Excess Returns He Wrote the Book on Bubbles | Edward Chancellor on If AI is Different
6 snips
May 12, 2026 
Edward Chancellor, financial historian and author on bubbles, traces capital-cycle echoes from railway mania to the AI data center boom. He spotlights why markets fund tech transitions but miss winners. Short takes cover AI capex races, GPU depreciation, hallucination limits for LLMs, anti-bubbles in beaten-up sectors, and why gold and old-economy value can shine when hype crowds out returns.
AI Snips
Chapters
Books
Transcript
Episode notes
Railway Mania’s Massive Overbuilding Example
- Railway mania peaked in 1843–45 with projected CapEx near 10% of UK GDP and massive duplicative lines.
- Chancellor notes three London–Peterborough and three Leeds–Manchester lines as concrete examples that crushed returns.
Markets Fund Transitions But Rarely Pick Winners
- Markets can fund technology transitions but often fail to pick the eventual winners, producing huge losses even for survivors.
- Chancellor cites the dot-com bust where Amazon fell >90% despite later dominance.
Hallucinations Limit LLMs’ Total Addressable Market
- Large language models (LLMs) are inherently probabilistic and prone to hallucinations, which limits where businesses can safely deploy them.
- Chancellor warns that error rates make LLMs unsuitable for high-stakes tasks and that agentic AI promises may be overhyped.
