Kevin Weil's departure from OpenAI arrived with little fanfare on April 17, but the timing spoke volumes. According to WIRED, Weil announced on X that it was his "last day at OpenAI, as OpenAI for Science is being decentralized into other research teams." The same day, the company quietly shuttered Prism, a web application for scientists that Weil had launched just three months earlier in January.
The move comes as venture capitalists have poured billions into AI-powered scientific research startups, pushing aggregate valuations past $20 billion across a handful of young companies. Whether Weil plans to launch his own venture in the space remains unclear, though his recent moves suggest he's staying close to the sector.
The sums involved are staggering, even by Silicon Valley's inflated standards. Periodic Labs was in talks this past March to raise hundreds of millions at a $7 billion valuation, Bloomberg reported. That represented a sharp jump from the company's previous raise of $235 million at an approximately $1.23 billion valuation, which came from Andreessen Horowitz, Felicis, and Accel. Around the same time, Lila Sciences was reportedly discussing a roughly $2 billion raise at an $8.5 billion pre-money valuation, according to Bloomberg Law. By July, CuspAI had closed a round led by Kleiner Perkins and NEA at a $2.6 billion valuation, Yahoo Finance reported, with Jeff Bezos's investment arm participating.
The frenzy hasn't been limited to startups. Anthropic, OpenAI's chief rival in large language models, acquired Coefficient Bio for just over $400 million in stock this April, Newcomer reported. Two months later, Anthropic launched Claude Science, a research workbench designed to help scientists leverage AI for experimental work.
Weil hasn't disappeared from the industry entirely. On July 8, he joined the board of Stoke Space, a rocket company, TechCrunch reported. In a June 26 appearance on an Andreessen Horowitz podcast, he spoke with notable enthusiasm about AI's scientific potential, arguing that models are "going beyond the frontier of human knowledge." His vision, he said, was bringing "the science of 2050…in 2030 instead."
That timeline would require more than incremental progress. It would demand the kind of focused effort that standalone ventures, rather than divisions within larger organizations, tend to produce. Whether Weil intends to build such a company himself remains an open question.
