Anthropic has named 20 foundations, institutions and research groups in an expansion of its AI for Science program. The program hands Claude credits to high-impact research, has run since May 2025, and until now mostly backed individual scientists. The discounted Team plan for scientists came alongside it recently.
Who passes the credits on
Six funders and societies will distribute credits to their own applicants: the Alfred P. Sloan Foundation, the Simons Foundation, Schmidt Sciences, the Open Source for Science Fund at Renaissance Philanthropy, the US National Science Foundation, and CAS, the chemistry division of the American Chemical Society. The NSF is the heavyweight here, a $9 billion federal agency that accounts for roughly a quarter of federal support for basic research at US colleges and universities.
Who uses them directly
On the institutional side sit organizations you recognize by their datasets and instruments. The Space Telescope Science Institute runs science operations for Hubble, James Webb and Roman, and holds one of the largest astronomy archives anywhere. The Wellcome Sanger Institute produces genomic data at scale, from human genetics to biodiversity. Add the Francis Crick Institute in London with more than 100 research groups, and the Department of Life Sciences at Imperial College London.
Two collaborations predate this announcement and show where it points. With the HHMI Janelia Research Campus, Anthropic is developing the Model Hardware Standard, a shared specification that lets AI agents safely operate lab instruments; it sits in research preview. And the Allen Institute is building teams of agents to analyze its own open datasets, turning months of manual analysis into hours.
The public goods are the part that lasts
Credits are a discount, and a discount is sales. What lifts this past a donation notice sits in a subordinate clause: many of these projects are meant to produce open datasets, open-source software, open standards and shared libraries.
That is the difference between a voucher and infrastructure. A standard like MHS only works if more than one lab adopts it, and an open archive outlives any credit balance. Anthropic already published its first progress report from six drug development partnerships this week. The two belong together. This isn’t compute being given away; it’s a position in science being built.