Snowflake is one of the fastest growing tech stocks on the market, and at a quick glance, it doesn’t appear cheap. Its market cap is around $100B and its 1-yr fwd P/S multiple is 46x. However, Snowflake is different from other SaaS stocks because the company bills customers based on their consumption rather than a subscription. This is a relatively new approach to software billing, which makes it harder to model and forecast near term sales. However, there are signs that Snowflake’s forward estimates are likely conservative, which artificially increases its multiple and makes it look expensive relative to peers. I explain why in more detail below.

Benefits of Snowflake’s consumption model:

As data creation, ingestion and storage soar in the cloud environment, cloud software providers are starting to migrate away from subscription agreements, which are fixed, to a consumption-based pricing model, which are uncapped. Snowflake is one of the few cloud providers that is nearly 100% consumption-based.

Consumption-based pricing has a few drawbacks. For example, its less predictable than subscription revenue and there isn’t a ‘floor’ on revenue, because if consumption declines then so will sales. However, the flip side is also true, consumption billing does not have a ‘ceiling’ on revenue, so if customer consumption rises, so does sales. This uncapped revenue potential, and Snowflake’s leading market position, sets the company up well to execute in the near term.

What is unique about consumption billing is that it is non-linear, and its harder to predict. However, Snowflake’s sales have started to accelerate (Q3 sales increased 110% YoY up from 103% YoY in Q2) and there are signs that sales will continue to be robust in the near term. Snowflake states that customers generally accelerate spending once they are fully deployed on the platform. Specifically, Snowflake disclosed in its 10Q that:

 “Consumption for most customers accelerates from the beginning of their usage to the end of their contract terms and often exceeds their initial capacity commitment amounts. When this occurs, our customers have the option to amend their existing agreement with us to purchase additional capacity or request early renewals”accelerates from the beginning of their usage to the end of their contract terms and often exceeds their initial capacity commitment amounts. When this occurs, our customers have the option to amend their existing agreement with us to purchase additional capacity or request early renewals”

The company’s NRR metric supports management’s claim that customer spending ramps overtime. For instance, NRR recently reached 173% in the latest quarter, the highest level since it went public and above the level disclosed in its S-1 (169%). The improvement in Snow’s NRR metric highlights that customer spending continues to ramp and that there isn’t a ceiling on total spend. The company’s NRR of 173% is also well above other cloud (SaaS) leaders at around 130%, further highlighting the uncapped nature of consumption based spending relative to subscriptions.

We can also see that growth in enterprise customers has outpaced sales growth. In the most recent quarter, enterprise customers spending >$1m per year increased 128% YoY, faster than Snowflake’s 110% YoY rise in total sales. These customers are also getting bigger each year. For example, Snowflake disclosed that 53% of revenue came from enterprise customers, up from 46% in the year-ago quarter, implying that spending per enterprise customer increased 6% YoY to ~$3.4 million per enterprise customer. The fact that enterprise customers are growing faster than sales and also increasing their spending highlights the uncapped revenue potential of Snowflake’s consumption model.and also increasing their spending highlights the uncapped revenue potential of Snowflake’s consumption model.  

As new customers join the platform and ramp spending, this will be a tailwind for sales going forward. At the same time, existing customer spending continues to rise, leading to multiple tailwinds for topline expansion.

Another key metric that highlights Snowflake’s revenue potential is the amount of future sales under contract. When customers sign onto the platform, they purchase consumption at specified prices, which gets recorded as remaining performance obligations (RPO). RPO increased 94% YoY to $1.8 billion, with a weighted contract length of ~2.5 years.

Looking forward, RPO is equal to a third of the aggregate FY2023 and FY2024 sales estimate (~$5.1 billion). In other words, Snowflake has a third of its forward sales estimates already under contract.

Furthermore, Snowflake’s RPO metric is backed by cash, giving us more visibility into future sales. Since customers are paying for their contracts a year in advance with cash, this demonstrates the pricing power Snowflake has and the strong demand for its products. Generally, customers paying for a service upfront is a sign of strength.

As shown below, deferred revenue is rising with RPO. However, deferred revenue relative to RPO did decline YoY from 48% of RPO to 43% of RPO in the latest quarter. Mgmt said this was due to customers migrating to quarterly billings, away from annual billings. This is a form of payment term extensions, which can temporarily juice sales (if you require less upfront cash, it’s a better deal for customers and incentivizes them to sign up).

While the lower deferred revenue to RPO is a trend to watch, it is not yet concerning in my opinion. In fact, Snowflake is a unique position of commanding upfront cash payments for consumption-based spending. For example, New Relic (NEWR) recently transitioned to a consumption-based billing model and is paid mostly in arrears (or after the fact), while Snowflake is pre-funded with cash. Being paid upfront helps pay for working capital and can be a significant advantage in the long run.

Snowflake is effectively getting the best of both worlds; uncapped revenue potential with consumption spending and upfront cash payments (usually reserved for subscription billing models). This trend demonstrates the demand for its products and improves the quality of revenue, which deserves a premium multiple.

Finally, there are signs that Snowflake’s RPO metric may be understated.understated. As mentioned above, Snowflake states that new customers accelerate their usage after deploying and “often exceeds their initial capacity commitment amounts”. often exceeds their initial capacity commitment amounts”. This statement, which is backed up by the company’s robust NRR metric of 173%, and rising spend per enterprise customer, suggests that there is upside potential in Snow’s RPO metric and future sales growth.

The fact that RPO growth is back-end loaded supports our thesis that forward estimates are likely conservative. This is because customer’s often go over their initial contractual amount, meaning that 1) RPO is likely understated and 2) analysts estimates are likely conservative (assuming they extrapolate forward estimates from trends in RPO). Because of the uncapped nature of Snowflake’s revenue model and the tendency of its customers to accelerate overtime, this makes it difficult to compare Snowflake to SaaS peers, which report more linear growth. I discuss the company’s valuation in more detail next.

Valuation

Snowflake has guided to $10B in sales by FY2029 and its expected share count dilution is forecasted to be <3% per year. This implies a ~10x P/S multiple on FY2029 sales. Sales are expected to rise at a CAGR of 36.5% for the next seven years to reach $10B, and the company’s multiple compression is expected to be ~26.5% per year. There is upside to the company’s valuation if sales grow faster than 36.5% through FY2029 (assuming a 26.5% multiple compression per year).

As of today, Snowflake trades at a 84 fwd P/S multiple (based on FY2022 sales ending in January 2022) and a 46 fwd P/S multiple (based on FY2023 sales ending in January 2023), which is a premium in the tech space. On a two-year forward basis, Snowflake trades at a 29x fwd P/S multiple (FY2024 sales), highlighting the large multiple compression forecasted by the Street going forward. Annual sales growth is expected to slow from 104% in FY2022 down to 66% in FY2023 and 56% in FY2024.

Given the non-linear nature of consumption spending, comparing Snowflake to SaaS peers may not be totally appropriate. Analyst estimates are mostly linear, since they have to be prudent with their estimates and its difficult to predict consumption. However, Snowflake’s robust NRR metric and likely understated RPO metric suggest that there is upside to forward estimates and that current estimates may be too low.

Are analyst estimates conservative?

The company’s strong metrics discussed above highlight the potential upside in future sales growth. For instance, the company’s contracted revenue (RPO) is already equal to a third of the aggregate FY2023 and FY2024 sales estimate, and RPO is backed by cash, further increasing the quality of the metric. With evidence that existing customers are ramping spending (with NRR rising above 170% and spending per enterprise customer also increasing), the argument can be made that RPO is likely understated. This is impressive, considering RPO grew by nearly 100% in the most recent quarter. 

Furthermore, consumption based spending is inherently unpredictable, which makes it difficult to model near term revenues. I believe there is a degree of conservatism priced into forward estimates due to the unpredictable nature of consumption spending, which makes Snowflake appear more expensive. Yet, there are trends that improve the quality of Snowflake’s forward sales, such as its RPO metric discussed above, which may be understated, its cash support backing RPO, and the rapid expansion in customer spending over time.

Looking forward, the Street is pricing in a rapid multiple compression. If growth can remain above trend for the next few years, there is upside potential to its valuation. It is important to remember that consumption growth is non-linear and uncapped, and the company’s metrics suggest that growth will remain robust in the near term. For instance, sales accelerated, NRR increased to over 170% and spending per enterprise customer also rose. Finally, the company is paid upfront for its consumption contracts, which is unique and highlights the strong demand for its platform.

With the fundamental explosion in data creation, ingestion and storage in the cloud environment as tailwinds, Snowflake’s uncapped revenue model is well positioned to benefit from these massive secular trends. The company’s key metrics suggests that sales will remain robust in the near term and we believe that Snowflake is well positioned to outperform going forward.

*Here's Beth's most recent editorial below*

Snowflake Accelerates in Revenue while Tech Growth Sells Off

The company was listed in September 2020 and the shares more than doubled on the day of the listing. It was one of the biggest tech IPOs of all time raising roughly $3 billion with a road show that attracted risk-adverse Warren Buffet.

In my deep dive published on Forbes, I noted Snowflake’s sky-high revenue growth of 173% in the year prior to the IPO. Another key metric that led to the success was the net retention rate of 158%, which was the highest for any cloud company at the time of listing; this metric is even higher now. Snowflake closed the opening day with a market cap of $70.3 billion that was more than five times its last private valuation of about $12.5 billion. Snowflake has been public for over a year and now trades at a market cap of $92.5 billion for a gain of roughly 33%, at time of writing.

Below, we revisit the product and the company’s financials now that it’s been a public company for a decent length of time. The information in this analysis is partly why we have decided to build a position with more information on what makes Snowflake stand apart provided to our premium members.

Snowflake’s Rare Acceleration in Net Retention

There are a few key metrics that Snowflake discloses that help investors better understand the demand for its products and the cadence of its growth going forward. One of the key metrics is its net retention ratio (NRR), which increased YoY from 162% to 173%, the highest level since it went public and a notable (rare) acceleration.

Since Snowflake uses a ‘land-and-expand’ sales strategy, growth with existing customers is critical to scale its business. An increase in its NRR metric demonstrates that customers sign on and then rapidly ramp spending as they fully deploy on Snowflake’s platform.

However, CFO Michael Scarpelli, cautioned investors that NRR will decline going forward, but nevertheless will still remain well above 140% for a “very long time”. Specifically, he stated that, “I'm not going to guide long term. It's hard to do that. I'm just going to reiterate again what I said to Derrick is we will [keep NRR] above 160% for this year. And I do expect longer term as our customer base gets bigger and bigger and more mature, that number will come down, but I still think it will be well above 130%, 140% for a very long time”.

Source: Investor Presentation

While Snowflake uses a “land -and-expand” sales strategy, it also uses a consumption billing model. For instance, Snowflake bills customers based on the amount of data they store and transfer and what resources they use.  Accruing revenue based on consumption rather than a ratable subscription model decreases the predictability of quarterly revenue, but it leaves revenue uncapped. This provides revenue upside, because if consumption soars, then so will revenue. We can see with the NRR metric discussed above that existing customers are ramping consumption. Furthermore, this consumption is also contracted, meaning that a portion of forward topline growth is locked in which provides visibility into future sales.

The company’s robust NRR metric of 173% discussed above also backs up management’s claim that customers often exceed their original contract amounts. Furthermore, since sales are accrued under a consumption model rather than a subscription model, there doesn’t appear to be a ceiling on customer spending. For instance, Snowflake disclosed that customers spending over $1 million (enterprise customers) grew 128% YoY to 148. Moreover, these customers accounted for 53% of total revenue, up from 46% in the year ago quarter. This implies that spending per enterprise customer increased 6% YoY to $3.4 million. While the outsized growth in enterprise count is impressive, it is also great to see spending per enterprise customer rise as well, signaling that large customers keep getting bigger.

As a result of the improvement in the metrics disclosed above, the company’s revenue accelerated by 110% YoY to $334.4 million. The topline growth was very strong and has been above 100% for at least five quarters in a row (since the company went public). Growth was led by financial, media, technology, and retail customers.

