In 2010, Demis Hassabis gave DeepMind 20 years to build artificial general intelligence. Fifteen years in, he says they are on track.
Who is he?
The CEO of Google DeepMind and a 2024 Nobel laureate in Chemistry for AlphaFold. Before that he was a chess master at 13, lead programmer on the video game Theme Park at 17, and a computer science student at Cambridge.
He went back to Cambridge in March 2025 to give a lecture on where AI is heading, and this week a repost of it passed 1.8M views on X. In one hour he covered:
An AI that went from random moves to world champion in under 24 hours
A billion years of PhD work done in a single year
Drugs designed in months instead of a decade
AI that plans, and robots, within two to three years
📚 In this newsletter, I break down what Demis Hassabis said about the road to AGI, AI for science and what comes after chatbots, with clips from the lecture.
High-signal analysis on markets, capital and the companies shaping the next decade
Inside you will find:
Why AGI Could Arrive Around 2030
See the two-step mission DeepMind pitched in 2010 and why he says it’s on schedule.How AI Taught Itself to Beat the World Champion in 24 Hours
Learn how self-play took AlphaGo from random moves to superhuman, and what move 37 proved.The Three Tests That Decide Which Problems AI Solves Next
Find out the criteria DeepMind uses to choose its next targets in science and industry.How AlphaFold Did a Billion Years of PhD Work in One Year
See how one model mapped 200 million proteins and became a standard tool for 2M researchers.Why Drug Discovery Could Shrink From 10 Years to Months
Learn what “digital biology” means, why he built Isomorphic Labs and the virtual cell he wants.Why Video Models Learned Physics From YouTube
See the “Turing test for video” and why generating a world is a sign of understanding it.What Comes After Chatbots: AI That Plans and Robots in Two to Three Years
Find out how AlphaGo-style planning on top of Gemini leads to agents and robotics.Why DeepMind Couldn’t Score a Single Point at Pong for Two Years
Learn why the bet on deep learning almost came 20 years too early, and the Babbage lesson.What AI Means for Coding and Careers
See why he still says learn to code, and the one skill he thinks matters most for the next 10 years.
This issue is available exclusively to paid subscribers!
If you’re not subscribed yet, you can unlock this and our full library by upgrading below:
Let’s begin!