Palace/Workshop

Bubble Detection

investingbubblespsychologyvaluationaihoward-marks

Bubble Detection

Bubble detection is the practice of judging market behavior around a theme, not merely deciding whether the theme itself is real.

Core Idea

Howard Marks' bubble framework separates two questions:

  1. Is the underlying development real and important?
  2. Are investors paying prices and making assumptions that cannot be justified by likely future cash flows?

A technology can be transformative and still produce terrible returns for many investors. Railroads, autos, the internet, and AI can change the world while overcapitalized or overpriced companies destroy capital.

Behavioral Signs

Useful warning signs:

  • "No price too high" thinking.
  • Huge early gains that create envy and FOMO.
  • Newness that frees imagination from historical restraint.
  • Narratives replacing valuation.
  • Lottery-ticket math: tiny probabilities of gigantic payoffs used to justify almost any price.
  • Circular deals or transactions that make progress look larger than it is.
  • Startups raising enormous capital before products or business models are clear.
  • Rising use of debt to finance speculative infrastructure.

Debt in Bubbles

Marks treats debt as especially dangerous in speculative technological races. Equity can work when outcomes are highly skewed: a few huge winners can offset many losers. Debt is different. The upside is capped at the coupon and principal repayment, while the downside includes impairment if the borrower or project fails.

That means debt-financed AI infrastructure deserves special scrutiny:

  • Are cash flows predictable enough to support debt?
  • Are useful asset lives long enough to repay 30-year obligations?
  • Could newer chips or models obsolete today's assets?
  • Are SPVs or vendor financing hiding leverage?
  • Are circular transactions making demand look stronger than it is?

Inflection Bubbles

Marks distinguishes mean-reversion bubbles from inflection bubbles. Mean-reversion bubbles are financial manias with little lasting social value. Inflection bubbles may finance real infrastructure and accelerate technological progress. But both can destroy investor capital. The investor's problem is to benefit from progress without becoming the money that gets incinerated to build it.

Connections

  • investment-fashion-cycles - Popularity changes future returns by attracting too much capital.
  • confidence-cycle - Optimism and risk tolerance feed bubbles.
  • second-order-thinking - The question is not "is AI important?" but "what expectations and prices already reflect that?"
  • ergodicity - Avoiding ruin matters even when a theme has positive expected social impact.

Sources