The AI Boom (and Bust?) Cycle: Lessons from the Gold Rush. Freestyle Fridays (June 19, 2026)

TL;DR

• 1

  AI looks a lot like a gold rush: Standing near South Pass City, which once had about 6,000 people and now has roughly 10, Joe Reis uses abandoned mines and wagon trails to frame AI as the latest in a long line of boom-bust cycles.

• 2

  The likely winners are infrastructure owners, not wrapper makers: He argues that data centers, frontier model companies, and other core providers resemble the people who sold picks and shovels, while thin GPT wrappers can be copied by the platform underneath them.

• 3

  'This time is different' is the dangerous phrase: With AI company valuations at extreme levels, Joe says history rarely rewards the belief that expectations will somehow all be met just because a technology feels unprecedented.

• 4

  Hard work is not the same as safe work: Even if you build the equivalent of a gold mine, you can still lose if the bubble pops, which is why he keeps coming back to the gap between expectations and what the market can actually support.

• 5

  Participate, but hedge the downside: He points to the housing bubble as a reminder that people get wiped out when they assume prices only go up, and says anyone chasing AI upside should make sure failure does not leave them broke.

• 6

  The boom can still be useful even if it ends badly: Joe expects real benefits from today's spending on models and infrastructure, much like past manias left behind lasting assets, even if many participants never got rich.