The company has a growing customer base. As mentioned above, customers with trailing 12 months product revenue greater than $1 million were 148, up 128% YoY.  Furthermore, total customers increased 52% YoY to 5,416 customers, while Fortune 500 customers grew by 30% YoY to 223.

Source: Company Website

The company is also growing internationally and the growth is higher than the company’s total growth. International revenue which was 14% of the total revenue in the Q3 FY21 has increased to 18% in the Q3 FY22.

In the earnings call, Frank Slootman said, “We continued our international expansion with product revenue from EMEA and Asia-Pacific outstripping the company's year-on-year growth, up 174% and 219% respectively. We recently launched operations in three new countries Israel, Korea, and the United Arab Emirates.”

Another key metric, remaining performance obligation grew by 94% YoY to $1.8 billion. This represents revenue that is contracted but not yet realized.

Of the total $1.8 billion the management expects about 55% to be recognized as revenue in the next one year. Some of the notable large multi-year deals in the recent quarter include a $100 million three-year deal to an existing customer and additional five eight-figure multi-year deals. This is a positive trend that the company has been able to win large contracts.

The company’s margins continue to show improvement. Total gross margin is 64% and adjusted gross margin is 71% when compared to 58% and 67% respectively, for Q3 FY2021. Adjusted product gross margin came in at 74.6% when compared to 73.6% in the previous quarter and 70% in the same period last year.

Net loss came in at $154.9 million or ($0.51) per share compared to $168.9 million or ($1.01) per share for the same period last year.

The company’s free cash flow improved to $9.5 million from a free cash outflow of $37.9 million for Q3 FY2021. Adjusted free cash flow came in at $21.5 million compared to adjusted free cash outflow of $37.1 million in the same period last year. The company has maintained a strong balance sheet as it has cash and investments of about $5.1 billion.

Source: Investor PresentationInvestor Presentation

Product:

Snowflake’s decoupled architecture allows for compute and storage to scale separately with the storage provided from any cloud provider the customer chooses. By processing queries using massively parallel processing (MPP), where each node in the cluster stores a portion of the data set locally, the virtual warehouses can access the storage layer independently so as not to compete for compute power. With the competitors, such as Redshift, where compute and storage are coupled, more time is spent reconfiguring the cluster. 

Snowflake calls this offering a virtual data warehouse where workloads share the same data but can run independently. This is crucial because Snowflake’s competitors combine compute and storage and require customers to size and pay based on the largest workload.

Data warehouses are centralized data repositories that collect and store information across many sources that are both internal and external. The raw data is ingested into the data warehouse and processed to answer queries. One key product differentiator is that Snowflake is not built on Hadoop, rather the company uses a new SQL database engine with cloud-optimized architecture. Overall, this translates to faster queries and also reduces costs by scaling up or down for both capacity and performance. This also allows the shift to the cloud while still honoring traditional relational database tools. Just like cloud infrastructure does not require you to hold server space for peak times year-round, a cloud data warehouse does not require you to plan, acquire or manage resources for peak data demand (i.e. elasticity).

The need for resources could change by either increasing or decreasing (scaling up or down). Customers that have a need for storage but less of a need for CPU computations do not have to pay up front and can shrink the environment dynamically. Users either pay for terabytes or are billed on a per-second basis for computations. As discussed above, Snowflake charges by execution-based usage and is not a cloud SaaS-company that charges by subscription.

Snowflake has a multi-cluster architecture which is unique from single cluster databases. The multi-cluster approach allows the clusters to access the same underlying data yet to run independently. This allows for heavy queries and operations to run very quickly and with fewer errors because the queries are not accessing the same data warehouse.

Beyond the value proposition of separating storage from compute for speed, and also scaling up or down to reduce costs, the third takeaway is that Snowflake is also much easier for customers to use as it’s designed to remove the role of a database administrator for monitoring and/or to tune query performance.

The end goal of choosing Snowflake is that you load data, run queries, and do little else – which is an immense value proposition due to the amount of time wasted prepping, balancing, tuning and monitoring traditional data warehouses originally built for on-premise.

Snowflake is capitalizing on the multi-cloud trend and growing rapidly with customers who want a choice in public cloud provider despite the cloud giants having their own data warehouse systems, such as Amazon Redshift, Azure Synapse and Google Big Query.

In our first article written at the time of Snowflake’s public listing, we discussed competitors Google’s Big Query and Amazon’s RedShift. Big Query has a strong following of about 2X customers compared to Snowflake, growing at 40% and also offers separate storage and compute. The differences between BigQuery and Snowflake include pricing structure where Snowflake is a time-based pricing model where users are charged for execution time and BigQuery is a query-based pricing model, where users are charged for the amount of data returned from the queries. Redshift has growth of 6.5% and is not as competitive due to coupling compute and storage.

In 2020, The Enterprise Technology Research study showed 80% of AWS accounts plan to spend more on Snowflake in 2020 relative to 2019 with 35% adding Snowflake as new compared to 12% adding Redshift as new. In Azure, 78% plan to spend more on Snowflake with 41% adding new. On Google Cloud, 80% plan to increase spending on Snowflake.

Granted, this study was in 2020 but this helps drive home why Big Tech owning the data centers is not a deterrent for Snowflake’s rapid adoption. Judging by Snowflake’s revenue growth, these preferences are likely still intact.

The company also launched support for unstructured data earlier this year, which is another strength compared to the SQL legacy competitors. Due to the increasing use of unstructured data, there is demand to support unstructured data for big data analytics.

Data Sharing and Data Marketplaces

Snowflake allows businesses to share their data with other external businesses on the platform. Data Marketplace allows free or monetized data sets to be exchanged. This has helped Snowflake break into new industries with use cases that other data lakes and competitors do not currently offer.

For example, earlier this year, Snowflake announced support for Unified 2.0, an open sourced and transparent identity framework that will help publishers, advertisers, and its partners identify users. When browser providers like Google plan to eliminate third party cookies, Unified 2.0 is seen as one of the potential replacements by ad tech firms.

In the Q2 earnings call, Jeff Green, the CEO and Founder of The Trade Desk, mentioned, “I think Snowflake adopting UID2 is one of the biggest headlines that has happened for UID to date and not enough has been said about it. I don't think most people understand why this is so big.” He further added, “So in the same way that Wix made it really easy for companies to build websites, Snowflake makes it really easy for companies to put their data to work.”

The management has maintained since its IPO that the opportunity in data sharing is substantial and largely untapped. In the recent earnings call, the CEO mentioned, “Generally I agree with what your assessment that we are just seeing the tip of the iceberg. Snowflake was built from the ground up as a data sharing platform and we've been at it from the beginning. You see a lot of other players following our lead in this regard, but we are in the beginning.”

The company also follows a consumption model, which makes investment decisions easier for its customers to decide which business units need the workloads. The management gave an example of the financial sector in the earnings call. The CEO mentioned, “That really mitigates the sticker shock, people can make investment decisions as they go along and as it warms it, we're seeing with some of our large banking customers as they went from recomputing loan rates on a monthly basis to doing it every night, while they had a business case for.”

In the most recent quarter, Data Marketplace grew 41%, which is “steady” but expected to could expand at a “meteoric rate” due to the non-linear way data sharing expands. The company recently introduced two industry data clouds: Financial Services Data Cloud and Media Data Cloud. The customers include companies like Allianz, Blackrock, New York Stock Exchange, State Street, Disney Advertising Sales, The Trade Desk, and Experian, among others.

Developers Building Apps with Snowpark

Snowpark offers the ability to migrate business logic with popular programming languages Python, Scala/Java Virtual Machine or Java. The library and DataFrame API allow querying and processing data without having to move data to where the application code runs. This extends programming functionality for ML model training and allows data processing to run natively in the data cloud.

Prior to Snowpark, code deployment required separate infrastructure. Building applications that interact with Snowflake’s virtual warehouses minimizes processing time and lowers the learning curve/broadens adoption of complex data pipelines by removing the need to move or copy data into other systems to overcome working with SQL.

The recent announcement of adding Snowpark for Python is key because of Python’s widespread popularity among developers. With the Snowpark Accelerator, Snowflake is courting developers to build more applications and this is likely to help Snowflake maintain a competitive advantage with a newer class of machine learning startups. The company had 23,000 developers register for the last Snow Day event.

As stated, unstructured data has recently become available in public preview, and this is being leveraged through Snowflake’s newer programmability as customers can now store new data types.

Risks

The company’s revenue growth has been exceptional. However, the company is undergoing losses. There is no clarity as to when the company will be profitable on a GAAP basis. In last year’s Investor Day presentation, the company has laid its roadmap to reach $10 billion in annual product revenue in the FY 2029 and adjusted operating income margin of 10%. So, it suggests that the competition is very high for its bottom line to improve significantly.

The company’s current revenue growth rates might not sustain long-term. The management expects long-term product revenue to grow by 30%. Overall revenue growth is down from 174% in FY 2020 to 124% in FY 2021, and for this year, analysts expect revenue to grow about 104% YoY.

Another key risk we will be monitoring is the reduction in payment terms, as Snowflake is migrating from annual upfront invoicing to quarterly upfront invoicing. This reduces the amount of cash customers have to pay upfront, which can temporarily juice sales. We will need to monitor this trend going forward to ensure that growth will be sustainable as customers fully migrate.

Another risk is the company’s consumption billing model, which is inherently unpredictable. This can make growth lumpy and some quarters may disappoint the Street. Investors should expect increased volatility in growth from Snowflake in the near term as new customers ramp consumption. However, management does expect revenue growth to smooth and become more predictable in the aggregate as customer consumption scales and matures on the platform.

Valuation

Considering the company’s strong metrics discussed further above, it makes sense that Snowflake trades at a premium multiple compared to other high growth companies. At time of writing, it’s 1-year fwd P/S multiple is 46x and Snowflake trades at a 29x 2-year forward multiple.

 

Looking forward, management guided Q4 product sales to increase 95% YoY to $348 million at the midpoint and for FY2022 product sales to increase 104% YoY to $1.1 billion. One year forward, the Street expects FY2023 sales to increase 66% YoY to $2.0 billion and then to further deaccelerate to 56% YoY growth in FY2023 and reach $3.1 billion. EBITDA is also expected to turn positive in FY2023 and then rapidly expand to over $260 million by FY2024.

Conclusion

Snowflake separates compute and storage which allows companies to store large amounts of data while running complex queries at high performance. The company also drives down costs for customers newly onboarded due to its pricing model where you pay for only what is used. Snowpark now allows data processing to occur natively on the data cloud instead of external Spark clusters and is opening up complex data pipelines with popular programming languages, such as Python.

The company is disrupting legacy databases while keeping a strong focus on how to lower the barrier of entry for data applications and machine learning workflows. The product is not where the company is often disputed by investors rather it’s the valuation. Those on the sidelines for the past 1.5 years have only given up 30% in gains (in terms of market cap) and have saved themselves a rather rocky, volatile ride. Snowflake has always been a strong company yet the last earnings report was a perfect 10. We think the company may be finally gathering its strength to truly earn its valuation once and for all. The I/O Fund is officially on board, per the disclosure below.

I/O Fund analysts Royston Roche and Bradley Cipriano 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. The I/O Fund plans to initiate a position in Snowflake in the next 72 hours.

SentinelOne: Exceptional Product at a Decent Valuation

 

At time of writing, SentinelOne is trading very close to its IPO opening price of $46.00 when the stock opened for trading on June 30th. We outlined how this price included a large 900% premium from its last private valuation round. Now that SentinelOne has more trading history and is doubling its revenue every year, the stock is catching up to its public market premium. By posting 128% revenue growth or more, the valuation has come down quite a bit. We are in the last quarter for fiscal 2022 and for fiscal year 2023, SentinelOne is trading at 33X expected revenue for 2023. We think this is a reasonable buy zone for a company with this level of growth that is expected to continue.

