Remember when everyone wrote off Snowflake during the 2023 cloud optimization panic? Yeah, that aged about as well as a carton of milk in July. The company that people dismissed as just a "data warehouse" has quietly transformed into something way bigger — the infrastructure layer for enterprise AI. And the numbers are absolutely wild.

We're talking $5 billion in trailing twelve-month revenue. A record $400 million-plus deal. 9,100+ customer accounts running AI workloads on the platform. This isn't the same Snowflake that went public back in 2020. This is an entirely different animal.

Let's start with the headline numbers from the latest quarter — Q2 FY2026, ended April 30. Revenue hit $1.39 billion, up 33.5% year over year. That's accelerating from the full-year FY2026 growth rate of 29.2%. At a time when most enterprise software companies are begging for 15% growth, Snowflake is out here doing 33% at a $5 billion run rate.

But the top line is just the appetizer. The real story is in the forward-looking metrics. Remaining Performance Obligations — the contracted revenue that hasn't hit the income statement yet — blew past $9.8 billion, up 42% year over year. That's 2.1 times trailing product revenue. In plain English: Snowflake has nearly two years' worth of revenue already locked in at current run rates. Not bad for a company that supposedly had a "demand problem."

Net Revenue Retention at 125% means every dollar a customer spent last year is now $1.25. That's elite territory by any standard. And it's not being driven by a few whales — the customer expansion data is staggering. Snowflake now has 733 customers spending over $1 million annually, up 27% year over year. The cohort spending over $10 million? 56 customers, up a jaw-dropping 56%. Enterprise land-and-expand is alive and well.

So what's driving all this? AI. But not in the way you might think. Snowflake isn't trying to build a better chatbot. Their bet is that the data layer — governed, catalogued, secure enterprise data — is the real AI moat. Models are commodities now. OpenAI, Anthropic, Llama — they're all interchangeable. But your company's proprietary data? That's unique. And whoever owns the platform where that data lives, gets governed, and gets queried by AI agents wins the enterprise AI race.

That's where Cortex AI comes in. It lets customers run LLM inference, vector search, and agentic workflows directly on their Snowflake data without ever moving it. Code ships to data, not the other way around. Then there's Snowflake Intelligence — enterprise AI search and agentic workflows over governed data — which hit 2,500+ accounts in just three months. Fastest product adoption in company history. Period.

Snowpark, the runtime for Python, Java, and Scala execution on Snowflake data, continues to be the bridge between "data warehouse" and "AI platform." And Apache Iceberg support — the open-source table format — is Snowflake's strategic play to become the governance and compute layer even when data lives outside its proprietary format. Counterintuitive? Sure. But it's working.

In total, Snowflake launched over 430 product capabilities in FY2026. That's more than one a day. Cortex Code lets you build data pipelines with natural language. Snowflake Postgres brings transactional workloads onto the platform. And the acquisitions of Observe Inc. and TensorStax are pushing the company into observability and AI-driven data engineering. The platform is expanding faster than most people can track.

Financially, the model is hitting its stride. Gross margin sits at 67.2%, up from 66.5% last year. Non-GAAP operating margin hit 10.5% and management is guiding to 12.5% for FY2027. Operating cash flow hit $1.22 billion for the year, with free cash flow at $1.12 billion. That's a company generating serious cash while still investing heavily — $1.97 billion in AI R&D — in the next wave.

Of course, nothing's perfect. Databricks is breathing down Snowflake's neck — the existential rival, competing head-to-head on data engineering, ML workloads, and the whole lakehouse narrative. The consumption model means macro weakness could compress revenue faster than a subscription model would. And GAAP losses of $1.33 billion, driven partly by $1.6 billion in stock-based compensation, mean traditional investors will have questions.

But here's the thing: 48 out of 48 analysts covering Snowflake rate it a Strong Buy. The largest deal in company history — over $400 million total contract value — closed in Q4. And with 9,100+ AI customer accounts already live, the flywheel is spinning faster than ever.

The data cloud isn't just a place to store your spreadsheets anymore. It's the control plane for enterprise AI. And Snowflake is running the show.

Disclosure: The Signal holds no position in SNOW. Positions may change. This is not financial advice.