Leadership1 distinct publisher3 min readPublished
Snowflake's quarter and its raised full-year target are both stronger than a year ago, yet the transcript puts no dollar figure on the AI attribution that anyone renewing a warehouse contract is being asked to pay for.
The Board Room · Leadership desk

Compiled by The Board RoomSomething wrong?How this is made
Start with the arithmetic inside the guide. A raised fiscal 2027 product revenue target of $6.07 billion at 36% growth implies a prior fiscal year of roughly $4.46 billion [5][14], and it sets the full-year rate one percentage point below the 37% the company just reported for the quarter [15]. Management described growth as accelerating over the past two quarters [1]. The figure it is willing to underwrite for the year does not assume that acceleration keeps going, and for anyone modeling a multi-year consumption commitment, that is the vendor's own stated view of its trajectory.
The AI attribution belongs to management, not to the disclosed numbers. Coco and Cowork are described as rapidly adopted and contributing significantly to revenue growth [7], and the call described a structural multiplier effect as customers put AI into operations [8]. No dollar figure for AI-attributable consumption appears in the transcript summary or the opening remarks, and neither does a bookings or remaining-performance-obligation number [20]. Recognized consumption revenue is not the same thing as contracted commitment, so the defensible reading is narrower than the narrative: revenue and margin both moved up, and the company says AI is the reason.
A skeptic would say the acceleration is a repricing of work customers were already doing, now carrying a Cortex label. The available counter is the margin line. Non-GAAP operating margin expanded by more than 400 basis points to 15% [2], which puts the year-ago figure near 11% [19]. A vendor buying consumption growth with discounted credits or subsidized inference would ordinarily show margin moving the other way. That does not confirm the AI attribution, but it makes the cheapest skeptical explanation harder to hold.
The customer figures point at the installed base rather than the sales funnel. 692 net new customers against 14,554 total is about 4.8% of the ending count added in three months [3][4][16], and 14 of those 692 came from the Global 2000, about 2% of the net adds [17]. Growth of 37% on that mix is mostly existing accounts consuming more.
Snowflake shipped more than 330 product capabilities in the first half [11], and it pitches model neutrality and the Cortex AI Gateway on cost optimization and flexibility [9][10]. That is the ground the vendor has already conceded in public, and it is the part of a renewal that only counts if it survives into the contract as routing rights and portability rather than staying a roadmap adjective. The surface being renewed is wider than the one priced at the last renewal.
The GAAP line sets the following quarter's context. Profitability on a GAAP basis is targeted for the fourth quarter of fiscal 2028 [6], six quarters after the one just reported [18], so the 15% margin a buyer is shown is non-GAAP and the vendor needs consumption to keep climbing across that window. A commitment signed this quarter is priced against a multiplier the disclosure does not yet quantify.
Ranked by verification strength, evidence, and original report placement.
Snowflake's non-GAAP operating margin expanded by more than 400 basis points year over year to 15%.
Snowflake reported 37% year-over-year growth in product revenue in the second quarter of fiscal 2027, reaching $1.49 billion, with what the company called a notable acceleration in growth over the past two quarters.
Snowflake added 692 net new customers in the quarter, including 14 from the Global 2000.
Snowflake raised its fiscal 2027 product revenue guidance to $6.07 billion, representing 36% year-over-year growth, and said it expects continued margin expansion.
Snowflake aims to achieve GAAP profitability by the fourth quarter of fiscal 2028.
Distinct publishers with included, body-backed reporting in this cluster.
1 article · September 4, 2026
Follow any of these and your For You feed starts watching them — no settings page required.
product
Snowflake books $370m less contracted revenue than analysts modelled2 distinct publishers
invest
Snowflake's backlog is growing faster than its revenue, and that is the AI story1 distinct publisher
invest
Salesforce bought $27.1B of its own stock in a quarter it grew 13%2 distinct publishers
invest
Snowflake adds $230m to its full-year product guide on a $70m quarterly beat4 distinct publishers
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
One issuer transcript, no second reading
The figures come from the company's own quarter, read aloud by its executives and reproduced unedited by Benzinga. That makes them a reliable record of what was said, and its accuracy stops there: the press release and the GAAP-to-non-GAAP reconciliation the operator points to are absent from this reporting, and the transcript breaks off before the Q&A where analysts would press on AI attribution.
Hard customer counts, soft AI counts
The platform side is countable and counted: 14,554 customers, 692 added in three months, and $1.49 billion of consumption revenue actually spent. The AI side is asserted. Coco and Cowork are described as rapidly adopted and significant to growth with no customer number, no revenue line and no bookings figure, so the adoption that the whole story turns on is the part left unmeasured.
Narration running ahead of disclosure
Management's own language, a once-in-a-lifetime technology shift and a claim that the agentic enterprise runs on Snowflake, carries weight the disclosure does not. Meanwhile the company guides the full year to 36% growth, a point under the 37% it just delivered, which is deceleration modelled quietly underneath an acceleration story. The framing is where the gap sits; the numbers themselves are stated plainly and are genuinely better than a year ago.
Seller-sourced from end to end
Every figure originates with the vendor on its own earnings call, including a margin measure that excludes stock compensation, and the AI framing is the same framing that underwrites renewal pricing for customers weighing next year's warehouse spend. Benzinga's own incentive is milder and different in kind: it publishes transcripts at volume as a feed product, which is why the text arrives complete, with no editorial scrutiny applied to it.
Solid arithmetic, thin corroboration
High for anything that follows from stated figures: implied prior-year product revenue near $4.46 billion, a year-ago margin around 11%, six quarters to the profitability date. Low for anything about what AI contributes, where a single unverified issuer account and a transcript cut short leave nothing to check against.