Analysis
Anthropic said on October 28 that its AI for Science programme, which has offered free Claude credits to researchers since May 2025, will now reach scientists through institutions and foundations. It named 20 partners across three routes: funders that distribute credits to their grantees, institutions that use credits internally, and research initiatives that receive credits directly.
The partners include the Sloan Foundation, Simons Foundation, Schmidt Sciences and the U.S. National Science Foundation; institutes such as the Space Telescope Science Institute, Wellcome Sanger Institute, the Francis Crick Institute and EMBL; and projects including FlyWire at Princeton and gnomAD at the Broad Institute. Anthropic says many projects are expected to produce public goods such as open datasets and software. No credit amounts are stated.
Our reading: routing credits through funders and institutes moves allocation decisions to organisations that already review scientific merit. That can scale support faster than individual applications and place it where peer review already happens. It also ties access to a single commercial model inside established institutions, which researchers outside those networks may not reach.
Anthropic notes that many collaborations are at an early stage, and the announcement reports no quantified results. The test will be whether the projects publish the open datasets and tools the programme anticipates, and whether those outputs remain usable independent of the credits that helped produce them.
Source: AI for Science expands to foundations and research institutions · 2026-10-28
Why it matters
Who distributes AI compute for science shapes which research gets accelerated and what remains openly available afterward.