The U.S. General Services Administration announced on October 27, 2026, that it has expanded its OneGov agreement with Anthropic, giving government buyers a 40 percent discount on token-based usage across Claude models available to the government, with no minimum purchase required. Eligibility now extends beyond federal agencies to state, local and tribal governments, in a deal that marks a shift from evaluating AI tools to using them in daily government work.
What The Expanded Deal Covers
Participating governments get discounted, consumption-based access to Claude, meaning agencies pay only for what they use, with no minimum order. The 40 percent discount applies to token-based usage across all Claude models available to the government, including FedRAMP-authorised environments for sensitive unclassified work. Agencies can buy directly, through resellers and on major cloud platforms.
Eligibility covers federal buyers in the executive, legislative and judicial branches, as well as state, local and tribal governments. Anthropic’s own announcement adds territorial governments to that list and says the discount applies to Claude Enterprise, Claude for Government and the Claude API, available for direct purchases now and through cloud providers in the coming months. It also says the discount now applies across every product surface for the first time.
That breadth matters for procurement. Government buying is often slowed by minimum commitments, separate vehicles for each product and different terms for different levels of government. A single consumption-based discount, with no minimum, lowers the barrier for a small agency or a local government to try Claude on a real workload, rather than committing to a large contract first.
The Savings Claim
The GSA says its OneGov strategy has generated more than $1.8 billion in savings for the federal government and taxpayers since its inception, including approximately $1.6 billion through its agreements for advanced AI tools. Anthropic separately estimated that the discounts offered through this agreement will save governments about $1 billion. Those are the agencies’ and the company’s own estimates, not independently audited figures, and the actual savings will depend on how much government usage grows at the discounted rate.
Anthropic says roughly 500 U.S. federal, state and local government organisations now use Claude, and that weekly active users of Claude for Government have grown about eightfold since February. Those figures, also company-reported, suggest the government market for Claude is already substantial and growing quickly, which helps explain why both sides are formalising the terms now rather than leaving them to individual agency deals.
The GSA says that as offers expire, it will renew agreements where appropriate and work to ensure continued access to critical tools while delivering the best value for taxpayers. Agencies can find ordering details through the GSA’s OneGov pages, and the agreement sits alongside other OneGov AI deals, including a September expansion of discounted, consumption-based access to OpenAI’s ChatGPT.
From Evaluation To Daily Use
The GSA frames the deal as a marker of a wider shift: the federal government is moving from evaluating advanced AI capabilities to applying them in daily mission work, while continuing OneGov’s move toward consumption-based pricing that scales with actual use. That shift raises practical questions that a discount alone does not answer, including how agencies govern AI use, protect sensitive data and check AI output before acting on it.
FedRAMP authorisation for sensitive unclassified work addresses part of that concern, by setting a security baseline for cloud services used by government. But authorisation is not the same as approval for every use case, and agencies will still need their own controls for the data they feed into Claude and the decisions they take from its output, especially where the work touches benefits, enforcement or public services.
The consumption-based model also changes the budgeting conversation. Instead of a fixed licence cost, agencies face a variable bill tied to token usage, which can rise with adoption. The 40 percent discount softens that, but finance teams will still need to track usage by department and project, or a successful pilot can quietly become a large line item.
Why It Matters
This matters because government procurement is becoming a major battleground for AI providers, and the terms set here will shape how public services use AI. A 40 percent, no-minimum, consumption-based deal makes Claude easy to adopt across every level of government, which could accelerate AI use in public administration, while also locking in a pricing model that rewards heavy use. It also sits in a wider policy debate about AI in government, where proposals to reshape AI policy and enterprise AI deals are moving at the same time. The test for this agreement will be whether agencies use Claude for work that genuinely improves public services, and whether the tendency of professionals to follow AI agents is matched by oversight that catches AI errors before they reach citizens.
Background
The General Services Administration is the U.S. federal agency that handles procurement and shared services for the government. Its OneGov programme negotiates government-wide terms with technology providers, and Anthropic is the AI company behind Claude, whose government offering is designed for public-sector use.
Source: GSA announcement
0 Comments