Snowflake Q2 FY2027: product revenue +37%, and $456M of stock comp

Snowflake sells a consumption-priced data platform — customers pay for the compute, storage and, increasingly, the AI they run on it rather than a fixed seat license — and this quarter the platform did something a decelerating cloud business is not supposed to do: it sped up. Product revenue grew 37% year over year, the third consecutive quarter of accelerating growth, net revenue retention rose to 126%, and management raised the full-year product-revenue target. On the profit line management cares about — non-GAAP operating margin — the company expanded to 15.3% from 11% a year ago while growing faster.

That is the story, and it is a good one. The counterweight, stated as plainly, is that the same three months produced a $263 million GAAP operating loss and a $192 million GAAP net loss, and the entire distance between “profitable” and “loss-making” is one line: $456 million of stock-based compensation, 29% of revenue. Understanding that single number — how it turns a GAAP loss into a non-GAAP profit, and what it costs shareholders — is the difference between reading Snowflake’s headline and understanding its economics.

	Q2 FY27	Consensus
Total revenue	$1.55B	~$1.48B	beat
Non-GAAP EPS (diluted)	$0.62	~$0.45	beat
Product revenue growth	+37%	—	accelerating
Net revenue retention	126%	—	—
FY27 product-revenue guide	$6.07B (+36%)	—	raised from $5.84B

Did Snowflake beat its own guidance in Q2 FY2027?

Yes. Product revenue of $1,491.9 million came in above the top of the $1,415–1,420 million range Snowflake guided to in its May 27, 2026 8-K — about $72 million above it (our arithmetic) — and that guide had implied 30% growth against the 37% delivered. Non-GAAP operating margin of 15.3% was above the 12.5% guided. Snowflake does not guide total revenue or EPS; against the year-ago quarter, total revenue grew 35% to $1,546.8 million, non-GAAP diluted EPS rose to $0.62 from $0.35, and the GAAP loss per share narrowed to $(0.55) from $(0.89).

The top line: acceleration is the headline, and it is real

Start with the number Snowflake tells you to watch. Product revenue was $1,491.9 million, up 37% year over year (from $1,090.5 million). Because Snowflake recognizes revenue on consumption, not on contracted seats, this line is the closest thing it has to a real-time read on how much customers are actually using the platform — and it accelerated for a third straight quarter.

Revenue line	Q2 FY27	Q2 FY26	YoY
Product revenue	$1,491.9M	$1,090.5M	+37%
Professional services & other	$54.9M	$54.5M	+1%
Total revenue	$1,546.8M	$1,145.0M	+35%

The acceleration is visible in the company’s own trajectory: product revenue grew roughly 34% in Q1 FY27 and 37% in Q2, and management called it the third consecutive quarter of acceleration. Two supporting metrics say the same thing from the customer angle. Net revenue retention was 126% — the trailing-two-year cohort of customers spent 26% more than it did a year earlier, meaning existing accounts are expanding, not just new logos filling a leaky bucket. And remaining performance obligations (RPO) reached $9.00 billion, up 30% — contracted future revenue not yet recognized. RPO is more than three times the $2.60 billion of deferred revenue on the balance sheet, which is the shape you want in a consumption model: customers have committed to far more than they have yet been billed for.

Underneath, the customer base widened too: 828 customers now spend more than $1 million on product in a trailing twelve months, up 27%; the company added 692 net new customers (up 32%) to more than 14,500 total; and it counts 829 of the Forbes Global 2000. Management attributes the reacceleration to AI adoption — its CoCo assistant surpassed 9,100 accounts (adding more than 2,000 in the quarter) and CoWork reached 5,800 — the mechanism being that AI features pull new workloads onto the platform, and new workloads are consumption, and consumption is product revenue.

If you own Snowflake for the growth-durability question, this quarter answered it in the affirmative. The rest of this piece is about what that growth costs and what it does not yet produce.

Three ways to read one quarter’s profit

Here is the same quarter read at three levels of the income statement. Every figure is from the company’s own release.

