Limitless Podcast This Week in AI: The Big AI Short Failed
31 snips
Nov 14, 2025 Michael Burry's hefty short on NVIDIA raises questions about inflated tech valuations and the reliance on aging GPUs in AI. The hosts debate the timing and risks of shorting AI companies while highlighting persistent demand for older GPUs by leading labs. They explore Microsoft's strategic gains with OpenAI, including the implications of the new GPT-5.1 release. Additionally, they discuss the advances from DeepMind and the potential limits of energy on AI growth. It's a lively conversation about AI's future and market dynamics!
AI Snips
Chapters
Books
Transcript
Episode notes
Burry's Nvidia Short And The GPU Thesis
- Michael Burry shorted about $300M of Nvidia, claiming GPU depreciation inflates AI company valuations.
- The hosts argue his thesis misunderstands GPU lifecycles and current demand dynamics.
Older GPUs Remain Valuable And In Demand
- NeoClouds and providers report long GPU lifecycles of five to eight years, not two to three as Burry estimated.
- CoreWeave and others are selling older GPUs months ahead, showing persistent demand for aged hardware.
Supply Pressure Persists Despite Old GPU Use
- Frontier AI labs like Google use GPUs up to eight years old for production workloads.
- Nvidia itself is oversubscribed for new Blackwell GPUs, indicating sustained supply pressure.
