The savings come with no spending floor
About 500 U.S. federal, state and local organizations use Claude. Since February, weekly active users of Claude for Government have grown roughly eightfold; use across federal civilian agencies is up about 20 times year over year.
Anthropic estimates the discounts could save the government $1 billion. Agencies do not have to commit to minimum spending in advance and can pay for what they use—a useful feature for organizations with constrained budgets and long purchasing cycles.
Anthropic says agencies are using Claude on backlogs, documents, forms and legacy code, as well as security work. Its examples include:
Anthropic also says its cybersecurity program for state, local, tribal and territorial governments has involved more than half of U.S. states, several of the largest local governments and some of the country’s largest critical-infrastructure operators.
The security boundary
Agencies can use Claude with the most sensitive unclassified information. Claude for Government offers coding and AI-agent capabilities and is close in functionality to Anthropic’s commercial products. It received general availability on September 30 and has FedRAMP High authorization.
Agencies building their own tools can access Claude through Amazon Bedrock or Google Cloud’s Vertex AI. On Bedrock, Claude is authorized for FedRAMP High, while specific models are approved for DoD IL4 and IL5 workloads in AWS GovCloud. Claude on Vertex AI is also authorized for FedRAMP High.
Claude Enterprise is pursuing FedRAMP Moderate authorization through the FedRAMP 20x program. Anthropic describes that as the next step toward making the same advanced technology available to government as to commercial customers.
The harder question is adoption
The discount removes one procurement hurdle, and the lack of a spending minimum limits the upfront commitment. But access and authorization do not show whether agencies can turn pilots into durable operations. Anthropic’s examples are promising, though they are company-reported cases rather than a measure of results across the roughly 500 organizations using Claude.
I think the more consequential part of the deal may be the new implementation program, not the 40% price cut. Government agencies have long purchasing cycles and constrained budgets; helping them put the tools to work could matter more than making them cheaper. Anthropic has not said how it will measure that program’s success, leaving the central test not whether agencies can buy Claude, but whether they can make it part of the work.
Government organizations can buy Claude at the OneGov discount through GSA’s Multiple Award Schedule. Anthropic’s public-sector team will discuss options with agencies already using Claude and those considering it.
Daily AI news
Every day we pick what actually matters in AI and explain it plainly — no hype, no filler. Subscribe if you want to follow where the industry is going.
Only what matters — every day
Follow on X