Snowflake's 22% jump is a bet that AI is finally feeding the data platform

Snowflake closed the regular session down 4.4%, near its 52-week high at a premium valuation, then shot 22% higher after hours when it reported fiscal second-quarter results. On the face of it, this was a clean beat: revenue of $1.55 billion against roughly $1.48 billion expected, and adjusted earnings of 62 cents a share versus 45 cents. Good quarter, good reaction. But the size of the move tells you the market entered this print scared of something, and the beat alone does not explain it.

The market's fear was never the current quarter. It was whether the AI story Snowflake has been selling all year is durable — whether its artificial-intelligence features actually accelerate spending on the platform or simply get bypassed by cheaper, faster AI tools. That is the question the guidance raise answered, and it answered it more forcefully than the Street was pricing.

The number that did the work is the full-year guidance. Snowflake raised its fiscal 2027 product revenue outlook to $6.07 billion from $5.84 billion, representing roughly 36% year-over-year growth, up from 31% before. Analysts going into the print expected only a nudge into the 32%-33% range. A company that was guiding 27% growth as recently as January has now raised its target twice in two quarters — a ~$410 million cumulative lift — and each raise has outstripped the quarterly beat that set it up.

Raised guidance of that size is a signal about demand, not a bookkeeping artifact. Product revenue came in at $1.49 billion, up 37%, and the company is guiding next quarter's product revenue to $1.588-$1.593 billion, another 37%-38% jump. When the growth number for the year moves up by more than the quarterly results can account for, management is telling you the momentum is broad, not a one-quarter artifact.

The mechanism behind the acceleration matters more than the multiple. The driver Snowflake keeps pointing to is an AI coding agent it calls CoCo, alongside other AI products, and the reason that is significant is the direction of the trade. The bear case on Snowflake has been "disintermediation": that AI will let companies query their data directly, bypassing the pricier data-warehouse middleman. The bull case, which this quarter supports with data, is that AI embedded on top of governed, trusted data becomes a consumption accelerator — a tool that makes people use more of the platform, not less.

That is the same line between "AI eats the platform" and "AI feeds the platform," and Snowflake's year is evidence for the second reading. In the prior quarter, product revenue grew 34% to $1.33 billion and the stock rose 36% in a single session. Now it has followed that with 37% growth, and the customer-expansion metric, net revenue retention, held at 126%. That retention number is the tell: existing customers are spending 26% more than they did a year ago, rather than shrinking as AI shifts workloads elsewhere.

The cost side is where the new risk sits. Accelerating growth is not free. In the first quarter, Snowflake's product gross margin slipped to 75.1%, down about half a point, because AI products carry lower margins, and management's offset was partly a $6 billion deal to buy bandwidth from AWS. The pattern should be familiar to anyone watching AI infrastructure: the usage grows, but the incremental margin is thinner than legacy workloads, and the company is trading some profitability to secure the capacity to keep growing. That is buildout cost, not failure — until it is not. It is the metric to keep watching alongside the top-line acceleration.

Now the part that should make a would-be buyer pause. The after-hours jump to roughly $373 put Snowflake's market value around $130 billion, or more than 20 times forward product revenue, and pushed the stock above the $331 consensus analyst price target. The whole question this quarter asked — "is the AI acceleration real?" — has been answered in the market's view before an investor has to commit. The Street's own target now sits below the price. That does not make the thesis wrong; it means the market has accepted the AI story and you are paying for it rather than being paid to wait for it.

And remember that after-hours movement is emotion, not proof. The regular session Friday is the first real test of whether the jump was a durable rerating or a gasp.

The checkpoint that validates or breaks this framing is conversion. Watch three things. First, net revenue retention: a sustained push toward 128% would be the strongest re-rating signal, while slippage below 123% would mean the expansion story is cooling. Second, remaining performance obligations — the contracted backlog — which decelerated and missed consensus in the first quarter even as revenue accelerated; that gap is the one place the durability argument can still crack. Third, whether product gross margin stabilizes above 75% as AI products scale. If revenue keeps growing while those three hold, Snowflake's premium is the price of the best-performing data story in software. If the gap between acceleration and conversion widens, the 22% pop is the top of the tape, not the start of the run.