Biohub Leads $1.8 Billion Effort to Build AI Models That Predict Cell Behavior
The research organization backed by Mark Zuckerberg and Priscilla Chan is coordinating a $1.8 billion initiative to train AI models that predict how cells behave, with funding from Meta, Google DeepMind, Isomorphic Labs and the US Department of Energy. A first dataset is expected in about a year.
Biohub, the research organization backed by Mark Zuckerberg and Priscilla Chan, is coordinating a $1.8 billion initiative to train artificial intelligence models capable of predicting how cells behave, a project that brings together some of the largest players in both technology and the life sciences.
The effort draws funding and resources from Meta, Google DeepMind, Isomorphic Labs and the US Department of Energy, which are contributing data, laboratory equipment and computing power. The scale of the commitment reflects a growing conviction that AI can accelerate biological discovery in ways traditional research methods cannot match.
At the center of the project is the ambition to build models that do not merely describe cells but anticipate their actions. Such predictive capability could reshape how researchers approach disease, drug development and basic biology, turning vast quantities of experimental data into simulations that guide laboratory work.
A first dataset is expected to be ready in roughly a year, according to the plan. That initial release would mark a critical milestone, giving researchers a shared foundation on which to train and test models before the project expands further.
The involvement of Google DeepMind and Isomorphic Labs is notable. DeepMind has already demonstrated the potential of AI in biology through protein structure prediction, while Isomorphic Labs was established to apply computational methods to drug discovery. Their participation signals that the new initiative is intended to move beyond isolated breakthroughs toward a broader predictive infrastructure for cell biology.
Meta's role, alongside the Chan Zuckerberg Initiative's Biohub, places the project within a wider pattern of technology companies investing heavily in biomedical research. The US Department of Energy adds another dimension, contributing both computational resources and scientific expertise from its national laboratory system.
The $1.8 billion figure covers funding for data generation, laboratory equipment and compute, the three pillars on which the initiative rests. Each is essential: high-quality data to train models, experimental capacity to validate predictions, and sufficient computing power to handle the complexity of biological systems.
Predicting cell behavior is among the hardest problems in biology. Cells respond to their environment, communicate with neighbors and change over time in ways that are difficult to capture with conventional methods. AI models trained on large, well-curated datasets could help researchers identify patterns and test hypotheses faster than traditional experiments alone.
If successful, the initiative could influence fields ranging from immunology to oncology, where understanding how cells behave under different conditions is central to developing treatments. It could also lower the cost and time required for early-stage research, allowing scientists to prioritize the most promising avenues before committing to expensive laboratory work.
The project remains at an early stage, with the first dataset still about a year away. Much will depend on whether the participating organizations can coordinate effectively and whether the resulting models prove accurate enough to be useful in real research settings.
For now, the initiative stands as one of the largest coordinated efforts to apply AI to fundamental biology, uniting academic-style research goals with the resources of major technology companies and a government agency. Its progress over the coming years will be watched closely by scientists, investors and policymakers alike.
4
