“Treat all diseases”: Google AI will carry out its first clinical trials

Say that the startup Isomorphic Labs is ambitious is the euphemism. The subsidiary of Google Deepmind simply intends to eradicate ” All diseases ». She said it a few months ago, and seems not to have turned her thumbs since. After years of development, the company is preparing to launch the first clinical trials on humans for drugs entirely designed by artificial intelligence algorithms.

This was recently entrusted to Fortune Colin Murdoch, president of the young shoot and commercial director of Google Deepmind. An announcement that could obviously mark a fundamental step for the application of AI to the health sector, an area where failures are expensive and where successes are rare.

Alphafold, the cornerstone

To understand the approach of Isomorphic Labs, you have to go back to its genesis. The company, founded in 2021, was born from one of the biggest promises in Deepmind: Alphafold. This artificial intelligence program is literally changing the situation in biology by managing to solve a problem that has been defeating scientists in failure for decades: predict the 3D structure of proteins with precision from their sequence of amino acids.

Why is it so important? The shape of a protein determines its function. Understanding this structure is therefore fundamental to decipher the mechanisms of diseases and to design molecules capable of interacting with these proteins to correct them. The most recent versions of alphafold go further, by modeling the dynamic interactions between proteins and other cellular elements, such as DNA or potential drugs.

It is this capacity that served as a launch ramp at Isomorphic Labs. The idea is no longer just to understand biology, but to act on it in a targeted way.

A hybrid economic model

How can a young company, even supported by a giant as a alphabet, hope to penetrate a market as regulated and capitalist as that of the pharmacy? Isomorphic has chosen to operate on two fronts to get there. On the one hand, it puts its “drug design engine” at the service of large names in the sector. On the other, the startup is not content to be a luxury service provider. The company in parallel develops its own drug candidates, focusing on areas where needs are important, such as oncology or immunology. And it is precisely this part that interests us here.

The strategy consists in carrying out these projects up to the first phases of clinical trials, a step obviously essential to validate their potential, before seeking to sell them under license to pharmaceutical partners who will take over for the industrial development and marketing phases.

The means of his ambitions

Such a strategy requires considerable means. This is the reason why Isomorphic Labs carried out its very first external fundraising in April 2025, an operation which raised $ 600 million. This financial contribution will notably allow internal research programs to accelerate and calmly prepare the costly stage of human tests, which should start “in 2025”.

Clinical trials, time of truth

Because this is where the real issue is located. Can AI really do better than decades of research and traditional development? The pharmaceutical sector is today faced with a paradox: despite scientific advances, the development of a new drug is a obstacle course. It is often millions of dollars and many years for a treatment to arrive on the market, with a success rate that caps around 10 % once the clinical trials have been initiated.

Isomorphic Labs intends to shatter these statistics. By modeling biological interactions with unprecedented precision, the company hopes not only to accelerate the design phase, but above all radically improve the chances of success of the selected molecules. The ambition, ultimately, is to reach a quasi-certainty, 100 % conviction, that a drug designed by AI will work in humans before even starting tests.

The ultimate vision, as Murdoch describes it, is almost science fiction: ” One day, we hope to be able to say: “Here is a disease”, click on a button, and get the design of a drug to treat it ».

A dizzying promise therefore, which could have an impact on all of humanity.