Anthropic's GSA Deal: 40% Off Claude for Federal, State and Local Government

Anthropic announced on October 27, 2026 an expanded agreement with the U.S. General Services Administration (GSA) that gives government partners a 40% discount on Claude Enterprise, Claude for Government and the Claude API. All figures below come from Anthropic's announcement.

40%Discount across Claude products

$1BEstimated government savings, per Anthropic

About 500U.S. government organizations using Claude

FedRAMP HighClaude for Government authorization

What changed

Anthropic says this is the first time the discount covers every product surface, including direct purchases and, in the coming months, purchases through cloud providers. It also extends beyond federal agencies to state, local, tribal and territorial governments. Agencies face no minimum spend commitment and pay only for use, and can buy at the discount through GSA's Multiple Award Schedule. Anthropic estimates the discounts will save the government $1 billion and says it will launch a program to help agencies implement Claude, with details expected in the coming weeks.

Reported adoption

Anthropic says about 500 U.S. federal, state and local government organizations use Claude, weekly Claude for Government active users are up about 8x since February, and usage across federal civilian agencies grew about 20x in one year. These are company-reported multiples with no disclosed starting levels, so they indicate direction rather than scale.

Reported growth in government use of Claude

Approximate multiples, as reported by Anthropic. Starting levels were not disclosed.

Weekly active users, Claude for Government (since Feb)

Usage across federal civilian agencies (one year)

20x

Security authorizations

Government buyers decide on authorizations as much as on price. Per Anthropic, Claude for Government became generally available on September 30 and holds FedRAMP High authorization. Through Amazon Bedrock, Claude is FedRAMP High authorized, and select models are authorized for DoD Impact Level 4 and 5 workloads in AWS GovCloud. Through Google Cloud's Vertex AI, Claude is FedRAMP High authorized. Claude Enterprise is in process for FedRAMP Moderate through the FedRAMP 20x program.

Stone facade of the U.S. General Services Administration headquarters building in Washington, DC — The U.S. General Services Administration building in Washington, DC. Image: General Services Administration, public domain, via Wikimedia Commons.

The examples Anthropic cites

Anthropic lists several deployments, all self-reported and not independently verified here:

Utah gave Claude Code to 34 developers across 10 state agencies; in a four-week pilot, 80% had a working environment within 30 minutes and 71% reported saving 20 or more hours.
A state CIO said their team responded to a ransomware attack in 8 minutes against the usual 2 to 3 days.
A federal security team used Claude Mythos to find and patch a vulnerability in about two days that a human penetration test the previous month had missed.
A state IT developer moved nearly 900 API methods to a new framework in two days, a project scoped for a month.

How to read the deal

Price is one input. A flat 40% discount with no minimum lowers the barrier to pilots, but a pilot is not a deployment. The headline $1 billion is an estimate, and its assumptions were not published in the post.
Competition is active. Governments are also buying from other labs. See our coverage of federal AI services offered by Google and xAI.
Authorization is the moat. FedRAMP High and defense-level authorizations take time to obtain, and they determine which workloads a model can touch.
Lock-in is worth planning for. Agencies that standardize on one vendor's API may later want to compare models. Metir is model-agnostic, which is the general argument for keeping workflows portable across providers.

Sources:

Image credits

Header and in-body image: U.S. General Services Administration Building, Washington, DC, by the General Services Administration, public domain, Wikimedia Commons (source). It does not depict Anthropic or Claude.

What changed

Reported adoption

Reported growth in government use of Claude

Approximate multiples, as reported by Anthropic. Starting levels were not disclosed.

Weekly active users, Claude for Government (since Feb)

Usage across federal civilian agencies (one year)

20x

Security authorizations

The U.S. General Services Administration building in Washington, DC. Image: General Services Administration, public domain, via Wikimedia Commons.

The examples Anthropic cites

Anthropic lists several deployments, all self-reported and not independently verified here:

Utah gave Claude Code to 34 developers across 10 state agencies; in a four-week pilot, 80% had a working environment within 30 minutes and 71% reported saving 20 or more hours.

A state CIO said their team responded to a ransomware attack in 8 minutes against the usual 2 to 3 days.

A federal security team used Claude Mythos to find and patch a vulnerability in about two days that a human penetration test the previous month had missed.

A state IT developer moved nearly 900 API methods to a new framework in two days, a project scoped for a month.

How to read the deal

Price is one input. A flat 40% discount with no minimum lowers the barrier to pilots, but a pilot is not a deployment. The headline $1 billion is an estimate, and its assumptions were not published in the post.

Competition is active. Governments are also buying from other labs. See our coverage of federal AI services offered by Google and xAI.

Authorization is the moat. FedRAMP High and defense-level authorizations take time to obtain, and they determine which workloads a model can touch.

Lock-in is worth planning for. Agencies that standardize on one vendor's API may later want to compare models. Metir is model-agnostic, which is the general argument for keeping workflows portable across providers.