OpenAI researcher Miles Wang in talks to launch AI drug discovery startup valued at $2B

Miles Wang spent his time at OpenAI figuring out how well AI models could accelerate scientific discovery. Now he's betting he can do it himself, outside the lab. According to four people familiar with his plans, Wang is leaving OpenAI to launch a startup that builds AI models for drug discovery, and he's reportedly in talks to raise around $200 million at a $2 billion valuation, with Lightspeed said to be circling as lead investor. Wang has pushed back on those numbers without offering alternatives, so treat the specifics as fluid until a deal actually closes.

What makes this interesting isn't just the price tag. Sources say the new company may focus on repurposing drugs that already exist, including ones that flopped in clinical trials, rather than chasing novel compounds from zero. That's a shrewd starting point: a drug that already cleared safety testing has a much shorter, cheaper path to a second life than something built from scratch. If AI can spot hidden uses for shelved or approved molecules, the payoff could arrive years faster than the traditional decade-plus drug development slog.

Wang isn't wandering into an empty field. Chai Discovery, another outfit with OpenAI DNA (co-founder Josh Meier is also an alum), just raised $400 million at a $3.8 billion valuation for its molecular-interaction models. Google DeepMind's Isomorphic Labs pulled in a $2.1 billion Series B back in May. Money is clearly flowing toward the bet that large models can shortcut biology the way they've shortcut writing code or drafting emails.

Wang's own path fits a pattern that's become familiar in this AI cycle. He dropped out of Harvard's computer science program, joined OpenAI in 2024, and co-authored papers on using AI to speed up scientific research before deciding to build the thing himself. A few years ago, venture capitalists might have hesitated to back a college dropout with no biotech company under his belt. Now that hesitation seems to have evaporated, especially when the founder's résumé includes OpenAI research credits and a handful of colleagues willing to follow him out the door.