Google DeepMind, Meta and a drug-discovery startup are pouring money into an audacious effort backed by Mark Zuckerberg to build software that acts like a living cell. The project, revealed Tuesday, unites Silicon Valley heavyweights with federal agencies in a bid to change how biologists study disease.

The Chan Zuckerberg Biohub announced a major expansion of its Virtual Biology Initiative. Total commitments now reach $1.8 billion in cash, existing data resources, computing power and new measurement tools. The goal remains a predictive model accurate enough that researchers can run experiments inside a computer rather than a wet lab.

But, the scale just grew. Biohub itself committed $500 million back in April. On Tuesday it added partners that bring serious resources. Google DeepMind, Meta and Isomorphic Labs together pledged $300 million. The Department of Energy will spend more than $500 million over five years. The National Institutes of Health will contribute datasets and repositories built with over $500 million in prior federal spending. Biohub will standardize that material for AI training.

The numbers sound large. They reflect the hunger for high-quality biological data. Current AI systems excel at language and images because those domains offer abundant, cheap examples. Biology lacks equivalent training material. Cells respond to drugs, mutations and environments in ways that remain poorly mapped at scale.

Alex Rives, Biohub’s head of science, captured the stakes. “An accurate predictive model of biology could dramatically accelerate scientific discovery by enabling scientists to perform experiments digitally,” he said in the organization’s announcement. He added that the creation of a virtual cell stands as one of the most important challenges for the next era of science.

Pushmeet Kohli, vice president of AI for science at Google DeepMind, echoed the point. “The quest to build a virtual cell is one of the great collective scientific challenges and key to understanding the mechanisms of life. We will not solve this challenge without open, experimental biological data at an unprecedented scale, showing how living cells behave and respond to changes,” Kohli stated in the same release.

Priscilla Chan, who founded Biohub with her husband, the Meta chief executive, offered a broader view in an interview with Reuters. “Biology has been just sort of a clever discovery-based science until this point,” Chan said. “We have always held this as a community asset, not just for one group, so that it can build upon itself over time.”

The initiative builds on earlier work. Biohub already helped produce a dataset with more than 120 million single cells and 225,000 perturbation interactions in partnership with Tahoe Therapeutics and the Arc Institute. That release, described as four times richer than a previous benchmark, offers a taste of what coordinated data collection can achieve. Yet participants acknowledge much more remains needed.

New measurement technologies sit at the center. The $400 million slice of Biohub’s original commitment funds cryo-electron tomography for near-atomic views inside cells, advanced microscopy capable of imaging millions to billions of cells in living tissue, and engineering methods to perturb biology at every scale from molecule to organism. The Department of Energy will contribute its exascale supercomputers, X-ray and neutron scattering facilities, and autonomous labs through its Genesis Mission.

Such tools generate the multimodal data that modern AI models crave. Combine transcriptomics, proteomics, structural imaging and perturbation responses. Train foundation models on the resulting pile. The hope is that scaling laws observed in language models will appear here too. More data, bigger models, better predictions. Rives had questioned that assumption when the project launched in spring. The new partnerships suggest confidence is growing.

Timelines reflect both ambition and realism. Organizers expect the first new dataset in roughly one year. Accurate predictive models could arrive within five years. That pace would compress work that might otherwise stretch across decades. Drug development timelines, which now run many years and cost billions, could shrink if virtual experiments reliably flag winners and losers early.

Commercial participants receive a temporary edge. Companies funding the data get a one-year head start before it becomes fully public, according to Rives in the Reuters report. After that, everything opens to the global scientific community. The approach attempts to balance private investment incentives with the open-science ethos that Biohub promotes.

Other partners round out the coalition. They include NVIDIA for computing expertise, the Arc Institute and Tahoe Therapeutics for research collaboration. The mix spans philanthropy, government, big tech and specialized institutes. No single player could shoulder the full cost or coordinate at this breadth.

Skeptics will note the history of grand biological promises. Previous cell atlases and single-cell projects delivered valuable catalogs but fell short of true prediction. This time the bet rests on AI’s recent progress and the volume of new data the partners promise to create. Success hinges on whether biology yields the same predictable improvements from scale that text and images did.

