Kevin Weil leaves OpenAI and raises $215 million for drug discovery startup
Warp Sciences spent six months building a system that it says can make the slowest and most expensive part of developing new medicines much faster. Kevin Weil, who recently left OpenAI's science division, is leading the company.
Warp Sciences announced its arrival on October 22, along with $215 million in funding, according to a report by Bloomberg. Lightspeed Venture Partners led the funding round, with IVP, Premji Invest and Lux Capital also taking part. The funding values the company at $865 million before it has released a product to the public. This shows how much investors still want to back people who worked closely on OpenAI's main AI models.
Weil knows that work well. He spent nearly two years at OpenAI. First, he was chief product officer, leading the teams that turned the company's most advanced AI models into ChatGPT, Codex and the OpenAI API. Later, he led a group called OpenAI for Science. That team built GPT-Rosalind, an AI model trained specifically on chemistry, protein engineering and genomics. Its goal was to use AI to carry out scientific experiments and understand their results, rather than simply answer questions about science. Weil announced the science group's launch himself on X in September 2025. He described the effort as an attempt to build "the next great scientific instrument."
The effort did not last. Weil and Bill Peebles, who led Sora, both left OpenAI on April 17, 2026, TechCrunch reported. Their departures were part of a larger reorganization. OpenAI moved its science work into its main model, product and infrastructure teams and shut down Prism, a web app it had built for researchers. An OpenAI spokesperson described the changes as a way to bring teams together ahead of a possible IPO, rather than a sign that the company had given up on its science plans. Weil's team, which had about ten people, was moved into the group led by Thibault Sottiaux, OpenAI's head of Codex.
Warp Sciences has a more focused goal than the broad idea of using AI for science. According to Bloomberg's report, the company is targeting one specific problem in drug development: using AI to speed up a step that can currently take years and cost hundreds of millions of dollars before a drug compound reaches testing in humans. This is a smaller, more business-focused goal than the wide-ranging research platform Weil described at OpenAI. It may also be easier to attract funding for. Investors can put money into a startup that promises to cut a known business cost, even when it is valued at $865 million and has not yet released a product.
Earlier reports by Quartz and Inc. in August said Weil was raising money for the same project, but the original target was lower: about $150 million at a valuation above $750 million. The final funding round was larger on both measures. The company raised $215 million at a valuation of $865 million. This suggests that interest from Lightspeed and the other investors was greater than the amount Weil had first planned to raise.
This is happening as other AI companies compete for top researchers. Anthropic, Google DeepMind and several independent companies with large amounts of funding are all trying to hire experienced researchers who can show that they helped build and release an advanced AI model. Each company would prefer these people to work for it rather than start their own businesses. Weil is instead joining a growing group of senior AI company leaders who have chosen to build their own companies. One example is former OpenAI chief technology officer Mira Murati, who started Thinking Machines Lab. A valuation of $865 million on the day Warp Sciences made its public debut shows how much venture capital investors may be willing to pay for this kind of decision.
Warp Sciences has not yet announced what product it will release or whether it will use any of the training methods behind GPT-Rosalind. Bloomberg's report did not give a product name or launch date. It only said the company planned to announce its funding round this week. The fact that a startup with so much money still has no public product is notable in itself. Investors backed Weil's past work and the size of the problem he wants to solve, rather than a product demonstration.