Sir Demis Hassabis, the founder and long-serving chief executive of Google DeepMind, announced last week that he is stepping down from his role to focus on advancing AI-driven drug discovery. Hassabis, 50, who has been instrumental in developing some of the most groundbreaking artificial intelligence technologies, including the protein-folding model AlphaFold, will transition to a less hands-on position as chairman of DeepMind and chief scientist at Alphabet, Google’s parent company.

The move marks a significant shift in Hassabis’s career from leading the AI research division of the $4.4 trillion Alphabet conglomerate toward concentrating on Isomorphic Labs, a spin-off startup he launched from within Google five years ago. Isomorphic Labs secured $2.1 billion in funding this May, aiming to revolutionize drug discovery by dramatically shortening the development timeline for new compounds. Hassabis envisions AI’s potential to cure “all disease” within the next two decades, predicting numerous breakthroughs on the scale of AlphaFold’s success, which earned him a Nobel Prize in 2024.

Hassabis’s departure from DeepMind follows a recent wave of high-profile exits from Alphabet’s AI division. Jeff Dean, Google’s chief scientist, along with senior executives Oriol Vinyals, Sanjay Ghemawat, and Quoc Le, also left last week to found Discovery Loop, a new company focused on applying AI to expedite scientific research. These departures contributed to a significant decline in Alphabet’s market capitalization, with the company’s valuation dropping by approximately $300 billion.

In prior months, other key figures tied to Alphabet’s AI efforts, including Noam Shazeer—one of the developers of the Gemini AI model—and John Jumper, Hassabis’s Nobel co-recipient, moved to rival firms OpenAI and Anthropic, respectively. Despite these exits, Alphabet maintains a dominant position in the AI landscape, supported by its lucrative advertising business and vast infrastructure encompassing data centers and specialized processors.

Hassabis has expressed optimism about the future trajectory of artificial general intelligence (AGI), forecasting its arrival around 2030. However, his current focus lies on the specialized AI models needed to understand chemistry and biochemistry critical for drug design. He highlighted that some AI-generated test compounds have already reached the pre-clinical stage, marking an important milestone in developing real-world cures.

While the AI field remains intensely competitive, Hassabis’s decision underscores a broader trend among leading researchers shifting toward applied scientific challenges rather than solely chasing fundamental AI advancements. As AI becomes increasingly integrated and accessible, Hassabis believes the coming years will eclipse early developments, ushering in transformative applications in medicine and beyond.