Skip to main content

The Dangerous Part Of The AI Boom Is Here

Adam Taggart | Thoughtful Money® Aug 20, 2026

Summary

Hyperscalers like Microsoft, Alphabet, and Amazon can weather an AI investment slowdown thanks to diversified businesses, while neoclouds like CoreWeave cannot.

CoreWeave illustrates concentrated-customer risk: its business depends heavily on one major client, leaving little fallback if AI demand softens or that client pulls back.

The AI infrastructure build-out may run its course within roughly 18 months, after which data centers must generate sustainable revenue rather than rely on capex.

Lance frames AI as analogous to the dot-com era: the technology may transform the economy while many AI-dependent stocks still fail.

Investors are urged to focus on companies with durable moats, real revenues, and enough existing cash generation to survive when the AI investment cycle turns.

Adam Taggart's takeaways

  • Top takeaway Adam Taggart's central thesis is that AI is real and we are still very early, but investors must separate moated hyperscalers (Microsoft, Google, Amazon) from exposed neo-clouds (CoreWeave) — long the former, wary of the latter.
  • He argues AI skepticism is overdone because the income is verifiable — hyperscaler operating revenues are real, visible, and trackable today.
  • He frames AI as analogous to the early internet, expecting both dot-com-style blowups (Meta as pets.com, Google as Global Crossing, Microsoft as Lucent) and 10x survivors; betting against the top hyperscalers means betting against the best capital allocators in history.
  • He expects the AI infrastructure build-out phase to end within roughly the next year and a half, after which data centers shift from cost centers to revenue generators — a timeline investors should plan around.
  • His actionable advice: own companies with moats and existing revenue generation, and watch operating revenues as the leading indicator that will tell you if the AI thesis has broken.