse revenue impact from derivate losses.
Net margin was 11.4% in Q2, which does not compare to the asset-sale impacted margin from the year ago quarter of 92.8%. Q1’s net margin was (34.6%), dragged down by derivatives.
Cash Cash
Talen’s operating cash flow was ($184 million) in Q2 for a (29.2%) margin, bringing 1H operating cash flow to ($65 million) for a (6.4%) margin, down from $150 million for a 15% margin in the year ago period.
Adjusted FCF was ($78 million) in Q2 for a (7.6%) margin, with 1H adjusted FCF just $9 million. Talen maintained its guidance for $450 to $540 million in adjusted FCF for the year, implying a significantly stronger second half of the year.
For fiscal 2026, Talen is guiding to $980 million to $1.18 billion in adjusted FCF, more than doubling YoY, and for fiscal 2027, adjusted FCF of $1.055 billion to $1.425 billion. This would represent over 250% growth in just two years.
Talen’s cash balance is extremely thin at $122 million versus its reported debt at $2.97 billion in Q2, though pro-forma debt is actually much higher at $6.56 billion, including $3.8 billion for the acquisitions of the Freedom and Guernsey natural gas plants to add >2.8GW capacity.
Valuation Valuation
Talen is trading at peak multiples on the top-line at 8.6x forward revenue, more than 2x its average multiple of 3.6x. On the bottom-line, Talen trades at 80.9x estimated FY25 EPS, at peak levels, but for FY26 EPS, Talen trades at a more reasonable 22.2x multiple, far below its peak 1-year forward multiple of nearly 41x. On an adjusted FCF basis, Talen trades at 18.8x FY26’s adjusted FCF per share guidance of $23.60 at midpoint, versus 43.1x FY25’s adjusted FCF of $10.30 at midpoint.
Notable Risks Notable Risks
Talen has a thin balance sheet with substantial debt, and has launched multiple financings via term loans, revolving credit facilities and senior notes to fund its recent acquisitions, though it still may need more cash over the next few quarters. The Amazon deal does de-risk the future growth story by locking in substantial revenue and free cash flow growth, but not for its immediate-term cash needs. Talen had emerged from Chapter 11 bankruptcy in 2023 with a significantly reduced balance sheet, though its cash balance has returned to extremely thin levels; the Amazon deal beckons a strong ramp in FCF that may be able to pad the balance sheet in the future.
Nebius: Microsoft Deal Worth up to $19.4 Billion Supports Hypergrowth Phase
Nebius is similar to CoreWeave, dubbing itself as AI native cloud infrastructure, which means the infrastructure was built specifically for AI workloads with architectures built on bare metal servers instead of hypervisor layers, for example. This approach along with a few other optimizations can result in significantly faster training.
Nebius offers an Nvidia-optimized cloud platform for teams that need to adjust compute resources, want high-performance storage and an easy-to-use AI environment yet do not want to manage the infrastructure or AI operations. The stock surged earlier this month off the announcement of a mega deal with Microsoft worth up to $19.4 billion. This deal signals just how supply constrained the market is, given Microsoft is willing to partner with a neocloud to scale quickly.
Given Nebius’ roots are from Yandex, one might question if Nebius is truly AI-native as some of the Finnish data center in Europe was built in 2014. However, that is also where one of Nebius’ strength lies, which is the company offers European data centers which helps a company like Microsoft to expand quickly overseas.
To serve the AI-native market, Nebius has been building out colocation sites to increase capacity and lower latency for the incoming inference market. Companies like Cloudflare and Shopify are early customers, both of which need to power inference at the edge. The company is also expanding beyond Europe with data center expansions in New Jersey and Kansas.
Overall Revenue Growth Overall Revenue Growth
Nebius reported a slight beat in Q2 with revenue of $105.1 million, up 625% YoY and 106% QoQ, versus estimates for $101.2 million. AI cloud infrastructure revenue rose more than 9x YoY in the quarter, fueled by strong demand for Hopper GPUs and almost peak GPU utilization.
Looking ahead, Q2 is marking an inflection point for revenue, with growth accelerating sequentially on a dollar basis. Q3 is estimated to see revenue up more than $52 million QoQ to $157.1 million, before rising another $106 million QoQ to $263.8 million in Q4. For FY25, Nebius has maintained its guidance for $450 million to $630 million in revenue, excluding $50 million to $70 million in revenue attributed to one of its subsidiaries, Toloka.
Nebius is one of the more unique names in the AI universe as one of the few, if not only, AI-focused stock to have multiple years of triple-digit revenue growth ahead. Nebius had already laid out expectations for a meaningful acceleration next year prior to the Microsoft deal, based on timing of capacity ramp in Q4 2025 and throughout 2026. The recent deal with Microsoft is extending this acceleration, with revenue revised more than $3 billion higher by 2029, or 24% to 65% above prior expectations.
Prior to the Microsoft deal, Nebius was expected to surpass $5 billion in annual revenue in 2029, doubling from $2.5 billion in 2027. After the deal, revenue estimates have been raised 24% to $3.1 billion in 2027, and 65% higher to $8.44 billion by 2029.
AI Revenue Growth AI Revenue Growth
In Q2, Nebius boosted its AI Cloud annualized run rate guidance by 14%, supported by its data center expansion, more power coming online in 2H and the rollout of more Blackwell and soon Blackwell Ultra GPUs. To note, the increased ARR guidance preceded the Microsoft deal, although it may be unlikely that ARR guidance will be raised much this year considering that capacity roll-out for Microsoft is more geared towards 2026.
Q2 ARR rose ~73% QoQ and 438% YoY to $439 million, marking a significant two-quarter expansion from $90 million in Q4. Nebius is targeting ARR to more than double over the next two quarters, raising its guidance to $900 million to $1.1 billion, up from its prior view for $750 million to $1 billion.
Nebius explained that the “increase in our ARR guidance reflects the strong demand we are seeing and the expected delivery of additional GPU capacity later this year, particularly the Blackwell Ultras. Because much of this capacity will come online by the end of the year, the impact will show up more in ARR than within the year’s revenue.” Much of the ARR growth is likely to be weighted towards Q4, considering the majority of GPU installations will take place that quarter, such as the ~4,000 Blackwell Ultras in the UK.
Earnings Earnings
Nebius is expected to report negative EPS through all of fiscal 2025 and 2026, with no clear indication yet of when the company could or will shift to profitability. Current estimates show Nebius losing ($1.47) in 2025 and minimal improvement to ($1.34) in 2026.
There are no analyst estimates beyond 2026, but given the capital intensity of greenfield data center development, continuous scaling of capacity, not to mention potential investments into the autonomous driving unit Avride, Nebius may face a long road to profitability.
Margins Margins
While gross margin is expanding rapidly, up from 26% in Q4 to over 70% in Q2, operating margins remain deeply negative as due to the high costs of aggressively expanding data center capacity and GPUs on hand.
Excluding Toloka’s contribution, its subsidiary that was deconsolidated in Q2, gross margin was 71.3%, up nearly 20 points from 51.5% in Q1 and improving nearly 25 points from 46.9% in the year ago quarter. Nebius is now slightly below CoreWeave’s Q2 gross margin of 74.2%, and it remains to be seen how much further upside there is to gross margin as new capacity comes online.
Q2 operating margin was (105.8%), a notable improvement from (236.4%) in Q1 and (773.8%) in the year-ago quarter. Nebius is exhibiting initial signs of operating leverage, with total operating expenses up just 71% YoY versus the 625% YoY growth in revenue; this was driven mainly by 560% YoY growth in depreciation from increased server capacity, as SG&A decreased (10%) YoY.
GAAP net margin was 556%, as Nebius benefited from a one-time $597.4 million gain from Toloka. Adjusted net margin offers a clearer view of Nebius’ trajectory, coming in at (87.1%) in Q2, up from (164.4%) in Q1 and (424.8%) in the year ago quarter.
Cash Cash
Aggressive capacity expansion plans and a 33% increase in FY25 capex expectations from $1.5 billion to $2 billion, or nearly 4x expected revenue for the year, mean Nebius is quickly burning through cash.
Operating cash flow was ($167.7) million in Q2, for a (159.6%) margin, improving from a (357.7%) margin in Q1. This represented just a $30 million sequential improvement from ($197.8) million, suggesting that the path to positive cash flows may be prolonged. Free cash flow was ($678.3) million in Q2 for a (645.4%) margin, versus ($741.8 million) for a (1341.4%) margin in Q1.
Capex was $510.6 million in Q2, or nearly 5x of revenue, down slightly from $544 million in Q1. 1H capex surpassed $1.05 billion, meaning $0.5 billion to $1 billion is on deck for 2H.
Cash and equivalents totaled $1.68 billion, up only slightly from $1.45 billion in Q1 as Nebius deployed much of June’s $1 billion raise to capex and operations. Including the recent $2.75 billion raise and $1 billion share sale, cash is likely around or above $5 billion. At current burn rates, Nebius would have nearly 10 quarters of cash, but it is likely that capex will accelerate (think of CoreWeave as a leading indicator) to support growth in power and capacity. Debt was $0.98 billion as of Q2, not including the recently priced $2.75 billion in notes.
Valuation Valuation
Nebius also trades at a substantial premium to CoreWeave despite lagging its competitor in terms of active power, contracted power and revenue scale. Based on FY26’s revenue estimate, Nebius trades at a forward 18.2x multiple, more than triple CoreWeave’s 5.9x multiple. However, given revenue growth is expected to be triple-digits through 2028, the market is pricing Nebius to quickly grow into these multiples.
Notable Risks Notable Risks
Nebius is even more complicated than CoreWeave given they also own a capital-intensive autonomous driving division, among other investments, and was formally the company Yandex. Nebius is high-risk given its success depends on how much capital the company can raise, and the likelihood it remains in CoreWeave’s shadow is high. Regardless of how Nebius competes with CoreWeave, it remains an AI bubble stock as the company has to hope the stock prices goes up to raise cash, which will dilute shareholders (or raise debt). It’s a vicious cycle as during any months/quarters that AI stocks are soft, Nebius stock will carry outsized execution risk.
