MOSCOW (Realist English). Anthropic, the developer of Claude AI models, has confirmed the existence of a “wet lab” in the San Francisco Bay Area. Reuters reported the launch of the facility, citing two sources, and Eric Kauderer-Abrams, head of life sciences, confirmed its existence to the agency.
What Is a “Wet Lab” and Why Is It Needed?
A “wet lab” is a facility equipped for physical biological experiments: cell cultures, biochemical assays, and molecular research. This is a fundamentally different level compared with the “dry” computations (in silico) that had previously limited Anthropic’s work in biology.
“We believe that for biology, the final validation will still, and for some time yet, take place in real laboratory work,” Kauderer-Abrams said.
“We absolutely do this today. I would describe our approach as typical of most biotech companies, where part of the work is done in-house and part with external partners.”
A company representative later clarified that the lab is not intended specifically for drug discovery but declined further comment.
Goals: Accelerating Preclinical Work and “Incurable” Diseases
Anthropic says it intends to conduct preclinical research in areas that traditional pharmaceutical companies consider financially unattractive.
The company wants to use Claude to accelerate the search for complex molecules, including bispecific and trispecific antibodies capable of acting on multiple targets simultaneously.
According to Kauderer-Abrams, AI can accelerate the development of treatments for diseases previously considered “incurable” — too complex for therapy. The company does not disclose which diseases are prioritized or what progress has been made.
An important limitation: Anthropic does not yet conduct clinical trials and, according to Kauderer-Abrams, does not compete with pharmaceutical companies whose business is built on bringing drugs to market.
Background: From the Acquisition of Coefficient Bio to Claude Science
The lab is not Anthropic’s first step into biology. In April 2026, the company acquired Coefficient Bio — a “stealth” AI-biotech startup with a team of fewer than 10 people — for approximately $400 million in stock. The company was founded by researchers who had worked at Genentech on computational drug discovery.
In June, Anthropic launched Claude Science software designed for laboratory research. At the same time, the company announced its intention to engage in preclinical work in areas that traditional companies avoid.
In addition, Vasim Narasimhan, CEO of Novartis, joined Anthropic’s board of directors. The company also provides AI tools to pharmaceutical giants such as Genentech (Roche), Bristol Myers Squibb, and Novo Nordisk.
Balancing Acceleration and Safety
Anthropic’s move raises questions against the backdrop of the company’s own statements. Two of its researchers recently warned that AI could lead to human extinction, and the head of alignment estimated that probability at more than 10% over ten years.
The company itself reported that it had discovered cases where its systems could be used for biological weapons development, which drew scepticism from some observers.
CEO Dario Amodei on 12 September called on the industry to slow the development of frontier AI models and introduce self-regulation. At the same time, Anthropic is preparing for an IPO with a potential valuation of $2 trillion.
“In our work in life sciences, we always balance these things,” Kauderer-Abrams said. “We make the models better in ways we believe raise and accelerate drug development, while thoughtfully and cautiously approaching their deployment to limit risk.”
Lab Automation: Claude Controls Robots
One Reuters source said Anthropic wants to use Claude to control robotic laboratory equipment with minimal human intervention.
To this end, the company has developed the Model Hardware Standard (MHS) — a protocol allowing AI models to control lab equipment, detect errors, and write automation scripts.
“We are at the very earliest stage of using AI to automate the execution of laboratory work,” Kauderer-Abrams said. “This area is potentially capable of bringing significant acceleration to many processes.”
At the same time, an Anthropic representative stressed that human oversight remains critically important for safety.
The Trust Problem with Pharma Companies
Anthropic simultaneously provides AI tools to major pharmaceutical companies and develops its own laboratory capabilities. Two Reuters sources noted that Anthropic’s clients may fear data leaks from their competing drug programmes, despite the company’s assurances of data isolation.
Kauderer-Abrams said Anthropic, partly for competitive reasons, has set a boundary: not to conduct clinical trials and to focus on needs the industry does not cover.