Profit measure	Q2 FY27	Margin	Q2 FY26
GAAP operating income (loss)	−$263.0M	−17.0%	−$340.3M
GAAP net income (loss)	−$191.7M	−12.4%	−$297.9M
Non-GAAP operating income	+$237.0M	+15.3%	+$127.6M
Non-GAAP net income	+$235.3M	+15.2%	+$129.0M

Two of those numbers are losses and two are profits, in the same three months, and the gap between them is almost entirely one item. Work the bridge from the GAAP operating loss to the non-GAAP operating profit:

• Start: GAAP operating loss of −$263.0M
• Add back stock-based compensation-related charges of $456.4M
• Add back amortization of acquired intangibles of $42.2M, plus $1.4M of acquisition costs and rounding
• Arrive: non-GAAP operating income of $237.0M

Read the first add-back twice. The $456.4 million of stock compensation is larger than the entire $237.0 million non-GAAP operating profit it produces — 1.9 times as large. Without it, there is no non-GAAP profit at all; there is a GAAP loss. This is not an accusation of impropriety. GAAP counts stock comp as a real cost because it is one; non-GAAP excludes it because it is non-cash and lumpy; both conventions are legitimate and every software company uses them. But it means the “profitable” Snowflake and the “loss-making” Snowflake are the same company viewed with and without a single $456 million line, and that line is the analysis.

Stock comp is 29% of revenue, and it is the shareholder’s bill

The reason the add-back matters is not accounting aesthetics — it is dilution. Stock compensation is a cost even though no cash leaves the building, because what does leave is a slice of every existing owner’s stake.

Snowflake expensed $456.4 million of stock-based compensation this quarter — 29% of revenue. For scale: that is more than a quarter of every dollar of sales handed to employees in equity, it is nearly double the non-GAAP operating profit, and it is more than five times the quarter’s $83.8 million of free cash flow. The dilution shows up directly in the share count. GAAP weighted-average shares rose to 349.3 million from 335.2 million a year ago — about 4% dilution in a year — and that is after the company spent $300 million buying back stock in the first half. The buyback is running just to slow the treadmill, not to move backward on it.

There is a second place the dilution hides: the gap between the GAAP and non-GAAP share counts. Non-GAAP diluted EPS of $0.62 is calculated on 377.9 million shares, not the 349.3 million in the GAAP loss-per-share line. The extra ~29 million shares are dilutive equity awards (18.3M) and shares issuable on the convertible notes (14.6M), netted against the capped-call hedge — potential ownership that non-GAAP EPS already spreads its profit across. So the non-GAAP profit is both created by adding back stock comp and divided by a share count inflated by that same equity issuance. Neither is wrong; together they are the full price of the growth.

A useful discipline: whenever you read “$0.62 of non-GAAP EPS, up 77%,” hold it next to “$456 million of stock issued to employees this quarter, and 4% more shares outstanding than a year ago.” Both are true. The first is the reward; the second is the bill.

The margin trajectory is the genuinely good news

Set the profit-lens debate aside and there is a real improvement worth crediting: Snowflake is scaling its non-GAAP operating margin while accelerating growth, which is the hard combination.

Non-GAAP operating margin	Q2 FY27	Q2 FY26
Sales & marketing (% of revenue)	32%	34%
Research & development (% of revenue)	20%	22%
General & administrative (% of revenue)	5%	6%
Non-GAAP operating margin	15.3%	11.1%

Every operating-expense line fell as a percentage of revenue year over year — the definition of operating leverage. Management raised the full-year non-GAAP operating-margin guide to 14.5%, from 13.5%, alongside the product-revenue raise, which is the more credible of the two raises: it is easier to guide revenue up in a consumption boom than to promise the margin comes with it.

One honest caveat sits at the gross line. Non-GAAP product gross margin was 74.7%, down from about 76% a year ago, and the full-year guide is 74.0%. The compression is small but directional, and the likely cause is the AI story itself: AI and GPU-heavy workloads carry more compute cost than classic SQL analytics, so the same consumption that drives the top line is modestly heavier to serve. Growth is reaccelerating and costing a little more per dollar at the gross margin — both are true, and the operating-margin expansion above is coming from spending discipline, not from a richer cost of revenue.