Cloudflare: Multi-Faceted AI Positioning, Steady Growth
Thematic: 9/10
Fundamentals: 5/10
Valuation: 1/10
Brief Overview: Brief Overview:
There is a quiet strength in Cloudflare’s fundamentals and key metrics. For example, Cloudflare passed a $2B run rate for the first time, signed their first $100M deal, dollar-based net retention (DBNRR) seems to have bottomed along with a slight 1.3% acceleration in revenue. Regarding the bottom line, Cloudflare is certainly stronger than many cloud peers yet tends to walk a razor’s edge due to capex.
Cloudflare references its business units as “Acts” – Act 1, Act 2 and Act 3. The company defines Act 1 as application security, Act 2 as Zero Trust and Act 3 as the Workers Platform. For our purposes as stock investors, it’s Act 3 we are most interested in.
Regarding AI inference and the Workers Platform, management connected some important dots on the earnings call as to why agentic AI will drive forward the massive inference trend. The I/O Fund team recently dug up a stat inference is expected to account for 60% to 70% of AI workloads by 2030. In particular, Cloudflare emphasizes their position is what will help the company win this market: “The fact that we sit in front of so much of the web and that more than half of our dynamic traffic is already between APIs means that we are strategically positioned to deliver the agentic web of the future.”
Cloudflare also introduced Act 4 – a new product that will help AI search engines connect with (and potentially) pay publishers for using derivatives of their copyrighted works. Although the amount of demand for this and exactly how Cloudflare will monetize this new product is not clear, it is interesting management feels confident enough to call the new use case its fourth act.
Overall Revenue Growth Overall Revenue Growth
Cloudflare reported its largest revenue beat in the last six quarters at 2.1% above consensus, with Q2 revenue up 27.8% YoY to $512.3 million. This also marked a slight 1.3 point acceleration on the top-line from 26.5% growth in Q1.
For Q3, Cloudflare guided for revenue of $543.5 to $544.5 million, ahead of estimates at the time for $538.9 million. This corresponds to a slight deceleration to the mid-to-high 26% YoY growth range, where Cloudflare is expected to remain through Q4. This provides no clear indication yet that the company is able to drive a sustained revenue acceleration aided by AI. However, consensus estimates are now above the high-end of management’s guidance at $544.9 million, essentially already pricing in stronger momentum fueling a beat for Q3.
For the full-year, Cloudflare raised its outlook to $2,113.5 million to $2,115.5 million, for YoY growth of 26.7%. This is a $22.5 million increase at midpoint from Cloudflare’s prior outlook for $2,090 million to $2,094 million for growth of 25.3%.
Key AI Metric Key AI Metric
Although management has been optimistic about AI driving a re-acceleration on the top-line, Cloudflare has not broken out AI revenue or contribution to growth. Other key metrics have remained strong in Q2.
RPO increased 39% YoY and 6% QoQ to $1.98 billion. Current RPO accounted for 66% of total RPO, or ~$1.30 billion, increasing 33% YoY in Q2, a four point acceleration from 29% growth in Q1.This is also a notable uplift from 26% growth in the year ago quarter. Billings also increased 33% YoY to $559.2 million, a third straight quarter with growth above 30% YoY.
Paying customers increased 27.5% YoY to 267,929 in Q2, the second quarter in a row with 27%+ growth. This is a notable improvement from 17% and 21% growth in Q1 and Q2 2024. Cloudflare stated that it added a record number of customers YoY spending over $1M and over $5M.
Earnings Earnings
Cloudflare topped estimates in Q2 driven by the revenue beat and stronger adjusted margins, and boosted its FY25 adjusted EPS outlook as a result. GAAP EPS was ($0.15), missing estimates for ($0.08) as GAAP margins drifted lower. Adjusted EPS was $0.21, beating estimates for $0.18, fueled the outperformance in adjusted operating margin in the quarter.
For Q3, Cloudflare guided for $0.23 in adjusted EPS, a slight uptick sequentially, while for FY25, the company raised its forecast from $0.79-$0.80 to $0.85-$0.86. This corresponds to growth of ~14.5% YoY, up from the mid-6% range previously. Growth is expected to be much stronger in FY26 at ~30% YoY to $1.12.
Margins Margins
Gross margins drifted lower in Q2, driven by both an increase in depreciation expenses and in allocated costs from higher network traffic from paying customers. GAAP operating margins followed, moving further away from reaching break-even.
Cloudflare had an interesting comment on long-term margins, stating that it expects to remain comfortably in its 75% to 77% adjusted gross margin target despite passing on substantial savings to Workers’ customers. This suggests that upside to operating margins will be driven by expenditures, such as moderating higher sales & marketing spending, at 36% of revenue versus its target range of 27-29%, and high SBC at 24% of revenue.
GAAP gross margin was 74.9% in Q2, down nearly 3 points YoY and 1 point QoQ. Adjusted gross margin was 76.3%, down 2.7 points YoY and 0.8 points QoQ.
GAAP operating margin was (13.1%), down 4.4 points YoY and 2 points QoQ. Adjusted operating margin was 14.1%, approximately flat YoY and up 2.4 points QoQ; this was also ahead of guidance for 12.6%. For Q3, Cloudflare guided for adjusted operating income of $75-76 million, pointing to adjusted operating margin of 13.9%, down nearly 1 point YoY and moderating slightly QoQ.
GAAP net margin was (9.8%), down 6 points YoY and 1.8 points QoQ. Adjusted net margin was 14.7%, down 2.6 points YoY but up 2.5 points QoQ.
Cash Cash
Cash flow margins contracted sequentially, while Cloudflare significantly bolstered its cash pile after a large convertible note issuance.
Operating cash flow was $99.8 million for a 19% margin, flat YoY but down from a 30% margin in Q1. Free cash flow was $33.3 million for a 6% margin, down 4 points YoY and 5 points QoQ. Network capex was 11% of revenue in Q2, down from 17% of revenue in Q2. Cloudflare stuck to its guidance for network capex to be 12-13% of revenue for the year, suggesting slight moderation in 2H.
In June, Cloudflare raised $1.97 billion in new convertible notes due 2030, raising its cash on hand to $3.96 billion while convertible notes outstanding rose to $3.26 billion.
Valuation Valuation
Cloudflare is trading at peak multiples, and is the second-most expensive name in the software universe behind Palantir. Cloudflare trades at 36.2x forward revenue, slightly below its late September peak of 37.6x but nearly double its average 19.7x multiple.
On the bottom-line, Cloudflare is not yet GAAP profitable, but on an adjusted basis, the company trades at 256x forward earnings, again far above its 144.5x average multiple.
Notable Risks Notable Risks
Given Cloudflare’s valuation, the entry is probably the most important aspect of this stock right now – whereas in the medium to long-term the most important aspect is timing for the broader inference market. The market looks to already be pricing in a sustainable AI inference-aided reacceleration on the topline despite the fact that this has not appeared concretely, amplifying risk if this does not visibly pan out over the next couple of quarters.
Conclusion:
We are thrilled about our new tier Discovery as the results are able to deliver new ideas to enthusiastic AI investors, such as ourselves. We quickly spotted the limitations around running an active portfolio that does not dedicate a separate effort to new idea generation as the market moves fast with new winners emerging every year. As we look at Q4 and beyond, we believe this quarterly analysis combined with an actively managed Top 10 list will become a strong offering. Our cumulative record proves we are one of the strongest teams in the world on AI stocks. Moving forward, our goal is to use our proven methodology to deliver additional value add as we participate heavily in the once-in-a-lifetime trend of AI.
In the coming weeks, we expect things to shift rapidly as new information is published daily during earnings season. Please reference our Top 10 Watchlist spreadsheet and incoming analysis as critical tools for staying on top of the Must Know stocks in the space.
Our goal is to update this list weekly, so stay tuned for frequent updates!
Damien Robbins, Equity Analyst at I/O Fund contributed to this analysis.
Every Thursday at 4:30 pm Eastern, the I/O Fund team holds a webinar for premium members to discuss how to navigate the broad market, as well as various stock and crypto entries and exits. Beth Kindig offers weekly deep dives including lesser-known cryptocurrencies and AI stocks, plus the team offers trade alerts. The I/O Fund team is one of the only audited portfolios available to individual investors. If you’d like to subscribe to the Advanced Market Signals plan, email us at premium@io-fund.compremium@io-fund.com.
Please note: The I/O Fund conducts research and draws conclusions for the Fund’s positions. We then share that information with our readers. This is not a guarantee of a stock’s performance. Please consult your personal financial advisor before buying any stock in the companies mentioned in this analysis.
Palantir leads the AI software pack in terms of strong earnings reports this past quarter as the company achieved significant milestones, the most impressive being US commercial revenue grew 93% YoY and 20% sequentially. You will be hard pressed to find this kind of QoQ growth across AI’s biggest players.
The Artificial Intelligence Platform (AIP) is driving most of the Commercial growth as there was a clear and obvious revenue inflection when AIP launched in mid-2023. AIP is a cloud-agnostic and model-agnostic platform that connects AI with existing systems and operations. AIP goes beyond what LLMs can deliver on their own by embedding models in workflows and logic. The value creation comes from being able to work with incomplete data sets through the ontology layer while also offering a level of reasoning that goes far beyond just analysing the data itself.
AIP’s scalability and flexibility continues to attract larger and more ambitious commercial engagements. As you’ll see below, key metrics such as total contract value (TCV), US commercial customer count, US commercial remaining deal value (RDV) and RPO are supportive of continued growth in future quarters.
There are also robust cash flows and expanding margins to strengthen the story. However, what is this Perfect 10 worth? In fact, Palantir is well beyond a Perfect 10 and is rather a Perfect 100 as its forward P/S sits at an astonishing 104. This means taking FY2025 revenue of $4.2B , it would take over 100 years to pay back its current market cap.
