Anthropic commits $150M to put Claude inside federal science projects

Why it matters

Anthropic is committing resources to become part of federal research workflows. Adoption will depend on whether scientists can integrate Claude into existing systems and verify its results.

Anthropic is committing $150 million over three years to bring Claude into hundreds of research projects under the federal Genesis Mission, the company said on October 8th. Co-founder and CEO Dario Amodei trained as a physicist before building AI systems; the commitment puts Anthropic's models and engineering support into a government science program spanning more than 15 agencies, including NASA, the National Institutes of Health and the National Science Foundation.

Anthropic describes a three-year program to provide Claude, Claude Code, API credits and support to researchers. The announcement does not disclose how the $150 million will be divided among credits, staffing or other services, so the headline figure does not show how much computing capacity researchers will receive.

For Amodei, the government work connects directly to his scientific background. He studied at Stanford, earned a PhD in physics at Princeton focused on neural circuits, and later worked as a postdoctoral scholar at Stanford School of Medicine. Before co-founding Anthropic, he was a research leader at OpenAI and worked at Google Brain, according to the Hertz Foundation. The deployment places Claude in institutions where researchers already work with specialized data, instruments and computing systems.

From access to research workflows

The program's test will be whether Claude can fit into that existing work. Access to a model can help researchers search literature, write code or explore hypotheses; scientific progress depends on researchers being able to check results against data, experiments and domain expertise. Anthropic's announcement describes training, onboarding and technical support alongside access to Claude, but offers no named projects or outcome measures for the hundreds of projects it says will participate.

Anthropic has been building toward this kind of deployment. In July 2025, Lawrence Livermore National Laboratory said it was expanding Claude access to about 10,000 scientists, researchers and staff. In December 2025, Anthropic announced a partnership with the Department of Energy as part of Genesis, describing work on energy, biological and life sciences, and scientific productivity. The latest commitment broadens that effort from DOE-linked research to a mission expected to reach agencies beyond the national laboratories.

Anthropic has also assembled products and programs for researchers outside government. It launched Claude Science, a workbench built around research tools, and said in August it would open 10,000 free or discounted seats for academic scientists. Its AI for Science program offers credits to research projects. Those efforts give Anthropic more ways to put Claude into scientific workflows; the Genesis commitment adds a federal route to adoption and a direct relationship with agencies setting research priorities.

The federal initiative already includes the company's largest AI competitors. When DOE announced 24 collaboration agreements in December, the list included Anthropic alongside OpenAI, Google, Microsoft, NVIDIA and AWS. DOE describes Genesis as an effort to connect national laboratories, supercomputers, experimental facilities, datasets and AI systems. Anthropic is competing for a place in that infrastructure, where a successful deployment could establish its products as familiar tools for researchers and government agencies.

That push comes amid uneven federal relationships. On October 5th, the Pentagon said it had stopped using Claude after a six-month phaseout, as RuntimeWire reported. The Genesis commitment concerns scientific research through DOE and participating agencies; it does not reverse the Pentagon's decision. It does show Anthropic continuing to pursue federal deployments across parts of government where the work and institutional relationships differ.

Anthropic is expanding access through products, credits and federal partnerships. Whether those deployments produce research scientists can independently verify remains an open question.