Snowflake is spending up to $200M over multiple years to integrate OpenAI models directly into its platform, eliminating its dependency on Microsoft Azure for access to GPT capabilities.
The deal brings OpenAI's models—including the latest GPT-5.2—natively into Snowflake's Cortex AI and Snowflake Intelligence products. Enterprise customers like Canva and WHOOP can now run natural language queries and build AI agents on their proprietary data without moving it off-platform.
What changed: Previously, Snowflake customers accessed OpenAI through Azure integration. This direct partnership establishes first-party access and tighter engineering alignment between the companies. Baris Gultekin, Snowflake's VP of AI, frames it as "a commercial commitment anchored in real AI consumption," not speculative positioning.
The context matters: This is Snowflake's second major AI partnership in three months. In December 2025, it committed $200M to Anthropic for Claude integration. The company is clearly betting that embedded AI capabilities will drive platform stickiness and justify its investment amid competitive pressure from hyperscalers and Databricks.
Bank of America maintains a Buy rating with a $275 price target (down from $310), expecting "high-20% product revenue growth" from AI features. The stock rose 2% on the announcement.
The competitive angle: Snowflake's dual-vendor AI strategy creates optionality but also complexity. Customers can now choose between OpenAI and Anthropic models within the same platform, running across AWS, Azure, and Google Cloud. That multi-cloud flexibility addresses vendor lock-in concerns—though it also multiplies integration and governance challenges for data teams.
Analysts flag the obvious risk: AWS, Azure, and Google all have native AI database offerings. Snowflake is betting that governed, multi-model AI on a neutral platform beats the convenience of hyperscaler bundling. We'll see if enterprises agree when renewal time comes.
What to watch: How Snowflake prices per-query AI consumption compared to Databricks Genie and hyperscaler alternatives. The $200M commitment suggests significant expected usage, but enterprise AI costs remain a moving target.