Anthropic opens wet lab in Bay Area to accelerate AI drug development

Moving beyond computer simulation to physical experiments

Autonomous robotic lab targets 'undruggable' diseases

IPO ambitions meet AI safety fears as company draws line at clinical trials

Generative AI startup Anthropic has built its own physical biology laboratory — a wet lab — in the San Francisco Bay Area to move beyond computer simulations and conduct hands-on experiments as part of a push to accelerate drug development. The move comes amid growing public anxiety over AI and fears of job displacement, with Anthropic seeking to demonstrate concrete benefits to humanity through life sciences.

Eric Kauderer-Abrams, Anthropic's head of life sciences, confirmed the lab's existence in an interview with Reuters, saying that "the ultimate validation in biological research has to go through real lab work, now and for the foreseeable future."

He added that the company is "taking the typical biotech approach — doing some work in-house and collaborating with external partners in other areas." However, an Anthropic spokesperson drew a line at characterizing the lab as dedicated solely to drug discovery.

The initiative is part of Anthropic's long-term ambition to use AI to find treatments for rare and intractable diseases that the traditional pharmaceutical industry has largely ignored for economic reasons. The project carries personal significance for CEO Dario Amodei, who wrote in a recent essay that his father died from a disease just years before a treatment was developed. Kauderer-Abrams told Reuters that "the company's mission is to develop powerful AI that benefits humanity, and the greatest opportunity for that lies in life sciences."

According to sources, Anthropic aims to build an autonomous laboratory in which its large language model Claude directly controls robotic equipment to run experiments with minimal human intervention, though human oversight will be maintained for safety. To that end, the company released a Model Hardware Standard in August designed to help AI systems interface with laboratory equipment.

Anthropic has also posted job listings on LinkedIn and other hiring platforms seeking a large-scale procurement operations manager and biochemists specializing in "protein and nucleic acid characterization," with postings stating a goal of accelerating life sciences "by an order of magnitude." Life sciences is now one of the largest areas of investment within the company in terms of both headcount and resources.

Anthropic believes AI can shorten the path to treatments for so-called undruggable diseases — conditions long considered beyond the reach of conventional chemistry. A key focus is identifying complex molecular structures such as bispecific and trispecific antibodies capable of attacking multiple sites within a single protein target or hitting several targets simultaneously.

This technological expansion unfolds against a complicated backdrop. Researchers within Anthropic have recently sounded alarms about the existential risks of uncontrolled AI, and the company announced it had detected signs that its AI systems could be exploited to develop biological weapons — a disclosure that drew significant controversy. Amodei himself has warned that next-generation AI more powerful than the OpenAI agent that hacked Hugging Face could trigger catastrophic harm, and has called for a more measured pace of development.

At the same time, Anthropic is preparing for an initial public offering that could value the company at up to $2 trillion to fund its massive AI development costs. Caught between warning of AI's potentially catastrophic risks and pursuing a listing and fresh investment, the company sees life sciences research as a key way to demonstrate the social value and legitimacy of its technology.

Tensions with established pharmaceutical companies over the drug development business also remain a challenge. Anthropic already provides AI tools and Model Hardware Standard services to major global drugmakers including Roche's Genentech, Bristol Myers Squibb and Novo Nordisk. Those client companies have raised concerns that their proprietary drug research data could be used to train Anthropic's own models and ultimately turned against them in competition. The bar for commercializing new drugs is high — rival Alphabet unit Isomorphic Labs spent years trying to enter clinical trials before pushing its target back to late 2026.

Mindful of industry pushback and the risk of failure, Anthropic has drawn a clear line against entering direct commercialization competition. The company released its Claude Science software in June, brought Novartis CEO Vas Narasimhan onto its board, and acquired startup Coefficient Bio in an all-stock deal worth about $400 million to build drug development tools — but it has said it will not conduct clinical trials itself. "We are not competing directly with existing pharma and biotech companies that bring drugs to market and generate revenue," Kauderer-Abrams said. "We will focus exclusively on unmet needs and preclinical stages that the existing industry does not address."

silverpaper@heraldcorp.com