Biohub brings Google DeepMind, Isomorphic Labs, Meta and the US government into its plan to build a predictive model of a living cell, with commitments reaching $1,8 billion
Biohub brings Google DeepMind, Isomorphic Labs, Meta and the US government into its plan to build a predictive model of a living cell, with commitments reaching $1,8 billion
On 7 October 2026, the nonprofit institute Biohub, the US Department of Energy, the US National Institutes of Health, Google DeepMind, Isomorphic Labs and Meta announced an expansion of the Virtual Biology Initiative, a programme to collect data for training AI models of cells. Total commitments rose from $500 million in April to $1,8 billion, and the government joined the initiative for the first time.
Biohub is a nonprofit research institute founded by Mark Zuckerberg and Priscilla Chan that uses artificial intelligence to study biology. For ten years, the institute developed projects to measure living cells separately, including the human cell atlas Tabula Sapiens and the shared single-cell data repository CELLxGENE, each with its own funding and leadership. In April 2026, Biohub brought these efforts together under one programme and committed $500 million to it. The Virtual Biology Initiative aims to collect data on how different types of living cells respond to drugs and other interventions. These data could train an AI model to predict a cell’s response to an experiment, replacing the need to test each response in the laboratory. Biohub acknowledged a limitation from the outset: the volume of data required is orders of magnitude greater than any single institute can produce. It therefore invited others to join.
Within six months, others had taken up the invitation.
The US Department of Energy will invest more than $500 million over five years through Genesis Mission, a nationwide programme to accelerate science through AI that Trump launched by executive order in November 2025. Federal commitments to that programme have already exceeded $5 billion. The department is contributing its research infrastructure to the project, including exascale supercomputers (among the fastest in the world), X-ray and neutron sources, cryo-electron microscopes and autonomous laboratories at its national laboratories.
The US National Institutes of Health are contributing existing resources without providing new funding. Their participation comes through their own Bio Genesis Mission, launched in July as the NIH contribution to the broader Genesis Mission, which identifies predicting the behaviour of living systems as one of six objectives. NIH will convert existing databases and repositories, developed with more than $500 million in federal funding, into a common format for AI training. These include the infrastructure of the National Center for Biotechnology Information, which supports nearly all biomedical research worldwide. Google DeepMind, the drug discovery startup Isomorphic Labs (a separate Alphabet company that grew out of Google DeepMind and designs drugs using AlphaFold, a neural network that predicts protein structure from amino acid sequences), and Meta are jointly contributing another $300 million, even as Google DeepMind and Meta compete elsewhere to build the best AI models. “We cannot solve this problem without open experimental biological data on an unprecedented scale,” explained Pushmeet Kohli, Google DeepMind’s vice president of AI for Science.
“Building a virtual cell is one of the most important challenges of the coming era of science,” said Biohub’s chief scientist, Alex Rives. Isomorphic Labs president Max Jaderberg described participation in the initiative as a step towards the next breakthrough in biology. Nicole Kleinstreuer, deputy director of the relevant NIH office, called it a way to “substantially shorten the time to medical discoveries compared with trying to achieve the same results through laboratory experiments alone”.
The data collected will become an open resource for the entire scientific community.