Snowflake Analyst Day 2024 Takeaways: It’s All About Revenue Per Query (NYSE:SNOW)
Investment thesis
snowflake (New York Stock Exchange: Snow) She finds herself in an unfamiliar place these days.
The manufacturer of cloud data services and products has seen a slowdown in the pace of growth in its widely appreciated cloud data product platform of the same name. Significantly from the days of nearly doubling revenues with each passing quarter. Then, in a shock announcement, the company’s former CEO, Frank Slootman, announced his sudden retirement, angering investors. The markets did not take kindly to the recent developments, as the stock has fallen by 34% since the beginning of the year.
The company announced that Sridhar Ramaswamy, founder of Neeva, an LLM-based search technology product startup that was acquired by Snowflake about a year ago, will now lead the company in its next phase. But investors appear to be ignoring the CEO transition as the stock declines, as shown in the chart above.
Compounding Snowflake’s plight was a cybersecurity incident that occurred last week.
There has been a lot for me to digest and analyze over the past few months. Snowflake’s recent FY24 analysis day provided sufficient information about its product roadmap and strategy. The commentary from Analyst Day last week was enough to support my analysis.
I recommend buying Snowflake stock here.
Snowflake’s transition to a true data service enterprise
In my editorial coverage of Snowflake, I recommended a neutral view on the stock and stated so “Snowflake has a lot to prove or lose in its next stage of growth.” Ironically, this happened two weeks before the sudden CEO shift. In that post I said:
With automation and artificial intelligence now taking center stage, every company in the world has reviewed its long-term strategy to stay relevant in the new world. Snowflake appears to be moving towards increasing ways to access data on its platform.”
The Analyst Day presentation seemed to address my previous expectations about increased access to data that I talked about in my previous post.
First, I should reiterate Snowflake’s revenue generation mechanism, which forms the basis of Snowflake’s product strategy outlined at Analyst Day and its expected revenue impact.
Snowflake is in the business of providing cloud-based solutions for collecting customer data, which is made available to the customer on demand and regardless of the consumer’s location or infrastructure. But the important aspect to remember is that a company makes money not only from the volume of data stored, but also from the volume of data that is queried or transferred. The more data Snowflake servers consume, the more revenue Snowflake generates from its customers. Here is an excerpt from the most recent 10-K:
We generate the vast majority of our revenue from fees charged to our customers based on the compute, storage and data transfer resources consumed on our platform as a single integrated offering. For computing resources, consumption charges depend on the type of computing resource used, the duration of use, or, for some features, the volume of data processed. For storage resources, consumption fees are based on the average terabyte per month of all customer data stored in our platform. For data transfer resources, consumption charges depend on the number of terabytes of data transferred, the public cloud provider used, and the region to and from which the transfer is performed.
Therefore, I find it important for Snowflake to increase the ways for customers to access their data stored on Snowflake.
So far, as Snowflake wins customers from legacy database systems, customers will mostly focus on cloud migration-type workloads — moving data from their legacy on-premise systems to Snowflake’s cloud storage. But Snowflake’s product lineup for FY25 looks very encouraging to me, considering its consumption revenue model.
What really impressed me was some of management’s comments about how products are designed around data consumption. Products such as Cortex AI, Iceberg, Unstructured Data, and Snowpark, as shown in Figure A above, are all designed for heavy query-type workloads, resulting in increased usage volume, while others are designed for lighter queries, expanding the dimensionality of use cases that can Through which we use customer data on Snowflake. I wouldn’t be surprised if the company’s industry-leading net retention rates (NRR) start to move back up as the company’s products move through the adoption curve of their product life cycles.
So far, the workload types on Snowflake have been mostly dominated by cloud migration workloads as the cloud data company continues to attract customers from legacy database software. But with the launch of the new Snowflake products, I believe Snowflake is expanding the range of use cases and adding options for customers to run a broader range of workloads, especially those oriented around AI, applications, and collaborative workloads.
To me, this suggests that the Land-and-Expand model can actually start to deliver higher results for the company, as the revenue-per-query strategy is set to pave the way for stronger consumption revenue growth. I also note that the current product roadmap is very much in line with the new CEO’s vision. The new CEO is a former Google executive who also founded his own search company, Neeva, based on GenAI, which was acquired by Snowflake last year.
To date, Snowflake has shown a CAGR of over 50% in revenue, as shown in Figure C below. But as the company scales, I expect this growth to return to around 25-26% CAGR over the next three years, which is slightly higher than the 24% CAGR I originally assumed based on my previous coverage of Snowflake.
