Anthropic has built its first wet lab in the San Francisco Bay Area, giving the AI company a physical place to run biology experiments rather than only modeling them in software. The lab is designed to let the company’s Claude AI direct robots through real bench work, and the early focus is on treatments for rare diseases that traditional drug development has largely ignored.<\/p>\n
A wet lab is a workspace set up for hands-on experiments with liquids, cells, proteins, and other biological materials, as distinct from “in silico” work that happens entirely inside a computer. Anthropic has said publicly that it wants to move beyond computer evaluations of biology, and the Bay Area facility is the first concrete step. The lab is built around automation: Claude issues instructions to robotic systems that then carry out the bench steps.<\/p>\n
The company’s head of life sciences described the approach as standard for biotech, with some experiments run inside Anthropic’s own facilities and others handled through external partners. That mix gives the team a way to validate AI-directed workflows on real biology while still drawing on established contract research capacity for scale-up work.<\/p>\n
Anthropic has framed the lab as a way to chase conditions the pharmaceutical industry has historically passed over. Several disease classes are considered “undruggable” because the underlying biology is hard to reach with conventional small-molecule chemistry, and the patient populations are often too small to make traditional clinical trials financially attractive.<\/p>\n
AI-driven lab automation is well suited to that problem in two ways. Automated systems can run thousands of tightly controlled experiments in parallel, which lowers the cost of exploring targets that have a low probability of success. And language models can mine the scientific literature for connections a human team might miss, narrowing the list of targets worth testing in the first place.<\/p>\n
The wet lab is part of a wider build-out rather than a standalone project. In recent months Anthropic has released science-focused software, added the chief executive of Novartis to its board, and acquired a biotech startup for $400 million in stock. Together, those moves suggest an AI company that is no longer content to stay on the software side of drug discovery.<\/p>\n
The board addition is a signal of intent: bringing a major pharmaceutical chief executive into governance gives Anthropic direct access to the regulatory, clinical, and commercial realities of drug development. The acquisition gives the company an existing scientific team and pipeline to integrate with the new lab infrastructure.<\/p>\n
Even as the lab opens, Anthropic’s leadership has been pushing for new rules around frontier AI. The chief executive has called for a framework to govern advanced AI systems and has urged the United States government to tighten export controls on advanced chips to China. In a recent television interview he described the risk of rogue AI agent swarms potentially taking over the internet in as little as six months, and called that prospect a warning sign that development needs to slow down.<\/p>\n
Other major AI chief executives publicly backed the call for a new governance framework on social media. US lawmakers have also been pressing for tighter rules. The US and China are scheduled to hold an AI safety summit this month, and a meeting between the two heads of state on 24 September is expected to include AI on the agenda.<\/p>\n
The US president has pushed back on what he described as exaggerated warnings, arguing instead that the United States needs to stay ahead of China in AI to remain secure and competitive. That tension between calls for restraint and pressure to maintain a lead is likely to shape the policy environment around any AI-driven lab work in the United States for the foreseeable future.<\/p>\n
Anthropic is now one of a small group of AI companies that operate their own biology labs rather than only partnering with outside ones. Owning the lab gives the company a faster feedback loop: an AI model can propose an experiment in the morning, a robot can run it by afternoon, and the results feed back into the model overnight. That loop is what most AI drug discovery companies are trying to build, and most still have to assemble it through external partnerships.<\/p>\n
For researchers working on rare diseases, the practical question is whether AI-directed labs can produce candidates quickly enough to matter clinically. The early work will be measured in published targets, validated assays, and partnerships with outside groups that can take leads into clinical trials. If the lab delivers on that pipeline, it becomes a template for how frontier AI labs might absorb parts of the biotech stack that have until now lived outside the technology industry.<\/p>\n
It is a physical laboratory in the San Francisco Bay Area where Anthropic runs biological experiments, complementing its earlier “in silico” computer-based biology work. Claude AI directs robotic systems to carry out the experiments.<\/p>\n
The lab is targeting rare diseases and conditions considered “undruggable,” meaning they are too expensive or scientifically difficult for the traditional pharmaceutical industry to pursue.<\/p>\n
The company’s leadership has warned about the risks of advanced AI systems, including the possibility of rogue AI agent swarms, and has urged the US government and other countries to agree on a framework that keeps those risks under control.<\/p>\n