Biohub Leads $1.8 Billion Push to Build AI Models of Human Biology
A major new collaboration involving Biohub, the U.S. Department of Energy, the National Institutes of Health, Google and Meta is putting $1.8 billion behind one of the most ambitious artificial intelligence projects in biology.
The initiative aims to generate enormous amounts of biological data that can be used to train AI models capable of predicting how living cells respond to different conditions and interventions.
The long-term objective is to create increasingly accurate digital models of biology, potentially allowing scientists to conduct some experiments virtually before testing the most promising possibilities in physical laboratories.
The project is part of Biohub's Virtual Biology Initiative, which was initially launched with a $500 million commitment earlier in 2026.
What Is the Virtual Biology Initiative?
The Virtual Biology Initiative is an international effort to create large, standardized and AI-ready datasets describing how biological systems work.
Modern AI systems require enormous amounts of high-quality data. While language models can learn from billions of pages of text, biological research does not yet have an equivalent comprehensive dataset describing how cells behave under different conditions.
Biohub and its partners are attempting to build that missing data foundation.
The goal is to collect measurements across different cell types, biological conditions and interventions and make the resulting information useful for training predictive AI systems.
Nearly $2 Billion Is Being Mobilized
The expanded initiative represents approximately $1.8 billion in combined funding, data, computing resources and measurement technology.
Biohub's original commitment is $500 million.
The U.S. Department of Energy will contribute more than $500 million over five years toward laboratory measurements, modelling and computing.
The National Institutes of Health will coordinate relevant datasets, repositories and research resources developed through more than $500 million in previous federal investment.
Meanwhile, Google DeepMind, Isomorphic Labs and Meta are collectively investing another $300 million in the Virtual Biology Initiative.
Mark Zuckerberg's Biohub Wants to Create a Virtual Cell
The most ambitious part of the project is the effort to build a predictive model of the cell.
A sufficiently advanced virtual cell could allow researchers to simulate biological processes digitally and study how cells may respond to drugs, genetic changes or environmental conditions.
Instead of testing every possibility physically, scientists could potentially use AI models to identify the most promising experiments first.
This could significantly reduce the number of unsuccessful experiments and accelerate early-stage research.
However, creating a reliable virtual representation of a living cell remains a major scientific challenge. The initiative is intended to build the data and technology required to move toward that goal rather than claiming that a complete virtual cell already exists.
Google and Meta Join the Biological AI Race
Google DeepMind is participating through its AI-for-science research capabilities, while Meta is supporting the project through its broader investment in AI-powered biology.
Drug discovery company Isomorphic Labs is also joining as a founding member of the initiative.
The participation of technology companies is significant because modern biological research increasingly requires large-scale computing, machine learning and sophisticated data-processing systems.
AI companies can contribute model development and computing expertise, while biomedical institutions provide experimental data and scientific knowledge.
Massive Biological Datasets Are the Key
The initiative is focused heavily on data generation.
Researchers plan to collect information about cellular responses across a much wider range of cell types and conditions than has previously been possible at comparable scale.
Technologies including advanced microscopy, cryo-electron tomography and other measurement techniques will be used to capture biological information at increasingly detailed levels.
These datasets will then be standardized so that researchers and AI systems can use information generated by different laboratories and institutions in a compatible way.
DOE Will Bring Supercomputing and Laboratory Resources
The U.S. Department of Energy will contribute its national laboratory infrastructure, including advanced computing and scientific measurement facilities.
The department's contribution includes work involving exascale supercomputing, imaging, modelling and autonomous laboratories.
These resources could provide the computing power needed to process extremely large biological datasets and train sophisticated scientific AI models.
The combination of experimental biology and high-performance computing is expected to become an important part of the initiative.
NIH Data Will Strengthen the Project
The National Institutes of Health will coordinate existing biomedical datasets and research resources relevant to the initiative.
These resources include biomedical repositories, genomic information and datasets developed through federally funded research.
Biohub plans to work with NIH to standardize relevant information so that it can be more effectively used for AI training.
This could provide researchers with access to a much broader biological information base than would be possible through a single institution.
AI Could Change Drug Discovery
Drug development is one of the areas where predictive biological AI could have a major impact.
Developing a new medicine typically involves years of laboratory research, testing and clinical development, with many potential candidates failing at different stages.
If AI systems can accurately predict how cells respond to different molecules or biological interventions, researchers could potentially prioritize the most promising candidates earlier.
That does not eliminate laboratory experiments or clinical trials, but it could help scientists make better decisions about where to invest research time and resources.
The Project Will Not Replace Human Scientists
The Virtual Biology Initiative is designed to support scientists rather than eliminate the need for physical experiments.
AI models can generate predictions, but those predictions still need to be tested against real biological systems.
The long-term vision is to create a cycle in which experiments generate data, AI models learn from the data, the models generate new predictions and scientists test the most important predictions in the laboratory.
This combination could allow biological research to move faster while maintaining experimental validation.
Open Data Will Eventually Become a Major Part of the Project
One of the defining features of the initiative is its focus on creating an open biological data resource.
Biohub says the resulting datasets are intended to become accessible to the wider research community, helping scientists around the world develop their own biological AI models.
Some data may initially have restricted access arrangements associated with the participating organizations before broader availability.
The ultimate goal is to create a shared scientific foundation rather than a dataset controlled by a single company.
A New Race for AI-Powered Biology
The Biohub initiative comes as major AI companies increasingly explore biology and scientific research.
AI systems are already being used to analyze proteins, predict molecular structures and assist with drug discovery.
The next challenge is significantly broader: understanding how entire biological systems behave and predicting how they change when researchers intervene.
That requires much more data and far more complex models than many existing AI biology applications.
What Happens Next?
Biohub and its partners plan to begin generating and integrating large-scale biological datasets under the expanded initiative.
The first major datasets are expected to emerge as the new measurement and data-generation programs scale.
Biohub's broader objective is to build increasingly accurate predictive models capable of representing biological systems digitally.
The project is expected to take years, and scientists will need to determine whether increasingly large datasets and more powerful AI models can produce reliable predictions about real biological systems.
Could AI Create a Digital Model of Life?
The $1.8 billion initiative represents one of the largest coordinated efforts yet to combine artificial intelligence, biology and large-scale scientific computing.
If successful, the project could give researchers new tools for understanding diseases, testing biological hypotheses and identifying potential treatments.
A fully predictive virtual cell remains a long-term scientific ambition rather than an existing technology.
But the scale of the new investment shows how seriously technology companies, governments and biomedical researchers are approaching the idea of using AI to model living systems.
The next major AI breakthrough may therefore not come from another chatbot. It could come from an AI system capable of understanding what happens inside a living cell.
Journalist: Vijay Singh