Anthropic wants to develop its own drugs
TL;DR
Anthropic announced Claude Science: a workbench meant to combine scientific datasets, analysis tools and figure generation for researchers. The company points to existing biotech and pharma customers, but offers few concrete proof points on actual drug-development wins. Much of this still reads like a platform pitch. The bigger move is that Anthropic says it wants to develop drugs of its own, initially focused on neglected diseases.
Nauti's Take
The interesting part is not another „AI workbench“. Platforms have promised to clean up fragmented research for years.
The hard part starts when a model does more than summarize papers and produce charts, and instead generates hypotheses that survive the lab and later clinical trials. Anthropic is making a big claim here, but not yet showing a big proof.
For pharma, the signal is clear enough: whoever controls the models, workflows and research data will eventually want more than SaaS fees.
Briefingshow
If AI labs only build tools for pharma, they sell software. If they develop drugs themselves, they enter a field where data, IP, regulation and clinical evidence create the real value. Anthropic is testing whether model companies can become more than infrastructure providers: actual players in the research pipeline.
Sources