Eric Schmidt is building an OpenAI-style moonshot for science

Silicon Valley spent the last two decades putting its best minds to work on getting people to click ads. Former Google CEO Eric Schmidt seems ready to fund something else.

Schmidt is building an ambitious new nonprofit organization to tackle major scientific challenges with AI, according to people briefed on the plans reported by Semafor. The initiative focuses on breakthroughs in areas like drug discovery and materials science.

To lead the project, Schmidt has recruited two rising stars: Samuel Rodriques, founder of the Applied Biotechnology Laboratory at the Francis Crick Institute, and Andrew White, a University of Rochester professor pioneering AI in chemistry.

Breaking down the lab silos

Those familiar with the initiative say it takes direct aim at what academic labs lack most: massive compute and competitive pay.

Schmidt plans to fund the venture primarily from his personal fortune, dangling tech-grade compensation and server power that university faculties can rarely afford. Outside funding might also be tapped as the project expands.

Last month, Schmidt penned an essay in MIT Technology Review arguing that AI will soon make science faster and almost unrecognizable, with fallout felt far beyond academic walls.

His two recruits are already deep in the trenches. White, who authored an online textbook on deep learning for molecules, wrote in May that GPT-4 can tackle complex chemistry purely from plain English. He even campaigned the Unicode Consortium for an official protein emoji, though it got rejected.

Rodriques has pointed out the biggest roadblock holding AI biology back: AI researchers get terrified by messy biological complexity, while biologists quickly decide AI cannot be trusted. His solution is simple: put both camps under one roof with rapid iteration cycles.

Borrowing the OpenAI playbook

People familiar with the effort say it is modeled after the early days of OpenAI, which launched as a starry-eyed philanthropic outfit to attract world-class brains.

While currently planned as a nonprofit, the new entity could eventually license intellectual property or spin off commercial startups, just like major research universities do.

Schmidt's post-Google moves, however, come with baggage. His aggressive push into U.S. defense policy and military technology has earned him vocal critics, meaning any growing footprint in federally adjacent science is bound to draw close scrutiny.

Why it matters

For researchers stranded in academic bureaucracies, this setup looks like a dream ticket. If you are a scientist bottlenecked by grant cycles and scarce GPU clusters, joining an elite, well-capitalized hybrid lab is an obvious leap worth taking right now.

For the tech ecosystem, it accelerates a long-overdue migration from web apps to the physical world. DeepMind proved what AI could do for biology with AlphaFold, and Schmidt is building a dedicated engine to push that logic into mainstream chemistry and medicine.

We think launching as a nonprofit is clever because it bypasses venture capital's obsession with quick revenue, letting them lure top researchers who would otherwise shun corporate labs.

That said, the model has familiar pitfalls. Hard science burns cash faster than code, and personal wealth only stretches so far. If this moonshot eventually pivots to commercialization to pay its compute bills, the lofty scientific charity pitch will ring just as hollow as OpenAI's original promise.