From the Nobel Prize in Chemistry to the Bet on Solving All Diseases with Artificial Intelligence

In October 2024, the Royal Swedish Academy of Sciences announced that year's Nobel Prize in Chemistry winners: half of the prize was awarded to David Baker, of the University of Washington, "for computational protein design," and the other half, shared between Demis Hassabis and John Jumper, of Google DeepMind, "for protein structure prediction." It's worth emphasizing: the prize was awarded to the three scientists as individuals, not to the company they work for. It is not accurate to say that "Google won the Nobel Prize" — it was the researchers who received it, individually recognized by the Swedish academy.

To understand why this work was considered so important, it's worth briefly explaining what proteins are. They are molecules made up of sequences of amino acids, a kind of chemical "alphabet" of life, and they underlie nearly everything that happens inside an organism: digesting food, carrying oxygen in the blood, defending the body against infections. The crucial detail is that a protein doesn't function solely because of the sequence of amino acids that makes it up, but mainly because of the three-dimensional shape it takes on as it folds in space. That shape determines how the protein interacts with other molecules — and therefore determines its function. Experimentally determining that three-dimensional structure, using techniques such as X-ray crystallography, used to take years of laboratory work for a single protein.

It was in this context that AlphaFold emerged, an artificial intelligence system developed by Google DeepMind. In 2020, Hassabis and Jumper presented AlphaFold2, an AI model that represented a significant advance in protein structure prediction, making it possible to predict, with good accuracy, the structures of a large number of already-catalogued proteins. It's important not to overstate the scope of this achievement: AlphaFold did not "solve all proteins," nor did it eliminate the need for laboratory experiments. What it did was offer computational structure predictions with unprecedented accuracy, serving as a starting point and a complement to experimental research — not a substitute for it. Structures predicted by AI are still typically validated, refined, and tested using traditional experimental methods.

This ability to predict protein structures has a direct connection to drug discovery. Many drugs work precisely because they fit, like a key into a lock, into specific regions of target proteins, blocking or activating their function. Knowing the three-dimensional shape of these proteins more precisely and more quickly can help scientists identify, more efficiently, which molecules have the potential to interact with them therapeutically. This is an early and important step in the long process of developing a drug — but it is far from being the whole process.

It was to explore this potential that Demis Hassabis founded, in 2021, Isomorphic Labs, a company of which he is founder and CEO, separate from Google DeepMind, though linked to it by its origins and by the backing of Alphabet, Google's parent company. The company is specifically dedicated to applying artificial intelligence to the design and discovery of new drugs. According to Isomorphic Labs, its mission is to use AI technologies to accelerate and rethink the drug discovery process "from first principles" — an ambition stated by the company itself, not a result that has already been proven.

This ambition has attracted significant investment. In March 2025, Isomorphic Labs announced it had raised $600 million in its first external funding round, led by Thrive Capital, with participation from GV (Google's investment arm) and additional funding from Alphabet itself, already an investor in the company since its founding. More recently, the company announced a $2.1 billion Series B funding round, also led by Thrive Capital, with participation from investors such as Alphabet, GV, MGX, Temasek, CapitalG, and the UK Sovereign AI Fund, significantly expanding its capital base. These figures show the degree of confidence investors place in the company's proposition — but the size of an investment does not, by itself, guarantee scientific or clinical success.

It is in this context that Hassabis has repeatedly stated publicly an ambitious claim: that Isomorphic Labs' mission is, in his own words and the company's, to help "solve all disease" through artificial intelligence. It's essential to understand this statement for what it is — a long-term vision stated by the founder and the company — and not as a fulfilled promise or an already-achieved result. No drug reaches the market simply because an algorithm has flagged a promising molecule.

Even drug candidates identified with the help of AI tools remain subject to all the traditional, rigorous stages of pharmaceutical development: preclinical research in the lab and in biological models, safety testing, successive phases of clinical trials in humans, efficacy evaluation compared with existing treatments, and, finally, review and approval by regulatory authorities such as the FDA, in the United States, or the EMA, in Europe. This process typically takes years and has historically low success rates — artificial intelligence can help optimize certain early stages, but it does not eliminate or replace this path.

About two years after the Nobel Prize was awarded, what exists in concrete terms is this: a real technical advance, recognized by the scientific community, in the computational prediction of protein structures; a newly founded, well-funded company dedicated to applying this kind of technology to drug discovery; and a stated, ambitious mission that is still in the process of being proven in practice. Between the promise and a proven therapeutic outcome lies a long road of experimental science, human clinical trials, and regulatory scrutiny. Even so, artificial intelligence's ability to accelerate early stages of drug discovery already represents, on its own, a meaningful gain for biology and medicine.