Biohub Leads $1.8 Billion Open-Data Drive to Build Predictive AI Models of Biology

The Chan Zuckerberg-backed Biohub and its partners have committed $1.8 billion to create open data, computing capacity and measurement technologies for predictive AI models of biology. The initiative brings together the U.S. Department of Energy, the National Institutes of Health, Google, Meta and other organizations in an international effort focused on how cells respond to interventions.

The funding combines cash, data, computing resources and new experimental infrastructure rather than representing a single grant or product budget. The scale places biological data alongside model development as a central investment target, a distinction reflected in coverage from Reuters, The Wall Street Journal and The Pharmacist Letter.

The project aims to let researchers ask and answer biological questions digitally by modeling cellular behavior. Its central objective is predictive: connecting an intervention with a measurable cellular response, such as changes associated with disease or treatment. The work is framed around building a digital representation of biology, often described in coverage as a “virtual cell.”

That goal requires datasets that capture more than static genetic sequences. Researchers need standardized measurements of cells, perturbations and outcomes so AI systems can learn relationships between biological actions and effects. The Biohub’s stated focus is generating and making those datasets accessible, while Google DeepMind contributes AI expertise and Meta is part of the broader corporate investment effort, according to The Verge, CNBC and The Next Web.

The partnership joins federal science agencies with major technology companies and a privately backed research organization. DOE and NIH provide institutional reach across national laboratories, biomedical research and public funding, while Google and Meta bring computing, machine learning and industry research capabilities.

The arrangement reflects a shared infrastructure model rather than a conventional commercial AI launch. Coverage from Quartz, Yahoo Finance and Reuters identifies Google and Meta as contributing a combined $300 million, while other commitments include data, computation and measurement capabilities. The mix gives the initiative access to laboratories, researchers and technical systems that are difficult for any single participant to assemble.

Open, standardized data is the initiative’s defining contribution because biological AI systems depend on experiments that can be compared across laboratories. Consistent formats and measurement methods can make datasets easier to combine, audit and reuse, helping researchers train models on a wider range of cellular responses.

The emphasis also addresses a recurring constraint in scientific AI: powerful models require large volumes of high-quality experimental data, while many biological datasets remain fragmented across institutions. Biohub’s planned data generation gives the effort a public research dimension, even as technology companies participate. The Wall Street Journal, American Bazaar Online and Technology describe the program as a major expansion of biological data infrastructure for AI.

Predictive cell models could help researchers test biological hypotheses before running every experiment in a laboratory, with applications in disease research and treatment development. The initiative’s stated purpose includes preventing and treating disease, although the announcement describes a research platform rather than a clinical system or approved medical product.

The practical value will depend on whether the resulting models predict responses across different cell types, diseases and experimental settings. Standardized data can improve reproducibility, but biology remains highly variable and experiments still determine whether a prediction holds. Coverage from The Verge, AA.com and The Pharma Letter connects the effort to human-cell modeling and biomedical research without presenting a timetable for clinical use.

The immediate task is building the data pipeline: selecting measurements, running experiments, standardizing results and making the outputs available for model training. The $1.8 billion commitment gives the effort substantial resources, but its impact will be measured by the quality, accessibility and predictive performance of the resulting datasets.

The partnership also creates a test for open scientific infrastructure. Researchers will watch how data access, intellectual-property terms, privacy safeguards and validation are handled as public agencies and private companies work together. Google DeepMind, Meta and federal institutions now have a common stake in producing a shared foundation for predictive biology, while the wider research community will judge whether that foundation supports results beyond the partners’ own systems.