Few people can claim to have solved one of science's grand challenges. Demis Hassabis is one of them. As the co-founder of DeepMind and the architect behind AlphaFold, Hassabis led the effort that cracked a problem biologists had wrestled with for half a century: predicting how proteins fold into their three-dimensional shapes. The achievement earned him a Nobel Prize in Chemistry in 2024 and opened entirely new frontiers in drug discovery, gene editing, and our understanding of life itself.
From Chess Prodigy to AI Visionary
Demis Hassabis was a prodigy long before he entered the world of artificial intelligence. Born in London in 1976 to a Greek-Cypriot father and a Singaporean-Chinese mother, he became a chess master at age 13, ranking second in the world for his age group. He went on to study computer science at Cambridge, co-designed the legendary video game Theme Park at just 17, and earned a PhD in cognitive neuroscience from University College London, where he studied the neural basis of imagination and memory.
But Hassabis had a grander ambition than games or academia. He wanted to build artificial general intelligence — a machine that could learn to solve any problem. In 2010, he co-founded DeepMind Technologies with Shane Legg and Mustafa Suleyman, setting up shop in London with a mission statement that most investors thought was science fiction: "Solve intelligence, then use that to solve everything else."
DeepMind and the Road Through AlphaGo
Google acquired DeepMind in 2014 for a reported 500 million dollars, giving Hassabis the resources to pursue his vision at scale. The first landmark came in 2016, when DeepMind's AlphaGo program defeated Lee Sedol, the world champion of Go — a game so complex that brute-force computation could never master it. The victory stunned the AI community and proved that deep reinforcement learning could tackle problems once thought to require human intuition.
But Hassabis was already looking beyond games. He had long believed that AI's true purpose was to accelerate scientific discovery, and he had identified a specific target: the protein folding problem.
The 50-Year Problem: Levinthal's Paradox
Proteins are the molecular machines of life. They start as simple chains of amino acids, then fold into intricate three-dimensional shapes that determine their function. Understanding a protein's structure is essential for understanding how it works — and for designing drugs that interact with it.
The problem is that predicting a protein's structure from its amino acid sequence is staggeringly difficult. In 1969, the molecular biologist Cyrus Levinthal calculated that a typical protein could theoretically fold into an astronomical number of possible configurations — more than the number of atoms in the universe. Yet real proteins fold into their correct shape in milliseconds. This contradiction became known as Levinthal's paradox, and solving it became one of biology's defining challenges.
For decades, determining a protein's structure required painstaking experimental work — X-ray crystallography, cryo-electron microscopy, or nuclear magnetic resonance. Each structure could take years and millions of dollars to resolve. By 2018, scientists had determined the structures of only about 170,000 proteins, a tiny fraction of the hundreds of millions known to exist.
AlphaFold 1: A Signal at CASP13
Every two years, the protein structure prediction community holds a competition called CASP (Critical Assessment of protein Structure Prediction), in which teams attempt to predict the structures of proteins whose shapes have been experimentally determined but not yet published. For decades, progress had been incremental.
In 2018, at CASP13, AlphaFold 1 entered the competition and dominated. While other teams used traditional physics-based methods, DeepMind's approach used deep learning to predict the distances between pairs of amino acids, then reconstructed the 3D structure from those predictions. AlphaFold 1 was far from perfect, but its margin of victory signaled that something fundamental had shifted.
AlphaFold 2: The Breakthrough at CASP14
Two years later, AlphaFold 2 arrived at CASP14 and delivered one of the most stunning results in the history of computational biology. The system scored a median GDT (Global Distance Test) of 92.4 out of 100 — a level of accuracy that matched experimental methods. For many proteins, AlphaFold 2's predictions were essentially indistinguishable from structures determined in the laboratory.
