The era of "all-purpose chatbots" is facing a brain drain as the industry’s top architects migrate toward the hard sciences. Miles Wang, an OpenAI researcher whose work focused on accelerating scientific and biological breakthroughs, is leaving the ChatGPT creator to launch his own drug discovery startup. According to four sources familiar with the matter, several other OpenAI employees plan to join him. This exodus signals a major shift: a pivot from building universal language engines to vertically integrated systems for generative molecular design. In this space, valuation isn't built on essay-writing subscriptions, but on the real economy.
A $2 Billion Premium on Scientific Logic
Wang is currently in talks to raise approximately $200 million at a $2 billion valuation. Sources suggest the round could be led by Lightspeed Venture Partners. Wang, unsurprisingly, disputes the figures without offering alternatives. Such a valuation for a company without a finished product is a pure "talent premium" for those capable of translating LLM scaling laws to biological data. Investors are no longer just buying software; they are betting that the logic which conquered human syntax will be equally successful at cracking the syntax of protein interactions.
The funding discussions highlight the feverish investor interest in applying AI to drive breakthroughs in life sciences.
This trend is already being validated by market activity from OpenAI and DeepMind alumni. Chai Discovery, founded by former OpenAI researcher Joshua Meier, recently raised $400 million at a $3.8 billion valuation for models predicting molecular interactions. Meanwhile, Isomorphic Labs (a Google DeepMind spin-off) closed its Series B in May. The industry is clearly drifting toward a model where the most valuable assets are systems capable of bypassing regulatory hurdles. Finding new applications for existing or "failed" drug candidates offers a much faster path to revenue than developing drugs from scratch.
OpenAI as a Scientific Incubator
For Sam Altman, the departure of Miles Wang and his colleagues highlights a strategic risk: OpenAI is becoming an elite incubator for its future competitors in applied niches. Wang joined the company in 2024 after dropping out of Harvard and co-authored papers on automating scientific discovery. He acquired the exact expertise needed to build a specialized AI player. While OpenAI maintains its lead in general-purpose models, it is losing the specialized talent in Scientific Discovery required to dominate high-margin biotech.
Investors are increasingly betting on young founders with "OpenAI pedigree" to reduce clinical trial failure rates. The valuation bubble for pre-product biotech AI startups continues to grow despite a lack of FDA-approved results. Vertical AI focus: The shift from general text generation to molecular design marks the next phase of the AI gold rush.
Investors may be eager to back founders who promise to revolutionize medicine, but the reality remains sobering. While AI can predict molecular hits at the speed of a GPU cluster, biological labs and the FDA continue to move at a pace best described as bureaucratic Paleolithic. We were promised a software revolution in medicine, but so far, we have mostly received a series of multi-billion dollar valuations for companies that have yet to put a single AI-designed molecule on a pharmacy shelf.