Below, the I/O Fund team weighs in on the future growth opportunities of Palantir, the spectacular earnings report and how we view it’s hotly debated valuation.
Palantir’s Platforms: The Foundation of Its AI Growth Strategy
Palantir’s first platform was Gotham for government purposes before many of the integrated features were expanded to Foundry, which launched around 2021 for commercial purposes (exact date is not available but generally understood to be around this time).
Gotham and Foundry create a unified data set for actionable insights across industries such as manufacturing, product development, and customer experience. Palantir primarily uses its customer's own databases for data but may also use publicly available information, like social media, for government clients. The system is traditionally deployed on-site at the customer's data center.
Its main competitive advantage is its ability to work with incomplete data sets, unlike traditional business intelligence tools that require complete data. Palantir describes this as offering "actionable depth"—focusing on the reasoning behind decision-making, not just analyzing the data itself.
The core platforms are designed to create a model of the real world by analyzing numerous data points. Unlike traditional SQL databases, they allow users to query data using natural language and receive results in real-time.
The Gotham platform enables users to identify patterns hidden deep within datasets using semantic, temporal, geospatial and full-text analysis.
Palantir Foundry is the commercial offering and has four layers of tooling: Foundry Core, Data Foundation, Ontology and Workflows. The Ontology layer offers a distinct, competitive advantage by allowing datasets to be turned into real-world concepts with the ability to accelerate on the company’s core ontology to reduce redundancy. Workflows are where it all comes together in an integrated environment. When a user has a question, it answers it using all layers and tools available
The Apollo layer provides continuous delivery and an automated configuration layer that enables Foundry and Gotham to operate across all cloud environments and in places where there is little to no connectivity. On top of Palantir being able to form conclusions from incomplete data sets, the company can also deploy its platform and applications anywhere.
Palantir’s marketing team says Apollo “goes where no SaaS has gone before” because it allows what is done on-premises to also run on multi-cloud SaaS with code that is deployed across all environments rather than written for a specific environment.
Where bandwidth is not an issue, the company transmits all raw inputs and enriched metadata from models. Where there are constraints, the platform transmits meta-data only which can reduce bitrate by 20X.
Apollo Edge AI creates a "meta-constellation" by linking up to 237 satellites to reduce decision-making delays. This network coordinates hundreds of sensors and AI models, enabling complex tasks like tracking submarines in areas without bandwidth. One example is tracking submarines that pose a threat to the U.S. and its allies. In this case, submarines are being tracked on a granular level in areas where there is no bandwidth available. A key feature is its ability to operate independently, without relying on a single cloud provider such as AWS or Azure.
Palantir’s Artificial Intelligence Platform (AIP)
The Artificial Intelligence Platform has helped the stock surge in recent years as it integrates generative AI with operational data and workflows. When AIP is combined with Foundry and Apollo, it provides an AI service mesh that can run hundreds of microservices, scale compute through its Rubix engine and orchestrate updates through Apollo. Like Apollo, AIP Is independent from any one cloud environment.
AIP Ontology is what Separates Palantir:
The knowledge graph referred to as Ontology is a distinct advantage. The graph offers better context than a large language model would on its own – or as Palantir states, it’s “the reasoning that goes into decision-making.”
You will often hear the management team state large language models will become commoditized, which is a way of saying the software that is on top of the LLM is where value creation comes from rather than the LLM alone. For this reason, AIP is designed to not only be cloud agnostic but to also be LLM-agnostic as it works with any large language model – for example, OpenAI, Anthropic, Meta’s Llama, etc.
The platform also offers an AI agent workflows for building AI agents that are further optimized for specific use cases and customized through additional tools. Autonomous agents can be built and tested on the platform.
When it comes to security and governance, Palantir’s roots in government contracts means the software company is exceptional compared to peers in this area.
Enterprise Growth in AI is Accelerating
Cloud and model providers are now pushing for greater enterprise AI adoption, a trend that arguably will benefit Palantir just as much from new use cases for AI, and increasing amounts of data generated from AI models and agents.
For example, OpenAI CEO Sam Altman told investors that they “should expect a huge focus from us on really leaning into enterprise,” as the company unveiled an array of integrations such as integrating Codex with Salesforce’s Slack, or using Figma in chats to create diagrams. OpenAI also launched AgentKit, aiming to provide its 4 million developers with a “unified suite” for AI agent development, a key area other enterprise-focused software firms like ServiceNow are focusing on.
IBM teamed up with Anthropic on Tuesday to integrate its Claude models on IBM’s platform to advance enterprise software development, with IBM noting that early testing across 6,000 clients found productivity increases of ~45% on average with recognizable cost savings. This follows Anthropic’s announcement on Monday that it would roll out its Claude models to Deloitte’s entire 500,000 global workforce.
… and Surging Token Growth Support Palantir’s Story
Surging token growth in a way also supports rising enterprise AI adoption and a tailwind for growth for Palantir, despite a majority of tokens likely being generated by the hundreds of millions of users for consumer-facing AI products. OpenAI disclosed that it is processing 6 billion tokens per minute, or ~260 trillion per month, while Google’s token generation more than doubled to 980 trillion in one month.
AI agents are expected to generate magnitudes more tokens, with Anthropic estimating agents process 4x more tokens and multi-agent systems 15x more than typical chatbot interactions. Over the long run, Dell projects data generated will more than triple from 2024 to 2029, from 174 ZB to 523 ZB. To put this in perspective, 1 ZB is approximately equivalent to nearly 4 billion 256GB iPhones, and over the next four years, 350x that.
For Palantir’s AIP, which connects frontier models directly to enterprise data streams, this creates a surge in data that Palantir can then contextualize and provide value for decision making for its customers. There is already more evidence below the headline figures that Palantir is benefiting from increasing enterprise AI adoption, such as Palantir’s quarterly deals closed, which not only shows that its deal flywheel remains healthy with 13% QoQ growth in Q2 (accelerating from 8% QoQ in Q1), but also an increasing amount of larger deals over the last three quarters.
Palantir’s quarterly deals closed rose 13% sequentially in Q2, accelerating from 8% growth in Q1, signaling strong AI-driven enterprise growth momentum.
Palantir Raises FY Guidance Following Strong Earnings Performance
Palantir's second-quarter performance surpassed a significant one billion revenue milestone for the first time. This achievement was driven by remarkable 48% YoY growth to $1.0 billion, a figure that outpaced consensus estimates by 6.8%. The result represents a dramatic inflection from the 12.7% growth reported in Q2 2023—the launch quarter of its Artificial Intelligence Platform (AIP), underscoring the platform's profound impact.
Most importantly, Q2 marked the eighth consecutive quarter of accelerating revenue growth, with the rate increasing by 870 basis points sequentially, signalling powerful and sustained business momentum driven by AI. The Q2 revenue growth rate also smashed the high end of the company’s prior guidance by nearly 10 percentage points and represents a 21 percentage points increase compared to the growth rate in the same period last year.
The company is witnessing strong growth in customers, which grew by 43% YoY and 10% sequentially to 849. Most importantly, revenue from the largest customers continues to expand. Second quarter trailing 12-month revenue from the top 20 customers increased 30% YoY to $75 million per customer, indicating strong market adoption and consistent expansion.
AIP remains the primary driver of the company’s revenue reacceleration. Its scalability, flexibility, and adaptability enable rapid integration across enterprises. Commercial customers leverage Palantir’s AI and machine learning capabilities to harness the power of cutting-edge large language models (LLMs) within Foundry and Gotham, delivering near-instant analytics and insights, as well as meaningful gains in productivity and efficiency. The company’s US commercial revenue was the main driver of growth, which accelerated 22 percentage points sequentially to 93% YoY growth to $306 million.
The company's Q3 revenue guide was also impressive in the range of $1.083 billion to $1.087 billion, representing a YoY growth of 49.5% at the midpoint. On a sequential basis it represents an 8-percentage growth and is the highest sequential revenue growth guide in the company’s history.
The current consensus estimates point to a 50.5% YoY growth to $1.09 billion. The company's Q3 revenue estimates have witnessed strong positive revisions, up from 34.8% during the first week of May, representing a solid 15.7 percentage point improvement within five months. While Palantir has been delivering tremendous results and consecutive acceleration on the top-line each quarter, a crucial moment to watch for will be when the company reaches its peak growth quarter. Analysts are currently modelling this to come as early as Q4 with revenue growth expected to be 44%, decelerating more than 6 points sequentially. This could be a tough spot for investors considering the company's stretched valuation, as there has yet to be a test of growth peaking and decelerating.
On the back of the solid Q2 results the management also raised the full year 2025 revenue guidance midpoint to $4.146 billion, representing a 44.7% YoY growth rate, a solid 880 basis points increase over the full year 2025 revenue guidance provided last quarter, and the largest ever full year revenue guidance raise.
Analysts expect 2025 revenue to grow 45.3% YoY to $4.16 billion, an increase of 10 percentage points from the 35.3% growth expected in the beginning of May. Looking forward, analysts expect 2026 revenue to grow 35% YoY to $5.62 billion and 34.4% YoY growth to $7.55 billion in 2027.
Palantir Stock Gains as U.S. Commercial Revenue Jumps 22%
AIP continues to drive strong results for the company’s US commercial segment. The Q2 US commercial revenue grew by a whopping 93% YoY and 20% sequential growth to $306 million. Revenue growth rate accelerated from 71% YoY growth and 19% sequential growth recorded in the previous quarter.
The strong acceleration also led to US commercial revenue to comprise 31% of the company’s Q2 revenue compared to 23% in the same period last year. If we exclude revenue from strategic commercial contracts, the Q2 US commercial revenue grew by 95% YoY and 20% sequentially. Management also emphasized during the earnings call that the company’s products are used in real-world applications and provide tangible benefits. The company is benefiting from larger new deals and the continued strong momentum with existing customers, driving strong revenue growth.
U.S. commercial revenue growth accelerated from 22% to 93% year-over-year in Q2 2025, reflecting surging enterprise adoption and AI-driven demand.
