Google, Meta and Isomorphic stake $300m on building a virtual cell

The News

Three of the biggest names in artificial intelligence have put money behind one scientific ambition: building a working computer model of a living cell. Google DeepMind, Meta and the drug-discovery firm Isomorphic Labs are jointly committing $300 million to Biohub, the non-profit biomedical research organisation set up in 2016 by Meta chief executive Mark Zuckerberg and the physician Priscilla Chan.

The cheque is one slice of a larger programme. Biohub's backers have framed the work as part of a $1.8 billion initiative to assemble the vast, high-quality biological datasets that machine-learning models need. The stated goal is a "virtual cell", software that lets scientists ask, predict and answer biological questions digitally rather than through slow, costly bench experiments.

If it works, researchers could test how a cell responds to a drug or a mutation on a screen before anyone touches a pipette, reshaping how medicines are found and how disease is prevented and managed.

Why It Matters

The commitment signals a shift in where frontier AI money is flowing. For three years the spending race has been dominated by language models and the data centres that train them. Routing $300 million, under a $1.8 billion umbrella, towards biology says the same players now see the life sciences as the next arena where data and compute compound into advantage.

There is precedent for the optimism. The last time AI reshaped biology at this scale was DeepMind's AlphaFold, whose 2021 breakthrough and later release of more than 200 million predicted protein structures collapsed work that once took years into an afternoon. A credible virtual cell is a far harder problem, because a cell is a dynamic system rather than a single folded molecule, but the prize is correspondingly larger.

The move also deepens the tie between Isomorphic Labs and Google DeepMind, both led by Demis Hassabis, and pulls Meta directly into the biomedical data race alongside them.

Indian Angle

For India, the first question is access. AlphaFold's open database became a quiet workhorse in Indian labs, at the IITs, IISc and CSIR institutes, precisely because it was free. Whether the datasets from this $1.8 billion programme are opened on similar terms will decide if cash-constrained Indian researchers are participants or spectators.

The timing is pointed. India approved its BioE3 policy in 2024 to push high-value biomanufacturing, and home-grown AI-drug-discovery firms such as Aganitha and Innoplexus compete in exactly this space. A validated virtual-cell platform owned by Western giants could hand those startups a powerful tool to build on or raise the barrier to entry sharply, depending on how licensing is set.

There is a governance dimension too. As biological and health data become the fuel for drug discovery, India's Digital Personal Data Protection Act and its rules on cross-border health data will shape how far Indian hospitals, biobanks and pharma majors such as Sun Pharma, Dr Reddy's and Biocon can plug into global efforts, or whether they must build sovereign equivalents at home.

FAQ

What is a "virtual cell"?

It is a proposed software model that simulates how a living cell behaves, letting scientists predict its response to drugs, mutations or stress on a computer rather than at the bench. The $1.8 billion initiative funds the biological datasets needed to train such a model.

Who is putting in the money?

Google DeepMind, Meta and Isomorphic Labs are jointly investing $300 million into Biohub, the non-profit founded in 2016 by Mark Zuckerberg and Priscilla Chan. That commitment sits within a wider $1.8 billion programme.

How does this compare to AlphaFold?

AlphaFold predicted the shape of individual proteins. A virtual cell is harder, modelling a whole living system in motion rather than one molecule, which is why the dataset-building effort behind it is so large and costly.

What does it mean for Indian researchers?

It depends on access. If the datasets are opened as AlphaFold's were, Indian labs and startups gain a major tool. If licensing is tight, the cost of competing in AI drug discovery rises sharply.

This story was reported by The Verge. Read the full original coverage at The Verge.