Snowflake’s results on May 27 landed as more than a routine software earnings beat. The company not only raised its outlook after another quarter of strong growth, it also committed $6 billion over five years to Amazon Web Services in its largest infrastructure agreement to date. Taken together, those moves suggest that the next phase of enterprise AI may be judged less by flashy demos and more by which software companies can secure the compute, data access, and cloud partnerships needed to turn experimentation into repeatable business usage.
Snowflake reported first-quarter revenue of $1.39 billion, up from $1.04 billion a year earlier, while product revenue rose 34% to $1.334 billion. It lifted its fiscal 2027 product revenue forecast to $5.84 billion from $5.66 billion and said second-quarter product revenue should come in between $1.415 billion and $1.420 billion, ahead of analyst expectations cited by Reuters. Investors responded quickly. Reuters reported that Snowflake shares surged 36% in extended trading on Wednesday after the company released results and announced the AWS agreement.
The financial numbers matter, but the infrastructure commitment is what gives the quarter broader significance. Snowflake said the new strategic collaboration agreement will deepen product integration with AWS around generative and agentic AI, expand joint go-to-market work through AWS Marketplace, and support workload migrations for customers trying to move AI projects into production. The company said it will spend the $6 billion on Graviton compute and AI infrastructure over five years, a sign that it sees enough durable customer demand to lock in capacity well ahead of time.
That is a notable shift in how the market should think about software economics in the AI era. For years, software companies were valued partly on the idea that they could scale without taking on the capital intensity of infrastructure providers. AI is starting to blur that distinction. If customers want models and agents to run directly against governed enterprise data, the software layer has to guarantee performance, availability, and cost discipline across enormous computing loads. In that world, access to cloud infrastructure becomes a strategic input, not just a vendor expense.
The Snowflake and AWS tie-up also shows how the cloud giants are tightening their grip on enterprise AI through custom silicon and commercial alignment, not only through headline model announcements. Snowflake said the majority of its customers already run on AWS, and the company has now surpassed $7 billion in lifetime AWS Marketplace sales, including more than $2 billion in calendar 2025 alone. Those numbers help explain why Snowflake was willing to expand its commitment so aggressively. The partnership is no longer simply about hosting a data platform. It is becoming part of the sales, procurement, and deployment machinery through which enterprise AI gets bought and used.
There is still plenty of reason for caution. Snowflake remains unprofitable on a GAAP basis, posting a net loss of $295.6 million in the quarter, even though that was an improvement from a year earlier. A large infrastructure commitment can look smart if AI consumption keeps rising, but it can also become an expensive promise if corporate customers slow spending or remain stuck in pilot mode. The company is effectively betting that demand for AI and data workloads will be deep enough and sticky enough to justify locking itself more tightly to its largest cloud partner.
Even so, the broader signal is hard to miss. Investors have spent much of the last year rewarding chipmakers, hyperscalers, and power suppliers as the clearest beneficiaries of AI spending. Snowflake’s quarter suggests the software layer may be entering a more consequential phase, one in which the winners are the platforms that can translate enterprise interest into recurring workload growth while managing the heavier infrastructure burden that comes with it. That does not make software less attractive. It makes the category more operationally demanding, and arguably more defensible for the companies that can meet that test.
If that reading is right, Snowflake’s update will matter beyond one strong quarter or one rally in the stock. It points to a market where software groups are no longer just renting infrastructure in the background. They are increasingly pre-buying it, shaping product strategy around it, and using those commitments to secure a place in the enterprise AI stack before customer demand fully settles into long-term winners.