Here are some of the top highlights from the quarter for US commercial:
The company reported the strongest quarter of US commercial TCV booked at $843 million, representing a YoY growth of 222%.
Over the trailing twelve months, the company closed $2.8 billion of US commercial TCV bookings, an increase of 141% from the prior 12 months period, demonstrating the demand for AI production use cases.
Total remaining deal value in the US commercial segment was up 145% YoY and 20% sequentially, reflecting continued momentum and expanding customer demand.
The US commercial customers grew by 64% YoY and 12% sequentially to 485 customers, adding 53 customers in the recent quarter. It also implies that the company’s customers increased 4x from Q2 2022.
Management shared a couple of TCV highlights during the earnings call, wherein a healthcare company completed a boot camp in April of this year and later signed a $88 million TCV deal a month later to coordinate and automate its patient care across facilities. Similarly, an American telecom company started working with the company in 2022 and since then has increased their contract 10x, projecting hundreds of millions in cost savings.
Management raised its full-year US commercial revenue guidance to over $1.302 billion, implying at least 85% YoY growth. This updated outlook is 17 percentage points higher than the prior quarter’s guidance, reflecting strong commercial business momentum.
Overall, Commercial revenue grew by 47% YoY and 14% sequentially to $451 million. Revenue accelerated by 14 percentage points from the previous quarter. When excluding the impact from strategic commercial contracts, second-quarter commercial revenue grew 49% YoY and 14% sequentially. The company closed $1.1 billion in commercial TCV bookings, up 185% YoY.
International Commercial revenue remains a weak spot, as Q2 revenue declined (-3%) YoY but grew just 2% sequentially to $144 million. Management noted in the earnings call that Palantir continues to “capitalize on targeted growth opportunities in Asia, the Middle East and beyond, but remain focused on accelerating the growth in our U.S. business.”
Strong Government Contracts Drive 49% Revenue Growth for Palantir
While Palantir’s success in the commercial sector is now quite evident, the government sector remains Palantir’s bread and butter. Government revenue accounted for 55% of the total revenue, and the commercial revenue accounted for the remaining 45%. Government outpaced commercial growth for the fourth consecutive quarter; however, if we exclude strategic commercial contracts, both government revenue and commercial revenue grew by 49% in the recent quarter.
In Q2, government revenue grew by 49% YoY and 14% sequentially to $553 million, accelerating four percentage points from the previous quarter. Within that, US government business grew by 53% YoY and 14% sequentially.
Management highlighted during the Q2 earnings call that the US Space Force's Space Systems Command had awarded the company a $218 million delivery order to support seamless, synchronized multi-domain warfighting for the space and air operational communities. Additionally, the ceiling for the Maven Smart System contract was increased by $795 million to prepare for what they expect will be significant demand from combatant commands for the AI-powered software capabilities over the next four years.
In early August, the company secured a 10-year enterprise agreement with the Army, valued at up to $10 billion, which consolidates 75 existing contracts into a single contract. More recently, the company also won the first billion-dollar deal outside the US. It signed a $1 billion AI deal with the UK’s Ministry of Defence.
What Key Metrics are Saying About Palantir’s Future Growth
Clearly, Palantir has been a winner up to this point – there is no denying that. However, what are the key metrics saying about future growth? Do the company’s fundamentals support the market’s lofty expectations—and what does a fair valuation look like from here?
Below, we examine the following
If the key metrics support continued growth – with one metric that you must not miss
The I/O Fund’s thoughts on Palantir’s valuation
The one risk that Palantir bulls are overlooking
Our Premium members received a 15,000 word report outlining the Top 15 AI Stocks for Q3 of 2025 with many stocks seeing higher returns last quarter than Palantir. Sign up now to find out which AI stocks we like better than Palantir.Top 15 AI Stocks for Q3 of 2025 with many stocks seeing higher returns last quarter than Palantir. Sign up now to find out which AI stocks we like better than Palantir.
There were a handful of other key metrics that performed quite strongly in Q2. This suggests that the underlying business momentum remains robust in the coming quarters. These include:
The company booked the highest TCV and ACV ever in the recent quarter with $2.3 billion in TCV and $684 million in ACV. The deal momentum was strong, as it closed 157 deals worth $1 million or more, including 66 deals worth $5 million or more and 42 deals worth $10 million or more.
The total remaining deal value for the company grew by 65% YoY and 20% sequentially to $7.1 billion, a stunning 20 points acceleration from the 45% YoY growth reported in the previous quarter.
RPO grew by 77% YoY and 27% sequentially to $2.4 billion, a whopping 31 percentage points acceleration from the previous quarter.
The net dollar retention rate accelerated 400 basis points sequentially to 128%. It is the highest rate since the first quarter of 2022. It was primarily driven by both expansions of existing customers and new customers acquired in Q2 of last year, implying that the company is a key beneficiary of the AI revolution.
Net Dollar Retention improved 400 basis points sequentially to 128% in Q2 2025, marking the highest level in over two years.
Strong Margin Expansion Supports Profitability
Revenue growth and key metrics were impressive, and management complemented this performance with solid bottom-line improvement, demonstrating disciplined execution.
The company’s gross profits grew by 47.5% YoY to $810.76 million
The operating profits grew by 155.7% YoY to $269.3 million, driven by strong operating leverage. Operating margin improved 11 percentage points YoY and 7 percentage points sequentially to 27%.
The adjusted operating profits grew by 83.1% YoY to $464.4 million. The adjusted operating margin improved by 9 percentage points YoY and 2 percentage points sequentially to 46%.
Palantir also delivered the highest Rule of 40 score of 94% (revenue growth + adjusted operating margin) in Q2. It also accelerated 11 percentage points from the previous quarter and was the eighth consecutive quarter of an expanding Rule of 40 score. Due to the increase in full-year revenue and adjusted operating margin guidance, management now expects a Rule of 40 score of 91% for the year 2025, up from 68% in 2024.
Q2 adjusted operating margin expanded 9 percentage points year-over-year to reach 46%, reflecting continued operating leverage and disciplined cost management.
Adjusted EPS Surges 78% Year-over-Year
The company’s adjusted EPS grew by a solid 77.8% YoY to $0.16, beating estimates by 15.6% driven by strong operating leverage.
Analysts expect adjusted EPS to grow 67.4% YoY to $0.17 in Q3 and 35.5% YoY growth to $0.19 in Q4.
Looking forward, analysts expect strong adjusted EPS growth of 56.7% YoY to $0.64 in 2025 and 31.8% YoY to $0.85 in 2026.
Adjusted EPS surged 78% YoY in Q2 2025, beating estimates by 16%, driven by strong operating leverage. Analysts project continued strength with 67% YoY growth in Q3 and 36% in Q4, underscoring robust earnings momentum through 2026.
Adjusted Free Cash Flow Margin Expands to 57%
The company’s strong revenue growth and profits are driving robust cash flows.
Q2 operating cash flows grew by 274.3% YoY to $539 million. Operating cash flow margin improved by 33 percentage points YoY and 19 percentage points sequentially to 54%.
Adjusted free cash flows grew by 282% YoY to $569 million. Adjusted free cash flow margin improved by 35 percentage points YoY and 15 percentage points sequentially to 57%.
The company has a strong balance sheet with cash & marketable securities of $6.0 billion, up from $5.43 billion in the previous quarter.
Valuation
To many investors on social media, Palantir’s valuation remains a hot topic, with it blowing past norms and reaching the upper echelons of the stratosphere for what is considered ‘typical’ for SaaS stocks. Put it this way — how often do you see a software company re-accelerate revenue from 13% growth a couple of years to 50% in nine quarters organically and sustainably, while both increasing profitability and posting cash flow margins more than >50%?
Palantir is now trading at about 105x forward P/S ratio, making other best-of-breed cloud names like Cloudflare and CrowdStrike look cheap at 35.5 and 25.2, respectively. On the bottom-line, it is trading at forward P/E ratio of 287, close to Cloudflare’s forward P/E ratio of 251.
Given Palantir is at 105 forward PS, there are certainly easier stocks to own going into the second half of the year. Within AI, specific niches are booming such as AI networking, AI energy and AI data plays that are trading far lower than Palantir yet will see an incremental lift from the incoming inference market.
Potential Risks and Challenges
The valuation with Palantir is a gamble. The bulls believe they’ve speculated correctly, while there’s likely to be short sellers who do well with this stock too. Palantir is attempting to set a new bar for AI software with the 105 forward valuations, it’s truly anyone’s guess if the stock can sustain the valuation or not.
Despite US Commercial serving as a crucial lever for Palantir’s tremendous eight-quarter acceleration on the top-line, government revenue cannot be overlooked as it remains Palantir’s bread & butter. Government still accounts for 55% of Palantir’s revenue, reinforcing the importance of the segment. Regarding the government, Palantir found itself in the hot seat on Oct 03rd, with shares down 9% as reports surfaced claiming that the Army stated in a memo that Palantir’s battlefield tech developed with Anduril posed a major security risk. Both companies have later clarified that the issues have been fixed.
Conclusion
Part of our process is to highlight stellar earnings reports and Palantir certainly qualifies. It’s hard to find a blemish in the company’s current quarter as it’s perhaps one of the best reports the company has reported yet – which is saying a lot. However, at the current valuation, it would take Palantir 100 years to pay back its market cap. Therefore, we feel there are easier AI stocks to own as Palantir’s current valuation does not leave much room compared to the dozen or so others also benefiting in AI’s enormous tailwinds.
Our Premium members received a 15,000 word report outlining the Top 15 AI Stocks for Q3 of 2025 with many stocks seeing higher returns last quarter than Palantir. Our Premium members will be receiving the Top 15 AI Stocks for Q4 next week. Sign up now to find out which AI stocks we like better than Palantir.Top 15 AI Stocks for Q3 of 2025 with many stocks seeing higher returns last quarter than Palantir. Our Premium members will be receiving the Top 15 AI Stocks for Q4 next week. Sign up now to find out which AI stocks we like better than Palantir.
