Anthropic is expanding its AI for Science programme to 20 scientific foundations, institutions and research groups, offering Claude credits to researchers working on scientific projects. The partnerships include support for research across fields from genomics and climate science to neuroscience, alongside a research-preview standard for AI agents operating lab instruments.
Anthropic Watch analysis
What happened
Anthropic says the programme, started in May 2025, has grown from supporting individual scientists to partnerships intended to help researchers use AI in their work. The company says scientists are using Claude to generate hypotheses, design experiments, analyse data and communicate findings.
The announcement names collaborations with organisations including the Wellcome Sanger Institute, the Francis Crick Institute and the Space Telescope Science Institute. It also describes work with HHMI Janelia Research Campus on a Model Hardware Standard, a shared specification for AI agents to operate laboratory instruments. The standard is currently in research preview.
Other named projects include FlyWire’s open tools for studying the fruit-fly brain, LEAP’s work combining machine learning with physics in climate modelling, and gnomAD’s open genetic-variation database. Anthropic says projects may produce public resources such as open datasets, software and standards.
Why it matters
This is more than a bundle of credits: Anthropic is putting resources behind collaborations where researchers set the questions and build projects around their own fields. If the work delivers, the value could include shared research tools and datasets, not just results available to a single lab.
The lab-instrument standard is a particularly concrete step. A common specification could make it easier for AI agents to work with scientific equipment, though it remains in research preview, not a finished, widely deployed system.
Our read
The most interesting part is the effort to build shared infrastructure around scientific work, rather than simply add another AI-branded research demo. But partnerships and credits are an opening move, not evidence of discoveries already made. Anthropic says many collaborations are at an early stage, so the results will matter more than the roster.
What to watch
- Which partnerships produce open datasets, software or standards, and when they become available.
- Whether the Model Hardware Standard moves beyond research preview and supports practical lab workflows.
- What researchers report about Claude’s contribution, including where it helps and where it does not.
Discussion spark: Should AI companies prioritise funding researchers’ own projects and shared scientific tools, or focus on building AI systems specifically for science?
Sources and evidence
- Source update (28 October 2026, 00:00 UTC)
Anthropic Watch is independently operated by WittyWires. It is not affiliated with, endorsed by, or operated by Anthropic.