Anthropic expands GSA deal: 40% Claude discount now covers all levels of US government

A much wider agreement

Anthropic has significantly widened its agreement with the US General Services Administration (GSA), extending a 40% discount on its Claude products to government buyers at every level — federal, state, local, tribal and territorial. According to the company's announcement, around 500 US government organizations already use Claude, and the discount structure is estimated to save the government roughly $1 billion.

What the deal covers

The 40% reduction applies to Claude Enterprise, Claude for Government and the Claude API. Anthropic says this is the first time the discount spans its entire product lineup, covering agencies that buy directly from the company today and, in the coming months, purchases made through cloud providers.

The agreement removes minimum spending requirements: agencies pay only for what they consume. Buying happens through GSA's Multiple Award Schedule under the OneGov discount. Anthropic is also introducing a dedicated program to help agencies implement Claude and get value from it, adapting practices it already uses with private-sector customers.

Adoption and usage figures

All of the numbers are self-reported by Anthropic. Beyond the roughly 500 organizations cited, the company says weekly active users of Claude for Government have grown about eightfold since February, while usage across federal civilian agencies has risen roughly twentyfold in a year.

How agencies are reportedly using Claude

The announcement lists several examples, most of them from state IT teams:

Utah gave Claude Code to 34 developers across 10 state agencies. Anthropic reports that 80% had a working environment within 30 minutes, and 71% said they saved at least 20 hours by the end of a four-week pilot.

A state CIO, unnamed in the announcement, said their team responded to a ransomware attack in eight minutes instead of the usual two to three days. A federal security team reportedly used a tool Anthropic calls Claude Mythos to find and patch a vulnerability in a critical production system within about two days — after a human penetration test the previous month had missed it.

Other cited cases include a state developer who moved nearly 900 API methods to a new framework in two days on a project scoped for a month, and a state IT director who used Claude Code to document 12 legacy repositories, mapping 467 database tables and flagging 27 risks along the way.

Anthropic also says more than half of US states, along with some of the largest localities and major critical infrastructure operators, have taken part in its state, local, tribal and territorial cyber defense program.

Security and compliance posture

Claude for Government, which Anthropic positions for agencies' most sensitive unclassified work, became generally available on September 30. It operates at FedRAMP High authorization and offers near feature parity with the commercial products, including coding and agentic capabilities.

Agencies building their own tools can reach Claude through Amazon Bedrock or Google Cloud's Vertex AI. On Bedrock, Claude is authorized for FedRAMP High, with select models cleared for DoD IL4 and IL5 workloads in AWS GovCloud; on Vertex AI it is authorized at FedRAMP High. Claude Enterprise is working toward FedRAMP Moderate authorization through the FedRAMP 20x program.

Why it matters

The expansion is one of the clearest milestones yet in the adoption of frontier AI models inside US government operations, reaching from federal civilian agencies down to state IT departments. The combination of a flat 40% discount, no minimum spend and usage-based billing is aimed squarely at the two frictions that most often keep agencies from experimenting with new technology: tight budgets and long procurement cycles.

It also concentrates substantial public-sector workloads in a single vendor's hands. The striking efficiency claims — an eight-minute ransomware response, a month-long migration finished in two days — come from Anthropic and its customers rather than independent evaluation, so agencies weighing similar deployments will want their own validation. Still, with usage up eight- to twentyfold in roughly a year, the direction is clear: AI assistants are becoming routine infrastructure for government software maintenance, records work and cybersecurity.