Please note: The I/O Fund conducts research and draws conclusions for the Fund’s positions. We then share that information with our readers. This is not a guarantee of a stock’s performance. Please consult your personal financial advisor before buying any stock in the companies mentioned in this analysis.
GE Vernova is part of the spinoff that General Electric first announced in 2021 and later completed in 2024. The company broke up its three biggest segments into separate units: GE Healthcare was spun off first in 2023, GE Vernova for the Energy business was spun off in 2024 and began trading as GEV, and GE Aerospace was the business that remained with the existing stock ticker GE.
As you can imagine, coming from General Electric, GE Vernova is a massive company that can easily slip under the market’s radar given it does not have much history on the public markets. At the time the Energy segment was split up, it was seeing $33 billion in revenue and was helping to generate 30% of the world’s electricity with 55,000 wind turbines and 7,000 gas turbines.
This year, the company is expected to report $37 billion in revenue with strong earnings growth of 45.3%. The stock is not a hypergrowth profile, rather, is a quality, defensible position that could do well during any periods of doubt in the broader AI trend.
The defensibility is particularly attractive when you consider that gas turbines is the crux of the issue for expanding gas power plants. According to Bloomberg’s calculations, more than “$400 billion worth of gas-fired power plants through the end of the decade are in jeopardy of delay or cancellation because of the lack of capacity to meet future turbine orders.” The same article points toward GE Vernova filling orders as far out as 2030.
In the chart outlined below, GE Vernova is the world’s largest gas turbine supplier at 25% ahead of Schneider at 24%. Even still, GEV nearly tripled its gas turbine equipment this past quarter – a statement that has us sitting up in our seats. Per the earnings call: “Power orders grew 44%, led by Gas Power equipment nearly tripling year-over-year.”
Also, consider that we have been covering Bitcoin miners and other energy sources that can quickly help hyperscalers secure powered shells in the 1GW to 3GW range – yet GEV has 50 GW in backlog for gas equipment contracts with expectations the backlog will reach 60 GW by the end of this year. In other words, the chances that GEV is not a significant player in supplying energy to data centers for many years to come is nil.
In a bid to supply options quickly to alleviate bottlenecks, GEV is also shipping aeroderivative gas turbine packages and doing extensive R&D on a small modular reactor (SMR) design. As detailed below, how exactly GEV evolves to solve the crucial bottleneck around AI power consumption is not set in stone, rather the company is experimenting rapidly with how to leverage their deep experience in natural gas, electrification and renewables like wind to meet global demand.
Natural Gas Capacity to Reach 60GW Backlog by YE
The question of “how fast, and how much” sums up what any given management team in the energy space will be pressed on in the coming quarters.
In Q2, GEV signed 9GW of new gas equipment contracts with 2GW going directly to orders and 7GW going into what’s called a slot reservation. During the quarter, the company also converted 3GW into orders and shipped 5GW of equipment.
Management stated they would exit the year with 60GW “at better margins with significant momentum into ‘26.” Here is the breakdown from that comment:
29GW are in the backlog, same as last quarter
Slot Reservation Agreements (SRA) grew from 21GW to 25GW
Total backlog including SRAs is 55GW, up from 50GW guided in April
Given this, the company expects to add another 5GW to its backlog to reach 60GW by end of the year.
There was a discussion on the call that this represents 3 years of backlog: “And we've talked about the fact that we'll get to at least 60 gigawatts by the end of the year. So that's directionally 3 years of backlog.”
There was mention of eventually seeing 80GW to 100GW in the backlog but no date or other details were discussed, other than that’s the goal over time.
Across both Power and Electrification, the backlog is at $129 billion across equipment and services. Wind is lagging for now compared to Power first and foremost, and Electrification is also reporting strong growth (more below). Management stated, “We now maintain a total backlog of $129 billion.”
When it comes to the equipment, GEV stated they booked 20 heavy-duty gas turbines up by 6 units compared to the year ago quarter. Of the 20 units, seven were the HA models which produce up to 571MW.
In the earnings call, management also clarified that the second half of the year will be stronger as there will be a higher value tied to the booked GWs: The combined cycles have gas turbines, heat recovery steam generators (HRSG) and steam turbines for a more complex, capital-intensive equation which results in the higher dollar value per MW.
“So you're going to see an orders dollar connection to gigawatts that will be larger in the second half of the year. The first half of the year has been more simple cycle orders. So that's where you've got to think about the gigawatt dollar connections and the mix between whether it's a simple cycle or a combined cycle deal and we'll have substantially more combined cycle orders in the second half of the year.”
As stated, GEV is not a hypergrowth stock, rather its bottom line is where the impact is most visible. Management hinted they would be discussing strong margins resulting from the backlog in their fiscal year call in January:
“We are growing this backlog at improved margins and consistent with prior communications, look forward to showing you at fourth quarter earnings next January, the full change in margin in the equipment backlog.”
The company also raised free cash flow by $1 billion due to the strength of the backlog: “In addition, we're raising our full year free cash flow guidance by approximately $1 billion to be in the range of $3 billion to $3.5 billion due to higher down payments from increased orders and our updated adjusted EBITDA outlook.”
The company also gets paid in its Services segment based on auction pricing from PJM, a United States transmission organization. Management pointed out they will benefit from current pricing: “When you even look at the PJM pricing that was confirmed yesterday with the capacity market, that's driving incremental demand for incremental services and frankly, justifies incremental pricing into our services book for upgrades that can create incremental capacity for things like those capacity markets.”
Small Modular Reactors and Aeroderivative Gas Turbines
GEV is working on releasing a 300MW small modular reactor (SMR) and is under construction in Ontario for the first project. The company stated they expect “more customer announcements with our SMR technology in the second half of the year.”
The first major announcement on aeroderivative gas turbine packages came in June from purchaser Crusoe, a Bitcoin miner that retrofitted its operations to become an AI data center with experience in utilizing what’s known as “stranded natural gas” that is burned off from oil fields.
According to the press release, ten aero units can produce 1GW of power. In the earnings call, management stated they have secured orders for 27 aero units compared to 1 unit last year. There were discussions that the aero units are particularly in high demand: “Well, there's a need for incremental bridge power and the beauty of aeroderivatives is, they can be commissioned faster and that's needed in the environment today and our customers are able to price at a premium for expedited power […] But in the near term, demand for aeroderivatives is very strong and that's in the U.S. but it's also in global markets”
Under the Electrification department, GEV also supplies synchronous condensers, which provide voltage support and frequency regulation during periods when the grid is unreliable. The company sees the market opportunity growing to $5 billion in a year’s time.
Financials
Revenue acceleration in 2026
GE Vernova is on a path to revenue acceleration in 2026 as its benefiting from the spending surge of hyperscalers. The company is a major beneficiary of the increasing energy requirements from the global AI infrastructure build-out, positioning the company as a key beneficiary of this secular trend. Notably, 25% of global electricity was generated using the company’s equipment.
The company’s Q2 revenue grew by 11% to $9.11 billion, beating estimates by 3.6%. Organically, revenue grew by 12% YoY to $9.04 billion, primarily due to higher equipment and services revenue. Analysts expect revenue to grow by 2.7% YoY in the next two quarters and revenue growth to accelerate to 8.8% in Q1 2026.
The company’s CEO, Scott Strazik, said in the Q2 earnings call, “We had a productive second quarter, positioning us well to continue to accelerate our growth and margin expansion. This era of accelerated electrification is driving unprecedented investments in reliable power, grid infrastructure and decarbonization solutions.”
On the back of strong demand for power and equipment, management has raised its full-year revenue guidance and expects 2025 organic revenue to come at the high end of the guidance of $36 billion to $37 billion. The power segment organic revenue guide is increased to 6-7%, up from the previous single-digit guide, and the electrification segment is expected to grow 20%, up from the previous mid-high teens percentage. On a side note, the wind segment is expected to be down mid-single digits due to the more challenging market conditions.
Revenue growth is set to accelerate over the next three years. Analysts expect a 6.4% increase in 2025, bringing the total revenue to $37.17 billion, a 5.1% growth in 2024. Momentum is projected to build further, with revenue climbing to $40.7 billion in 2026, up 9.5% and to $45.5 billion in 2027, up 11.9% YoY.
Adjusted EBITDA growth of 25%
Q2 adjusted EBITDA grew by 25% YoY to $770 million, driven by strong growth in the electrification and power segments. Adjusted EBITDA margin improved 210 basis points YoY to 8.5% driven by profitable volume, better pricing, and productivity gains. The company is witnessing an annual EBITDA margin expansion, increasing from 2.4% in 2023 to 5.8% in 2024, and management has further guided expansion in the range of 8-9% for 2025.
Q2 operating margin was 4.2%, an improvement from 0.5% in Q1, and was down from 6.4% in the same period last year due to the nonrecurrence of $300 million received related to an arbitration refund last year.
Strong EPS Growth, Firmly GAAP Profitable
Q2 GAAP EPS came at $1.86, beating estimates by a solid 14.3% driven by profitable volume, better pricing, and productivity gains. Analysts expect GAAP EPS of $1.85 for Q3 2025 compared to (-$0.35) in the same period last year. They expect GAAP EPS to grow 95.4% YoY to $3.38 in Q4 and 137.4% YoY to $2.16 in Q1 2026.
Analysts continue to expect strong EPS growth in the coming years. For the year 2025, analysts expect GAAP EPS to grow 45.3% YoY to $8.11, and 58.9% and 41% YoY in the subsequent two years, reaching $18.16 in 2027.
Cash Flow Guidance Raised by $1 Billion
The company’s free cash flows are expected to increase significantly in the year 2025. The strong cash flows are driven by growth in revenue, profits, and improvements in working capital management.
Most importantly, management has raised the full-year free cash flow guidance from a range of $2.0 billion to $2.5 billion to a new range of $3.0 billion to $3.5 billion. It was primarily driven by a higher profit outlook and increased down payments due to rising orders. Year-to-date, the company has generated $1.17 billion in free cash flow, implying a strong acceleration of free cash flow in the second half of the year to $2.1 billion.
