DeepMind’s Isomorphic Labs has grand ambitions to ‘cure all diseases’ with AI

“There are people sitting in our office in Kings Cross, London, working, and they cooperate with artificial intelligence to design cancer medications,” Murdoch said. During an interview in Paris. “This is happening now.”

After years of development, Murdoch says the human clinical experiences of drugs that are with the help of AI from ISOMORPHIC are finally on the horizon.

He said: “The next big teacher is actually in clinical trials, and she began to put these things in humans.” “We are working now. We are very close.”

The company, which was removed from DeepMind in 2021, was born from one of the most famous DeepMind, Alphafold, an AI system capable of predicting protein structures with a high level of accuracy.

The alphafold reactions of the ability to carefully predict individual protein structures have advanced to modeling how proteins interact with other particles such as DNA and drugs.

These jumps have made much more useful for detecting drugs, helping researchers to design medicines faster and more accurate, and convert the tool into a much larger ambition launcher.

“This was an inspiration for the similar laboratories,” Murdoch said of Alfafold. “This really indicates that we can do something very essential in artificial intelligence that can help open a drug detection.”

In 2024, in the same year that alphafold 3 was released, ISOMORPHIC signed a great cooperation in the field of research with Pharma Novartis and Eli Lilly.

A year later, in April 2025, ISOMORPHIC LABS raised $ 600 million in the first external financing round ever, led by Thrive Capital.

Dealms are part of the Isomorphic plan to build a “global drug design engine”, a system that combines automatic learning researchers with veterans in drugs to design new drugs faster, more licensed and successful.

As part of the deals with the main Pharma players, Isomorphic supports current pharmaceutical programs, but it also designs their internal drug candidates in areas such as tumors and immunity, with the aim of ultimately licensing them after early stage experiments.

He said: “We define an incomplete need, and we start our drug design programs. We develop them, put them in human clinical trials … We have not yet got it, but we make good progress.”

Today, pharmaceutical companies often spend millions of trying to provide one drug to the market, and sometimes with an opportunity to succeed in only 10 % once experiments start. Marduk believes that Isomorphic technology can radically improve these difficulties.

“We try to do all these things: accelerate them, reduce cost, but also improve the chance to be successful,” he says. He wants to harness alphafold to reach a point where researchers are reading 100 % that the drugs they develop will work in human experiences.

Murdoch said: “One day, we hope that we are able to say – well, here is a disease, then click a button and bypass the design of the drug to treat this disease.” “All are supported by these amazing artificial intelligence tools.”