Schmidhuber: current GPU spend could be a temporary AI bubble
TL;DR
- Jürgen Schmidhuber tells the Europe on Edge podcast that current GPU spending could be part of a temporary AI bubble.
- He argues AGI may require machines that master the physical world, not just text-based systems.
- The episode, hosted by ETH Zurich's Filipe Barata, runs one hour and forty minutes and was published September 8, 2026.
Jürgen Schmidhuber, described in the show notes as "one of the pioneers of modern AI," tells host Filipe Barata that current spending on GPUs "could be part of a temporary AI bubble" in a new hundred-minute interview on the Europe on Edge podcast.
The episode, published September 8, 2026, is billed as covering what today's AI systems are "still missing," why AGI "may require machines to master the physical world," and "the origins of deep learning, self-improving AI, robotics, and the future of artificial intelligence." Barata is a researcher at ETH Zurich who leads the AI & Digital Biomarker core at the Center for Digital Health Interventions.
The public notes carry the framings, not the numbers. No dollar figure, timeline, or named lab is attached to the bubble claim in the description, and no transcript has been posted. ETH Juniors and the ETH Student Project House are listed as sponsors.
Shared on Bluesky by 1 AI expert
Originally reported by podcasts.apple.com
Read the original article →Original headline: #5 — Jürgen Schmidhuber: The GOAT of AI on Deep Learning, Robotics, the AI Bubble, AGI, and What Comes Next