The company generated free cash flows of $194 million in Q2 compared to $821 million in the same quarter last year. The lower free cash flows in Q2 were due to tough comparables, as the company had benefited from a $300 million arbitration refund the previous year.
The company had a working capital benefit of $600 million in the recent quarter, driven by strong down payments and slot reservation agreements in the Power segment. They were partly offset by cash taxes and capital investments to support the strong future growth.
The management efforts to improve the billing and collections processes, thereby enhancing cash management, are proving successful. In the recent quarter, the day sales outstanding decreased by 2 days sequentially, resulting in an additional $200 million in free cash flow. Similarly, the company was able to increase its free cash flow by $150 million in Q1 due to the implementation of stronger daily cash management, which improved the timely payment of invoices.
The company has a strong balance sheet. At the end of Q2, the company had cash of $7.9 billion and no debt. The financial strength was further validated during the first half of this year when both S&P and Fitch upgraded GEV’s credit ratings outlook to positive from stable and reaffirmed the company’s investment-grade credit rating.
Demonstrating its commitment to the capital return strategy, the board of directors had authorized a share-repurchase plan of $6.0 billion in December 2024. Year-to-date, the company repurchased shares worth $1.6 billion, leaving $4.4 billion in remaining capacity for future buybacks.
Key Segments
Equipment and services drive power segment growth
The power segment revenue grew organically by 9% YoY to $4.76 billion in Q2. It was driven by power equipment revenue, which grew by 23% YoY and the increase in power services revenue, which benefited from higher transactional services volume and favorable prices. Management expects mid-single digit organic revenue growth in Q3, driven by higher equipment deliveries and continued services growth.
Power segment witnessed robust orders, which grew by 44% YoY to $7.1 billion, driven by strong demand for Gas Power equipment, which nearly tripled YoY. Management expects continued growth in gas equipment orders in Q3. The company is witnessing robust demand for its aeroderivative technology to support data centers, securing 27 aeroderivative units in the recent quarter compared to only one in the same period last year.
Power services orders increased by a mid-single-digit percentage during the quarter, led by strong performance in Steam Power. The growth was primarily driven by higher demand for life extension and upgradation at existing nuclear facilities. The hydro end market also witnessed an uptick in orders driven by higher demand for upgrades.
EBITDA margin improved 260 basis points YoY to 16.4% driven by increased price, productivity and higher volume, partially offset by additional expenses to support R&D, higher capacity, and inflation. Due to seasonality of services outages, both power revenue and EBITDA margin is expected to be lower sequentially in Q3.
Electrification segment grew 20%
Q2 electrification revenue grew by 20% organically YoY to $2.2 billion, primarily driven by strong volume and higher prices at Grid Solutions. Management expects Q3 revenue to grow 20% YoY, driven by growth in Grid Solutions and Power Conversion & Storage. Total orders were down (-31%) YoY to $3.3 billion due to the tough comparable and due to large equipment orders recorded in the second quarter of last year. Equipment orders continued to outpace revenue, expanding the equipment backlog to approximately $24 billion, an increase of over $6 billion YoY. Data center related demand also remained strong in this segment. The company received $500 million in orders in the first half of 2025 compared to $600 million for the full year 2024.
Most importantly, EBITDA margin was up 740 basis points YoY to 14.6%, primarily driven by profitable volume, increased productivity, and favorable pricing, primarily at Grid Solutions. Management expects significant YoY EBITDA margin expansion in Q3 with a margin rate slightly above Q2, mainly driven by higher volume, productivity gains, and favorable prices.
Wind segment EBITDA to approach breakeven in Q3
Q2 wind revenue grew by 9% organically to $2.25 billion due to higher onshore wind equipment volume in North America and partially offset by lower offshore wind revenue. Wind orders were down (-5%) YoY, driven by lower onshore wind equipment orders outside of North America. Sequentially, onshore orders improved led by equipment growth in North America.
EBITDA losses increased to (-$165 million) compared to (-$117 million) in the same period last year due to higher onshore wind services costs as the company deploys crew and cranes to improve the installed fleet performance, partially offset by more profitable onshore wind equipment volume. Offshore, the company incurred additional costs primarily due to the impact of tariffs.
Management expects the wind segment revenue to be down mid-teens percentage YoY in Q3. The company benefited from a one-time settlement from an offshore contract termination in Q3 last year; excluding this one-time adjustment implies wind revenue to be down in the low single digits. Most importantly, EBITDA losses are expected to improve substantially YoY and approach breakeven in Q3 driven by further improvement in the profitable onshore wind volume.
Key Metrics
Strong demand for gas power equipment driving orders
The company is the largest gas turbine production supplier in the world, with a 25% global market share. The top three suppliers account for 71% of the global production capacity. Due to the sudden surge in AI-related electricity demand, the orders for turbines are vastly outpacing demand. This raises concerns about whether current production capacity will be sufficient to meet the anticipated growth in global electricity needs.
Q2 orders grew by 4% YoY to $12.4 billion, primarily driven by continued strong demand in the power and electrification segments. Equipment orders grew by 5%, primarily driven by the Power segment, where orders doubled YoY. Electricity equipment orders, even though they were strong, were down YoY due to tough comparable. Service orders increased by 3%, driven by growth in Power and Onshore Wind.
Robust backlog to support long-term growth potential
Q2 backlog increased by 11.5% YoY to $128.7 billion, driven by strong orders. The growth has accelerated from 6.1% in the previous quarter. The equipment backlog grew from $45 billion in Q1 to $50 billion in Q2 and was up 16.3% YoY. The company’s CEO had earlier this year said the equipment capacity is effectively sold out through 2027, implying potential long-term revenue conversion.
Conclusion:
Looking ahead, energy stocks like GE Vernova could deliver attractive returns with relatively lower risk over the next several years as demand for power of all kinds – but especially natural gas – accelerates. We’ve outlined predictions around what kind of demand capex investors can expect in the coming year in more detail here: “Why Power is Critical for Data Centers and Their Hyperscaler Customers.”
GE Vernova is one of the largest companies powering the AI economy with 60GW in backlog. The company stated the backlog is likely to grow quite a bit in 2026, which forecasts “the incremental growth we expect in this company in 2029 and 2030.” How big will this backlog get, what percentage of capex will Big Tech spend on energy versus compute, and what will the government do to make sure companies like GEV can delivery quickly. I would imagine all of the above far exceeds current expectations and becomes a matter of national importance if the United States is going to beat out energy-plentiful China. Therefore, for these reasons, the chances that GEV is a quality, resilient stock over the next few years is higher than most stocks in the AI universe.
Equity Analyst, Royston Roche, contributed to this article.
Please note: The I/O Fund conducts research and draws conclusions for the Fund’s positions. We then share that information with our readers. This is not a guarantee of a stock’s performance. Please consult your personal financial advisor before buying any stock in the companies mentioned in this analysis.
CoreWeave saw muted price action following the latest earnings report; yet the soft price action is rare for the AI darling. The company went public in March and has stood out as the premier IPO among AI stocks given the stock is up over 200%. The lockup expired last month, which begs the question – can CoreWeave continue to defy the odds and overcome the typical insider selling that is seen around a lockup expiration? Investors typically fare better waiting for anxious insiders to sell, yet CoreWeave has been anything but ordinary.
In terms of timing, CoreWeave has hinted the second half of the year will be stronger. We look at why CoreWeave could end the year on a high note, yet to be prudent, we also look at why there was a negative reaction to the most recent earnings report. We end with a buy plan strategy the I/O Fund is eyeing for weighing the puts and takes on what promises to be a highly volatile, fast moving stock.
CoreWeave Rivals the Big 3 as the First “AI Hyperscaler”
CoreWeave brands itself as the world’s first “AI hyperscaler” as they offer both infrastructure and a software platform for developing large language models and deploying them. Being dubbed an AI infrastructure player means CoreWeave must offer a compelling value proposition to attract business from arguably the largest competitors in the world – AWS, Microsoft Azure and Google Cloud.
In the S-1 filing, the company points out it was built for AI workloads as opposed to the legacy cloud infrastructure-as-a-service providers that were primarily optimized for the cloud software era and e-commerce era. CoreWeave also asserts that outdated cloud infrastructure leads to lower utilization rates when you factor in usage.
One of their primary value propositions is offering bare metal servers, as the company does not need to offer shared GPUs like the hyperscalers. By stripping away the virtualization layer, raw performance goes up for R&D labs, who do not need workload flexibility. Although CoreWeave offers shared infrastructure in terms of storage and networking, one of the company’s key differentiations from the Big 3 is by offering dedicated bare-metal access.
The company also offers proprietary software to help achieve higher total system performance and more favorable uptime relative to competitors. According to the S-1 filing, “by delivering more compute cycles to AI workloads and thereby reducing the time required to train models, our capabilities can significantly accelerate the time to solution for customers […].”
CoreWeave Competes with Big 3 on Higher Usage Utilization Rates (MFUs)
To further understand CoreWeave’s competitive advantage, it’s important to discuss the model FLOPs utilization gap. The “MFU gap” is a metric that describes the gap between compute capacity and usage, which today often ranges between 30% and 40%. Cloud providers are often at 100% GPU utilization, yet there is a much lower utilization rate for GPUs when factoring in maximum floating-point operations per second (FLOPs). Initially, when MFU was coined by Google’s PaLM Paper, model training was running at 20% MFUs.
According to Google’s PaLM paper, they came up with the metric to better gauge a more realistic utilization rate: “Given these problems, we recognize that HFU (hardware FLOPs utilization) is not a consistent and meaningful metric for LLM training efficiency. We propose a new metric for efficiency that is implementation-independent and permits a cleaner comparison of system efficiency, called model FLOPs utilization (MFU).”
