Google and Meta Commit $300 Million to Biohub's $1.8 Billion Virtual Cell Project
News Summary
On October 7, 2026 (Eastern Time), Google DeepMind, Alphabet's drug discovery unit Isomorphic Labs and Meta agreed to put a combined $300 million into Biohub's Virtual Biology Initiative, an effort to build AI models that predict how human cells behave. Biohub is the science organization backed by Mark Zuckerberg and Priscilla Chan. Together with commitments from Biohub itself and two U.S. federal agencies, the project is now described as a roughly $1.8 billion push to generate open biological data at a scale never attempted before.
What Is a "Virtual Cell"?
A virtual cell is a computational model that can predict how a living cell responds when something changes, such as a gene being switched off, a drug being added or an infection taking hold. Instead of running every experiment in a lab, researchers could ask the model first and test only the most promising answers at the bench. The idea is similar to how large language models learn patterns from huge collections of text. Biology, however, has lacked an equivalent collection of consistent, large-scale measurements, and that gap is what this initiative targets.
Who Is Putting In What
Reporting from CNBC, Quartz and others describes a layered funding structure. Google DeepMind, Isomorphic Labs and Meta together commit $300 million. Biohub committed $500 million in April 2026, split into $100 million for external research and coordinated data generation and $400 million for technology development and internal data generation. The U.S. Department of Energy plans to invest more than $500 million over five years in biological measurement, modeling and computing. The National Institutes of Health is contributing datasets and resources drawn from more than $500 million of earlier federal investment. Adding these pieces gives the $1.8 billion headline figure. Note that Biohub's own page refers to a larger $2 billion commitment, so totals vary slightly depending on how contributions are counted.
How the Data Will Be Generated
The initiative aims to measure how cells respond across far more conditions than scientists have ever tested. Planned data types include genomic, transcriptomic and proteomic measurements, cell and tissue-level imaging, cryo-electron tomography that reveals structures at near-atomic resolution, and next-generation spatial omics and microscopy. Biohub's partner list includes the Allen Institute, Arc Institute, Broad Institute, Human Cell Atlas, Human Protein Atlas, the Wellcome Sanger Institute, NVIDIA as a technology partner, and the Billion Cells Project, a collaboration of 17 institutions including MIT, Stanford and UCSF.
Timeline and Goals
The first dataset is expected in about one year, with the goal of accurate predictive models within five years. Alex Rives, Biohub's head of science, has said that "to build artificial intelligence that can accurately represent the full complexity of biology, we need orders of magnitude more data than exists today."
Open Data With a Commercial Incentive
Data will be shared openly, but with a twist. Data funded by commercial partners carries a one-year embargo before public release, while government-funded data has no embargo. Rives explained: "We have to have some incentive for commercial players to be a part of this, and the embargo period creates that." The model lets companies benefit early while the wider research community gains access afterward.
Why Google and Meta Are Involved
Isomorphic Labs already works with pharmaceutical companies including Eli Lilly and Novartis, and raised $600 million from external investors in March 2025, so better cell models align directly with drug discovery. For Meta and Google DeepMind, large and well-labeled biological datasets could become training material for the next generation of scientific AI. Biohub had earlier signaled this direction when Zuckerberg and Chan said in November 2025 that they would shift the bulk of their philanthropy toward science and AI-driven biology, and that Biohub would acquire the AI lab EvolutionaryScale.
What to Watch Next
The first milestone is the initial dataset expected around late 2027. Researchers will be watching how quickly it is released, how consistent the measurements are across labs, and whether early models can predict cell responses to conditions they were not trained on. If they can, the approach could shorten the path from basic biology to new therapies. The Verge's original report was not directly accessible during research, so details here are cross-checked from CNBC, Quartz, Axios headlines, TIKR and Biohub's own announcement.