The AI-Powered Biohub: Why Mark Zuckerberg & Priscilla Chan are Investing in Data, from Latent.Space

What It Covers

Mark Zuckerberg and Priscilla Chan discuss the Chan Zuckerberg Initiative's ten-year evolution and future focus on AI-powered biology through the Biohub network. They announce the acquisition of EvolutionaryScale and detail their strategy to build frontier biology labs paired with frontier AI labs, creating massive datasets and models toward a virtual cell capable of enabling precision medicine.

Key Questions Answered

• •Frontier Biology Plus Frontier AI: CZI combines cutting-edge biological tool development with advanced AI modeling in synchronized fashion, rather than having AI researchers work with existing datasets. This integrated approach designs new microscopes and data collection techniques specifically to generate the types of data needed to train better biological models, creating a virtuous cycle between wet lab experimentation and computational modeling that traditional grant-funded research cannot achieve.
• •Human Cell Atlas to Billion Cell Project: The initial Cell Atlas took ten years and significant funding to catalog 125 million cells, with CZI contributing 25 percent of data while the broader ecosystem added 75 percent. The billion cell project now completes in months at a fraction of the cost, demonstrating the acceleration pattern of slow initial data collection followed by rapid scaling once methodologies and models mature through iterative improvement.
• •Virtual Cell Development Strategy: Building biological models requires hierarchical understanding from molecules to proteins to cells to organ systems like the immune system. Models must incorporate multiple dimensions including spatial data from cryo-electron microscopy, temporal dynamics, transcriptome expression patterns, and cross-species conservation analysis. Each level of abstraction requires different scientific disciplines working together rather than in isolation, which traditional funding models fail to enable effectively.
• •EvolutionaryScale Acquisition and Leadership: CZI acquired EvolutionaryScale, creators of the ESM3 protein model, with CEO Alex Rivas leading the combined AI and biology program. This signals AI research as fundamental rather than supplementary to the mission. CZI operates one of the first large-scale compute clusters dedicated to biological research and commits to releasing frontier models, positioning the organization as both a leading biology lab and AI lab simultaneously.
• •Precision Medicine Through Genetic Variants: Current medicine treats variants of unknown significance as diagnostic mysteries, leaving patients uncertain about genetic findings that may or may not indicate disease risk. Future models will simulate how individual genetic variants affect cellular behavior and disease pathways, enabling true n-of-one treatments. This applies beyond rare diseases to common conditions like depression, where treatment currently relies on empirical trial-and-error over months rather than biology-based predictions.
• •Engineered Immune Cells as Diagnostic Tools: The New York Biohub develops cellular engineering approaches where immune cells enter organs like the heart, detect problems such as arterial plaques, record findings into their DNA, self-lyse, and release cell-free DNA readable as binary diagnostic signals. Subsequent engineered immune cells could then clear detected plaques. This leverages the immune system's natural mobility and privileged access throughout the body for both diagnosis and treatment.