Snowflake (NYSE: SNOW) reported second-quarter fiscal 2027 results after the close on Tuesday, September 2, 2026, and the print was the clearest evidence yet that the company’s AI product cycle is translating into consumption on the data platform: product revenue of $1.491 billion, up 37% year over year, remaining performance obligations of $9.00 billion, up 30%, and a full-year outlook lifted by $230 million at the top line and 100 basis points on operating margin. Shares initially rallied more than 20% in after-hours trading before giving back part of the pop into Wednesday’s open as a fresh leg higher in Treasury yields pressured long-duration growth names across the tape.
The report matters beyond Snowflake for two reasons. It is one of the first clean data points on whether enterprise AI spending is expanding beyond model training into the data-preparation, retrieval, and inference workloads that Snowflake sells. And it is a rare guide raise into a Q3 in which the sell-side had been quietly worrying that consumption growth might decelerate as customers digest recent price changes.
The beat vs. Snowflake’s own guide
Because Snowflake’s consumption model makes near-term revenue a function of customer usage rather than committed contract renewals, the cleanest way to size the print is versus what management told investors 90 days ago on the Q1 call. Every number in the “prior guide” column below comes from Snowflake’s May 21, 2026 Q1 FY27 press release; every actual comes from the September 2, 2026 Q2 release.
| Q2 FY27 metric | Prior guide (May 21) | Actual (Sep 2) | Beat |
|---|---|---|---|
| Product revenue ($M) | $1,435 – $1,440 | $1,491.9 | +$54.4M vs. mid |
| Product revenue growth YoY | 32% – 33% | 37% | +~450 bps |
| Non-GAAP operating margin | ~11% | 15.3% | +~430 bps |
| Non-GAAP diluted EPS | n/a (implied ~$0.45) | $0.62 | +~$0.17 |
Product revenue printed roughly $52 million above the top of the range. Growth accelerated four points sequentially, from 33% in Q1 to 37% in Q2 — the first time Snowflake has re-accelerated product revenue for two consecutive quarters since fiscal 2024, according to figures compiled from company disclosures. The non-GAAP operating margin came in more than 400 basis points ahead of the guide, evidence that operating leverage is showing up even as Snowflake continues to invest heavily in Cortex AI, Snowflake Intelligence, and its Marketplace listings.
Backlog, retention, and customer mix
The revenue beat matters less than the shape of the backlog and the retention profile for a subscription-plus-consumption business. Both moved the right way.
| Metric | Q2 FY26 | Q2 FY27 | YoY change |
|---|---|---|---|
| Remaining Performance Obligations | $6.92B | $9.00B | +30% |
| Net Revenue Retention | 124% | 126% | +2 pts |
| Customers with > $1M TTM product revenue | 652 | 828 | +27% |
| Forbes Global 2000 customers | ~770 | 829 | +~8% |
The $1M-plus customer cohort is the most watched slice of Snowflake’s book because those accounts contribute a disproportionate share of platform consumption. Adding 176 customers to that cohort year over year is the fastest absolute expansion Snowflake has disclosed since the company began publishing the metric. Net revenue retention of 126% is up two points sequentially from Q1’s 124% — the first sequential rise in more than two years.
Product revenue trajectory
The AI flywheel commentary
On the earnings call, CEO Sridhar Ramaswamy said “AI continues to compound our advantages, creating a flywheel effect,” framing the beat as a function of two dynamics. First, more customer data landing on the platform because AI features (Cortex AI, Snowflake Intelligence, Cortex Sense, Cortex AI Gateway) require the data to be governed and ready. Second, more compute against that data because those AI features run inference natively inside Snowflake compute credits rather than through an outside serving stack.
The company did not disclose an AI-specific revenue number, but it did report that consumption from customers using Cortex products was among the fastest-growing segments in the quarter. That echoes what Databricks and MongoDB have said in recent quarters — enterprises are consolidating data-plus-inference workloads on a smaller number of platforms.
Raised guidance
Snowflake lifted both the Q3 outlook and the full-year FY27 guide.
| Metric | Prior FY27 guide | New FY27 guide | Change |
|---|---|---|---|
| Product revenue ($M) | $5,840 | $6,070 | +$230M |
| Product revenue growth YoY | 31% | 36% | +5 pts |
| Non-GAAP operating margin | 13.5% | 14.5% | +100 bps |
| Non-GAAP FCF margin | ~22% | 23.0% | +~100 bps |
For Q3 FY27 specifically, Snowflake guided product revenue of $1,588–$1,593 million, implying 37–38% year-over-year growth — a rare quarter of guidance that implies re-acceleration for a third consecutive period. The Q3 non-GAAP operating margin guide is 15.5%, above the 15.3% actual print in Q2.
The bond-yield overhang
Despite the print, SNOW gave back a portion of its after-hours gains on Wednesday as the U.S. 10-year Treasury yield climbed to its highest level since 2023, roughly 4.60%. Rising long-end yields typically compress the multiples of unprofitable-on-a-GAAP-basis growth names most, and Snowflake still runs a large GAAP operating loss — $263 million in Q2 — driven primarily by $456 million of stock-based compensation expense.
The GAAP-vs-non-GAAP gap is not new for Snowflake, but it becomes more relevant when the discount rate rises. Investors weighing the print into Wednesday’s open faced two competing signals: the fundamentals accelerated, and the multiple on those fundamentals compressed. The initial 20%+ after-hours reaction reflected the former; the fade reflected the latter.
What to watch
- Cortex disclosures. Snowflake has not yet broken out an AI-specific revenue line. Analysts will press management for a percent-of-consumption disclosure on the next call.
- Sequential NRR. The first sequential uptick in more than two years is meaningful. Watch whether Q3 holds above 125%.
- Free-cash-flow margin. The new 23% FY27 FCF-margin guide implies roughly $1.4B of free cash flow. That is the number that anchors any valuation model based on cash rather than reported earnings.
- Bond yields. If the 10-year keeps grinding higher, SNOW’s multiple compression will continue to fight the fundamentals. Watch the ratio of SNOW to the iShares 20+ Year Treasury Bond ETF (TLT) as a rough gauge of when the tape is finally rewarding the earnings acceleration.
Sources
- Snowflake Q2 FY27 press release, September 2, 2026 — investors.snowflake.com
- Snowflake Q1 FY27 press release, May 21, 2026 — investors.snowflake.com
- 10-year Treasury yield chart — Yahoo Finance ^TNX
- Databricks newsroom — databricks.com
- MongoDB investor relations — ir.mongodb.com
Disclosure: This article is for informational purposes only and is not investment advice.