Mark Zuckerberg’s Biohub gets millions from Google, US govt to build virtual cells

Mark Zuckerberg (Image: ANI)

Mark Zuckerberg and Priscilla Chan's nonprofit Biohub has brought Google, Meta and the US government into its plan to build a "virtual cell". A virtual cell is an AI model of human biology that can predict how cells behave, fall sick and respond to treatment.

With the new commitments, total funding for the effort has reached $1.8 billion. Biohub says this is the largest coordinated investment in AI-ready biological data so far.Google DeepMind, Meta and Alphabet-owned drug discovery startup Isomorphic Labs are jointly putting in $300 million. The US Department of Energy (DOE) and the National Institutes of Health (NIH) have also signed on. The money will build open datasets that could let researchers "ask, predict, and answer biological questions digitally".

In practice, scientists could test ideas on a computer before taking them to a lab bench.

How the $1.8 billion Biohub funding is split

DOE will invest more than $500 million over five years through its Genesis Mission. That money covers lab measurement, modelling and computation. It will use exascale supercomputers, cryo-electron microscopy and autonomous labs across the US National Laboratory system.NIH is not writing a fresh cheque. Under its Bio Genesis Mission, the agency will pool datasets and repositories built with over $500 million in earlier federal funding.

Biohub will then standardise them for AI training. NIH's own announcement calls the target "SI-ready" data, short for super intelligence. Biohub uses the term AI-ready.Biohub anchored the Virtual Biology Initiative with $500 million in April 2026. Of that, $400 million goes into new measurement technology. This includes cryo-electron tomography, which resolves near-atomic detail inside a cell, and microscopy that can image millions to billions of cells in living tissue.

The other $100 million funds research outside Biohub. Nvidia is offering computing infrastructure and software.

The Broad Institute, Allen Institute, Wellcome Sanger Institute and the Human Cell Atlas are among the scientific partners.

Virtual cell models need billions of cells to learn from

Today's cell datasets run to hundreds of millions of cells, Biohub's head of science Alex Rives told Reuters. He said an accurate predictive model will need billions and eventually trillions.

The new measurements will come from techniques such as spatial transcriptomics, which maps molecular activity inside intact tissue. They will also come from screens that record how cells react when their environment changes.The timeline is tight. Rives said work like this would normally take decades. The partners want to finish it in five years, and the first dataset should be ready in about a year."An accurate predictive model of biology could dramatically accelerate scientific discovery by enabling scientists to perform experiments digitally," Rives said in a statement.

He called the virtual cell "one of the most important challenges for the next era of science."

Google and Meta get early access before the data goes public

Biohub calls the project open science, but corporate funders get a head start. Companies that fund the project can work on the datasets for a fixed period before anyone else, Rives told Reuters. After that, the data becomes publicly available. The government-funded work will carry no such restriction. Biohub plans to approach pharmaceutical companies and philanthropies next.Other AI companies are moving into biology as well. Anthropic has set up its own wet lab. The OpenAI Foundation has launched a grant programme worth more than $125 million for biological and medical datasets.For Zuckerberg and Chan, the virtual cell has been a long-running goal. Biohub launched in 2016 with $600 million from the Chan Zuckerberg Initiative's $3 billion pledge. That pledge aimed to cure or manage all diseases within their children's lifetime. One of Biohub's first projects was a cell atlas. In 2023, it started building a cluster of over 1,000 Nvidia H100 GPUs to model cells. Nearly a decade later, the goal remains the same, and Google, Meta and two US government agencies are now helping pay for the data it needs.