I’ll review the product which was first covered here in my Forbes write-up. Below, Bradley also discusses the operating losses and other key points regarding the financials losses that are being front-loaded from customer acquisitions as it captures market share. However, he outlines in more detail below that there are signs of leverage in its model. Customer growth is especially strong with SentinelOne and the net retention rate is healthy. We also discuss SentinelOne’s products and why cloud is a key area of strength. 

SentinelOne Product Overview: Fight Machine-with-Machine

 

SentinelOne is an AI-powered cybersecurity company at the forefront of autonomous threat detection and prevention. The company is one of the first to introduce autonomous threat detection and prevention. It has developed an AI-powered XDR platform to make cybersecurity protection truly autonomous from the endpoint and beyond. Endpoint security refers to protecting the endpoints or entry points of the end-user devices such as desktop PCs, laptops, mobile devices, and servers from being exploited. SentinelOne expands this definition (hence XDR for “extended” instead of EDR) to include more data points.

Overview of SentinelOne

The product differentiation is best summed up by the fact other vendors require data to be sent to the cloud for analysis and often have many humans monitoring the alerts to take action. Meanwhile, SentinelOne uses automation to find the threat which reduces the number of false positives. Instead of getting every piece of telemetry that requires the security team to investigate, SentinelOne’s endpoint detection and response solution eliminates the noise so that the security team is only responding to those that have the potential to be critical.

According to SentinelOne’s S1, “Cyberattacks have become the output of military-grade, highly resourced, and automated nation-state and cybercrime operations. We envisioned a revolutionary data and artificial intelligence paradigm where technology alone could autonomously prevent, detect, and respond to cyberattacks. It is time to fight machine with machine.”

They emphasize that legacy antivirus powered by human-generated signatures still remains a widely used security technology. This is in spite of the fact that they are ineffective and reactive. Human-powered endpoint detection and response, or EDR, emerged as an alternative where people became the detection and response crew.

This approach led to the “1-10-60” rule which claims the best achievable cybersecurity outcome was capped at one minute to detect an attack, 10 minutes to investigate, and 60 minutes to respond. Recent ransomware attacks have proved that it only takes milliseconds to breach an organization and cause damage. 

SentinelOne: Singularity XDR Platform

SentinelOne launched the XDR solution in the first half of 2020 prior to going public in 2021. This platform offers Active EDR, which allows for more visibility and automated responses for Endpoint Detection and Response (EDR). SentinelOne has many competitors in the EDR space while XDR extends the definition of “endpoint” to not only include devices and workstations, but to also include other data points on the network, such as containers and cloud-native applications, and also across the entire stack, such as email, the network, and identity. Extended detection and response (XDR) is cross-layered detection and response. XDR collects and automatically correlates data across multiple security layers – email, server, cloud workloads, and network – so threats are detected faster and security analysts improve investigation and response times.

SentinelOne uses many data sources to create a data lake. The single pool of raw data is built across a wide range of sources, including other vendors or internal data sources. Automation works best with a lot of data and SentinelOne is compatible with AWS, Azure or Okta, Splunk, Zendesk, and Slack. What matters to customers is that every threat is detected very quickly, and SentinelOne proposes a solution that is able to do both because automation and AI is best done at the data level rather than managing thousands of user endpoints to mitigate attacks.

The company’s Singularity Platform ingests, correlates, and queries petabytes of structured and unstructured data from ever-expanding disparate external and internal sources in real-time. It builds rich context and delivers greater visibility by constructing a dynamic representation of data across an organization. As a result, the company’s AI models are highly accurate. The company’s distributed AI models run both locally on every endpoint and every cloud workload, as well as on the company’s cloud platform and the AI models predict threats in milliseconds. The behavioral AI model maps and links all behaviors on the endpoint to create Storylines. When an activity is deemed to be a threat, the system automatically takes action to kill the attack.

Although SentinelOne and the XDR Platform is listed behind Crowdstrike and Microsoft on Gartner’s Magic Quadrant, SentinelOne leads on peer reviews. This was discussed in the earnings call with 97% of reviewers saying they would recommend SentinelOne in the 2021 Gartner Voice of the Customer Report. We had also noted the company’s strength on peer reviews in our first write-up. The company also scores high on the highly respected MITRE ATT&CK evaluations with 100% visibility and zero missed detections.

Cloud is a Growth Lever for SentinelOne

Cloud is a growth lever for SentinelOne as the company leverages a microservices architecture for rapid and frequent updates. The company offers support for Kubernetes workloads with additional runtime protection and simplified deployment. Kubernetes is automation orchestration for containers and allows for scaling of a container rather than an entire application. Kubernetes was created by Google and is used by 78% of companies managing containers with this open-source system.

This was probably the most important thing said on the call: “Cloud still remains our fastest-growing module. About 10% of endpoints are covered by cloud and servers. It has been our fastest-growing module for some time. Cloud is a piece of the business, I think that we think will expand greatly in the future. We anticipate that at some point, it will be the similar size to the endpoint market.”

According to the earnings call, cloud was the fastest growing segment: “In particular, our cloud workload protection product delivered the highest growth during the quarter, a testament to the demand for our real-time run-time protection for cloud workloads and containerized environments.” This was expanded on later to say: “The vast majority of what we sold this quarter was the Complete package. I think that we’re seeing just overall standardization on the Complete platform. People are opting for our complete EDR package. I think what I can also say on top of that is just increased adoption of our cloud modules. We’re just seeing increased demand for cloud workload protection.”

In terms of cloud being a growth driver competitively speaking, the company stated the following: “And obviously, if you look at our mix today, also going into the cloud security opportunity, kind of further compounds it, and it’s something that the incumbent vendors never had to offer.”

The company stated its biggest competitor here is Palo Alto Networks and a few startups.

Last February, SentinelOne acquired Scalyr, a leading cloud-native data analytics platform that serves as a big data engine for the XDR platform. This helps SentinelOne ingest “massive amounts” of data real-time for the XDR platform by eliminating data schema requirements and also reduces index limitations. This speeds up the process and drives down costs by ingesting and correlating terabytes of data at machine speed. This also makes SentinelOne more competitive against SIEM tactics for data correlation and response.

In August, the company released SentinelOne Storyline Active Response (STAR) which is a cloud-based automation engine that allows security teams to create custom detection and response rules. STAR requires security teams to turn queries into rules for detection, and this challenges legacy providers. SentinelOne’s platform aggregates Storylines, which is essentially behavioral AI. The textbook definition of behavioral AI is to track behavior on a device to reveal insights. SentinelOne leverages behavioral AI to make a decision without relying on sending signals to the cloud or to security engineers before a decision is made. Instead, SentinelOne uses ActiveEDR to sift through alerts and anomalies and to form storylines. The machine helps to identify the threat and then automates a response. This is differentiated from other EDR products that are only used to detect rather than to respond.

Ranger for Agent Deployment and IoT

SentinelOne is able to find any device connected to a network through a ML device fingerprinting engine (FPE) by running an inventory of IP-enabled devices. This helps to identify unsecured endpoints and to close the security gap in agent deployment. This is what is meant by “limited visibility” or lacking full visibility of every device where just one unknown device can run malware or host ransomware and compromise a network. Other cases of unsecured endpoints could be a new server that doesn’t have an agent or new employees who are onboarded without protection yet installed. Ranger and Ranger Pro detect and notify IT teams of these unsecured endpoints. This is especially important for the internet of things (IoT) where the number of devices connected to the internet proliferates and is hard to track. For example, hospitals are becoming smart hospitals where there are thousands of devices connected to the internet. In this example, Ranger would notify the IT department if one device was unprotected.

In the earnings call, it was stated that Ranger grew triple-digits and that “In Q3, two of our Fortune 10 customers renewed with multiyear deals, and both expanded their use of the Singularity platform, adding modules such as Ranger and remote script orchestration.” Adding modules like Ranger help to keep net retention rate strong, which reached a record 130% in Q3.

Remote Script Orchestration (RSO)

RSO is a new product released this past quarter. The goal is to increase the speed in response to cyberattacks. This is done by executing scripts and commands remotely across thousands of endpoints. The company provides a script library to run scripts for all platforms from a console to find single endpoints or multiple endpoints that are compromised. This allows security teams to collect whatever is needed from remote machines. This allows the security team to terminate processes, remove files, delete directories and other responses very quickly. With STAR, this can also be automated and RSO is built for users of all technical abilities due to the script library.

SentinelOne also supports Zero Trust which eliminates the need for perimeter-based security for better protection in remote work scenarios.

Product Differentiation

Cybercrime will cost companies $10.5 trillion annually by 2025 with the cybersecurity market worth $345 billion-$400 billion. SentinelOne’s addressable market is expected to reach $40.2 billion in 2024 with $12 billion from endpoint security and $17 billion in analytics, intelligence and response.

According to SentinelOne, using their products can produce cost savings can be up to 353% – granted this number is a marketing department, however, the point is that any company increasing ROI in cybersecurity has a real chance of taking market share if their product improves the results. The savings quoted is achieved by reducing the amount of cybersecurity tools a company needs by standardizing endpoint security across more data types. The consolidation in this case saves up to $3 million over a three-year period and the enhanced threat detection saves $671K over three years. Due to automation, $1.2 million can be saved over three years by reducing time and employee hours across the IT team.

This breakdown is important to look at because SentinelOne’s main value proposition is actually consolidation of cybersecurity tools, and secondly, its automation/reduced hours. This is a different argument then relying only on enhanced threat detection alone, which is the main argument for many of the competitors (debate on whose product is better).

We see real evidence of this in the financials with 4 quarters of revenue acceleration. Here’s how the company compares to other high growth cybersecurity names in terms of acceleration.

The product differentiation is best summed up by the fact other vendors are on the endpoint and require data to be sent to the cloud for analysis and often have many humans monitoring the alerts to take action. Meanwhile, SentinelOne uses automation to find the threat which reduces the number of false positives by leveraging a data lake. Instead of getting every piece of telemetry that requires the security team to investigate, SentinelOne’s endpoint detection and response solution eliminates the noise so that the security team is only responding to those that have the potential to be critical. Per SentinelOne: “What enterprises need is automated security, not repackaged legacy AV and crowd-powered protection.”

SentinelOne is not breaking ground in a new market rather its goal is be a superior product to take business away from legacy vendors. Here’s a quote from management: “I think it’s safe to assume that about over 50% of it is still in the hands of the incumbents. Looking at our pipeline for Q4 and the out quarters, that doesn’t seem to change. So to us, that cycle is still ongoing. It’s a pretty big TAM that we’re serving. And obviously, if you look at our mix today, also going into the cloud security opportunity, kind of further compounds it, and it’s something that the incumbent vendors never had to offer. So that makes the entire buying cycle really more sticky, more inclusive and just overall more important for the enterprise. So it becomes part of the picture. But again, in almost every account that we go into, call it high 90s, we see an incumbent vendor. So we don’t see that tapering away anytime soon.”

Financial Overview:

By Bradley Cipriano

 

SentinelOne has pioneered a new approach to endpoint cybersecurity and the company is quickly capturing market share.

We can see this in recent results, as annualized recurring revenue (ARR) has accelerated for four consecutive quarters. Furthermore, the acceleration in ARR helps explain the large losses incurred by SentinelOne, as customer acquisition costs are front loaded. However, as these new customers renew their contracts, the firm’s topline will continue to grow but expenses will normalize, leading to strong profitability in the future. I outline why in more detail below.

Accelerating growth drives large losses but losses are temporary

SentinelOne has reported four consecutive quarters of accelerating ARR growth. Specifically, ARR most recently increased 131% YoY in Q3 FY2022, an acceleration from the 127%, 116% and 96% YoY increase in Q2 FY2022, Q1 FY2022 and Q4 2 FY2021, respectively. As of the most recent quarter, ARR increased $37 million QoQ to $237 million, which marked the 10th consecutive quarter of QoQ increases in ARR (there are only 10 quarters disclosed). It is noteworthy that the most recent sequential increase in SentinelOne’s ARR was as large as the firm’s entire ARR metric in Q1 FY2020.

