We have been watching the AI-attached consumption thesis on Snowflake for two quarters and this is the cleanest confirmation yet. The combination of 34 percent core growth, raised guidance, and a multi-year hyperscaler deal makes the platform harder to displace at the warehouse layer, even as Databricks pulls ahead on training workloads. For teams in renewal cycles this quarter, the negotiation window just narrowed.
Snowflake reported Q1 FY2027 revenue of $1.39B on Thursday afternoon, with year-over-year growth of 33 percent and core product growth of 34 percent. Profitability came in materially above consensus, and the company raised its full-year growth guide to 31 percent. The market response was unusually large for a company at this scale: the stock closed Wednesday near $175, finished Thursday at $239.20, and ran to roughly $255 by Friday's close, a roughly 46 percent move in three sessions and a 38 percent single-day jump on the print itself.
A $6B AWS prepay that resets Snowflake's gross margin math
The earnings story sat alongside two structural disclosures that matter more than the quarterly beat. The first is a five-year, $6B commitment with Amazon Web Services. We read that as a capacity prepay rather than a pure marketing alliance: at Snowflake's current growth rate it implies AWS confidence that Snowflake's compute footprint will keep expanding fast enough to consume that capacity, and it locks in unit economics on the most expensive line item on Snowflake's P&L. For customers, it also signals that AWS-region capacity headaches that surfaced during last year's AI rush are unlikely to repeat through 2028.
The gross margin implication is the line item we are watching. Snowflake's product gross margin sat at roughly 76 percent in Q4 FY2026, with hyperscaler cost of revenue the single largest drag. A $6B prepay at negotiated multi-year rates should add 150 to 250 basis points of structural margin lift as the volume tier kicks in across FY2028 and FY2029, assuming consumption tracks the raised 31 percent guide. That gap matters competitively. Databricks runs a lower published gross margin profile because its training workloads are GPU-heavy and its Mosaic acquisition still carries integration cost. If Snowflake can compound free cash flow at 30 percent plus while expanding margin, the IPO comparison shifts from growth rate to durable unit economics. For procurement teams, the practical signal is that Snowflake now has room to absorb deeper multi-year discounts without breaking its own margin story.
The second disclosure was an indicated near-term acquisition aimed at strengthening the AI offering for enterprises. Management did not name a target, but the direction of travel is consistent with the past month of release activity: AI Function Studio went into public preview on May 20, Gemini 3.5 Flash arrived as a multimodal Cortex function on May 28, and Cortex AI Guardrails went GA on May 14. The shape of the platform is clearly pivoting from a SQL warehouse with bolt-on AI to a data plane with first-class AI primitives. A tuck-in of an inference, agent, or evaluation specialist would round that out.
What the Databricks IPO pressure looks like in negotiation
The competitive subtext is Databricks. The private company is running at roughly $5.4B annualised on 65 percent growth, with $1.4B coming from AI products, and continues to win the lion's share of incremental AI training budget in head-to-head bake-offs. Databricks is also expected to file its S-1 in Q3 2026, and the rotation risk into a public Databricks ticker has been a hanging threat over Snowflake's multiple. Thursday's print is the first time in several quarters that Snowflake materially answered that pressure with operating numbers rather than narrative.
For platform leaders running both, the practical implication is that the two vendors are converging on different sides of the same problem. Snowflake is pushing AI inference, semantic views, and Cortex agents down into the warehouse where governed data already lives. Databricks is pushing operational data and serverless Postgres up into the lakehouse via Lakebase and Genie. We expect most large enterprises to keep both for at least the next 24 months, with the split increasingly determined by where the governance boundary lives rather than by workload type.
What this changes for European retail data teams
For operators in our patch, the read-through is concrete. At MediaMarktSaturn-scale retail, where pricing, inventory, and CRM are already governed in Snowflake, the cost-to-serve case for keeping AI features inside the warehouse just got stronger. AI_CLASSIFY for product taxonomy work, AI_PARSE_DOCUMENT at the new 2,000 page limit for supplier paperwork, and Cortex Search going GA in batch mode all reduce the volume of data that needs to leave the perimeter to power retail AI use cases. For REWE digital, where the Iceberg-first lakehouse pattern is more entrenched, Snowflake's May 26 GA of Iceberg write support via external query engines means the warehouse can finally participate as a peer in a Spark or Trino-led architecture rather than as a destination.
At MediaMarktSaturn specifically, the renewal lands in Q3 calendar 2026 and the recent shift of returns-fraud scoring into Cortex AI has roughly $400K of annual inference spend on the table that would otherwise route to a standalone vector database. Pinning AI_CLASSIFY pricing at current rates through FY2027 protects that case. For REWE digital, the Iceberg write GA changes a December architecture decision: the team can now keep the Trino-led query layer for the loyalty data product while letting Snowflake own governed writes back to the shared catalog, saving a planned six-week migration window.
There are two risks worth flagging. The first is the run-up itself: a 46 percent move in three sessions creates an obvious pullback target on any negative macro print, and procurement teams should not let an inflated stock chart reset their negotiating posture. List prices and credit pack discounts are still under pressure, and the AWS prepay actually gives Snowflake more room to be flexible on multi-year deals, not less. The second risk is concentration: $6B of committed AWS spend is efficient but it tightens the operational coupling between Snowflake and a single hyperscaler at exactly the moment when European customers are pushing harder on sovereignty and exit clauses. We would expect to see a parallel commitment with Azure or a sovereign-region carve-out within the next two quarters.
The Q2 print is the real test
Two prints frame the next 120 days. Snowflake's Q2 FY2027 report is expected in the final week of August, and the number we are watching is net revenue retention: anything above 128 percent confirms that the AI-attached consumption flywheel is durable rather than a one-quarter pull-forward. Below 124 percent and the AWS prepay starts to look defensive. The Databricks S-1 is expected in Q3 calendar 2026, and the disclosure that matters is the split between training and inference revenue inside the $1.4B AI line. A heavy training mix supports the bear case on Snowflake. A balanced split says both platforms can compound through the next budget cycle.
The action item for this week is straightforward. If a Snowflake renewal lands in the next 90 days, request the AI credit pack pricing and pin language around Cortex pricing stability through the end of FY2027. If a parallel Databricks evaluation is running, hold the line on a late-2026 decision: Lakebase pricing is still in flux and the IPO-window product roadmap will be more visible by September.
