Zuckerberg's Biohub bets $1.8B on virtual cells: the bottleneck is data, not AI
HarperSCarroll · x · 2026-11-03
Alex Rives, Head of Science at Zuckerberg and Priscilla Chan's nonprofit Biohub, explains in a long interview why the biggest holdup for AI curing disease is not models but a 'fundamental data gap': we have barely measured how cells behave. The $1.8B effort aims to generate that data to build a 'virtual cell' that predicts cellular behavior.
Key takeaways
- Protein AI (ESM, AlphaFold) was built on 50 years of data; nothing comparable exists for cells.
- A virtual cell is judged by whether its predictions match real experiments.
- ESM learned structure and function on its own by guessing hidden amino acids.
- Zero-shot protein design narrows hundreds of thousands of candidates to 100 lab tests: '24 hours of computation, seven days of lab work.'
- Custom medicines aren't here yet: safety, developability, and regulation aren't built for one-patient therapies (see baby KJ).
- Biorisk today comes down to the lab (gene synthesis screening), not the AI; he floats the idea of a universal immune system.
Biohub open-sources its main protein models, including the ESM family Rives started at Meta FAIR.
More from AGI Musings
- Is consciousness computable? The debate that decides immortality in 20 years — fkasummer · 2026-11-03
- Jack Clark's weekly Tech Tales: children play god in 2040s city-building game — jackclarkSF · 2026-11-03
- MATS researcher studies who captures AI's economic gains, seeking interviews — luke_drago_ · 2026-11-03
- Researcher: I want one novel discovery to my name before AI sprints past us — generativist · 2026-11-03
- AI isn't more expensive than humans: capability cost curves crash, retail prices mislead — Nawozane · 2026-11-03
- Giraffe evolution meme: AI's low-hanging fruit keeps hanging higher every year — itsOmSarraf_ · 2026-11-03