News

Alphabet's Isomorphic Labs Preps First Human Trials for AI-Designed Drugs

Company aims to be first to test fully AI-generated compounds in humans after $600M funding round and Nobel Prize validation

• By John K. Waters
• 07/16/2025

Alphabet's Isomorphic Labs says it's nearly ready to begin clinical trials on drugs designed entirely with artificial intelligence, a move that could reshape how new medicines are developed, but still faces a long and uncertain road ahead.

The company, a sibling to Google under the Alphabet umbrella, was spun out of DeepMind in 2021 to take the lab's celebrated AI breakthroughs, like protein-folding model AlphaFold, and turn them into something practical: actual drugs for actual diseases. Now, after a string of high-profile partnerships, a hefty $600 million funding round, and a Nobel Prize for the science behind it, Isomorphic says it's "very close" to testing its AI-generated compounds on humans.

"There are people in our London office right now working with AI to design drugs for cancer," Colin Murdoch, president of Isomorphic Labs and chief business officer of Google DeepMind, told Fortune. "The next big milestone is actually going out to clinical trials."

A General-Purpose Drug Design Engine?
The promise is ambitious, bordering on science fiction: use AI not just to assist human researchers, but to run the entire drug discovery process, from modeling disease-related proteins to designing custom compounds and predicting how they'll behave in the body. The company's long-term vision, as described by Murdoch, is to eventually "click a button" and have a drug blueprint pop out. That blueprint would then move quickly into lab testing and trials.

That dream is powered by AlphaFold, DeepMind's flagship AI system, which stunned the biomedical world in 2020 by predicting the structures of hundreds of thousands of human proteins. The current version, AlphaFold 3, goes even further, simulating how proteins interact with DNA, RNA, and potential drug molecules. It's this foundational model that Isomorphic Labs is betting on to build what it calls a "world-class drug design engine."

But Murdoch insists this isn't about targeting just one disease or protein. "It's really thinking about, how do we create a general drug design engine with AI—something we can apply again and again across any disease area?" he said.

Big Pharma Partnerships, Big Expectations
To make that vision real, Isomorphic Labs hasn't gone it alone. Since releasing AlphaFold 3 in 2024, it's signed research deals with drugmakers Novartis and Eli Lilly, collaborating on both new discoveries and ongoing pharma pipelines. It also raised its first outside capital—$600 million led by Thrive Capital—in April 2025. Thrive's founder, Joshua Kushner, called the company "one of the most consequential" in its space.

That backing is helping Isomorphic scale up, hiring across biotech and pharma disciplines as it readies for trials. Its internal programs are focused on oncology and immunology, high-risk, high-reward targets that could validate its approach if early results look promising.

Still, this is a moonshot. Most drugs fail in human testing, and AI doesn't change the brutal odds: fewer than one in ten experimental treatments make it to market. Murdoch argues Isomorphic's tech can beat those odds, saying it could "really improve the chance that we can be successful."

Hype or Inflection Point?
The pharmaceutical industry is watching closely, as are AI skeptics. Success for Isomorphic could validate the broader thesis that machine learning can do more than just assist R&D—it can replace significant portions of it. Failure, or even a stumble, could cast doubt on AI's ability to leap from simulation to biology.

For now, the company is still in the preclinical phase. No trial sites have been announced. No patients have been enrolled. No AI-designed drug has yet proven itself outside a lab.

But something fundamental may be shifting. Just this month, Microsoft claimed its new diagnostic AI beats most doctors on complex cases. Researchers at Northwestern unveiled an AI tool for cancer screening. And now, Isomorphic is lining up for the next big test: can AI design a drug that actually works in humans?

It's a question with no answer—yet. But for an industry notorious for being slow, expensive, and error-prone, even a small shift in the odds could be transformative.