Anthropic has confirmed it is running a wet lab in the San Francisco Bay Area, giving the Claude maker its own physical biology capacity to test AI-generated hypotheses against real experiments. The confirmation, reported by Reuters and Quartz on September 18, comes as Anthropic pushes its Claude Science stack deeper into pharma and rebrands life sciences as one of its most heavily resourced units.
From Software To Test Tubes
Anthropic head of life sciences Eric Kauderer-Abrams told Reuters the company is doing real bench work today, arguing that "to do biology, the final test is still and will be for a while in real lab work." The Bay Area lab is aimed at rare diseases traditionally neglected by big pharma, and it is being used to validate binder-molecule and target predictions generated by Claude. The company recently disclosed that Claude Mythos 5.1 hit high-affinity binders in about 50% of attempts against a benchmark where the industry standard is 10–15%.
A Coefficient Bio-Sized Move
The wet lab expansion follows Anthropic's roughly $400 million acquisition of Coefficient Bio earlier this year, which supplied the initial biology team and infrastructure. Anthropic frames the lab as complementary to, rather than a replacement for, its enterprise Claude Science partnerships with Novo Nordisk, Bristol Myers Squibb and Roche's Genentech; the company still plans to license validated discoveries rather than run its own drug pipeline.
Robots On The Bench
The lab is also a proving ground for Anthropic's Model Hardware Standard (MHS), a protocol that lets Claude drive lab robots and instruments directly. Howard Hughes Medical Institute is separately piloting MHS-driven Claude workflows in Virginia, hinting at where Anthropic wants its "AI scientist" stack to land next. The rollout underlines how frontier AI labs are drifting into physical R&D, previously the exclusive turf of large biotechs and CROs.
Reporting based on coverage from Reuters, Quartz, SiliconANGLE and The Next Web.