Drug discovery is undergoing a quiet but profound transformation. For decades, the pharmaceutical industry has relied on a slow and laborious cycle: a researcher conceives a molecule, a chemist synthesizes it, an analyst tests its effects, and then waits weeks or months for the results. Each iteration advances knowledge by millimeters. To find a viable treatment, thousands of these millimeters are needed — and years of exhausting work. This is precisely the scenario Kevin Weil, a former OpenAI executive responsible for AI-for-science initiatives, set out to tackle.
Weil announced today the launch of Warp Sciences, a $215 million startup aimed at automating wet-lab work — the technical term for classical experimental chemistry, where compounds are actually mixed, heated, and analyzed rather than merely simulated on computers — using robots and artificial intelligence. The company was valued at $865 million, according to Bloomberg, with a round led by Lightspeed Venture Partners and participation from IVP, Mayfield, Lux Capital, Premji Invest, and Menlo Ventures. But what truly stands out is not the money, but a roster of angel investors that includes Sarah Friar herself (OpenAI's CFO), Fidji Simo (ex-OpenAI AGI Deployment), Brian Armstrong (Coinbase CEO), Reid Hoffman (LinkedIn co-founder), and Sébastien Bubeck (OpenAI research lead).
Warp Sciences' proposition is ambitious on two fronts. First, it aims to automate the drug candidate optimization stage — the moment researchers identify a promising molecule and must create and test hundreds or thousands of variations to improve its activity, reduce side effects, or refine other properties. This is one of the longest and most expensive phases of drug development, and also one of the most repetitive. Secondly, and here lies the geopolitical play, Weil wants to bring it back to the United States. For years, biotech companies have outsourced the most labor-intensive work of synthesizing thousands of molecule variations to contractors in China, such as WuXi AppTec. The pandemic and trade tensions between Washington and Beijing have already made clear the risks of this dependency. Warp Sciences wants to offer a domestic alternative, powered by automation.
The company's competitive advantage lies in how it trains its robots. Rather than programming predefined movements, Warp's chemists wear cameras and gloves with pressure sensors to record every gesture they perform in the lab. These data are used to train AI models that will eventually learn to perform complex physical tasks — pipetting solutions, mixing compounds, preparing samples — as well as an experienced human. Once trained, robots can operate continuously, without breaks, accelerating the experimental cycle in ways that are simply not physically possible with human teams. It is the same reasoning that led Boston Dynamics to build robots capable of operating in dangerous industrial environments, but applied to a radically different domain: pharmaceutical chemistry.
The Warp Sciences founding team combines expertise in science, robotics, and artificial intelligence. In addition to Weil, the team includes Manohar Paluri (former vice president of Meta's Superintelligence Labs), DJ Patil (the White House's first chief data scientist), and Solomon Reisberg (former executive at Terremoto Biosciences). The company already has about 20 employees and has acquired equipment including robotic arms and AI chips. And, unlike many biotech startups trying to develop their own drugs, Warp Sciences will not compete with pharmaceutical companies: it will be a service company, offering discovery capacity as a contract (CRO), like WuXi AppTec, but automated and based in the United States.
What makes Warp Sciences particularly interesting is that it represents an unexplored frontier of applied AI: physical intelligence. In recent years, we have seen language models write texts, generate images, and even make complex decisions. But the ability to understand and manipulate the physical world — a classic objective of robotics — remains a monumental challenge. Warp Sciences is trying to solve this problem not as academic research, but as a commercial product, with real customers paying for real results. It is a risky bet. AI models need to generalize to tasks that humans took years to master, and chemical unpredictability is brutal: humidity, temperature, impurities in reagents can completely alter a result predicted in simulation.
But if Warp Sciences can scale, the impact will go far beyond pharmaceuticals. The same autonomous laboratory architecture can be adapted for materials, energy, agriculture, and materials science. Each successful experiment generates data to train better models; better models guide better experiments. It is a virtuous cycle of accelerated discovery that, if it works at scale, could reduce drug discovery time from years to months.
Warp Sciences is betting that the physical automation of science is not just another efficiency tool, but a fundamental shift in how discovery itself happens. The companies that master autonomous laboratories first will have a structural advantage in an era where the pace of scientific progress determines competitive advantage.
Sources: TechCrunch, Bloomberg, Drug Development
✓ Independent sources cross-checked and verified before publishing
