Hassabis sticks to 2030 AGI call at Sequoia AI Ascent

Hassabis's talk at Sequoia AI Ascent was framed as the latest chapter of a single long plan: he says he mapped out a route to AGI at 15, then went on to found DeepMind. He recalls that at the time of founding, nobody in academia or industry took the goal seriously, and he pairs that memory with a lesson for founders — go 50 years ahead and you fail. In the same talk he ties the AGI goal to the practical: an extension of AlphaFold, in which new drug development that takes 10 years could be shortened to months or weeks.

The talk also carries a philosopher's side. He explains why economics cannot become physics and suggests AI simulation could break through that wall. He offers the analogy that machine learning is to biology what mathematics is, and speculates that the root of the universe may be information — meaning AI, as information processing, may connect deeply to the universe itself.

What the outcome hinges on is the gap between this conviction and the wider field, where AGI timing predictions vary widely; Hassabis's consistency gives his 2030 call weight, but it remains one executive's forecast rather than a consensus. For researchers and engineers, the more durable takeaway is likely the framing of AI as more than a tool — a new language — which may shape how they approach problems rather than when AGI lands.