AI Radar

Noteworthy AI conversations and writing, summarized from publisher notes.

Zuckerberg & Chan: AI, Virtual Cells, and CZI's Decade-Long Bet to Cure Disease

· The a16z Show

Mark Zuckerberg and Priscilla Chan join a16z to discuss the Chan Zuckerberg Initiative's (CZI) 10-year journey building infrastructure for modern biological research, and why they are now consolidating their efforts into the Biohub as the central pillar of their philanthropy. They explain the origin story of the Cell Atlas project—what began as a simple annotation tool (CZ Cell×Gene) inadvertently became the de facto data standard for single-cell biology, with ~75% of its millions of cells contributed by the broader scientific community. Zuckerberg argues that breakthroughs in science are typically preceded by new observational tools, and that biology is still missing a "periodic table of elements"-equivalent foundation. The core of the conversation centers on virtual cell models—a grand challenge they believe could compress the disease-curing timeline from a century to something significantly sooner. They describe a hierarchical approach: state-of-the-art protein models (built with the EvolutionaryScale team joining from Meta) feeding into cellular models, eventually enabling virtual immune systems and personalized simulations. The goal isn't 100% accuracy but enough directional signal to de-risk hypotheses in silico before expensive wet-lab work. Zuckerberg announces that the Biohub will now operate as a unified entity under Alex Reeves (formerly of EvolutionaryScale and Meta's protein-folding work), combining frontier biology teams in SF, Chicago, and New York with a central AI team. They highlight their compute cluster expansion from ~1,000 to 10,000 GPUs as the new "lab space," and emphasize their open-source philosophy—giving away tools that others won't build because the credit accrues to users, not builders. Chan adds that the user interfaces are intentionally designed for low barrier of entry so immunologists can contribute to neurodegeneration research, biologists can engage with engineering, and interdisciplinary collaboration can flourish organically.

AI Radar wrote this summary from the publisher’s notes with AI, not from a transcript. How it works.