Biohub, Meta, Google DeepMind and the US pool $1.8bn for AI biology data

Biohub, the US Department of Energy and the National Institutes of Health are putting $1.8bn into data to train AI models that predict how cells behave. Meta, Google DeepMind and Isomorphic Labs are joining with $300M between them, Biohub announced on Wednesday.

Biohub is the nonprofit research institute backed by Mark Zuckerberg and Priscilla Chan. The $1.8bn covers funding, data, computing and new measurement technology. Biohub called it the largest coordinated commitment to AI-ready biology data so far.

“An accurate predictive model of biology could dramatically accelerate scientific discovery by enabling scientists to perform experiments digitally,” said Alex Rives, Biohub’s head of science.

Who is paying for what

The Department of Energy will spend more than $500M over five years on lab measurement, modelling and computing. Its share runs through the Genesis Mission science push. It will draw on exascale supercomputers, X-ray and neutron scattering, cryo-electron microscopy and self-running labs.

The NIH will bring datasets and repositories built with more than $500M in earlier federal funding. Biohub will standardise them for AI training.

The money joins Biohub’s own $500M, pledged in April when it launched the Virtual Biology Initiative. Of that, $400M goes to new tools for measuring cells, and $100M funds research outside Biohub.

The Allen Institute, the Broad Institute, the Gladstone Institutes and the UK’s Wellcome Sanger Institute are also taking part. So are the Human Cell Atlas and Human Protein Atlas consortia. Nvidia will provide computing and software, and Renaissance Philanthropy is helping to raise more money.

Open data, after a head start

The datasets will become a public resource. The commercial funders get one year of exclusive access to the data first, Rives told Axios.

“We have to have some incentive for commercial players to be a part of this, and the embargo period creates that,” Rives told Axios.

The government-funded work will carry no such restriction, Rives told Reuters. Biohub plans to approach drug companies and philanthropies next.

Current cell datasets hold hundreds of millions of cells, Rives told Reuters. An accurate model will need billions, and later trillions. The partners aim to have a first dataset in about a year, and accurate predictive models within five years.

Other AI labs are working on biology data too. Anthropic has built its own biology lab, and the OpenAI Foundation has started a grant programme of more than $125M for biology datasets, Reuters reported. In Paris, Rivercell raised $25M on Wednesday to build an AI virtual cell.