Better Offline The Enshittifinancial Crisis: Part Four
112 snips
Jan 23, 2026 In the final discussion of this series, the host explores the dark side of the AI industry. He highlights how banks are hedging risks related to AI loans and details the alarming cycle of debt fueling unprofitable AI startups. Concerns mount as real demand is overshadowed by an oversupply of computing power. Predictions of price collapses and potential market failure raise alarms. The discussion emphasizes the need for accountability in the face of hype surrounding AI, setting the stage for future inquiries.
AI Snips
Chapters
Transcript
Episode notes
Hedge Data-Center Loan Exposure
- Investors and banks should hedge exposure to AI data-center loans because tenant demand may not materialize.
- Ed Zitron highlights banks exploring synthetic risk transfers and short positions to reduce risk.
Data-Center Builds Likely Outstrip Real Demand
- Massive data-center builds may far exceed real demand for GPU compute, creating an oversupply that collapses rental prices.
- Zitron estimates much of the $85B GPU portion of data-center debt could be underwater within two years.
GPU Rental Prices Are Already Dropping
- Rental prices for top GPUs have already fallen as more capacity comes online, undercutting data-center economics.
- Zitron cites Blackwell GPU rental averages declining to roughly $4.45/hour before major capacity arrives.