Demis Hassabis, Co-Founder and CEO of Google DeepMind and Nobel Prize Laureate in Chemistry 2024, attends the Artificial Intelligence (AI) Action Summit at the Grand Palais in Paris, France, February 10, 2025.
Benoit Tessier | Reuters
On a rare sunny day in London at the Google DeepMind office, Colin Murdoch saw the bank deposit from his employer come through.
It wasn’t his paycheck. Google was investing in his startup.
The fledgling company was Isomorphic Labs, the secretive artificial intelligence life sciences startup spun out of Google DeepMind after one of the biggest biological breakthroughs in the last 50 years. The check came in November 2021, and Murdoch was on the founding team.
As DeepMind’s chief business officer, Murdoch was tasked with finding business applications for DeepMind’s innovations in AI research. With Isomorphic, Murdoch wanted to zero in on the possibilities of AlphaFold2, an AI system with the capability of predicting the structures of proteins.
The viral program was developed by DeepMind in 2020 and helped solve a mystery that had flummoxed biologists for 50 years. If applied correctly, it had the potential to find new treatments for diseases like Alzheimer’s, Parkinson’s and cancer. The system would go on to win a Nobel Prize in chemistry.
It also helped Murdoch, and Isomorphic’s founder Demis Hassabis, decide on an ultimate goal: to “solve all disease” through AI-powered drug discovery.
“When AlphaFold came along, I think Demis and I looked at this and we thought, ‘Well, what could we do with AlphaFold?'” Murdoch, now Isomorphic’s president, told CNBC in an interview by video chat. “Really, that was a signal that we could begin to apply AI in this world of drug discovery.”
Hassabis remains with Isomorphic as CEO, and is simultaneously running DeepMind, the Google division at the forefront of the search company’s efforts to compete in the tech industry’s ongoing AI race. While Isomorphic is an independent unit of Google parent Alphabet, Hassabis’ dual roles reflect the intertwined relationship between the two parties.
Isomorphic has never revealed how much Google initially invested in 2021. But last month, the company locked down its first external funding in a $600 million round led by Thrive Capital with participation from GV, formerly known as Google Ventures, as well additional capital from Alphabet. With its fresh capital, Isomorphic plans to expand into biologics and to add compute to power its work in AI.
The drug discovery market is expected to reach $71 billion in value by 2025 as AI continues to change the industry.
On average, it can take anywhere between 10 and 15 years — and sometimes more than $2 billion — for a drug to make it from the bench to final approval from the Food and Drug Administration. It also costs hundreds of millions of dollars to conduct clinical trials on potential new drugs. For drugs that fail those trials, that’s a sunk cost. But using AI in drug discovery could make the process more efficient and less costly, leading to more successful drugs that make it to market.
Since its founding, Isomorphic has grown to more than 200 people and has established partnerships with Eli Lilly and Novartis, two of the world’s largest pharmaceutical companies. They pay Isomorphic, in part, to help make sure their long-term bets pay off down the line. The two partnerships include both upfront and milestone payments and “have the potential to be worth nearly $3 billion to Isomorphic Labs,” per the company, minus potential royalties resulting from future drug sales.
Google DeepMind and Isomorphic on Wednesday made AlphaFold3 available to the scientific community, for non-commercial use. The system can “predict the structures of and interactions between all life’s molecules,” Isomorphic said. “This goes beyond predicting proteins to DNA, RNA, and ligands and much more which is important for advancing drug discovery and biological understanding.”
Using AI to ‘solve’ disease
Murdoch, 46, wanted to be a doctor growing up, but his squeamishness led him to abandon the idea in favor of tech and engineering. DeepMind’s research into protein structures rekindled his interest in medicine.
The announcement of AlphaFold2 in 2020 — a step up from the initial version in 2018 — took the life sciences world by storm. That’s because proteins are one of the main building blocks of the cell, driving all chemical reactions that make up life as we know it. Whichever shape they fold into, of virtually infinite possibilities, decides their function.
If you can predict a protein’s shape, you can predict its function, and since research suggests that many diseases are influenced by incorrectly folded proteins, the research made waves.
Shortly after AlphaFold2’s release, Murdoch and Hassabis assembled a group of DeepMind employees to explore the ways the AI system could be put to use. After several months, they identified areas where they thought AI could make a difference in drug discovery, and put together a business plan for Isomorphic that they pitched to Alphabet.
The Google parent “thought it was a phenomenal idea,” Murdoch said.
Colin Murdoch, Chief Business Officer, DeepMind, on Centre Stage during day one of Web Summit 2022 at the Altice Arena in Lisbon, Portugal.
Eóin Noonan | Sportsfile | Getty Images
Alphabet decided to spin off Isomorphic from DeepMind, tasking the startup with using the AlphaFold technology to deliver biomedical breakthroughs and advance drug design programs.