When factoring in FLOPs, the best possible (realistic) MFU is in the range of 50% to 60%, as this translates to raw compute being the bottleneck. Lower MFUs indicate inefficiencies, which CoreWeave specializes in solving. This could involve optimizing memory bandwidth, improving communication between GPUs, clearing data input bottlenecks, and other ways in which to fix batch size, enable faster data loading, and/or better ways to balance the compute.
Popular large language models do not publicly report their MFUs, but internally, this utilization rate is a dominant factor in competitiveness and time to market. R&D labs with a higher MFU rate have an important advantage as even an incremental increase in single digits to low double digits can result in a 25% to 50% increase in training speed and cost.
Due to going public, CoreWeave has published its MFU rate of 35% to 45%, stating it is 20% higher than competitors, which means other AI data centers have MFU rates more in the 30% range. Due to FLOPs performing an astronomical number of calculations, small percentages translate to an important advantage.
To put it simply, efficiency equals money and time in large-scale AI projects — training huge models can cost millions of dollars and weeks of time, so even a few percentage points of MFU improvement can translate to a significant advantage.
CoreWeave is Adding Capacity Hand over Fist Supported by OpenAI, Meta Agreements Totaling $36.6B
Before looking too granular at the financials, it’s important to note that CoreWeave is adding new capacity at a rapid clip. At a high level, the company is the most competitive outside of the Big 3 for new AI workloads, evidenced by the strong partnerships it’s securing with Meta and R&D labs like Open AI.
Last week, CoreWeave announced an expansion of its agreement with OpenAI, worth an additional $6.5 billion. The extension now takes CoreWeave’s total deal value with OpenAI up to $22.4 billion, building on to its initial $11.4 billion deal in March and the first $4 billion expansion in May. The $6.5 billion deal extends through May 2031, representing average annual revenue of more than $1.2 billion.
On Tuesday, CoreWeave signed a $14.2 billion deal with Meta, also lasting through 2031, marking one of its largest single deals to date. The deal represents average annual revenue of more than $2.3 billion.
This provides CoreWeave with other major revenue anchors and additional diversification away from its largest customer Microsoft (72% of revenue in 1H), as the deals could represent nearly 30% of CoreWeave’s current 2026 revenue estimate of $12.1 billion.
When discussing its customer concentration, CoreWeave made it crystal clear that there is no better customer to have at the moment: “And that when you have a company like OpenAI or an entity like OpenAI consuming compute, they're just doing it at an order of magnitude that these other companies have not achieved yet.”
Looking beyond Open AI, CoreWeave has a contracted backlog of $30.1 billion, up $4 billion from Q1 and has doubled year-to-date, with expectations for 50% (or ~$15 billion) to convert to revenue over the next 24 months. The most recent deals with OpenAI and Meta bring this to over $50 billion in contracted backlog.
There are many pieces that must come together to deliver this backlog in the coming years including raising capital to build the infrastructure. However, keep in mind the company has a market cap of about $60 billion or roughly 1.3X the contracted backlog. Although I’m not suggesting a valuation be based off backlog, it’s certainly convincing the stock has room to run given the sheer size of the contracts it’s securing from the largest players in AI.
In early September, CoreWeave announced that key partner and investor Nvidia had entered a new order worth up to $6.3 billion under the duo’s pre-existing 2023 master services agreement.
With the new order, whenever CoreWeave’s compute capacity is not fully utilized, Nvidia will be obligated to purchase the unsold capacity. The deal extends through April 2032 and carries a total value up to $6.3 billion. While demand is currently strong and capacity is likely sold out for the near future, having Nvidia backstop future capacity helps alleviate concerns related to high customer concentration and de-risks its future growth story by providing some degree of guaranteed revenue. CoreWeave will disclose the entire MSA in its upcoming quarterly report.
Expanding Footprint with $6B Pennsylvania Data Center, UK Investment
In mid-July, CoreWeave announced a $6 billion data center project in Pennsylvania, with an initial capacity of 100MW with potential to expand to 300MW, though the company has not been upfront about delivery timelines for the initial phase or subsequent phases. This project is expected to be the cornerstone of CoreWeave’s vision of creating a mid-Atlantic hub from New York to Virginia.
Under its first phase, it likely will become one of CoreWeave’s largest data center facilities, considering its current active power footprint averages just over 14MW per data center. At full capacity, the facility will represent nearly 14% of CoreWeave’s current contracted power. On the capex side, CoreWeave has secured $4 billion in funding for the project, including a $200 million investment into grid infrastructure to support the facility. This will limit the amount the company will have to fund out of pocket, lessening its capex needs, which will already be quite elevated come Q4.
CoreWeave is also expanding its presence in the UK with a new ~$2.0 billion investment, working with Nvidia and data center operator DataVita in Scotland to deploy Blackwell Ultra GPUs. CoreWeave’s international presence is rather limited, with just five data centers across Europe and two in the UK, and investing to expand its international presence will help the company tap into growing demand across Europe.
Capex Weighted Heavily Towards Q4
One of the larger risks for investors coming into year-end is CoreWeave’s planned capex, with the company currently expecting the majority to land in Q4 due to timing of when infrastructure will go live. CoreWeave maintained its full-year capex guide of $20 to $23 billion, though in the first half, reported capex was only $4.8 billion (total PP&E increase of $5.7 billion minus $0.9 billion related to construction progress). To put this in perspective, this is 4-5x of the company’s guided revenue for the year.
Given that CoreWeave guided for just $2.9 to $3.4 billion in capex in Q3, or YTD spend of just $7.7 to $8.2 billion, Q4 capex is implied to be in the range of $12 to $15 billion, or nearly 7-9x estimated quarterly revenue of $1.8 billion.
The first caveat here is that CoreWeave does not have nearly enough cash to fund this capex entirely by itself. CoreWeave likely will have close to $5 billion in cash and equivalents after raising $2 billion in 9.25% senior notes and upsizing another 9.0% note raise to $1.75 billion, with additional access to a third delayed draw term loan facility for $2.6 billion to fund capex and GPU acquisitions for customer contractions. Combined, this still will not fully cover projected capex for Q3 and Q4.
The second caveat is that turning to the debt market will be costly, as CoreWeave has currently been pricing senior notes at or above a 9% rate, meaning raising $10 billion in fresh debt (such as what analysts from DA Davidson expect) at a similar rate could cost nearly $1 billion annually in interest payments.
CoreWeave Only Has 20% of Contracted Power Active
The reason why CoreWeave must spend aggressively on capex is directly tied to capacity, which is needed to convert its backlog to revenue and maintain hypergrowth. As of Q2, CoreWeave had ~470MW of active power across 33 data centers, or nearly 20% of its 2.2GW of contracted power, leaving an additional ~1.73GW to be developed. At its current scale, CoreWeave has energized >250,000 GPUs, suggesting that at 2.2GW, it can easily energize well over one million leading-edge GPUs.
Bringing the rest of its power footprint online will not only be expensive but necessary to support its growth story. Management stated they are aiming to nearly double active power by year-end to 900MW, hence the substantial increase in capex in 2H:
“We are aggressively expanding our footprint on the back of intensifying demand signals from our customers, ensuring that we maintain a durable multiyear runway for growth. We are now on track to deliver over 900 megawatts of active power before the end of the year.”
CoreWeave operates 33 data centers, primarily in the U.S. with five in Europe, supported by ~470MW active power and ~2.2GW contracted power. Source: CoreWeaveCoreWeave
Core Scientific’s acquisition has provided an extra leg for growth in power: “Upon closing, CoreWeave would own approximately 1.3 gigawatts of gross power capacity across Core Scientific's national data center footprint with an incremental 1 gigawatt or more available for future expansion. This scale enhances our flexibility to take on new projects and meet accelerated customer demand.”
Looking ahead, and assuming power and revenue remain correlated linearly, CoreWeave’s current contracted power footprint may be able to sustain a $25 billion revenue run rate, or enough capacity to take it to 2029’s current estimate.
Unleashing CoreWeave’s Monetization Path and Stock Buy Plan
CoreWeave’s suite of software is another area the company extends its value proposition beyond the Big 3 as the company reduces the need for specialized orchestration frameworks, engineering resources component failures and the need to constantly monitor for downtime.
Sign up below to continue reading:
How CoreWeave plans to monetize every chatbot response, API call and application
The I/O Fund’s buy plan and risk management strategy to help participate in this stock while protecting to the downside
Details from the recent earnings report that help clarify why the stock sold off
Important commentary from management that hints the end of the year will be strong for CoreWeave
According to Uvation, by focusing only on GPUs and software optimizations (detailed below), CoreWeave offers bare metal servers at a cost that is up to 20% to 50% cheaper than hyperscalers. This helps to explain why the company’s rapid and aggressive growth.
The company is able to scale quickly with new GPUs due to the Mission Control automation layer that provides automated deployments of systems like the GB300 NVL72s. The company stated: “Mission Control continues to be the cornerstone of CoreWeave's ability to scale at breakneck speed, building a fully automated and rigorous process for cluster life cycle management with unmatched visibility for our customers.”
CoreWeave also offers a Virtual Private Cloud for a private network space. By combining an isolated virtual private cloud with Nvidia’s Quantum InfiniBand, customers get ultra-low latency with enhanced security. Customers can connect workloads to other clouds like AWS, Azure and Google Cloud. In the most recent earnings call, CoreWeave stated: “We also saw significant growth in our backbone and networking service as one of our largest AI lab customers leveraged our networking backbone to connect its multi-cloud inference infrastructure.”
The company's Kubernetes Service is an AI-optimized Kubernetes environment for scheduling AI workloads and scaling up/down for the right mix of CPU, GPU, memory and storage (known as elasticity). SUNK, known for Slurm on Kubernetes, combines container orchestration with a job scheduler to manage large batch jobs. AI labs use this service to combine scheduling for high performance computing with a cloud-native environment.
Local access object transport accelerator (LOTA) for AI object storage is another feature that is optimized for AI workloads by focusing on performance and cost efficiency. The company recently added archive tier object storage, which allows data to move between hot and cold storage based on access patterns, which optimizes costs. In the recent earnings call, the company stated they are seeing customers “shifting petabytes fo their core storage to CoreWeave in the form of multiyear contracts.”