It is also notable that the acceleration in ARR started after the October 2020 quarter. In December 2020, the high-profile SolarWinds cyberattack was identified, which had exploited key vulnerabilities in numerous service providers such as SolarWinds, Microsoft products (Office 365) and VMware. SentinelOne outperformed the competition during this period and disclosed in its S-1 that none of its customers were impacted by the SolarWinds cyberattack. This event may have been a catalyst that identified SentinelOne as a leading cybersecurity platform. Furthermore, the company launched its XDR platform in early 2020 and covid lead to a general acceleration in software and cybersecurity usage during this period, each of which likely contributed to the acceleration in ARR shown below.

The growth in ARR also flowed to the income statement, as Q3 sales increased 128% YoY to $56 million, which marked the fourth consecutive quarter of accelerating YoY growth. Growth was driven by new customers, as SentinelOne disclosed in its 10Q that new customers accounted for 46% of its topline expansion in the most recent quarter, while existing customers contributed 37% and channel partners accounted for the remaining 17%. Acquisitions provided $4 million in sales, and absent the impact of M&A, organic sales increased 113% YoY.

During the quarter, revenue from international markets grew 159% year-over-year to $19 million, and represented 33% of total revenue, up from 29% a year ago. International markets will be a key area of growth for the company going forward. Furthermore, SentinelOne’s international growth was similar to CrowdStrike’s international growth when it was a similar size as SentinelOne (~$56 million in quarterly sales in FY2019). Specifically, CrowdStrike’s international sales increased 196% YoY to 23% of sales in FY2019, highlighting the similar path that SentinelOne is following. I compare SentinelOne and CrowdStrike in more detail further below.

Gross margin improved from 58% in the year-ago quarter to 64%, which represented an all-time high (10 quarters of public information). However, despite the improvement in gross margins, operating margins remained deeply negative. For instance, Q3 operating margin was -120%, a slight improvement from the year-ago quarter of -121% and an improvement from the 10-quarter average of -141%.

While it is concerning to see operating losses larger than sales, this is due to the rapid growth in new customers. As mentioned above, new customers accounted for the majority of topline growth, a favorable trend. Furthermore, acquiring customers front-loads expenses in the early years, but SentinelOne recognizes sales ratably, which makes losses appear outsized. As customers renew their contracts, these one-time customer acquisition expenses will decline, while the topline will expand as customers adopt more products. This trend will lead to an improvement in SentinelOne’s bottom-line going forward. 

Evidence of leverage in SentinelOne’s business

What is critical for SentinelOne’s story going forward is that there are signs that its subscription service is sticky, and that customers are increasing their spending. This would provide a light at the end of the tunnel that losses will turn into profits and cashflows. While SentinelOne is still a few years out from breaking even, there are positive signs that customers are both sticky and expanding their usage of its products. 

We can see this with net retention ratio (NRR), which improved to 130% in Q3, an all-time high, and was up from 115% in the year ago quarter. The improvement in NRR showcases that customers are expanding the amount of products they use each year, highlighting the success of SentinelOne’s ‘land-and-expand’ model. Furthermore, gross retention ratio, which only considers customer attrition, was 97% as of Q3, signaling that SentinelOne’s customers are sticking with the platform beyond one year.

Furthermore, SentinelOne’s customer metrics are also high quality. For instance, no single customer accounted for more than 3% of sales in the most recent quarter and the company disclosed that it has over 6,000 customers as of Q3, up 79% YoY. Furthermore, SentinelOne counts three Fortune 10 companies as customers, two of which recently renewed with multiyear deals in Q3. In its S-1, SentinelOne disclosed that it also counted 37 out of the Fortune 500 companies as customers, highlighting the large opportunity in front of it as there are still a plethora of enterprise customers yet to sign on.

Moreover, customers with ARR over $100,00 grew 141% YoY to 416, an acceleration from the 140% and 127% YoY growth rates in Q2 and Q1, respectively. SentinelOne’s success with enterprise customers suggests that the firm is rapidly capturing market share in the cybersecurity market.

Another example that highlights the leverage in SentinelOne’s model is the improvement in sales and marketing (S&M) expense. S&M expense increased just 1% QoQ in Q3, while Q3 sales increased 22% QoQ. This drove S&M margin down from 90% in Q2 to 74% in Q3, an all-time low. The improvement in S&M margin highlights that SeninelOne is spending less to attract customers, which is impressive considering that sales have been accelerating. As adoption grows, the company’s ability to expand the amount of products customers use will drive S&M margin lower, further improving its bottom-line in the future.

SentinelOne relative to CrowdStrike

SentinelOne’s metrics appear in-line when viewed relative to other cybersecurity platforms such as CrowdStrike. For instance, CrowdStrike’s S&M margin was 70% when its quarterly sales were around $56 million, This compares to SentinelOne’s S&M margin of 74% with $56 million in sales. Furthermore, SentinelOne’s sales grew 22% QoQ, which was faster than CrowdStrike’s 18% QoQ growth when its quarterly sales were $56 million. The faster growth rate helps explain the higher S&M margin.

However, SentinelOne has reported a steeper operating loss relative to Crowdstrike at $56 million in quarterly sales. This is likely due to timing, as SentinelOne went public with a rich valuation, which increases stock-based compensation expense. Expense items such as G&A expense may be inflated relative to historical periods for other tech stocks (like Crowdstrike), when tech valuations were lower.  Nonetheless, we expect the outsized SBC expense to normalize going forward. This will also improve SentinelOne’s bottom-line and bring it more in-line with peers in the future.

As shown below, SentinelOne and CrowdStrike had similar S&M margins when they were the similar sizes. However, CrowdStrike’s ARR was growing faster, as was its customer base. SentinelOne’s ARR per customer grew faster and its NRR was more robust. However, SentinelOne’s operating margin was considerably lower than CrowdStrike’s was. It should be noted that CrowdStrike was not public in 2018, so unrecognized SBC was not included in operating expenses. As mentioned above, we expect that SentinelOne’s earnings will improve as SBC from its recent IPO normalizes.

Outlook and Valuation

Looking forward, management expects Q4 sales to increase 103% at the midpoint to $61 million and raised their full-year guide for sales to $200 million, which increased the implied growth rate from 103% to 115% at the midpoint. Gross margin is expected to be 62%, up from the prior guide of 59% and an improvement of 100 bps YoY from FY2021. Finally, operating margin is expected to be -81% at the mid-point in Q4, demonstrating continued leverage in SentinelOne’s business model. 

Analysts expect growth to remain robust for the foreseeable future and FY2024 sales are forecasted to rise 185% from FY2022 levels to $570 million. Losses are also expected to persist throughout this time period, but are anticipated to materially improve by 2024. We are still early in SentinelOne’s growth story, but the opportunity in front of the company is large as its total addressable market was estimated to be around $30 billion in FY2021 and is expected to grow to $50 billion by 2024 (S-1).

The company’s market cap is below $12 billion and it currently trades at a 53x P/S multiple and a fwd (1-yr) P/S multiple of 33x. This is a premium relative to other cybersecurity competitors listed in its S-1, such as CRWD, which trade at a fwd P/S multiple of 22x. However, the company is clearly capturing market share from competitors, evident in its accelerating ARR metric discussed above, which warrants a premium valuation.

Furthermore, SentinelOne is unique in its growth as sales have accelerated for four consecutive quarters and the firm’s topline is growing over 100% YoY. Relative to other rapidly growing SaaS firms such as Snowflake, SentinelOne’s FY2023 fwd P/S ratio of 33x appears more in-line. There is also room for share price appreciation at this valuation. For instance, if SentinelOne’s sales grow to $570 million in FY2024 as expected and its fwd P/S multiple contracts to 30x (similar to CrowdStrike’s fwd P/S multiple when annual sales were at ~$500 million), the company’s share price will appreciate by 49% (assuming a constant share count).

Risks and Conclusions

There are some key risks going forward. SentinelOne’s approach to endpoint security is new and the market may not fully accept its approach of utilizing A.I. to combat cyber threats. Furthermore, the company has also reported large losses and these losses are expected to persist for the next few years. This may require SentinelOne to issue more shares, diluting shareholders. However, the company currently has $1.7 billion in cash on balance, which provides ample liquidity in the near term. Moreover, SentinelOne has limited financial information, which makes it difficult to thoroughly analyze the company’s financials and identify anomalies.

Despite the risks and limited financial history, SentinelOne appears to be well positioned in the cybersecurity market. ARR has accelerated for four consecutive quarters and it appears that the company is capturing market share, especially after demonstrating its success during the major SolarWinds hack in late 2020.

While losses are steep, there are signs of leverage in its business as S&M margin is improving. Furthermore, new customers are driving topline growth, which is favorable but also front loads customer acquisition costs. There is also evidence that SentinelOne’s customers are sticky, which suggests that the losses today are setting the company up well to report profits in the future. The company has a premium valuation relative to peers, which is warranted due to its elevated growth rate. Lastly, there is still room for capital appreciation even if the company’s multiple declines and mimics peers in the future. SentinelOne is early in its growth story and the market in front of it is massive, if it can continue to rapidly capture market share, it will likely reach profitability sooner than the Street currently expects.

Equipment for semiconductor manufacturing usually falls into one of the three categories: wafer fabrication equipment, assembly, or testing equipment. Water fabrication equipment (WFE) is a primary segment for Lam Research, specifically for memory and storage chips. The deposition process creates layers of insulating and conducting materials with techniques like chemical vapor deposition or atomic layer deposition, which allows for thin films of atomic layers to be coated onto surfaces.

The excess material is then etched away. Deposition and etch are processes that require complex machines for wafers to be built into integrated circuits. Lam Research has done well by specializing in memory chips. In order for Lam to do well, the WFE market must be growing and Lam must create new equipment and processes to maintain or grow market share. Our goal is to participate in memory with less cyclical risk through Lam’s specialization, especially with 3D NAND, which is an emerging market where Lam leads.  

We pointed this out that Lam lets us participate in the memory market with reduced risk in our original analysis when we stated:

“Analysts covering Lam Research like to point out that the company is protected from supply and demand as memory manufacturers will continue to buy from Lam Research even during a low point in the cycle. This was proven during 2015 when Lam Research did not feel the effects of the memory trough.”

The price action below since we covered Lam helps to illustrate what we mean:

Lam has significant business in supplying equipment for leading edge nodes and this is the leading growth market for Lam. Two years ago, we discussed how Qualcomm will sell up to 50% more dollar chip content per device versus 4G generations, which refers to the dollar value of chips the device holds. Something similar is happening at the equipment level as complexity increases.

Management said the following in the most recent earnings call,

“At the leading-edge, semiconductor content growth, large die, and rising capital intensity are fueling increased wafer starts and strong WFE spending. In Foundry/Logic for instance, the next-generation processor chip for a top smartphone maker is more than 20% larger than its prior iteration. In DRAM, higher capital intensity is being driven by the increasing need to correct single bit errors through the addition of an extra on-chip bit. In 3D NAND, increasing device layer counts and the resulting higher degree of manufacturing difficulty is requiring the addition of new deposition and etch processes to address stress management, defect control, and multi-stack integration challenges.”semiconductor content growth, large die, and rising capital intensity are fueling increased wafer starts and strong WFE spending. In Foundry/Logic for instance, the next-generation processor chip for a top smartphone maker is more than 20% larger than its prior iteration. In DRAM, higher capital intensity is being driven by the increasing need to correct single bit errors through the addition of an extra on-chip bit. In 3D NAND, increasing device layer counts and the resulting higher degree of manufacturing difficulty is requiring the addition of new deposition and etch processes to address stress management, defect control, and multi-stack integration challenges.”

Lam’s Reliant equipment has also posted 11 quarters of record revenue. This equipment provides a lower cost of ownership for non-traditional chip markets, such as micro electromechnical systems (MEMS), power chips, radio frequency (RF) filters and CMOS image sensors for better connectivity and more powerful imaging. This particular segment refers to trailing edge nodes, which means larger nodes, such as 24nm, 28nm or 90nm processes. Although we’ve covered leading edge nodes, such as 5nm in the past when discussing AMD, Lam has also found success in supplying equipment for larger nodes as these are used in automotive and medical equipment. In fact, the emphasis on leading edge nodes may be why Lam has found success in the overlooked trailing edge market. Management stated that demand is exceeding WFE equipment.

