OpenAI Foundation Bets $125M on Health Data Grants

The OpenAI Foundation launched Public Data for Health, a $125M grant program funding health datasets, including files from failed biotechs.

On September 15, 2026, the OpenAI Foundation introduced Public Data for Health, its second science program, with more than $125 million in initial grants to create and preserve high-quality scientific datasets made broadly available to researchers. Three initial grantees were named: OpenADMET, which runs blinded competitions testing whether AI models can predict how small molecules move through the body; CTD Commons, led by 1Day Sooner, which received $500,000 to test acquiring Common Technical Documents from failed or shelved drug programs; and the University of North Carolina at Chapel Hill, which received $40 million to collect data about novel cancer vaccines through its Initiative for Generative Immunotherapy.

The biotech archive grant turns an unusual idea into funded reality: last year, policy analyst Ruxandra Teslo proposed bidding at bankruptcy proceedings for failed biotech companies to obtain regulatory filings, manufacturing strategies, and safety data usually treated as trade secrets — what she called 'biotech's lost archive.' According to MIT Technology Review, the OpenAI Foundation is now funding that idea, pursued by 1Day Sooner. unite.ai notes the premise is counterintuitive: the foundation argues that in biology, unlike mathematics, AI breakthroughs require more observations of the world, and some datasets of enormous public value may never exist because no single institution has the incentive or capacity to fund them.

MIT Technology Review reports that 1Day Sooner already holds three datasets, two donated by Lumen Bioscience, a biotech that previously used a Chapter 11 strategy to gain insights into another company's drug development. President Josh Morrison said nonexclusive copies of company datasets could be acquired for 'a few tens of thousands of dollars' each, though two other bid attempts this year failed. Common Technical Documents compile an investigational drug's full journey — animal toxicology, manufacturing details, and FDA correspondence — of which only a small fraction appears in published papers, per the foundation's announcement described by unite.ai. unite.ai also cites OpenADMET's rationale: 90% of drug candidates fail in clinical trials, often because absorption and distribution are hard to predict.

The foundation, holding a 26% equity stake in OpenAI ahead of a potential IPO that MIT Technology Review says could value the company at $1 trillion, is on track to become the richest charitable organization on the planet, potentially sitting on $250 billion in stock value — ahead of the Gates Foundation's roughly $180 billion at the end of 2025. According to MIT Technology Review, executive Jacob Trefethen said the foundation hopes to give away $1 billion by the end of the year, with grantmaking ramping up only this year. unite.ai reports the foundation expects funded data to be broadly available, encourages preprints and regular sharing, and is inviting ideas at science@openaifoundation.org while hiring for four life sciences roles.