In early 2022, the team was made up of just over a dozen people, working out of a temporary office space in King’s Cross, London’s tech hub.
Since then, Isomorphic has been operating in quasi-stealth mode, keeping mostly quiet about its work and advancements. In recent years, representatives for Isomorphic have said the company wasn’t “quite ready to share our progress yet,” and that executives were shying away from media interviews as the company built its technology and added talent.
The DeepMind employees who had been working on Isomorphic didn’t all stay after the spinout. Hassabis and Murdoch had to hire a team from scratch. They started with the C-Suite: a chief technology officer, chief AI officer, chief scientific officer and chief people officer. From there, they hired experts in therapeutics and machine learning.
“We’ve put a huge amount of energy and effort into building what we call an interdisciplinary culture that combines these two sorts of expertise,” Murdoch said. “That blend of AI and drug discovery expertise is absolutely essential to doing what we’re doing.”
Chess prodigy, knight and AI ‘rockstar’
Hassabis, 48, is one of the most important people at Google.
He co-founded DeepMind in 2010 with the goal of building “general-purpose” AI, and he sold the startup to Google in 2014, reportedly for between $400 million and $650 million. DeepMind served as one of Google’s key AI research divisions for years.
But that was upended in late 2022, when OpenAI released its ChatGPT chatbot, ushering in the age of generative AI and forcing Google to step on the accelerator.
DeepMind and Hassabis’ importance within Google have only risen since.
Google merged DeepMind with Google Brain, another AI research unit at the search company, to become Google DeepMind in April 2023. Hassabis was tapped to lead the combined division. More power shifted his way last October, when the Gemini app, which includes Google’s direct-to-consumer AI products, was moved under the Google DeepMind umbrella.
Through the drastic changes at DeepMind since 2023, Hassabis has continued in his other CEO role at Isomorphic.
“All the AI talent that’s at the company, led by Demis, they have cutting-edge research talent, [and] a lot of these folks could go work at any AI lab — it didn’t have to be a bio-focused one,” said Vince Hankes, a partner at Thrive Capital.
Hankes calls Hassabis “an extraordinarily special entrepreneur.”
People who have known Hassabis for years describe him the way they would a celebrity, expressing awe, respect and a slight air of incredulity that he knows who they are. He’s a chess prodigy, a Nobel Prize winner and, as of 2023, a knight — Sir Demis Hassabis.
Nobel laureate in Chemistry Demis Hassabis receives his award from Sweden’s King Carl Gustaf at the Nobel Prize ceremony in the Konserthuset in Stockholm, Sweden December 10, 2024.
Pontus Lundahl | TT | Via Reuters
Krishna Yeshwant, GV managing partner, said meeting Hassabis at Oxford University years ago was like “meeting a rockstar.” Already, Hassabis’ accomplishments were well known, but he remained approachable and didn’t resort to jargon to talk about his work in AI.
“Before DeepMind was acquired, even back then, he was thinking with a mindset towards breakthroughs,” Yeshwant said.
GV invested in Isomorphic because of the startup’s strength in both AI and therapeutics. Although GV invests across many areas of tech and biotech, Isomorphic struck Yeshwant as unique because of how Hassabis thinks.
“In biotech, you run these experiments, and sometimes experiments don’t work, and sometimes they do. And when they do, you move humanity forward in some way,” Yeshwant said. “In tech, it’s not always set up that way. It’s often, ‘Did we sign the next customer? Did we develop this feature?'”
“It’s unusual to see a person taking somewhat of that same [biotech] approach in technology,” Yeshwant said. “Demis, to his credit, has been doing it that way for a long time.”
Hassabis told CNBC in a statement that he believes improving human health is one of the most important use cases for AI.
“We’ve got this amazing tool that could finally unlock some of the biggest mysteries around disease and help us overcome them,” Hassabis said in the statement. “The idea of a future where people are healthier for longer, that’s what really excites me.”
Expanding operations and adding compute
Now that Isomorphic is beginning to open up, the startup says it plans to use its first external funding to expand into the field of biologics.
Currently, Isomorphic can use its general-purpose engine to design small-molecule drugs quickly — imagine a pill that you can take as a treatment — but the expansion into biologics opens up the potential range of diseases that the startup will be able to tackle, since biologics are typically injections rather than pills, Murdoch said.
“We’re primarily focused on cancer and immunology with our own internal programs, and so we’ll be able to take those further, run those faster and add programs,” Murdoch said.
Isomorphic will also focus on finding use cases for AlphaFold3.
The company’s new funding will also be used for compute a major issue in the AI space, where companies need all the graphics processing units (GPUs) they can access. Isomorphic will need to build more models to run at a greater scale with more data attached, both on the AI and therapeutics side.
“It’s expensive to build a full-stack life science company — and this one, they have to invest in compute, they have to invest in talent, they’ve got to invest in data, they’ve got to invest in developing assets,” Thrive’s Hankes said.