As Bradley points out below, the following statement was quite encouraging in regards to Lam’s business overall: “As a result, we see the WFE investment required to achieve the same bit growth percentage over the next 5 years to be notably higherbe notably higher than the 5-year period just completed.” Lam has grown sales at a CAGR of 28% over the past five years. Consequently, the stock has seen over 550% gains in five years from a share price of $105 to $724.

While these comments are encouraging, we need to watch Lam Research closely into 2022 as Gartner is predicting a slowdown in WFE equipment in 2023-2024 as integrated circuit manufacturers and foundries “pause to digest the new capacity.”

There are other forecasts predicting a slowdown to occur sooner with 6% growth in equipment for 2022 following an estimated 34% in 2021. The foundry and logic segments, which are more than half the WFE sales, are forecast to grow 8% in 2022 following 39% in 2021. Similar forecasts are provided for NAND and DRAM equipment with a marked slowdown in 2022.

On the earnings call, however, the idea a slowdown would occur in 2022 was negated when an analyst asked about the potential for a slowdown:

Good afternoon and great job on the quarterly execution, guys. You know, the market is concerned that we're heading into a multi-quarter downturn in Memory, kind of similar to the 2018/2019 Memory downturn, which is a pretty severe 6-quarter downturn, but the one thing I clearly remember was that ahead of that downturn, your memory customers proactively cut their CapEx very, very rapidly.

Now, if I look at it this time around, there's some near-term pricing weakness in memory. But the overall memory demand environment remains pretty strong, and I think most memory companies seem optimistic right under the baton outlook for next year.

So I guess the question is, has the Lam team seen any signs similar to the 2018 downturn of customers either getting concerned or canceling or slight pushing out of shipments due to a concern on our projected memory downturn next year? -Harlan Sur, JP MorganHarlan Sur, JP Morgan

Here was management’s response:

“Yeah. Harlan, let me take that first. I think the simple answer is no. When the vast majority of our conversations with customers today is still about delivering equipment that they feel they badly need to meet their near-term requirements. And as Doug mentioned in his prepared remarks, I would say lead times have stretched out to the point where our visibility into demand in '22 is better than usual.. When the vast majority of our conversations with customers today is still about delivering equipment that they feel they badly need to meet their near-term requirements. And as Doug mentioned in his prepared remarks, I would say lead times have stretched out to the point where our visibility into demand in '22 is better than usual.

So I don't think that the hypes of initial indicators that you're talking about are things we're seeing right now. We feel much more constrained by supply chain challenges and ability to meet shipments and an over shipping situation.” -Tim Archer, CEO

Regarding the comment on supply chain challenges, Lam stated on the call that this is the biggest challenge the company faces right now. There are hundreds of parts for WFE and many are now supporting new process flows and 3D architectures. It only takes a delay on one of those parts to slow Lam’s delivery: “We're beginning to see constraints in the supply chain. So we have to work our way back up through some of those things. And that's the biggest thing we're dealing with right now.”

Lam’s Product & Growth Opportunities

Lam’s main competitors are ASML, Applied Materials, Tokyo Electron and KLA with Lam tied for third place. There are a few key products that Lam is developing and bringing to market that could help increase the company’s market share. Certainly, the manufacturing expansion in the United States, Korea, Taiwan and Malaysia hints towards Lam expecting it will need more capacity.

The first growth opportunity for Lam is 3D NAND. We covered 3D NAND in detail in our Micron analysis with the 176-layer release that is 40% higher than the nearest competitor, Samsung. The new NAND device is also 10 times denser than previous 3D NAND devices with increased power efficiency and capacity limitations removed. The data transfer rate is also very fast at 1,600 MT/s while maintaining the same height as the 64-layer device.

Here's an excerpt from the Micron analysis that will help frame Lam’s new etch solution:

“According to Micron, “current 3D NAND design has begun to reach the limits of its monolithic die-level maximum capacity. It will continue to fall short of the immense system-level storage capacities demanded by future data-driven applications. Cell-to-cell capacitive coupling complications and smaller etch requirements account for many of these limitations.” If Micron is correct, then this could be an opportunity for the company to see more market share on 3D NAND.”smaller etch requirements account for many of these limitations.” If Micron is correct, then this could be an opportunity for the company to see more market share on 3D NAND.”

Micron is trying to move very quickly with their new replacement-gate design which replaces the traditional floating-gate design before Samsung or others catchup while Lam is busy solving the issue from the front-end WFE perspective with a high-productivity cryogenic etch solution. The etch removes the material in devices at cold temperatures below 100 degrees Celsius for high-aspect ratios with 200+ layers. The cold temperatures are achieved with liquefied nitrogen gas. You can click here for an article that describes this process. 

According to management on the call, the cryogenic etch solution has already gone through QA and is being shipped this year. If the company is successful with this solution, it could extend to leading edge 3D DRAM and foundry/logic (3D DRAM is not on the product road map right now but management hinted that it will be put into production in the future).

“As one example, Lam has developed a new high-productivity cryo etch solution, which increases etch rates in high-aspect ratio features required for NAND devices with greater than 200 layers. We have installed this new capability at every major 3D NAND manufacturer for qualification with additional systems now shipping to support planned ramps to high-volume production next year.”We have installed this new capability at every major 3D NAND manufacturer for qualification with additional systems now shipping to support planned ramps to high-volume production next year.”

In the Micron report, we also discussed EUV or Extreme Ultraviolet Lithography where we stated the following:

“This manufacturing method uses smaller 13.5nm wavelengths of ultraviolet light to etch wafers as opposed to lasers from Deep Ultraviolet Lithography (DUV). You could argue that EUV is a point of weakness for Micron as Samsung is using this manufacturing method while Micron is delayed until 2024.”

EUV photomasks reflect light with alternating layers of molybdenum and silicon as opposed to conventional photomasks that block light with a quartz substrate or chromium layer. TSMC and Samsung are leaders with EUV for 5nm production. This process adds capital intensity, and this is good for Lam.

“In patterning, we're using the learning we have acquired over many years of multi-patterning etch leadership to win new applications as the industry adoption of EUV progresses. EUV requires use of special photoresist materials which, given the material composition, can amplify existing challenges with pattern roughness, and defectivity.

Unaddressed, these will lead to performance in yield loss, especially at smaller device dimensions. Lam has developed critical etch and deposition technologies to help solve these EUV implementation issues. In etch, we introduced earlier this year a new pulse plasma etch capability that has demonstrated an order of magnitude reduction in EUV-related pattern defectivity.” -Tim Archer, CEO opening remarksLam has developed critical etch and deposition technologies to help solve these EUV implementation issues. In etch, we introduced earlier this year a new pulse plasma etch capability that has demonstrated an order of magnitude reduction in EUV-related pattern defectivity.” -Tim Archer, CEO opening remarks

 Conclusion:

Lam’s management stated they feel confident that they have visibility into next year and that the company has “significant unmet demand.” The company also stated there are “tailwinds relative to the business” for 2022.

Does that mean every quarter will meet or exceed guidance? No, it could be lumpy and that’s the nature of semiconductor stocks. The analyst on the call mentioned a 6-quarter slowdown. If this doesn’t happen in 2022, it could happen in 2023, etcetera. Semis are especially challenging right now because they’re expected to be cyclical but are transitioning into a more secular trend. Therefore, a slowdown could actually be much further out due to the drivers we discussed in the Micron report. 

However, the key reason we think Lam could fare better than its peers is because as 3D layers increase, capital intensity also increases. The process does not scale linearly, instead it’s non-linear because it takes longer than 2X to etch a stack that is 2X high and requires more complex etch and deposition equipment.

Here’s one of the more important statements the company said on the call in regards to our thesis and stock position: “In 3D NAND, increasing device layer counts and the resulting higher degree of manufacturing difficulty is requiring the addition of new deposition and etch processes to address stress management, defect control, and multi-stack integration challenges.” This in turn, leads to increased investments in WFE to maintain percentages in bit growth.

As Micron, Samsung and others continue to compete on 3D NAND, and maybe even 3D DRAM in the future, we think Lam will become a clear winner. Cryo etch is leaving R&D for the first time and this is because Lam is pushing the envelope to serve this emerging trend. Lam is the leading equipment provider on 3D NAND and being a first mover here is key in serving the memory market moving forward. EUV patterning is another area where Lam leads and is seeing demand as the company aims to solve EUV implementation issues and the pattern defects that occur with equipment from its competitors.

Lam Q1 FY2022 Results

By Bradley Cipriano

Lam’s sales, earnings and cashflows all increased over 30% in the most recent quarter, signaling the unique position the company is in. We believe that Lam’s growth and earnings will continue to be robust going forward as the memory market nears an inflection point. Furthermore, capex from key customers signals that sales will continue to be strong in the near term.

In the latest quarter ending in September, Lam’s Q1 FY22 sales grew 36% YoY to $4.3 billion, which met the Street’s estimate. System revenue, which includes Lam’s leading edge equipment in deposition, etch and clean markets, increased 36% YoY to $2.9 billion while customer support and other increased 34% YoY to $1.4 billion. Management guided for Q2 sales to grow 39% YoY to $4.4 billion, which would mark the ninth consecutive quarter of YoY topline growth.

Gross margin declined 150 bps YoY to 45.9% as supply chain issues impacted margins. GAAP earnings increased 48% YoY to $8.27 per share, while non-GAAP earnings were $8.16/share, which beat estimates by 2%. TTM free cashflow increased 51% YoY to $3.2 billion and cash on hand remained high at $4.9 billion, but declined QoQ due to $1 billion in share repurchases during the quarter.

Demand was driven by memory, as 64% of equipment sales were for memory, up from 58% in the year-ago quarter. Lam described the key drivers for its products in its 10Q as “3D device scaling, multiple patterning, process flow, and advanced packaging chip integration, [which] will lead to an increase in the served addressable market for our products and services in the deposition, etch, and clean businesses.”

Lam is benefitting from a secular tailwind in semiconductor demand, and there are signals that memory is becoming less cyclical as the boom and bust cycles of years past are smoothing out due to rising demand from AI, 5G, IoT and edge computing.

Since Lam provides equipment that is used by its customers to manufacture semiconductors, we can measure their capex levels to get an understanding of where the market is moving. As shown below, quarterly capex trends from our sample group of semiconductors (n = 72) have accelerated during 2021. Capex grew 10% QoQ both in Q1 and Q2 and then increased 9% QoQ in Q3 to ~$78 billion.

Aggregate capex is up 32% YTD in 2021, well above the prior five-year Q3 YTD average of 13%.

The above capex trends add support to CEO Tim Archer’s comments on the Q1 call that “[Lam is] exiting this year with significant unmet demand… on the supply side, rising capital intensity, different architectures, new processes that need to be inserted into process flows to deal with increased manufacturing complexity. And those will be drivers for WFE structurally for a very long time. So I think there are a lot of things that will be positives for WFE in 2022, from an equipment perspective.”

The strong capex outlined above and demand for WFE provides more visibility into Lam’s 2022 sales. However, the caveat of the strong capex is that some of the capex should turn into output next year, which could lead to overcapacity. However, there are signs that the memory market is becoming less cyclical as technological innovations drive structural demand. I discuss these innovations in more detail next. 

Lam and 3D NAND

 

Looking forward, Lam is also preparing for new technologies in the memory market. One of the new technologies is 3D NAND – which Beth had previously discussed in Lam’s premium analysis here.

3D NAND moves memory from a 2D plane to a 3D plane, which dramatically improves the storage capacity.  It also requires a lot more equipment to manufacturer, which Lam provides.

