The Infinity Machine: Demis Hassabis, DeepMind, and the Quest for Superintelligence

Author: Sebastian Mallaby Year: 2026 Genre: Artificial Intelligence

About This Book

Mallaby tells the story of modern AI through one life. Demis Hassabis was beating adults at chess at four and was a master by thirteen. At seventeen he was coding Theme Park at Bullfrog, then came a Cambridge computer science degree, his own games studio, and a neuroscience PhD, because he figured brains were the best existing proof that general intelligence could be built. In 2010 he founded DeepMind with Shane Legg and Mustafa Suleyman. Google bought the lab in 2014, AlphaGo beat Lee Se-dol in front of 200 million people in 2016, AlphaFold cracked protein structure prediction and earned a Nobel Prize in chemistry in 2024, and today Hassabis runs Google DeepMind.

Mallaby is not an AI writer. He is a Council on Foreign Relations fellow who wrote More Money Than God and The Power Law, so the funding rounds and the geopolitics get as much attention as the algorithms. He sat through more than thirty hours of interviews with Hassabis, then went looking for counterweights: Suleyman, the cofounder who left under a cloud; Ilya Sutskever at OpenAI; and Geoffrey Hinton, who now fears the field he helped build.

Key Insights

• The mission predates the hype: Hassabis was pitching AGI in 2010, years before most investors knew what a large language model was. It sounded deranged then. It just looks early now.
• DeepMind set the template: in 2013 one network taught itself to play dozens of Atari games from raw pixels. The core bet, that scale plus learning beats hand-coded expertise, is still most of what the field does.
• Google’s 2014 acquisition bought a research agenda: DeepMind kept its independence and its habit of publishing until ChatGPT forced everyone to ship products. Mallaby’s account of that pivot is the best part of the book.
• Two speeds, one lab: AlphaFold solved a fifty-year-old problem in biology and charged nothing for it. The same lab now burns billions a year on compute to keep pace with OpenAI. Mallaby lets that tension sit unresolved.
• The prophecy is also the pitch: one investor describes AGI as “finding God’s algorithm.” Mallaby quotes this material deadpan and lets you wince on your own.

Why I Recommend It

I read it in two sittings, which I did not expect from a 480-page corporate biography. The reporting carries it: Mallaby got access nobody else gets, and he still plays the Suleyman exit and the OpenAI rivalry straight. The AlphaGo chapters alone are worth the price of the book.

It has a flaw, and you should know it going in. Mallaby repeats Hassabis’s grander claims about superintelligence and new physics with too little pushback, and his dialogue tags get purple; the Guardian’s review takes him apart on exactly this. Read those passages with your guard up. Even so, this is the closest thing to a definitive record of the lab that started the modern AI race, written while the race is still running.