Squawk Pod Davos 2026: Google DeepMind CEO Demis Hassabis 1/24/26
18 snips
Jan 24, 2026 
Demis Hassabis, co-founder and CEO of DeepMind and architect of AlphaGo and the Gemini models, explains how DeepMind unified Google’s AI stack. He discusses rolling Gemini into products, Apple’s Siri choosing Gemini, paths toward AGI, model specialization, compute needs, competition, and AI’s effects on jobs and enterprise winners.
AI Snips
Chapters
Transcript
Episode notes
Organizing Assets Fueled Gemini's Leap
- Demis Hassabis says DeepMind combined Google's research bench, TPUs, and products to accelerate model quality deployment.
- He credits a year of organizing these assets for Gemini's recent performance and product integrations.
Model Quality First, Then Product Rollout
- Demis Hassabis explains product integration follows model quality and infrastructure improvements.
- Expect Gemini features to roll into Search, the Gemini app, and more Google services this year.
AGI Likely Needs More Than Scaling
- Hassabis takes an empirical view: scaling yields gains but some capabilities are still missing for AGI.
- He lists continual learning, true creativity, and long-term planning as gaps that may need breakthroughs.