Snowflake bypasses Microsoft to strike multiyear deal with OpenAI – News

Snowflake Inc. today said it has struck a $200 million, multiyear partnership with OpenAI Group PBC to bring enterprise-grade generative artificial intelligence models directly into its data platform, deepening a relationship that until now had largely flowed through Microsoft Corp.’s Azure ecosystem.

Under the agreement, OpenAI’s models will be natively available inside Snowflake Cortex AI managed service and Snowflake Intelligence, giving Snowflake’s 12,600 customers the ability to build and deploy AI applications and agents directly on governed enterprise data across all three major public clouds. The deal positions OpenAI as one of the primary model providers on Snowflake’s platform, alongside Anthropic PBC, Meta Platforms Inc. and Mistral AI SAS.

Snowflake Chief Executive Officer Sridhar Ramaswamy said the partnership clears the way for Snowflake customers to move beyond experimentation to production-ready AI. “Customers can now harness all their enterprise knowledge in Snowflake together with the world-class intelligence of OpenAI models, enabling them to build AI agents that are powerful, responsible and trustworthy,” he said in a prepared statement.

The companies said joint customers including Canva Pty Ltd. and Whoop Inc. plan to use OpenAI models within Snowflake to deploy context-aware AI applications and agents across their businesses. OpenAI models such as GPT-5.2 will be accessible within Snowflake Intelligence, an enterprise agent that allows employees to query and act on structured and unstructured data using natural language.

Direct relationship

Executives framed the deal as a shift from Snowflake’s earlier, Microsoft-enabled access to OpenAI toward a direct, first-party relationship. “The partnership, or the ability to host OpenAI models, has been mostly in a partnership with Microsoft,” Christian Kleinerman, Snowflake’s executive vice president of product, said during a media and analyst roundtable ahead of Snowflake’s Build London event this week. “This is directly with OpenAI.”

Kleinerman said the goal is to make advanced models available while preserving the security, governance and compliance controls enterprises expect from Snowflake. “All innovation at Snowflake is centered and rooted on the challenges customers have,” he said, pointing to issues such as data silos, infrastructure complexity and regulatory requirements.

As part of the collaboration, Snowflake and OpenAI will co-develop new features using OpenAI’s application programming interfaces and agent frameworks, with an emphasis on building interoperable AI agents that can reason over governed data and act across enterprise systems. Snowflake said agents will run directly on customer data inside its AI Data Cloud, rather than requiring data to be moved to external services.

The companies also highlighted Snowflake’s enterprise-readiness features, such as its 99.99% uptime service-level agreement, business continuity capabilities and governance controls provided through the Snowflake Horizon Catalog. Those safeguards are intended to address concerns among regulated industries about reliability and responsible AI use.

The partnership also extends to internal use. Snowflake said OpenAI uses Snowflake as a data platform for experiment tracking and analytics, while Snowflake uses ChatGPT Enterprise internally to support employee productivity and decision-making.

Image: News/Google Whisk