My current portfolio & performance vs. the S&P 500 can be found here.

37 earnings reviews from this season can be found here.

SNOW trades for 73x FCF & 151x EPS heading into tonight’s report. FCF is set to compound at a 30% clip over the next 2 years while EPS is expected to compound at a 48% clip for the next 2 years. Estimates are going to meaningfully rise while forward multiples use estimates another quarter into the future. That will help control the multiple expansion, but the stock is also up 20% after-hours and I do still think the forward multiple will expand modestly from here when taking everything together.

AI Positioning:

The vast majority of the call was spent working through Snowflake’s AI positioning and why this new age is so positive for the company’s overall business. While it’s always nice to hear that, it’s a lot more real when it’s accompanied by a sharp revenue growth acceleration. That’s what SNOW is currently delivering, as FY 2027 growth is set to accelerate to 36% compared to 29% last year. 

This impressive rise in growth rates isn’t based on anything weird, inorganic or one-off. There’s no massive deal with OpenAI or Anthropic temporarily propping up growth rates. This is a structural, broad-based, diverse acceleration that should persist for the foreseeable future. And it’s a byproduct of the elite value proposition Snowflake provides customers. Snowflake provides access to required data alongside integrated apps and models all from a single platform. As we hear virtually every company shift to mixing and matching cheaper models when possible and see model leaderboards remain a game of leap frog, it’s becoming increasingly clear that models are amazing technology, but not the source of differentiation. As I’ve been arguing since the “SaaSpocalypse” began, it’s the experience from scaled distribution. It’s the ability to harness models with this data and great applications to ensure they deliver reliably certain value at scale. It’s the edge case mastery & the intimate client understanding that can only be learned through years of hard work. It’s the data.

It blends complete and vendor-agnostic model choice with an AI gateway that routes each task to the right model based on the customer's own cost and performance rules. Together, these two features ensure the best model for the price is being used for each piece of agent-based work rather than one expensive model for everything. And to build on this, Snowflake lets customers post-train models on their own data and business context, which makes outputs more accurate and more relevant to that specific enterprise's goals.

From there, customers can enjoy two model harnesses or agents that help turn raw model intelligence and potential into reliable and certain outcomes at enterprise scale. This is done with structured governance and permission frameworks to make sure agents aren’t going rogue and that companies more quickly know if that’s happening. Cortex Code (CoCo) is its coding agent while CoWork (used to be Snowflake Intelligence) is positioned as an agentic assistant for knowledge workers to automate data analysis and other work while taking action on a user’s behalf. That action-taking runs through Natoma, the model context protocol (MCP) platform Snowflake bought at Summit, which lets these agents reach into outside tools like Slack.

It means the breadth of work that can be done is uniquely compelling. Just like Snowflake leverages its data custody to help customers make LLMs work better for them, this same data also makes agents better at doing work in desirable and compliant ways. That helps make both of these products best-in-class and is supporting impressive ramps as the agents complete SNOW’s end-to-end AI data platform. CoCo jumped from 7,100 accounts to 9,100 in just one quarter while CoWork moved from 5,200 to 5,800.

These two agents paired with all of the other tools SNOW brings to the table make the company confident in becoming a company’s agent control plane for data-focused agents. It already has the raw context these agents need to actually provide value and an ability to let customers each conversationally operate them with powerful guardrails to help easily optimize costs without any heavy lifting. Between CoCo, CoWork and the platform's agent observability tools, customers can build what they need, see what 1st and 3rd-party agents are doing and conversationally power complex task automation. They're an intuitive choice.

Sustainability & Health of This Acceleration:

A few years ago, Snowflake got into a tough spot when cloud-native darlings began consuming its platform in irrational ways and at unsustainable clips. This led to extremely difficult growth comps and people are worried that issues might recur as customers scramble to adopt AI. Snowflake isn’t letting that happen. Through tools like gateway and a revamped sales culture that praises helping customers use SNOW efficiently, they’re proactively making sure all usage is good usage. There is no unhealthy growth sugar high playing out like there was in 2021.

