Snowflake (NYSE:SNOW) delivered the sort of quarter investors needed to justify its huge AI premium. Revenue, product revenue and adjusted EPS all beat expectations, but the biggest catalyst was the $230m increase in FY2027 product revenue guidance.
And wow, the stock reacted well, surging more than 20% after-hours. At approximately $378, the shares are trading at peaks not seen since their post-IPO hype back in 2021, when they soared beyond $390.
| Snowflake (NYSE:SNOW) | Price: $377.50 (~+23% after-hours) | Market cap: ~$131bn |
The critical question now is no longer whether Snowflake has an AI opportunity. It is whether AI can turn Snowflake’s already strong data platform into sustained >30% growth while expanding margins.
The numbers that mattered
| $bn unless stated | Q2 FY2027 | YoY | Consensus | Takeaway |
| Revenue | 1.55 | +35% | 1.48 | Strong beat |
| Product revenue | 1.49 | +37% | ~1.42 | Key growth metric |
| Adj. EPS | $0.62 | +77% | $0.45 | Large beat |
| Net revenue retention | 126% | — | — | Expansion remains healthy |
| RPO | $9.0bn | +30% | — | Strong contracted demand |
| >$1m customers | 828 | +27% | — | Enterprise penetration |
| Forbes Global 2000 customers | 829 | — | — | Enterprise scale |
Most importantly, FY2027 product-revenue guidance jumped from $5.84bn to $6.07bn, implying roughly 36% growth. Q3 product revenue is expected at $1.588bn-$1.593bn, comfortably ahead of the roughly $1.57bn consensus. Full year adjusted operating margin guidance also rose to 14.5% from 13.5%.
That combination — higher growth and higher profitability — explains the violent share price response.
What management said
CEO Sridhar Ramaswamy described the quarter as evidence that Snowflake is becoming an important part of enterprise AI infrastructure:
‘AI continues to compound our advantages, creating a flywheel effect across the business.’
He said CoWork and CoCo were generating customer outcomes while driving ‘rapid adoption, user growth, new workloads, and overall platform consumption.’
That is strategically important. Snowflake’s AI products are not simply another software module. Management’s thesis is that AI creates more data consumption, which creates more Snowflake usage, which encourages customers to build more AI applications on the platform.
CoCo passed 9,100 accounts, while CoWork reached 5,800 accounts; together they represent almost 15,000 accounts using these AI capabilities.
Why Snowflake is becoming an AI infrastructure play
Snowflake sits in an increasingly valuable position between enterprise data and AI models.
Its opportunity is broader than competing with Databricks in data warehousing. Snowflake is trying to provide the governed data, development environment, AI models, agents and applications required to turn corporate information into automated decisions.
Its partnerships with OpenAI, Anthropic and other model providers, together with AWS, strengthen that proposition. The previously announced $6bn, five-year AWS agreement is particularly significant: it gives Snowflake access to substantial cloud compute while signalling AWS’s willingness to support Snowflake’s AI expansion.
The Natoma acquisition also extends Snowflake’s ambition into AI-agent governance, potentially allowing enterprises to control what autonomous agents can access and do across corporate systems.
Analyst reaction
Analysts had already become considerably more bullish before the results. TD Cowen raised its target to $370 from $300, while Cantor Fitzgerald increased its target to $405 from $282, citing AI momentum.
Earlier commentary captured the changing investment thesis. Rosenblatt’s Blair Abernethy said the company was benefiting from ‘core data cloud consumption and faster migrations’, while Wedbush’s Dan Ives argued that strong consumption and AI momentum were challenging the Databricks share-shift narrative. Needham described Snowflake as receiving a ‘meaningful uplift from AI.’
The crucial development after this quarter is that the earnings themselves now support the bullish narrative rather than AI being primarily a valuation story.
Valuation: the catch
The problem is that investors already knew Snowflake was a high-quality growth company.
Snowflake Analysis: What +37% stock surge means for UK retail investors
Before the results, Snowflake had a market capitalisation of roughly $111bn. At around $320, it was trading at approximately 20x FY2027 guided product revenue.
| Company | Approx. EV/LTM revenue |
| Snowflake | ~20.6x |
| Datadog | ~19.7x |
| MongoDB | ~12.2x |
| Oracle | ~7.7x |
| Palantir | ~63x |
Peer multiples vary with market prices and reporting periods, but the comparison illustrates Snowflake’s premium valuation.
Snowflake therefore needs to earn its premium through sustained growth. It also remains GAAP loss-making, with stock-based compensation a significant issue. Shares outstanding have increased about 2.5% year-on-year.
Bull 🐂 vs 🐻 bear case
| 🐂 Bull case | 🐻 Bear case |
| AI drives incremental Snowflake consumption | AI spending could prove cyclical |
| Product revenue still growing ~37% | Growth eventually returns toward 20% |
| 126% NRR shows customer expansion | Databricks/Microsoft intensify competition |
| AI products gaining rapid adoption | Premium ~20x sales leaves little room for disappointment |
| Rising operating margins | GAAP losses and SBC remain substantial |
| AWS/OpenAI/Anthropic partnerships strengthen ecosystem | Cloud infrastructure costs pressure margins |
Investor verdict
This was the sort of quarter investors needed to justify its huge AI premium, genuinely strong, not merely AI-hype. The combination of 37% product-revenue growth, a major guidance upgrade and improving margins materially strengthens the investment case.
But at more than 20x sales, investors are paying today for substantial future success. The stock’s next leg higher will require evidence that AI is accelerating consumption rather than merely improving the narrative.
For a long-term investor, Snowflake increasingly looks like a bet on the data layer of enterprise AI. The opportunity is enormous; the valuation means the margin for error is not.
You might also like:
