AI Is About to Run the Whole Damn Lab — and Anthropic Just Bet $400 Million on It
Anthropic is pushing into biotech with a $400 million acquisition and a new Bay Area lab, as AI systems begin running experiments on their own. The moves signal a shift from AI as a research helper to AI as the researcher itself.
Anthropic, the artificial intelligence company known for its Claude models, is moving beyond software and targeting the human body. In its S-1 filing for a major initial public offering (IPO) next month, the company outlined a plan to become a key player in biological and medical research, presenting the life-sciences strategy to investors as both a path for financial growth and a way to support its mission [275860]. The document states that AI "could rapidly accelerate scientific innovation" and "bring cures for major diseases within reach" [275860].
The company is already spending heavily on the effort. It launched a product called Claude Science in June and started a molecular biology lab in the Bay Area this year. It also acquired the biotech startup Coefficient Bio for $400 million in stock and added Vas Narasimhan, CEO of pharma giant Novartis, to its board [275860]. According to the filing, nine of the 10 largest pharmaceutical companies in the Fortune 500 spent at least $100,000 with Anthropic over a recent year. Eli Lilly, a $1.3 trillion company, provides Claude enterprise accounts to its 50,000 employees, with over 85% actively using the tool [275860].
CEO Dario Amodei has personally championed the focus. In a 2024 essay, he wrote that "Biology is probably the area where scientific progress has the greatest potential to directly and unambiguously improve the quality of human life," and he has set ambitious goals, including hopes that AI will help eliminate most cancers and double the human lifespan [275860]. Public optimism is lower, however. An Anthropic survey from August found that 91% of its staff believe AI will make a Nobel-level discovery by 2030, compared with 39% of US adults. To date, no AI-designed drug has received approval from the Food and Drug Administration (FDA) [275860].
The push comes as new biotech startups question whether drug discovery is even the best use of artificial intelligence, arguing that AI's largest impact may lie elsewhere in the industry [279167]. At the same time, the technology is moving from helper to lead researcher. A new biology laboratory operates with almost no human hands: AI "agents" — software programs that make decisions and complete tasks on their own — now run the entire facility, from sequencing DNA to carrying out complex experiments, planning experiments, operating equipment, and adjusting their approach based on results without direct human control [279100]. Scientists have also built an experimental "AI-native" laboratory designed for autonomous biomolecular operations, in which researchers set the goals and the AI chooses experiments, runs equipment, and reviews results on biological molecules such as DNA and proteins with little or no human input [279131]. Supporters say the approach could speed up discovery and reduce human error, while critics question how much oversight such systems truly need [279100].
The shift is not limited to biology. At the AI startup Anything, 15 human employees work alongside roughly 150 agents, a ratio founder Dhruv Amin believes will soon be the norm, with companies boasting about doing more with less — an approach creating tension with their human workers [274253]. A new paper in Science (Volume 394, Issue 6823, October 2026) warns that as AI systems increasingly solve problems and find patterns first, the loss of science's "aha!" moment could drain research of its excitement and purpose for human researchers, and calls for a serious discussion about how science should adapt [274605].