Yet the upside looks clear. An effective virtual cell would let scientists test thousands of genetic variants, drug combinations or environmental stressors in silico. It could reveal mechanisms behind rare diseases, predict patient-specific responses or accelerate design of cellular therapies. Cancer researchers might simulate tumor evolution. Immunologists could model immune cell behavior under different conditions.

The project also highlights shifting power in biomedical research. Philanthropy from tech fortunes, federal computing might and corporate AI labs now shape large-scale biology. Traditional academic grants still matter. They no longer dominate projects of this magnitude. Biohub’s model, which began with a $600 million commitment from Zuckerberg and Chan in 2016, has evolved into a convener that attracts additional billions.

Questions linger about data quality, standardization and interpretability. AI models trained on cellular data can hallucinate plausible but incorrect behaviors. Validation against real experiments remains essential. Biohub officials stress the need for continued wet-lab work alongside the virtual efforts. The virtual cell complements rather than replaces traditional science.

Even so, the announcement lands at a moment of intense interest. Earlier efforts from the Chan Zuckerberg Initiative produced tools like the Virtual Cell Platform and models trained on tens of millions of cells. Tuesday’s news scales those attempts dramatically. It also draws direct involvement from DeepMind, whose protein structure work with AlphaFold already transformed structural biology.

Isomorphic Labs, the Alphabet subsidiary focused on drug discovery, brings pharmaceutical expertise. Meta’s contribution signals broader interest in applying its AI infrastructure to scientific domains beyond social media. The Department of Energy sees an opportunity to apply its high-performance computing to problems of national interest in health and biotechnology.

Observers on X reacted with a mix of excitement and caution. Some called it a genuine moonshot. Others wondered whether the data would remain truly open or become another asset controlled by big tech. The one-year exclusivity period fuels that debate. Proponents argue it encourages investment. Critics fear it delays broad access.

Whatever the outcome, the initiative marks a departure. Biology has long been an experimental science limited by the cost and time of physical assays. Digital prediction at cellular resolution would mark a genuine break. It would let researchers ask what-if questions at a pace never before possible.

Rives and his colleagues recognize the coordination challenge. No institution alone can generate data across enough cell types, conditions and measurement modalities. The partnership pools strengths. Biohub supplies biomedical focus and standardization. Federal agencies offer unique facilities and legacy datasets. Tech firms contribute AI talent, computing and cash.

The result could influence drug pipelines, personalized medicine and basic research for years. Companies might screen compounds virtually before synthesis. Regulators could demand simulation data alongside traditional trials. Academic labs could explore hypotheses that budgets once ruled out.

Of course, biology retains its complexity. Cells operate in tissues, organs and organisms influenced by immune systems, microbiomes and environments. A perfect isolated cell model would still miss much. Organizers speak of expanding toward tissues and eventually whole organisms. The virtual cell represents a starting point, not an endpoint.

For now the focus stays concrete. Generate better data. Build better models. Release both openly after the initial period. Measure progress against concrete benchmarks rather than hype. The $1.8 billion buys time and resources to test whether AI can crack biological prediction the way it cracked other domains.

Success would validate the bet that data hunger, not just algorithmic cleverness, limits current AI in the life sciences. Failure would remind everyone that cells refuse to behave like tokens in a language model. Either result will shape research agendas for the rest of the decade.

The announcement also carries political weight. Federal investment in basic science faces scrutiny. Demonstrating concrete partnerships with industry and philanthropy may help justify continued spending. At the same time, reliance on private donors raises questions about priorities and control.

Biohub officials invite the worldwide scientific community to join. They emphasize open resources that build upon each other. If the project delivers on its promises, it could set a template for other grand challenges in science where data scale exceeds any single organization’s capacity.

So the virtual cell effort stands as both technical endeavor and organizational experiment. It tests whether rival tech companies, government agencies and philanthropic institutes can align around shared infrastructure for public benefit. Early signs suggest they see mutual advantage. The coming years will test how well that alignment holds when results emerge and commercial opportunities appear.