Biohub Leads $1.8 Billion Push for AI Virtual Cells
Biohub, Google, Meta and U.S. agencies are pooling funding and biological datasets to build models that predict cellular behavior.
What happened
The Chan Zuckerberg-backed Biohub said it is coordinating a $1.8 billion effort to build open datasets and AI models for biological research. The initiative, described as a Virtual Biology Initiative, brings together Biohub, Meta, Google DeepMind, Isomorphic Labs, the U.S. Department of Energy, the National Institutes of Health and scientific institutions.
Meta, Google DeepMind and Isomorphic Labs are jointly investing $300 million. The Department of Energy is committing more than $500 million over five years for laboratory measurement, modeling and computation. NIH will coordinate and standardize datasets and repositories built with more than $500 million in earlier federal funding, while Biohub has already supplied a separate $500 million philanthropic commitment.
The scientific objective is to measure how cells respond to many more conditions than researchers can test manually, then use those observations to train predictive models. In principle, such systems could help researchers prioritize experiments, identify promising drug candidates and reduce the number of expensive laboratory iterations.
Why it matters
The significance lies in the data and measurement infrastructure rather than a model announcement. Biology models often fail because training data are fragmented, inconsistent or too narrow to represent real cellular variation. A coordinated program combining open datasets, standardized repositories and new experiments could address that bottleneck at national scale. The risk is that “virtual cell” becomes an expansive label for systems that predict limited experimental settings. Drug discovery timelines will only change if predictions transfer reliably into wet-lab results and remain useful across different cell types, diseases and interventions.