CEO Archer added that “we see the WFE investment required to achieve the same bit growth percentage over the next 5 years to be notably higher than the 5-year period just completed. However, as the leading equipment supplier to the 3D NAND market, we are investing in new and differentiated capabilities to ensure scaling remains cost effective”

As 3D NAND nears an inflection point, there will be a structural increase in demand for WFE investments, benefitting Lam’s sales, earnings and cashflows. The equipment required to manufacturer 3D NAND is significantly higher. So, if the market increasingly adopts 3D NAND, then demand for Lam’s WFE should also increase at a relatively faster rate than prior years.

A risk to our thesis going forward is that Lam does not innovate fast enough to keep pace with changes to the memory market. While Lam’s research and development expense has increased YoY for sixth consecutive quarters, it fell to just 9% of sales in the most recent quarter. This is below the five-year seasonal average of 12% of sales.

On a TTM basis, R&D expense declined to 10% of TTM sales, which was also below the five-year average of 12% of TTM sales. However, R&D may seem low relative to sales due to the company’s rapid increase in sales recently. Nonetheless, this is a trend we will need to monitor going forward

It is likely that Lam’s prior R&D investments are now starting to pay dividends. Lam is well positioned to benefit from the rise in capital intensity from key customers. We had previously discussed in our Micron analysis that Micron was expected to continue to ramp capex, driven by investments in 176-NAND. Micron disclosed on its most recent conference call (12/20/21) that it expects capex to be around $11 to $12 billion in FY2022. This would represent a 15% increase in spending, which followed a 22% rise in FY2021. In the most recent quarter (q1 FY22), Micron’s quarterly net capex increased 22% to $3.3 billion, a record high. The growth in Micron’s capex is a forward looking metric that translates into demand for Lam’s WFE, driving topline growth at Lam.

Conclusion

Lam’s sales, earnings and cashflows all grew over 30% in the most recent quarter. Management guided for continued growth next quarter and stated that visibility into CY2022 was clear due to strong demand for semiconductors and memory. The company has tailwinds with new technological innovations such as 3D NAND, but needs to keep investing into R&D to ensure it can compete in the future. The company should continue to do well in the near term but we will need to monitor its investments in innovation going forward to make sure growth is sustainable. 

Recommended Reading:

 

Micron Deep Dive: Automotive, 5G and Data Centers

Lam Research Premium Analysis

5G Part 1 Premium Analysis

5G Premium Analysis: Semis Overview

Q2 2020 Semis Update

5G Update

Chip Shortage

AI Accelerator and 5G Chips

Outsourcing is a trend that is at the intersection of two major structural tailwinds:

1) the rise of the cloud environment and

 2) transition to hybrid and remote work.

The concurrent rise of these two microtrends allows companies to be decentralized and access talent on a global scale. This has created a unique environment that is well-suited for outsourcing firms, which we expect will grow strongly going forward.

Outsourcing helps solve labor supply imbalances, such as the need for skilled and/or cheap labor. Furthermore, not all outsourcing firms are the same and some are positioned better than others for the current environment.

In the following analysis, I discuss three different firms across the outsourcing market: Grid Dynamics, TaskUs and Fiverr. Each of these firms is uniquely positioned to address different types of labor demands, such as the need for highly technical labor (Grid Dynamics), lower cost labor (TaskUs) or freelance workers (Fiverr). I begin my discussion with Grid Dynamics, the firm we believe is best positioned to outperform in the current environment.  

Grid Dynamics: digital transformation trend drives demand

The outsourcing firm we like best is Grid Dynamics (Grid, ticker: GDYN), a $2.5 billion company that was founded in 2006 in Menlo Park, California. Grid is focused on outsourcing highly technical labor to work on big ideas such as NoSQL, cloud computing, and AI/machine learning. The firm’s sales are accelerating and its business structure is low risk, which we believe sets the company up well to capture share in enterprise-level digital transformation, a $700 billion market that is expected to grow at a CAGR 22% for the next decade.

Grid is capturing share in this massive market, demonstrated by its robust 120% YoY growth rate in the most recent quarter. As shown below, enterprises are rapidly migrating to the cloud and digitally transforming their businesses, and we are in the early stages of this massive secular shift impacting nearly every business.

We can proxy the pace of demand for digital transformation by looking at FAANG+M Capex. As enterprises migrate to the cloud and digitally transform their businesses, this spurs capacity expansion at FAANG+M. As shown below, the major cloud providers are ramping capacity expansion. For instance, FAANG+M Capex increased 34% in Q3 to $36 billion and has increased 30% or more for 4 consecutive quarters.

Furthermore, FAANG+M capex is up 42% in the last twelve months to $132 billion. To put this into perspective, this would rank 59th in global GDP if FAANG+M capex was its own country. This analysis helps illustrate how digital transformation is a massive trend that is growing very fast, and Grid is uniquely positioned to benefit from this secular tailwind.

Grid’s Opportunity

Grid Dynamic’s clients are primarily US based, but only 10% of Grid’s workforce is based in the US. In fact, 90% of its employees are based overseas in Central Eastern Europe (CEE), a region that is known for its STEM expertise, especially in programming and computer science.  

Grid explained in its 10K that its supply of CEE labor gives it a competitive advantage in the current environment due to its focus on STEM (emphasis added):

“CEE is increasingly known for the quality of its software development talent, enabled in part by decades of focus on fundamental STEM disciplines in higher education. CEE-based teams and individuals are frequent winners of programming contests such as the ones held by the Association for Computing Machinery, or ACM, TopCoder and Kaggle. Grid Dynamics believes that this disparity between the supply and demand for technical talent can be a significant opportunity for Grid DynamicsGrid Dynamics believes that this disparity between the supply and demand for technical talent can be a significant opportunity for Grid Dynamics”

According to DataArt, CEE-based programmers placed 1st, 2nd or 3rd 73 times in the five most prestigious programming competitions in the world, or nearly half of all podium spots available (2011-2020). These competitions are intense, and the most recent competitions had 182,000 registered participants. CEE’s success in these competitions highlights the regions expertise in programming, a skill that is in high demand due the rapid rise in digital transformation discussed above.

A growing digital economy, accelerated by the rise of cloud computing and remote work, is a driving force behind heightened demand for highly skilled programmers, regardless of nationality. Grid explained in its 10K that it “targets the top 10% of technical talent from top technical universities. Nearly 100% of Grid Dynamics’ engineering personnel have advanced degrees in computer science”. We can see that Grid’s labor supply is in high demand, as sales recently accelerated to 120% YoY growth, and there are signs that this acceleration may continue into Q4. I discuss Grid’s recent financial performance in more detail below.

On top of providing access to highly skilled (and cheaper) global talent, Grid is also benefitting from a unique dynamic as job openings in the US have reached record highs (shown below). For instance, a recent Gartner survey highlight that “IT executives see the talent shortage as the most significant adoption barrier to 64% of emerging technologies, compared with just 4% in 2020”.

Software Developer Salaries in USA v Eastern Europe  

Source

10-Year Trend in Number of US Job Openings (in thousands)

Grid is well positioned to benefit from the rapid rise of cloud computing, transition to remote work and the supply imbalance of labor in the US. Grid’s supply of highly technical labor should do well going forward, especially as companies accelerate their digital transformation. Highlighting its strength in digital transformation, Grid Dynamics recently earned Google Cloud Premier Partner Status, a status reserved for the top 3% of Google partners. Grid was also labeled a leader in midsize software development service providers by Forrester (below).

Midsize Software Development Service Providers

Grid’s topline growth is accelerating

Due to the tailwinds mentioned above, Grid is growing rapidly and the growth is accelerating. Q3 sales grew 120% YoY to $58 million, an acceleration from the 113% and 21% YoY growth rates in Q2 and Q1, respectively. Absent recent acquisitions, organic sales grew 68% YoY or 15% QoQ to $44 million, which is well above peers (shown below) but slightly below its organic growth rate of 72% in Q2.

It should be noted that Grid’s sales declined in 2020 as spending was cut by its largest vertical (retail/ecommerce). However, this was offset with a rapid rise in its Tech, media and telecom vertical, which has increased sales by 39%, 35% and 43% YoY in Q3, Q2 and Q1, respectively. Furthermore, sales have been growing strongly on a sequential basis, and have increased QoQ every quarter since bottoming in Q2 2020.

Looking forward, management guided for Q4 sales to grow 94.2% YoY and organic sales to increase 52.7% YoY. For reference, Grid guided Q3 sales to grow 93.7% YoY and organic sales to grow 51.9% YoY. This guide may be conservative, considering Q3 sales came in well above guidance, which is typical in the tech industry. Assuming Grid beats by a similar amount in Q4 as it did in Q3, then Q4 sales will accelerate, as well.

Earnings are also robust as gross margin increased 125 bps YoY to 44% which was well above the three-year average of 40%. EBITDA (not adjusted) turned positive in Q3 and YTD adjusted EBITDA is up 227% YoY from $8 million to $27 million, which resulted in an adjusted EBITDA margin of 19%. Q3 adjusted EPS increased 120% YoY to $0.11/share, which beat estimates by 40%.

Grid also has a strong balance sheet and cash is its largest asset at nearly $200 million, or 65% of total assets. The majority of Grid’s customer contracts are under low-risk master service agreements (MSA), which carry little to no risk of cost overruns. Contract types are often an overlooked area for service providers, but high-risk contracts such as fixed-price contracts, can temporarily juice sales but can result in large losses in the future. Grid’s ability to accelerate growth and capture market share while utilizing low-risk MSA contracts is a sign of strength.

Risk

A key risk with Grid Dynamics is its small size and high customer concentration. Grid’s top 5 customers accounted for 42% of sales in Q3, while its top 10 customers accounted for 58% of Q3 sales. However, this is improving, and its top 5 and 10 customer concentration is down from 60% and 78% in Q3 2020, respectively.

Furthermore, Grid is exposed to foreign currency risk as it is paid in US dollars but pays its employees in their local FX. However, Grid has agreements that pay employees in a US equivalent amount, which naturally hedges foreign currency risk to a degree.

Another risk is reputation risk. Providing a service (such as outsourcing) is highly dependent on having a strong reputation and any damages to Grid’s reputation could impact Grid’s ability to win new contracts. However, Grid’s management team appears sound and the CEO (who is from Eastern Europe) has been with the firm since 2014. It is noteworthy that Grid’s founder recently left the company in August 2021 and founded a new company. However, according to her LinkedIn, the company she founded is not an outsourcing firm and does not compete with Grid’s core market.

We really like Grid’s robust growth and low risk business model. Looking forward, Grid appears well positioned to capitalize on the digital transformation trend, a structural tailwind that is expected to grow strongly for the next decade. We will be watching this company closely and might initiate a position given the company’s strong fundamentals. Next, I discuss TaskUs, another fast-growing outsourcing firm that specializes in outsourcing labor to digital-native companies.

TaskUs: backend operator for the tech industry

TaskUs (Task) is a fast-growing, digitally native outsourcing firm that was founded in 2008 in the Philippines and went public this year. Task is increasingly becoming the operation infrastructure provider of choice for digitally native companies, such as Zoom, Coinbase, Facebook and others. Task’s sales recently accelerated to 64% YoY growth, and the company will likely continue to grow strongly going forward. However, Task has high exposure to risky contracts which diminishes the quality of recently reported growth, which is keeping us on the sidelines for now.

Task’s opportunity

Task focuses on providing non-voice customer service, content moderation, and annotations/transcriptions to companies in the digital economy. Task’s employees are primarily based in the Philippines and provide the operational infrastructure for its US-based digital customers.

Task explains that there is demand for its services because technology companies are focused on new products and services and “often lack the desire, expertise, scale and/or geographic presence to build the operational infrastructure to support their growth”. The company claims that since it was founded just 12 years ago, it “grew up” in the cloud environment, which allowed Task to enter the market without investing in expensive, legacy infrastructure.

The company’s model is also highly profitable and produces strong cashflows. Task was profitable throughout 2020 and sales have grown sequentially every quarter since at least Q3 2019 (earliest date of public info). This high profitability and robust cashflow generation are due to the company’s focus on non-voice, digital channels, which accounted for 94% of 2020 revenues. Task explained in its S-1 that non-voice channels allow the company to utilize resources more efficiently, driving higher profitability. 