The key innovation was the Evoformer, a novel neural network architecture that processed multiple sequence alignments and structural information simultaneously, allowing the system to reason about the relationships between amino acids at every level. The architecture was designed by John Jumper and the AlphaFold team, but it was Hassabis who had set the direction, assembled the team, and made the strategic bet that protein folding was the right problem for AI to tackle.
The CASP organizers declared the protein folding problem "largely solved." The scientific world took notice.
200 Million Structures and a Nobel Prize
Rather than lock AlphaFold behind a paywall, Hassabis made the decision to open it up. In partnership with the European Bioinformatics Institute (EMBL-EBI), DeepMind released the AlphaFold Protein Structure Database, which eventually grew to contain predicted structures for over 200 million proteins — nearly every known protein in existence. The database was made freely available to researchers worldwide, and within its first year it was accessed by over a million scientists in 190 countries.
In October 2024, Hassabis and John Jumper were awarded the Nobel Prize in Chemistry, sharing the prize with David Baker of the University of Washington, who was recognized for his pioneering work in computational protein design. It was a remarkable moment: an AI researcher standing alongside experimentalists on science's highest stage.
Impact on Gene Editing and Drug Discovery
AlphaFold's impact on gene editing has been profound. CRISPR-Cas systems depend on precise protein-DNA interactions, and understanding the three-dimensional structure of Cas proteins is essential for engineering better gene editors. With AlphaFold, researchers can now model how Cas9, Cas12, and other editing proteins interact with their guide RNAs and target DNA, enabling the rational design of variants with improved specificity and reduced off-target effects.
In drug discovery, AlphaFold has compressed timelines that once stretched over years into days. Pharmaceutical companies now routinely use AlphaFold-predicted structures as starting points for drug design, and several candidates informed by AlphaFold predictions have entered clinical trials.
AlphaFold 3 and Isomorphic Labs
Hassabis did not stop at proteins. In 2024, DeepMind released AlphaFold 3, which extended the system's capabilities to predict the structures of complexes involving proteins, DNA, RNA, and small molecules. This was a critical advance, because most biological processes involve interactions between multiple types of molecules, not proteins in isolation.
In parallel, Hassabis founded Isomorphic Labs, a commercial venture spun out of DeepMind and focused on using AI to transform drug discovery. The company has signed major partnerships with Eli Lilly and Novartis, each worth over a billion dollars, to apply AlphaFold-derived methods to real-world drug development.
Recent Developments (2025–2026)
Following his Nobel Prize in Chemistry in 2024, Hassabis has accelerated the translation of AI into drug discovery. Isomorphic Labs raised $600 million in a Series A in March 2025 led by Thrive Capital, and in February 2026 announced its Drug Design Engine — which doubled AlphaFold 3's performance on protein-ligand structure prediction and can identify new binding pockets using only amino acid sequences.
The partnerships with Eli Lilly ($1.7B potential) and Novartis ($1.2B potential) are advancing, with Novartis expanding the collaboration in February 2025 to tackle "undruggable" targets. At the World Economic Forum in Davos in January 2026, Hassabis announced that the first AI-designed cancer drug will enter Phase 1 clinical trials in early 2026 — a landmark moment for AI-driven drug discovery.
Research Lab & Companies
- Google DeepMind — Co-founder and CEO
- Isomorphic Labs — Founder and CEO (AI drug discovery, $3B+ in partnerships)
- AlphaFold Database — 200M+ protein structures freely available
- Nobel Prize in Chemistry 2024 — shared with John Jumper
A Legacy Still Unfolding
Demis Hassabis set out to build an intelligence that could accelerate science. With AlphaFold, he delivered on that promise in a way that few could have imagined. The protein folding problem — once considered intractable — is now largely solved. The structures of virtually every known protein are freely available to any researcher with an internet connection. And the tools Hassabis helped create are already reshaping gene editing, drug discovery, and our fundamental understanding of biology.
The revolution Hassabis started is far from over. But its first chapter is already one of the great scientific stories of the twenty-first century.