Cash flow: strong for the year, thin this quarter, and seasonal

Cash is the measure least sensitive to the GAAP-versus-non-GAAP argument, so it deserves a clear look — including its awkward seasonality.

Cash measure	Q2 FY27	H1 FY27
Operating cash flow	$91.4M	$334.6M
Free cash flow	$83.8M	$316.6M
Adjusted free cash flow	$92.3M	$357.8M
Adjusted FCF margin	6.0%	12.2%

Q2 free cash flow of $83.8 million — a 5.4% margin — looks thin next to the 23% full-year adjusted-FCF margin management reaffirmed, and the reason is billing seasonality, not weakness. Deferred revenue fell $278.6 million in the quarter as the balance built up in the January and April billing-heavy periods was consumed; Snowflake collects much of its cash in Q4 and Q1, then draws it down. Judge the cash on the year, not the quarter: H1 adjusted free cash flow was $357.8 million, and the balance sheet holds $4.33 billion of cash and investments against $2.28 billion of (zero-coupon) convertible notes — roughly $2 billion of net cash. The business is genuinely cash-generative across a year. It is also, at the same time, issuing $456 million of stock a quarter. Both facts are real, and holding only one of them is the mistake.

The bear case, stated fairly

Not “the growth is fake” — it plainly is not. The narrower version:

1. There is still no GAAP profit. Operations lost $263 million and the company lost $192 million on a GAAP basis. The path is clearly toward profitability — the GAAP operating margin improved from −30% to −17% in a year — but it is not there, and the only profit that exists depends on excluding a $456 million cost.
2. Stock comp is 29% of revenue and dilution is ~4% a year. The non-GAAP profit is, in substance, the add-back; the cost is ongoing dilution that a $300 million half-year buyback only partly offsets. This is the single most important thing the $0.62 EPS does not show.
3. Gross margin is drifting down as AI scales. Non-GAAP product gross margin fell ~1.8 points year over year and is guided lower still for the year. If AI workloads are structurally heavier to serve, some of the consumption tailwind is offset at the cost line.

None of that is refuted by the quarter. It is the price of admission to a business reaccelerating to 37% growth while expanding operating margin.

The bull case, from the same numbers

1. Acceleration with retention. 37% product growth, a third straight quarter of acceleration, and 126% net revenue retention together say the platform is expanding inside existing accounts, which is the most durable kind of growth.
2. Operating leverage is real. Every opex line fell as a share of revenue and the full-year margin guide went up, not just the revenue guide.
3. A raised full-year target. Product revenue guided to $6.07 billion (36% growth), up meaningfully from $5.84 billion — management put its forecast behind the reacceleration.
4. A fortress balance sheet. ~$2 billion net cash and real annual free cash flow fund the growth without needing the equity market.

What to check at the Q3 report

Specific and falsifiable, in rough order of importance:

• Product revenue growth against the guide. Q3 is guided to $1,588–1,593 million, 37–38%. Whether the three-quarter acceleration holds or 37% was the peak is the whole thesis.
• Net revenue retention. 126% this quarter; the cleanest single read on whether AI adoption is durably expanding existing customers.
• Stock-based comp as a percent of revenue. 29% now. Whether it trends down as the company scales is the dilution dial — and a stated management goal.
• Non-GAAP product gross margin against the 74.7% print and 74.0% full-year guide — the tell on whether AI workloads are structurally margin-dilutive.
• GAAP operating margin. −17% now, from −30%. Watch how fast it closes toward zero — the moment “profitable” stops needing the add-back.
• RPO and $1M+ customer count — $9.00 billion and 828 — for confirmation the committed pipeline keeps pace with recognized revenue.

The read

Snowflake delivered the quarter a decelerating cloud company cannot: product revenue reaccelerated to 37%, net revenue retention rose to 126%, RPO climbed to $9.00 billion, and non-GAAP operating margin expanded to 15.3% while management raised the full-year product-revenue and margin guides together. If the question is “is the platform healthy and is AI actually pulling new consumption onto it,” the answer this quarter was yes, and the rich multiple is a bet on exactly that.

