Solutions with Henry Blodget Why OpenAI Won’t Survive an AI Crash
14 snips
Jan 26, 2026 Paul Kedrosky, venture capitalist and tech-cycle analyst, predicts an imminent AI crash that reshapes infrastructure and markets. He explores parallels to past booms, why frontier model firms may lose while builders sometimes win, collapsing token economics, fragile data‑center finance, and long‑term hopes for cheap cognition. Short, sharp takes on who and what could survive the turmoil.
AI Snips
Chapters
Transcript
Episode notes
AI As A Capital-Intensive Infrastructure Build
- The AI boom is a rapid, capital-intensive build-out of an underlying fabric like railroads or rural electrification.
- Those who build the substrate often lose while builders on top capture most long-term gains.
Frontier Model Firms Likely Lose Out
- Most frontier AI model companies will be commoditized and many will disappear.
- Winners will likely be companies building on the stack, not the heavy-capex model builders themselves.
Collapsing Token Prices Hit Business Models
- Token units (the atomic units of model input/output) are collapsing rapidly in price, making inference much cheaper.
- Falling token prices transfer value to consumers but create a disastrous business model for producers.