Why Snowflake is Winning:

Aside from the compelling AI product bundle, Snowflake is winning because CEO Sridhar Ramaswamy has the overall innovation engine in a much better place. I’ve been praising him for this for nearly 2 years now, and I think these last two quarters have been the coming out party in terms of great product work translating into great financial trends. Product capabilities launched are up 35% Y/Y so far this year and total use cases deployed are up 89% Y/Y. Average use cases won per account executive is also up 43% Y/Y, as clients such as Indeed discover that using more of Snowflake leads to lower overall costs.

4 Specific Ways AI is Helping SNOW’s Business:

Snowflake took a bit longer than Palantir, but they’ve emerged as an app layer company obviously showing that AI is a clear business tailwind. Specifically, there are 4 AI tailwinds helping SNOW’s business that continue to rage. 

First, AI accelerated the pace of workload migration to modern platforms like SNOW. Data modernization is a prerequisite for successfully embracing AI, so the need to keep up with the times creates more urgency to make the move.  And the agents are now doing a lot of the migration work themselves, so the projects that used to take years take quarters.

Second, CoCo and CoWork are already material revenue drivers and were the main sources of AI’s contribution to company growth acceleration. Because they’re so easy to use, winning accounts with either agent also leads to a jump in the number of customer employees actively using SNOW. This makes the company stickier and invites more consumption too. 

And third, when customers use Snowflake’s AI tools, they end up using a lot more of the core too. If we think about it, an AI agent completing exponentially more work across data analytics use cases in sales, finance, marketing is great for the core business. The amazing and automated jobs these agents do rely heavily on the same tools and data that Snowflake has provided for years. Separately, this activity also produces a lot more data, which feeds demand in another way. While agent work looks magical, their processes mirror humans. They just do what we can do far more quickly and perpetually, which is a good thing for Snowflake’s consumption-based model.

Snowflake actually only talked about 3 tailwinds in this part of the call, but I wanted to add another. Snowflake is wholeheartedly infusing its AI innovation into as much of its own operations as possible. They’ve used it to build search optimization tools that help them cut 3rd party marketing costs, reduce tedious finance team work and automate 70% of initial outreach emails for its sales team.

This was an elite quarter. It shows exactly how well SNOW is positioned to capitalize on AI as a tailwind and exactly how well it's executing. As I’ve said in this section of my Snowflake reviews for over a year, Ramaswamy is a superstar. Snowflake had lost its way for a while before he took over. The company was flailing on innovation efficacy, speed and traction and was beginning to morph from a disruptive, trailblazing darling to a bit of a fading dinosaur. Databricks was eating its lunch and there were real questions about growth curve longevity.

They needed to bring in someone to kick product improvements and new category expansions into high gear. They needed someone with a vision to build a holistic platform that would thrive as AI took over and that leveraged Snowflake’s precious and massive data treasure chest to drive differentiation. That is now happening, as evidenced by a 7 point acceleration in Y/Y revenue growth at large scale based on its updated annual guide (which they’ll beat). To sum it up, the company is in a great spot, with its core business thriving and every single emerging product enjoying traction as hoped for. My… what a difference a world-class leader can make. I have a ton of respect for Frank Slootman (who Ramaswamy replaced), but this was a CEO upgrade that changed the increasingly murky path of this company for the better.

The only issue with this company right now is the valuation, and I understand why some are baffled by the 20%+ move despite entering the print at already lofty multiples. So what gives? Some companies are beat-and-raise machines. Through consistent execution and drama-less outperformance, they create a sense of reliability with investors and analysts. Over time, this pattern leads people to assume more outperformance is clearly on the way. That puts a ton of pressure on business models to consistently succeed, which is tough. But? Some can handle the pressure and I think Snowflake under this team is one of those companies. They delivered a quarter that will foster large positive estimate revisions. And they delivered a quarter that lets everyone assume a 3%-5% top-line beat with large bottom-line outperformance will come again in Q3 and Q4. It makes everyone comfortable with guessing that the forward estimates are too low and the true valuation is cheaper than it looks. Who knows if this pop will hold, but I wanted to explain why I don’t find it irrational.

While I love the company, there is a chance that I will be trimming in the near future due to hefty multiple expansion. I will keep you posted and update if anything happens.

Here is a 40 minute YouTube episode I did dissecting Zscaler’s current fundamental challenges and where we might go from here.