California Governor Gavin Newsom has signed a first-of-its-kind agreement with Anthropic that gives every state agency, city, and county in California access to Claude at a 50 percent discount. The partnership — announced on June 29 — makes Claude the first AI productivity tool available across all state agencies through the California Department of Technology’s newly launched SITeS (Statewide Information Technology Shared Services) portal, which centralises AI tool procurement in one place with transparent, standardised pricing.
What Happened
Under the agreement, any California state employee — and workers at local governments statewide — can access Claude for tasks ranging from drafting documents and summarising reports to analysing complex datasets and improving internal workflows. Anthropic is also providing free workforce training, technical assistance, and dedicated workflow consultation as part of the deal.
The SITeS portal, which serves as the operational backbone of the agreement, is designed to give public sector organisations a streamlined procurement path for AI tools without requiring each agency to negotiate individual enterprise contracts. California officials described it as a model for how governments can responsibly adopt commercial AI at scale without fragmenting procurement or creating uneven implementation across departments.
Several California agencies have already been using Claude in pilot programmes. The Department of Motor Vehicles deployed Claude for customer service improvements, while the California Department of Health Care Services used it to streamline internal administrative workflows. The new agreement expands those isolated pilots into a coordinated, statewide deployment.
Why It Matters
The California deal is notable for several reasons beyond its scale. It positions the world’s most populous US state as an early and prominent adopter of Anthropic’s technology at a moment when the federal government has been less receptive. The US Department of Defense clashed with Anthropic over contract terms — specifically Anthropic’s insistence on provisions preventing its models from being used for autonomous weapons or mass surveillance — and that tension has made California’s open embrace of Claude politically significant.
For Anthropic, a deal covering roughly 240,000 state employees plus millions of local government workers represents a major anchor customer that validates its enterprise strategy ahead of a widely anticipated IPO. The company has been on an upswing in recent weeks, restoring Fable 5 access globally after a brief export control suspension and launching Claude Sonnet 5 as a more affordable, high-capability model for agentic workloads.
The 50 percent discount also sets an interesting precedent for how AI companies approach government procurement. Rather than full-price enterprise licensing, Anthropic is apparently willing to accept lower per-seat revenue in exchange for deployment at extraordinary scale and the reputational value of a high-profile government partnership.
Background and Context
California has been unusually proactive about AI adoption at the state level. The state previously rolled out Poppy, its in-house AI assistant, to all state employees — making it one of the first governments anywhere to deploy a purpose-built government AI tool at that breadth. The Anthropic agreement builds on that foundation, giving agencies access to more powerful, general-purpose AI capabilities that can complement Poppy’s government-specific functions.
Newsom has positioned California as a counterweight to federal AI policy under the current administration, which has alternated between encouraging AI development and restricting exports of frontier models. The Governor’s office framed the Anthropic agreement as part of a broader strategy to make California a model for responsible, high-benefit AI deployment in government.
What Comes Next
Implementation will be the real test. Rolling out an AI assistant to hundreds of thousands of state and local employees involves significant change management, training, and security review work — even with Anthropic’s promised technical assistance. The SITeS portal is designed to simplify onboarding, but individual agencies will still need to adapt their internal processes and data handling procedures to use AI tools responsibly.
Other states and cities are likely watching closely. If California can demonstrate measurable productivity gains and cost savings from the Claude deployment, it could accelerate AI adoption across the US public sector more broadly. Anthropic is simultaneously pursuing an aggressive hardware strategy, in early talks with Samsung to manufacture a custom AI accelerator chip — suggesting the company is building the infrastructure capacity to support large-scale public and enterprise deployments well into the future.
