The U.S. General Services Administration signed a OneGov agreement with Snowflake on May 21, handing every federal agency a 20% discount on compute and a 26.67% cut on storage across Snowflake's AI Data Cloud — with compute discounts scaling to as much as 50% as consumption grows. The pricing runs through September 30, 2027.
The move makes Snowflake the latest data and AI vendor to swap volume-based federal demand for a single negotiated rate, and it lands as agencies race to stand up AI on top of governed data rather than shipping that data out to separate model endpoints.
The terms
The headline numbers are flat entry discounts — 20% off compute, 26.67% off storage — applied to new federal customers across the platform. The differentiator is the consumption tier: as an agency's overall usage climbs, it can qualify for compute reductions of up to 50%. That structure rewards exactly the workloads agencies are scaling fastest — large analytical queries and AI inference running directly against warehoused data. The agreement is a one-year deal available through the end of fiscal 2027.
Why FedRAMP is the unlock
The discounts only matter because Snowflake can legally hold the data. The company carries FedRAMP High authorization on AWS GovCloud, secured in 2023, and on Microsoft Azure Government as of June 2025. For practitioners, that authorization is the gating constraint on any government AI build — it determines whether sensitive datasets can sit in a managed cloud at all. Snowflake's multi-cloud posture also gives agencies a path that isn't bolted to a single hyperscaler's compute.
Where it sits in OneGov
OneGov treats the entire federal government as one centralized customer to extract enterprise-scale pricing. GSA says the program has identified more than $1.15 billion in savings since it launched in April 2025, with roughly 20 vendor agreements in place. AI tooling from OpenAI, Anthropic, and Perplexity is already available through the framework, alongside infrastructure and software deals with vendors including Microsoft, SAP, and Palo Alto Networks.
"Federal agencies are seeking efficiency in cost, enterprise scaled performance, intuitive design driven tools for the workforce and simplicity in contracting — we are the only multi-cloud data platform that can meet this charge on day one," Snowflake CEO Sridhar Ramaswamy said. GSA Administrator Edward Forst framed the payoff in capability terms: "With stronger cross-agency data capabilities, we can accelerate AI tools tailored to each agency's mission."
What changes for buyers
For enterprise and government AI teams, the practical signal is procurement gravity: the cheapest, fastest-to-contract path for federal AI increasingly runs through whoever owns the FedRAMP-authorized data layer. That pressures competing lakehouse and warehouse vendors to match both the authorization bar and the OneGov pricing, and it pushes more agency AI workloads toward platforms where the model runs next to the data rather than across an egress boundary.