Task’s recent results and outlook

Task’s three sources of revenue are Digital Customer Service, Content Security (Moderation) and A.I. operations. Digital Customer Service provides customer care services through non-voice channels. This is Task’s largest segment and grew 64% YoY and accounted for 62% of Q3 sales.

Content Security (Moderation) deals with misinformation, offensive content, and critical policy issues. This segment experienced strong demand in 2020 during the election cycle and sales grew at a CAGR of 157% between 2017 to 2020. Content Security sales growth has since deaccelerated in 2021 and grew 34% YoY in Q3 2021 and accounted for 23% of sales.

Task’s fastest growing segment is A.I operations, which grew 145% YoY to $30 million, or 15% of Q3 sales. A.I. operations consist of data labeling, annotating and transcription services for training AI and ML algorithms. Management explained that demand is being driven by autonomous driving, which requires annotations by humans down to the pixel level.

Task’s aggregate Q3 sales increased 64% YoY to $201 million, which beat by $9 million and represented an acceleration from the 57% and 49% YoY growth rates in Q2 and Q1, respectively. On the Q3 call, management highlighted that wage pressure in the US is “pushing clients to move more quickly to an overseas delivery mode” which is driving demand for Task’s outsourcing services. On a sequential basis, sales increased 12% QoQ, which slightly lagged the 13% QoQ rise in Task’s headcount. It will be important to monitor this trend going forward to ensure that Task is able to improve efficiencies and grow sales faster than headcount growth.

Gross margin declined 251 bps YoY to 44% and adjusted EBITDA margin also declined by 70 bps YoY to 24% but beat management’s initial Q3 guide by 50 bps. The decline in margins was driven by costs related to the IPO and investments in new initiatives (Q3 call). Despite the decline in margins, GAAP net income rose 2% YoY to $12 million and adjusted EPS increased 25% YoY to $0.30/share, which was in-line with the consensus estimate.

Looking forward, management raised their full-year guide and now expect 2021 sales to increase 57% YoY to $749 million, up from the prior guide of 48% YoY growth (at the midpoint). Q4 sales are also expected to grow 55% YoY to $215 million. The Q4 guide for 55% YoY growth was an acceleration from the Q3 guide of 50%, and if Task outperforms its guide similarly in Q4 as it did in Q3, then sales will accelerate in the upcoming quarter.

Task’s recent performance has been strong as sales accelerated, margins remained robust, and guidance suggests a further acceleration in sales growth. However, there are some key risks that investors should be aware of going forward, which I outline in more detail next.

Risks

While Task has strong operational metrics, there are some key concerns that are keeping us on the sidelines for now. For instance, Task’s customer agreements include risky fixed-price contracts, meaning that Task carries the risk of project cost overruns. While not directly disclosed, we can proxy the company’s exposure to fixed price contracts by looking at the balance sheet for unbilled receivables, which are a direct result of fixed-price accounting. Utilizing this approach, it appears that around 34% of Task’s sales are from fixed-price contracts. As the name implies, the contract amount is fixed, which introduces the risk of cost overruns in the future.

Customers prefer fixed-price contracts since it passes the risk of cost overruns onto the contractor. This tradeoff can make it easier for the contractor to win new contracts but increases the risk of losses going forward. Task has a relatively high exposure to fixed-price contracts relative to peers. For instance, Grid Dynamic’s fixed-price exposure was around 10% of total sales, as Grid primarily utilizes low-risk master service agreements instead of risky fixed-price contracts (discussed in more detail above).

Task also has significant fixed expenses, which may be supporting earnings as expenses are temporarily stored on the balance sheet. For instance, Task’s Q3 capex increased 393% YoY to $15 million, or 8% of three-month sales. Furthermore, Task’s net property, plant and equipment (PP&E) has increased 27% YTD to $72 million, or 10% of total assets. This is high relative to Grid, which reported that quarterly capex was just 1% of Q3 sales and that net PP&E was 1% of total assets. A rise in capex and net PP&E may be shielding expenses from the income statement by storing them on the balance sheet, which temporarily improves margins and earnings.

Included in Task’s net PP&E is a risky account called construction in progress (CIP), which increased $9 million YTD to $14 million. Construction in progress is a unique account that is not depreciated until it is placed into service and is usually utilized by project-based companies (i.e. construction companies) and is not typically reported by outsourcing companies. The YTD rise in CIP was material and accounted 9% of YTD adjusted net income. Furthermore, the account increased by $5 million QoQ, which provided an after-tax $0.04 benefit to earnings during the quarter. Absent the sequential rise in CIP, Task would have missed its Q3 EPS estimate.

Another risk is the company’s significant customer concentration. For instance, Facebook (27% of Q3 sales) and DoorDash (11%) accounted for 36% of Q3 sales, however this is down from 44% in the year ago quarter. Task’s top five and top ten customers accounted for 61% and 76% of total sales in Q3, which also improved YoY from 67% and 81% in Q3 2020, respectively.  

Task is growing its topline rapidly due to its exposure to fast-growing technology companies. However, there are risks, such as Task’s exposure to risky contract types and the rise in fixed costs such as construction in progress. The company also has higher customer concentrations relative to Grid. We will be monitoring Task going forward but will likely hold off on initiating a position until its financials de-risk. In the next section, I revisit Fiverr, a marketplace for freelancers that outperformed during 2020.

Fiverr: marketplace for freelancers

The I/O Fund has covered Fiverr in the past, here and here. Below, I will be providing an update on the company’s most recent results and how they compare to other outsourcing firms such as Grid Dynamics and TaskUs. The key takeaway is that while Fiverr has reported strong growth, there are signs that momentum in its business is slowing, leaving us on the sidelines for now.

What differentiates Fiverr from other outsourcing firms is that the firm is a marketplace for freelancers, and Fiverr does not employ the labor that it supplies. The company saw a rapid rise in demand for its marketplace during the covid-pandemic as unemployment surged and remote work took hold, which was an ideal environment for gig workers to capture share. However, engagement on Fiverr’s marketplace has slowed, which is a concerning trend for future growth. I discuss these trends in more detail below.   

Recent results and slowdown in topline growth

As shown below, Fiverr’s sales have grown strongly, especially during 2020, but there are signs that momentum is dissipating. In the latest quarter, Q3 sales increased 42% YoY to $74 million, yet sales declined on a sequential basis by 1%, the first time sales have declined QoQ since Fiverr went public. Moreover, the 42% YoY growth rate represented a deacceleration from the Q2 and Q1 YoY growth rates of 60% and 100%, respectively, and represented the slowest pace of YoY growth since Q2 2019. This trend compares unfavorably to Grid and Task, which have both reported accelerating YoY growth and strong QoQ growth in the most recent quarter.

However, it should be noted that Fiverr outperformed during 2020, so its comparables are tougher than Grid and Task, which struggled during 2020. Nonetheless, markets are forward looking and enterprises having strongly rebounded in 2021, which are Grid’s and Task’s main customer cohort, while small business have struggled post 2020, which is Fiverr’s main customer cohort. The outperformance of enterprise customers relative to small business owners helps explain the divergent growth trends between Fiverr and other outsourcing firms such as Grid and Task.

Looking forward, Fiverr’s Q4 guide implies a topline growth rate of 37% at the mid-point, which would represent the slowest pace of YoY growth in Fiverr’s history as a public company. Nonetheless, while Fiverr’s sales are deaccelerating, the Q4 guide still represents 157% growth from Q4 2019, highlighting the overall strength in both Fiverr’s business and the general outsourcing market.

Continuing down the income statement, Fiverr’s gross margin slightly declined by 10 bps YoY to 83%, yet this is well above other outsourcing firms such as Grid and Task. The different gross margin profiles are due to the fact that Fiverr does not employ the labor it supplies, so its gross margins are mostly related to maintaining its software and marketplace rather than employees.

Adjusted EBITDA margin improved 180 bps YoY to 9.8%, which was well below both Grid’s and Task’s 20%+ adjusted EBITDA margins discussed above. Furthermore, the $3 million YoY increase in Fiverr’s adjusted EBITDA was entirely driven by an $11 million rise in stock-based compensation (SBC), which is a low-quality trend. This is because SBC is still a cost to shareholders, and signals that true profitability has not improved.  Non-GAAP earnings increased by $3 million YoY and non-GAAP EPS of $0.19 beat estimates by $0.17, yet the beat was driven entirely by a rise in SBC, a low-quality trend.

Risks

The biggest risk for Fiverr going forward is the decline in engagement on its platform. According to similarweb.com, Fiverr’s website visits have declined 2% over the last six months, which is a concerning trend that might signal declining demand for its marketplace.

It is noteworthy that Fiverr’s sales and marketing expense as a percentage of TTM sales increased 300 bps YoY to 54% as of the latest quarter. It is concerning to see that engagement has declined despite the relatively higher levels of marketing spend. Fiverr disclosed in it 20-F that lower engagement is a key risk, stating that “if user engagement on our websites declines for any reason, our growth may slow or stall.”

Another risk to Fiverr’s growth, which applies to most gig companies such as Uber, Lyft and DoorDash, is the political headwinds around gig workers being reclassified as employees. Earlier in the year, there were headlines that the Labor Security supported classifying gig workers as employees. This development could reduce demand for gig workers and thus reduce demand.

Furthermore, as outlined in the I/O Fund’s prior analysis of Fiverr, we explained that “Fiverr benefits from high unemployment and low hiring numbers because companies are looking for ways to save money.  If companies need talent on a budget, freelancing becomes the most attractive option.” The rapid improvement in the labor market could reduce demand for gig workers going forward. This trend may be causing a deacceleration in Fiverr’s business.

We chose Fiverr as a momentum play because hiring environments change and the current environment appears to favor outsourcing firms focused on digital transformation. Fiverr will likely continue to grow going forward, but the decline in engagement is a concerning trend that will need to improve before we reenter the name. In the last section, I conclude my discussion with an analysis of valuations and reiterate why we favor Grid in the current environment.

Valuation and conclusion

Below are the market cap and sales and earnings multiples for key outsourcing companies. Grid has been awarded a premium valuation relative to the peer median, yet this appears appropriate given its stronger growth rate and its exposure to highly technical labor, which is in high demand. Task appears relatively cheap compared to peers, based both on sales and earnings multiples. However, Task has exposure to risky fixed-price contracts and relatively higher levels of fixed-costs, which are high risk and warrant a lower multiple. Finally, Fiverr has also been awarded a premium multiple by the market, but this is likely due to its different business model which is primarily software based. Software companies generally receive premium multiples due to their low overhead and ability to quickly scale.

I wanted to cover outsourcing broadly and horizontally because it provides a clearer picture for what we are positioning for and why. The key takeaways are that the digital transformation trend is a massive tailwind that is driving demand for highly technical labor. Furthermore, cloud computing and hybrid work environments set the stage for outsourcing firms to capture share going forward. Grid appears to be best positioned, given its outsized growth and exposure to low-risk contracts. Nonetheless, Task, Fiverr and other outsourcing firms will likely continue to grow strongly as companies look to access talent on a global scale. We favor Grid for the current environment and may decide to add it to the momentum portfolio, but will also be closely watching Task and Fiverr for improvements in their businesses. We will keep you in the loop as we weigh these decisions.

Going into February of 2021, we warned of a selloff based on weakening momentum trending into an important broad market time factor. We used this opportunity to raise some cash and even had a proactive hedge on QQQ at the time. This shift helped us remain competitive with the pros in the growth markets, as did our proactive move out of crypto (and back in again). Even when the growth selloff intensified into late Q1, many were calling for the end of the bull market, as we used that opportunity to add to some quality companies that were getting beaten up. Like many, we underestimated the difficulty in the high beta market, showing mixed results in our additions to high conviction/long-term plays, but we continued to hold the broader thesis that the bull market is not over, as we do today. I do believe we will see a return to 2020-style growth investing before this great bull market is over, but I also believe that we will get a chance to grab shares of targeted companies at lower prices first.