Read the profit with equal care. The same quarter is a $263 million GAAP operating loss, a $192 million GAAP net loss, and a $237 million non-GAAP operating profit — and the $456 million of stock compensation, 29% of revenue and nearly double that non-GAAP profit, is the entire distance between the two. It is a legitimate add-back and a genuine cost at the same time: legitimate because it is non-cash, a cost because it dilutes owners about 4% a year. The two numbers to watch next quarter are product-revenue growth against the 37–38% guide — the durability of the reacceleration — and stock comp as a percent of revenue, the rate at which the growth is paid for in ownership. The first is why you would own it; the second is what owning it costs per share.

---

Sources and method. Company figures in this piece — total, product, and professional-services revenue and their growth rates; net revenue retention of 126%; remaining performance obligations of $9.00 billion; the 828 customers over $1 million, 829 Forbes Global 2000, 692 net new customers, and 14,500+ total; the CoCo and CoWork account counts; GAAP and non-GAAP product and total gross profit and margins; GAAP operating loss of $263.0 million and non-GAAP operating income of $237.0 million; GAAP net loss of $191.7 million and non-GAAP net income of $235.3 million; GAAP net loss per share of $0.55 and non-GAAP diluted EPS of $0.62; the $456.4 million of stock-based compensation-related charges and the GAAP-to-non-GAAP bridge; the operating-expense percentages; operating cash flow, free cash flow, and adjusted free cash flow with their margins; the deferred-revenue movement, the $4.33 billion cash-and-investments and $2.28 billion convertible-notes balances, the $300 million of first-half buybacks, and the share counts; and all Q3 and full-year FY2027 guidance — are Snowflake’s own disclosures from its Q2 FY2027 results release (Form 8-K, Exhibit 99.1) filed 2026-09-02 for the quarter ended 2026-07-31, reported after the close. Prior-year comparatives are the year-ago columns in that same release. The Q2 FY2027 guidance — product revenue of $1,415–1,420 million (30% growth) and a 12.5% non-GAAP operating margin — comes from Exhibit 99.1 to Snowflake’s Form 8-K filed 2026-05-27. Whether the quarter beat or missed is judged in the guidance section only against the company’s own previously issued guidance (Form 8-K, filed 2026-05-27) and against the year-ago quarter; the third-party estimates cited below are context, not the basis of that verdict.

Not company figures, and attributed accordingly: the consensus estimates (adjusted EPS ~$0.45 and revenue ~$1.48 billion per the LSEG consensus reported by CNBC; Zacks consensus of ~$1.47 billion revenue and ~$0.45 EPS reported via Yahoo Finance). This piece makes no claim about the share-price reaction. The interpretation — that stock compensation is the entire distance between the GAAP loss and the non-GAAP profit, that the dilution is the shareholder’s bill only partly offset by the buyback, that the gross-margin drift is plausibly an AI-workload cost, and that the thin Q2 free cash flow is billing seasonality rather than weakness — is our analysis, not a company statement, and is not a forecast. Derived figures (the ~$72 million by which product revenue exceeded the top of the guided range, year-over-year and sequential growth rates, margins, the SBC-to-profit and SBC-to-FCF ratios, the ~4% dilution, and the net-cash figure) are our arithmetic off the disclosed dollars and tie to the tables in the release.

Related: CrowdStrike Q2 FY2027, the other high-growth security-software name where stock comp exceeds the non-GAAP operating profit and the GAAP-versus-non-GAAP gap is the whole debate; Palantir Q2 2026, where the same profit-lens question meets an even richer multiple; CoreWeave Q2 2026 and Nebius Q2 2026, the AI-infrastructure buildout whose compute Snowflake’s AI workloads increasingly consume.

Tradune shows what you actually hold across every account, and whether a position has beaten a plain index fund since you bought it. Educational analysis, not investment advice.