CoreWeave recently completed its acquisition of Weights&Biases in May, with reports placing the transaction at $1.7 billion, a 36% increase from the startup’s $1.25 billion valuation in 2025. The core product that W&B offers is observability, which means engineers can quickly diagnose a failure or inefficiency in the software layer and infrastructure layer. For example, if a model is training slowly, the observability platform will help an AI engineer identify and resolve this quickly.
More recently, CoreWeave integrated W&B for a joint launch of its Inference-as-a-service feature, which allows developers to use APIs to tap into AI models from OpenAI, Meta, DeepSeek, etcetera. Inference is key for CoreWeave to fully monetize its investments in capex-heavy infrastructure. For example, these popular LLMs combined with chain of reasoning inference, which means generating step-by-step reasoning, will become compute-intensive especially at scale. This will lead to CoreWeave monetizing every chatbot responses, API calls and applications to easily payback their initial investments plus some (in time).
Here is what was stated on the call: “In addition to that, the infrastructure that we're building has increasingly been used for chain of reasoning, which is driving a substantial amount of consumption on the inference level. And that's very exciting for us. As I always say, inference is the monetization of artificial intelligence. And we are extremely excited to see that use case expanding within our infrastructure.”
CoreWeave also acquired agentic AI training startup OpenPipe in early September for an undisclosed sum. OpenPipe aids enterprises in customizing AI agents via reinforcement learning with its open-source toolkit ART (agent reinforcement trainer), tying into W&B’s observability and evaluation frameworks for agents that can be built directly on CoreWeave’s infrastructure platform.
Financials
Strong Revenue Growth of 207%
CoreWeave reached a new milestone of over $1.0 billion in revenue in Q2 2025, growing 206.7% YoY to $1.21 billion. On a sequential basis, the Q2 revenue grew by 23.6%. The company beat analyst consensus estimates by 12.2%, driven by strong demand for the company’s AI cloud infrastructure services.
Revenue growth is expected to be strong in the coming quarters, driven by the robust demand due to training and inference workloads. Management revenue guidance for Q3 is in the range of $1.26 billion to $1.30 billion, representing YoY growth of 119.2% and 5.5% QoQ at the midpoint. While the underlying business momentum remains robust,
CoreWeave’s revenue growth is decelerating due to tough comparables. For example, the company put up sky-high growth of 420% in Q1 of 2025, thus making it challenging to sustain a higher growth rate a year later.
Revenue growth is expected to show a 20% acceleration QoQ in Q4 with revenue growing 139.5% YoY and a further 16 percent acceleration QoQ in Q1 2026, highlighting the strong deals signed in the recent quarters.
CoreWeave revenue growth is projected at 119% in Q3 2025, re-accelerating to 156% by Q1 2026.
Looking forward, revenue is expected to grow by 174% YoY to $5.26 billion in the year 2025 and 129.6% YoY to $12.08 billion in 2026 and 48.3% growth in 2027. Most importantly, management has increased the full-year revenue guidance for the second quarter in a row due to the strong customer demand. Management increased guidance by $250 million at the midpoint to a new range of $5.15 billion to $5.35 billion for the year 2025.
Robust Backlog
The company’s backlog was $30.1 billion at the end of Q2, up 86% YoY driven by the company’s strategic deal with OpenAI in March 2025 and the signing of subsequent expansion deals with the company. The company is signing new contracts with enterprise customers and AI startups along with expansion with its hyperscaler customers. More recently, the company also expanded its contract with OpenAI by $6.5 billion which brings the total contract value with the company to $22.4 billion. The company had signed an initial contract with OpenAI in March 2025 for $11.9 billion and an expanded contract in May for $4 billion.
The company’s CEO and co-founder, Michael N. Intrator, highlighted in the Q2 earnings call that the company is signing deals with a diverse customer base, of particular interest is sovereign customers: “We have a tremendous number of sovereigns that are beginning and discussing and talking through how to go about doing this, what technology to use, what software stack to use, where it should be placed right up and down the line. And we are very confident that we will continue to expand our footprint within the sovereign cloud universe.”
The key takeaways are that the company has strong future growth and it is also diversifying its customer base, as this helps to allay investor fears regarding the high customer concentration. The company derived 77% of 2024 revenue from its two largest customers, i.e., Microsoft and Nvidia. While in the recent quarter, Microsoft accounted for 71% of the total revenue. Goldman Sachs estimates that Microsoft’s share is expected to drop to 38% in 2026, followed by OpenAI at 21%, Nvidia at 6%, and the remaining 35% to be attributed to other customers.
Margins
The company is investing heavily in data center and server infrastructure to meet very strong AI demand from its customers. The management tried to explain in the Q2 earnings call that expenses are front-loaded and have a short-term impact on the margins. However, the Street sold the report as the top line raise did not flow through to the bottom line, causing a post-earnings sell-off.
Operating leverage will help the company improve margins in the coming years. Furthermore, the company is also expected to have $500 million of annual run rate cost savings by the end of 2027 once the Core Scientific acquisition is completed.
GAAP profitable in 2027
The company reported GAAP loss per share of (-$0.60) in Q2 compared to the analyst consensus estimate of (-$0.49), missing estimates by –21.7% due to the higher operating expenses, particularly the technology and infrastructure expenses.
Analysts expect GAAP loss per share of (-$2.67) for this year, followed by (-$0.90) for 2026. They expect a positive GAAP EPS of $1.59 in 2027.
Analysts expect CoreWeave to turn GAAP profitable by 2027 with estimated earnings of $1.59 per share.
Cash, Debt & Capex
CoreWeave’s business model is based on aggressive capacity expansion, currently fueled primarily by debt. As a result, cash is rather thin and gets spent quickly, and free cash flow is widely negative.
CoreWeave reported $1.15 billion in cash and equivalents (excluding $0.56 billion in restricted cash and equivalents), though CoreWeave updated in an 8-K related to its now upsized $1.75 billion raise that total cash will be closer to $5 billion.
Operating cash flow was ($251.3 million) for a (21%) margin in Q2, widening from ($117.8 million) in the year ago quarter. Free cash flow was ($2.7 billion) for a (223%) margin, widening slightly from ($2.36 billion) in the year ago quarter.
Debt was reported at $11.05 billion in Q2, with $3.62 billion being current. Current debt is likely closer to $12 billion now, with a majority (~$6.7 billion) tied to its two existing delayed draw term loan facilities; if the new DDTL is drawn upon, debt could rise to $14.6 billion. On the other hand, the company is trying to reduce its cost of capital by raising cash through secured debt, using its highly valuable GPUs as collateral, which is positive.
As discussed above, capex for the second half of the year is expected to be >$15 billion, with cash on hand only covering one-third of that at maximum. This will place the emphasis on finding alternative funding to finance this spending.
Management Hinted 2025 Will End Strong
As stated, CoreWeave currently operates 33 data centers for 470 megawatts of power, yet is going to deliver an additional 400-plus MW of power by the end of the year. This means the year will be back-half loaded, as management made abundantly clear:
“And so we are very comfortable with the ramp that we are seeing in front of us in order to deliver the 900 megawatts-plus of power as we go through Q4. It is going to be backloaded, as Nitin said. We knew that it was going to be backloaded as we came in. And we're watching the build-out and scaling of that infrastructure very systematically as we continue to move through the year.”
It was also helpful to hear in the last earnings all that the company has signed expansion contracts with both hyperscalers, with one of those contracts being reported in Q2 already but the other will be reported in the upcoming Q3 earnings report: “One of those contracts was signed in Q2 and is reflected in the Q2 revenue backlog number. The other one was signed in Q3 and will be reflected in our Q3 revenue backlog number.”
Given the use of the word hyperscaler, this would be in addition to the OpenAI $6.5B announcement.
CoreWeave’s Buy Plan
The near-vertical move from CoreWeave’s IPO low suggests the formation of a bullish long-term structure, taking the shape of a developing five-wave pattern. Since the June peak, price action has produced a three-wave pullback, which supports the potential for a larger breakout and a push to new highs.
Based on the current, limited price history, I am tracking two potential scenarios. Both imply that this correction could ultimately lead to new highs, provided that any additional weakness remains above $61.50.
Green – In this case, the market is in the early stages of a new uptrend. As long as any pullback holds above $96.75, we should expect a move back to retest the all-time highs. A decisive breakout above those highs would likely confirm a sustained continuation of the uptrend that originated at the IPO low.
The current pattern suggests a leading diagonal is in play. This is evident with the overlapping push higher from the September 12th low. We should see this push fail under $155 and then trend back into the $125 – $115 range but holds $114.65. If we instead see a strong push over $155 on expanding momentum and volume, we will then likely attack the $183 – $188 region. Over $188, and we should see the larger uptrend resume.
Red – If price breaks below $114.65, and then $96.75—ideally in a sharp, decisive decline—downside targets shift toward the $65 region before a meaningful bounce develops. Under this scenario, any long positions should be managed with protective stops near $61.50.
CoreWeave stock's buy plan by Knox Ridley, Portfolio Manager of the I/O Fund
For more Information:
Join us this Thursday at 4:30 p.m. Eastern Time as Portfolio Manager, Knox Ridley, goes through in detail the I/O Fund’s plans for buying CoreWeave with a strategy for risk managing the stock while also protecting to the downside. Every week, Ridley discusses the most promising AI stocks, and more.
The I/O Fund issued 22 buy alerts between March and April of this year, targeting the AI economy. We now have a handful of positions up over 80%, one position up over 100% and two entries up over 300% in 2025. Our cumulative returns of 210% over a five-year period would place us as #2 if we were a hedge fund and #5 if we were an ETF. Learn more here
Please note: The I/O Fund conducts research and draws conclusions for the Fund’s positions. We then share that information with our readers. This is not a guarantee of a stock’s performance. Please consult your personal financial advisor before buying any stock in the companies mentioned in this analysis.