I believe increased product adoption and expected rise in NRR should drive increased consumption and ultimately benefit the company as it grows at these rates.
TAM is increasing despite recent talk of a slowdown in software spending
Recent comments from cloud software companies like Salesforce (CRM) and MongoDB (MDB) added pressure to Snowflake shares in May, as the narrative of slowing enterprise cloud spending added to momentum.
At the same time, interestingly, Gartner updated its public cloud spending report just days before Salesforce’s Q1FY25 earnings report. Gartner’s updated H1 2024 cloud spending report indicated marginal gains in 2024 compared to the previous H2 2023 report. In fact, projected SaaS + PaaS spending is expected to remain mostly unchanged when compared to Gartner’s report for the second half of 2023.
Going back to the broader comments from other cloud companies’ earnings reports, I think investors in general may have become too optimistic about cloud companies in the first quarter. At the same time, Q1 generally tends to be a slower quarter for most cloud companies, in my view, and markets have been ignoring the trends given the optimism stemming from GenAI.
Additionally, on Snowflake’s analysis day, management updated its TAM based on the product roadmap and the general availability of these products, as detailed in Exhibit A.
I’ve added the updated TAM for management in Figure D below.
As shown in Figure D above, the company’s total turnover (TAM) is now expected to grow at a CAGR of 17.6% over the next five years. If you compare this to the company’s previous TAM report detailed in the FY23 Analysis Day slide deck (Slide 41), the TAM expanded by 1.9%.
To me, this suggests a larger market footprint for Snowflake, given the broader usage dimensions of its products.
Snowflake price target upgrade
I now have confidence based on management’s product roadmap and the overall market outlook to believe that Snowflake is poised for some upside.
Here are my assumptions about Snowflake:
- As for revenue, I already mentioned my 25-26% CAGR assumption in the previous section. These forecasts have raised my previous assumption of a CAGR of 24% over the long term.
- In terms of adjusted operating margins, management is guiding for adjusted margins to decline to 3% in FY25. This is because the company is cost effective on acquired GPUs, as opposed to cost leveraging as most of its peers do. But with ownership of these GPUs, management has indicated that it aims to deploy them for inference workloads. This one-time hike will lead to a significant margin increase going forward, in my view. According to my estimates, I believe the company will see adjusted operating income grow at 50% CAGR.
- I have assumed an equity dilution CAGR of approximately 2.5%, while discount rates are based on the calculations here.
Based on this, I believe the company now warrants a forward P/E of ~90x, more than half the forward P/E of +200x it trades at today.
This means an increase of at least 20% from current levels.
Risks and other factors to consider
Databricks is the most powerful counterpart to Snowflake. The company reported FY24 sales of $1.6 billion, up 50% year over year. While the startup’s operating leverage profile is unclear, Databricks has an edge over Snowflake only in terms of revenue growth if you compare the 36% revenue growth that Snowflake showed in the same reporting year. This will be a key area to watch moving forward as Databricks will be competing for investor capital while also targeting Snowflake’s target market. The data startup is widely expected to go public soon.
I also want to address the risks posed by the Snowflake hack I mentioned earlier. At this point, with the information I have, I don’t think this matters to Snowflake. The hack appears to be a mistake on the part of a Snowflake customer who did not have their MFA turned on. Google Mandiant also has a diagram on how an attacker could gain unauthorized access to an exposed Snowflake instance via a compromised device of a client’s end-user account. This indicates a vulnerability in the Snowflake customer or in security policies that Snowflake customers should have implemented. In my opinion, this incident does not indicate any lapse in Snowflake’s ending. However, this could impact the stock in the immediate short term, and assuming there are no further updates on this front, I don’t expect any material impact on Snowflake.
NB: Databricks is expected to launch its own products at a product unveiling event in San Francisco from June 10-13 this week.
He stays away
There are several reasons why investors are pessimistic about Snowflake. The company saw sales slow while the previous CEO abruptly left the company, which may have left investors feeling like they were left “high and dry” while holding the stock at relatively higher valuations.
I expect valuation premiums to decline further over time based on my model, but the decline in premiums is expected to be complemented by stable growth at the top line and stronger growth at the bottom line, driven by an impressive product portfolio that requires leveraging the company’s core consumption-based revenue model. .
The insights revealed at Snowflake’s Analyst Day encouraged me to upgrade the stock to Buy.