Demis Hassabis on Building DeepMind, AlphaFold, and the Final Stretch to AGI

Why It Matters

DeepMind’s roadmap shows how strategic AI research can turn once‑theoretical goals into practical breakthroughs, reshaping drug discovery, scientific inquiry, and the future of industry.

Key Takeaways

• Early gaming work taught Hassabis AI and GPU fundamentals.
• DeepMind founded to combine reinforcement learning, deep learning, neuroscience.
• Hiring strategy focused on being 5‑10 years ahead, not too futuristic.
• AlphaFold demonstrated AI’s power for solving long‑standing scientific problems.
• Future AI will enable new simulation‑driven sciences and accelerate drug discovery.

Summary

Demis Hassabis sat down to recount the evolution of DeepMind—from his teenage fascination with AI, through a stint as a game developer, to the creation of a company aimed at building artificial general intelligence. He describes how early games like Theme Park and Republic forced him to engineer large‑scale AI and graphics on the first GPUs, laying a technical foundation that later powered DeepMind’s research.

In 2009 Hassabis and a small team of neuroscientists and engineers combined deep learning breakthroughs from Geoffrey Hinton’s lab with reinforcement‑learning ideas, believing they were roughly a decade ahead of the field. Their hiring mantra—be five to ten years ahead, not fifty—attracted talent willing to pursue what many still dismissed as sci‑fi. The company’s mission, “solve intelligence, then use it to solve everything else,” guided projects from AlphaGo to AlphaFold.

AlphaFold, the protein‑folding system that cracked a 50‑year challenge, exemplifies the “AI for science” agenda. Hassabis highlighted the spin‑out Isomorphic Labs, which aims to automate drug‑design pipelines, and he mused that AI will eventually become a discipline of its own, enabling ultra‑accurate simulations for economics, social policy, and beyond.

The interview underscores that DeepMind’s blend of game‑based research, neuroscience inspiration, and ambitious timelines is reshaping how industry tackles fundamental scientific problems. As AI tools become more capable, they promise to accelerate drug discovery, create new simulation‑driven sciences, and fundamentally alter the pace of innovation across sectors.

Original Description

Demis Hassabis, co-founder and CEO of Google DeepMind and 2024 Nobel laureate in chemistry for AlphaFold, joins Sequoia partner Konstantine Buhler at AI Ascent 2026 for a wide-ranging conversation about the path to AGI and what comes after. He explains why he believes AGI is achievable by 2030, why drug discovery could collapse from ten years to days, and why we should think of information, not matter or energy, as the most fundamental substance in the universe. Also: what Einstein would tell us about the limits of today's models, and why the next year or two will be critical for humanity.

00:00 Introduction

00:38 Demis’ Origins and Common Thread

01:29 Games as AI Training Ground

02:59 Elixir Studios Startup Lessons

04:39 Founding DeepMind in 2009

07:25 DeepMind Mission and AGI Bet

08:52 AI for Science Culture

10:37 Biology Breakthroughs and Isomorphic

12:42 New Sciences via Simulations

20:29 Consciousness Philosophy