 

That being said, what I do want to point out is the difference between February 2021 and today. The economic environment tends to lend itself towards high beta being in favor or out of favor. Coming out of the 2020 bear market, we saw a large acceleration of economic growth with low inflation – ie, the perfect environment for our world. As inflation picked up in mid Q1 of 2021, growth investing became more difficult. We still saw numerous winners in this field coming out of the February/March selloff, so a more discerning filter really paid off. The reason for this is that we saw inflation and growth accelerating up at the same time. However, today's dip in growth assets and rotation into more deflation protection assets is signaling something different, in my opinion. I believe looking at the growth trends in the macro-environment will help clarify what that is…

 

First up, we'll look at our economic heat map. 

Note the difference between February of 2021 and today. In February, inflation was starting to pick up but wasn't a major concern (according to the bond market). Also, note how we were in the early stages of accelerated growth in the US economy. Now, compare that to today – growth is starting to slow as our models expect the rate of change in this deceleration to start turning soft red in mid-late Q1. With inflation in the red as well, we also are expecting this slowdown to lead to inflation slowing down as well. We are seeing signs of this slowdown globally with China leading the way. Much like we saw a global growth push in 2017, the macro environment is suggesting a global growth slowdown into 2022.

 

Our quadrant analysis, which measures the Rate of Change of both growth and inflation in the US, is supporting this thesis.

The October data is being uploaded soon, but it will have the Oct. numbers closer to the bottom right quadrant – inflation/stagflation. This is a risk-off macro environment where high beta struggles, and we see a flight to real estate, utilities, staples, long-duration bonds, and very liquid mega-cap growth companies (exactly what has been working over the last month). We do believe this quadrant will take us into the lower left quadrant (deflation) in 2022. 

 

With this in mind, think of the FED's position. They waited too long to raise rates so are under the gun to rush this process in order to have ammo to fight the next deep correction. So, with growth slowing, inflation peaking, and the FED in a rush to tighten conditions for the next reflation attempt, probabilities suggest that risk assets will continue to struggle until a floor is found in the economic slow-down.

 

But, don't take our word for it, just look what the bond market is saying.

Remember, long-duration yields are controlled by market expectations of growth and inflation (not the FED). As long-duration yields go down, it's signaling the same slow down our models are showing. The spread between the 10-year yield (not affected by FED policy) and 2-year yield (highly affected by Fed policy) has seen the sharpest flattening since 1994. 

 

Interestingly, this also lines up with our Elliott Wave analysis. I have always been amazed at how using Fibonacci/golden ratio analysis on charts tends to predict future outcomes. Anyone that has been with us for a while knows that we've been talking about 2022 being a tough year based on this analysis alone. We now have a combined macro analysis, coupled with inflation and FED policy catching up to what Elliott Wave was predicting over a year ago.

Trends move in 5 waves in the direction of the predominant trend. this is true in all markets and on all time frames. Since March of 2020, the predominant trend is up. We have a clean 1st and 2nd wave in place (in red). In order to complete the minimum targets for the 3rd wave, we need the current trend to move us into the SPX 4900-5200 region. This will lead to the larger 4th wave, which we expect to be between a 10%-15% drawdown in early/mid-2022. We also have two important broad market time factors coming up – Dec 27-Jan 3 and one in late January (I'll narrow down the focus as we approach this). Almost every chart I map is showing the same inflection points.

 

These time factors are marking inflection points and how the market trends into them will be crucial. Interestingly, they are also lining up with what our models are suggesting – an acceleration of the global economic slowdown in late Q1 (remember, markets are looking into the future, so expect a top prior to the data coming out). Furthermore, if we look at inter-market signals, a similar warning is still present.

Our risk-on and economically sensitive sectors are diverging from the broad market. When we see this pattern, only one of 2 things will happen: (1) either the broad market is leading the risk-on sectors; (2) the risk-on sectors are leading the broad markets. With the preponderance of evidence discussed, we believe 2 is more likely as we push into early 2022. 

Why This is Not the End of the Great Bull Market

In one word – liquidity. I've talked about this for several weeks, in various ways. Last week we showed the strange phenomenon that money market funds and the S&P 500 are both close to ATHs, which is one of many examples of the excess liquidity in the markets. This week, we will focus on how the flood of liquidity in the economy is affecting the banking system.

 

I, for one, do believe the central banking system will go down in history as a monumental failure. However, you have to play the market you are dealt – not the one that you want or think makes more sense. Many pundits have stubbornly ignored this reality, and as a result, missed out on one of the greatest bull markets in US history. That being said, this is a FED-driven, liquidity-fueled bull market, and the drastic actions taken by the FED (and global central banks) during the COVID crash last year is why we are continuing to push higher today.

 

In 2008, we saw a gradual tightening that ultimately put in motion the GFC. During this time we saw the first run on banks (and money markets) since the turn of the century. The fear was sudden and caused banks to shut the borrowing window for even qualified companies to refinance their debts. As a result, defaults intensified, and only exacerbated the problem. This liquidity/credit crunch was the primary driver behind the pain experienced in the GFC. 

 

Compared to 2007/2008, last year the FED flooded the economy with excess liquidity that was needed by struggling companies to stay afloat. In other words, they guaranteed that a credit crunch would not happen. this lead to the bizarro world of monthly economic data that literally went off the charts in a negative way, while the stock market kept powering higher. As a result, it's hard to look at the current landscape and see any hint of a liquidity crunch. In fact, the opposite seems to be true. 

 

Liquidity in Banks

 

We tend to see the beginning stages of a liquidity crunch in overnight markets. In other words, if a bank needs additional liquidity to meet their normal business demands, they will borrow this cash from other banks in the overnight markets. If these private banks see trouble on the horizon they tend to not want to participate in overnight loans, pushing the overnight borrowing rate up, and thus making liquidity even harder to come by. This lack of overnight liquidity begins to filter into the banking system and ultimately down into the economy. To combat this free-market phenomenon, the FED created the repurchase agreement program (Repos) to be the lender of last result. In other words, when no private bank wants to take on the counter-party risk of another bank needing cash to operate, the FED steps in.

Prior to the economic slowdowns, we tend to see the FED's repo program tick up, signaling that private banks are not willing to loan into the overnight markets. 

Note how these repo agreements preceded major market events. It's a signal that liquidity is drying up, and this removal of liquidity from the market by the banking system is what ultimately crashes the markets. 

 

Now, let's overlay this metric with inflation expectations (10 yr. breakeven rate) as well as economic growth (PMI – manufacturing).

Note how inflation peaked as did economic growth going into these liquidity crunches. What preceded was sharp bear markets in both instances. Now, note the 2011 and 2015 deep correction periods. We had a deceleration in both economic growth and inflation (much like we're projecting going into 2022), but there was no liquidity crunch. The result was either a sharp and quick selloff (2011), or a correction that was more in time than price (2015) before the bull market resumed. It's important to see that we are not in a similar liquidity crunch today. In fact, the opposite seems to be happening.

 

The FED also has a program designed to address this opposite problem banks can face, and it's called reverse-repo agreements (Reverse-Repos). Banks operate under strict regulations, one of which addresses the maximum amount of cash/credit they are allowed to hold relative to assets. If a bank is close to that limit, it will lend this excess liquidity to other banks for a small overnight rate. However, what happens when most banks are flushed with cash and cannot accept anymore? The FED steps in to be the buyer of last result, exchanging cash for bonds at a minuscule rate. 

 

Today, the reverse repo operation is at record highs. 

In other words, the FED is "borrowing" excess cash from banks on a nightly basis. This is what too much liquidity in the banking system looks like. Now, factor in that the FED is STILL adding ~$60 Billion of QE per month on top of the chart above! 

So, like 2019 and 2008, the economy is starting to slow as inflation is likely about to peak. However, unlike 2008 and 2019 there is tons of liquidity in the system that has to go somewhere, and with the CPI running at 6.8% and likely to move into the 7% region, money markets and bonds are not very attractive. For this single, yet crucial reason, I see 2022's volatility to be more like 2011/2015 – an extended road bump in a much larger bull market, which we want to prepare our readers for.

Remember, the FED and global governments need inflation. It is literally the only way out of the massive debt obligations taken on by governments, aside from a default. the same problem was seen in the post-war 1940's – large increase in money supply as inflation ran hot for over a decade. This allowed the US government to get out from under their war debts, while the stock market went on a multi-decade bull market. This, I believe, is what the FED is trying to do today – get everyone used to higher inflation, so that the US has a shot of unwinding its debt. So, creating a market crash would not be conducive to this end. 

 

Our Game Plan for 2022 and Beyond

 

Going back to our Elliott Wave analysis, if we are completing the 4th wave of the larger 3rd wave, that means that we have the final 5th wave to go. If everything moves as planned, which is always a BIG if, we will bottom into late December, and rally hard into the late January time factor, which is targeting the 4900 regions, at a minimum. We also know that we are in a stagflation/inflation style macro environment, which tends to favor mega-cap growth names. The type of mega-cap growth that is working right now is FAANG and semis. Since our recent entries in NVDA and AMD, we are sitting on 60% gains in just under 2 months, as LRCX and MRVL were almost at ATHs while the rest of tech sold off hard. You are seeing where the money is flowing in real-time. We plan to continue to ride this momentum as the SPX makes its minor 5th wave push, focusing on a potential MSFT play and continued adds to our semis. 

Regarding high beta, we are trending down into the first-time factor, which, as previously stated, I'm expecting to be a low. This should kick off the final minor 5th wave push. Also, even though we are focusing on mega-caps and semis, I do not think risk assets are being completely left for dead on this push, as of now. 

The options market is not as bearish on risk assets now as the popular narrative/Twitter suggests. Note how ARKK is showing negative implied volatility on a 30-day basis based on their recent realized volatility. When this pattern is present at market/asset highs, it's a big warning. However, when we see this closer to market/asset lows, it's a signal that the options market believes the volatility in this asset is likely to reverse. 

 

So, over the short term, our plan is to continue to rotate into lower beta positions, ride any momentum of our higher beta plays (CFLT, VYGVF, ASAN, AFRM, etc), and then build cash along the way. 

Over the longer-term time frame, we plan to have a reasonable cash position to accumulate shares of great companies for when the economic environment clears up and the bull market resumes. Our targeted themes will continue to be semis, as Beth's long-term thesis is starting to really pick up, as well as Cloud. As growth slows and rates move higher, borrowing costs will be affected as will revenue streams. In a slow-growth environment, we tend to see companies address margins in a much more aggressive fashion. We saw this in 2020 regarding cloud. The cloud migration intensified because companies saw that cutting costly IT infrastructure budgets and moving to cloud systems both reduced overhead and made their business more efficient. We believe the same will happen in the current environment. 

 

Arguably, the most cost-effective microtrend is the work-from-home trend. This allows companies to reduce COL allowances as well as expensive real estate. This trend is here to stay, and we are only in the early stages. For this reason, we like plays like ASAN, ZM, MNDY, etc. Regarding Zoom, it showed a 94% enterprise growth on top of triple-digit comps. Yet, the market continuously beats it down as a consumer play that peaked during COVID. This is a mistake and allows for a great chance to buy a quality company at low relative valuations. Though we did not expect the market to beat it down this much, we do believe a relative floor is under Zoom based on its current price to forward growth projections. 

 

Look for ideas like this as we move into 2022. If/when we start approaching any bottom from the coming correction we will shift our focus to more high beat/speculative plays. But, for now, defense is warranted for anyone looking to not be shocked by 2022. As always, we developed a thesis based on incoming data. If that data changes, so will our thesis and pivot. We are not looking to be right, only make money. Markets are fluid, and thus require the need for pivots If anything changes, you will be the first to know. 

Also, it's important to not overreact and sell everything or buy everything in an emotional rush. Instead, we prefer to tilt our portfolio into a defensive/offensive posture. We will target 10-15% cash if our plan for a bounce manifests.