Biohub Announces $1.8 Billion Initiative for AI Models in Biology

$1.8 Billion Commitment

The Chan Zuckerberg Biohub, the U.S. Department of Energy (DOE), the National Institutes of Health (NIH) and new investment partners announced on October 7, 2026, a combined $1.8 billion commitment to build open datasets for predictive artificial intelligence (AI) models in biology. The commitment takes the form of funding, data, compute capacity and measurement technologies. According to Reuters, the initiative aims to build AI models that accurately predict biological processes on top of open datasets. Government agencies, scientific organizations and major technology companies are taking part together.

The commitment has four components: funding, research data, computing power and modern measurement technologies. The initiative is called the Virtual Biology Initiative.

The Government Share: DOE and NIH

Under the Genesis Mission program, the U.S. Department of Energy will direct more than $500 million over the next five years toward laboratory measurements, modeling and computation. The National Institutes of Health will coordinate datasets created with earlier federal funding — worth more than $500 million — while Biohub standardizes the data for AI training. According to Biohub, the government agencies' contribution forms the scientific backbone of the initiative.

The Tech Companies' Contribution

Google DeepMind, Isomorphic Labs and Meta are jointly investing $300 million in the Virtual Biology Initiative. NVIDIA is contributing computing infrastructure, software and technical expertise. The companies aim to create a new generation of biological modeling built on open datasets.

The April Groundwork: a $500 Million Commitment

This initiative did not come out of nowhere: Biohub announced a $500 million commitment back in April 2026. Of that, $400 million went to measurement technologies such as cryo-electron tomography and $100 million to research outside the organization. The October expansion to $1.8 billion builds on that foundation.

Timelines and Openness Terms

"The first dataset will be ready in about a year, and accurate predictive models in five years — compressing a process that takes decades. Commercial sponsors get an embargo until the data is open, and government-funded work has no restrictions," said Alex Rives, Biohub's chief science officer.

Datasets will be published openly as they are completed; a limited embargo period is planned for commercial sponsors, while results from government-funded research will carry no such restrictions.