Demis Hassabis, and how AI just might wrangle our molecular universe
It was a brisk October night as Demis Hassabis and I talked about the stars.
There was good reason. For one, place: We were sitting in a London observatory opened in 1929. For another, metaphor: Talking about stars and constellations is a great way to talk about vastness, the wonders and limitations of what the human brain can process.
There’s vastness far closer to us that transcends even the stars. It may seem impossible but there are, in fact, more possible chemical compounds in our world than stars across the sky. And it’s not close: A conservative estimate suggests the number of small, drug-like molecules out there is somewhere around 10^60, while the number of stars in the observable universe lingers around 10^22 (perhaps 10^24 by some estimates).
I’ve been fixating on this for months, because it conveys the absolute impossibility of drug discovery. I always knew penicillin, for example, was an accident. But I didn’t realize every drug was effectively a miracle, a spectacular defiance of the scientific odds.
This week, Fortune published a magazine feature I’ve been thinking about since fall, an inside look at Hassabis’s Isomorphic Labs. Hassabis, Google DeepMind pioneer and Nobel Laureate, started Isomorphic in 2021 to turn the powers of AI on the vastness of chemical space and ultimately “solve all disease.”
It’s worth being clear about what that tagline actually means (and I certainly showed up in London prepared to give Hassabis a hard time about it). As I wrote:
When I bring this up to Hassabis, he outlines his philosophy: The idea of “solving disease” is broader and more practical than eliminating illness once and for all. There’s a reason that he doesn’t say “cure.” While you can’t promise no one will ever get sick again, he says, you can develop a systematic, repeatable, and scalable process—powered by advanced AI and technology platforms—for discovering, designing, and optimizing drugs or treatments as needs arise.
So, it’s not a tagline so much as a theory, mission, and in some sense a promise of what technology can make possible. Because a drug isn’t just a drug. It’s perhaps better treatment for something intractable, perhaps it’s a cure. And it’s most definitely more time for us or someone we love. Isomorphic doesn’t yet have a drug in clinical trials (and will say very little about exactly when that will happen). But make no mistake: Ultimately, the only success is a life-changing drug that reaches patients.
“To show the value of something like this, you have to really show it,” Google Ventures managing partner Krishna Yeshwant, a doctor-turned-investor who was involved in early conversations around Isomorphic, told me as I was reporting. “You have to discover your own drugs. You have to bring it through to people, and prove that it works.”
Isomorphic, like the entire AI drug discovery space, is entering a new era—one that demands results. If those results materialize, they’ll bring more than just better treatments for autoimmune disorders or cancer.
In the coming years, if Hassabis is right, an entirely new system will emerge. One that promises to, within reason, wrangle the vastness inside all of us.
See you Monday,
Allie Garfinkle
X: @agarfinks
Email: alexandra.garfinkle@fortune.com
Submit a deal for the Term Sheet newsletter here.
Joey Abrams curated the deals section of today’s newsletter. Subscribe here.
This story was originally featured on Fortune.com