Apollo calls AI a 'bubble' worse than even the dotcom era

Original Article ↗ Hacker News Discussion ↗

Whether today’s AI boom is a genuine revolution or a speculative bubble is fiercely contested. Commenters compare Nvidia’s soaring valuation and the rush to add “AI” features to everything with the dot‑com and crypto eras, noting real advances in machine learning alongside shaky business models, hype-driven investment, and limited proven monetization. Many expect AI to become foundational infrastructure like the web, but warn that most current AI startups may fail and that investors risk overpaying for uncertain future returns.

Is AI a Bubble or Just Overhyped?

  • Many argue current AI valuations, especially around large models and GPUs, are bubble-like: rapid price rises, speculative behavior, media frenzy, weak fundamentals for many startups.
  • Others say “bubble” only applies to investor expectations, not to the underlying tech, which they see as genuinely transformative.
  • Some distinguish between a speculative generative-AI bubble and a broader, durable ML/AI trend that predates this wave and will outlive it.

Comparisons to Dot‑Com and Crypto

  • Strong parallels drawn to the dot‑com era:
    • Overfunded companies built on vague or premature business models.
    • “Shovel sellers” (now Nvidia, then Cisco/Sun) profit heavily during the rush.
    • Survivorship bias: Amazon-style winners don’t erase masses of Pets.com‑style failures.
  • Unlike dot‑com, very few pure‑AI companies are public; much of the excess is private and not shortable.
  • Compared to crypto: AI is seen as having clear existing utility; crypto is often portrayed as still searching for a use case. Some note the same hype class has migrated from crypto to AI.

Nvidia, Valuation, and Addressable Market

  • Debate over whether Nvidia’s valuation is justified:
    • Bulls cite current revenue/profit, AI as a proxy for “all human labor,” and massive theoretical TAM.
    • Skeptics highlight heroic assumptions in discounted cash flow models, dependency on unprofitable AI customers, and parallels to past GPU/crypto booms.

Actual Utility vs Hype

  • Supporters cite concrete uses: recommendation systems, logistics, computer vision, language tools, code assistance, image editing, and future process optimization in enterprises.
  • Critics argue generative AI is still unreliable, hallucinates, and often serves as superficial “AI-washed” features (chatbots, summarizers) with little real demand.
  • Some foresee AI deeply embedded in back-office optimization; others say “corporate efficiency” is already maxed out.

Macro, Markets, and Long‑Term Impact

  • Multiple comments note excess global capital “sloshing around” chasing returns, amplifying hype cycles.
  • Even if this is a bubble, many expect AI to become as foundational as the web, with many firms dying but a few enormous winners.
  • Concerns raised about societal impacts: job displacement, bias, misuse in welfare/immigration/proptech, and overreliance on “magic” systems without proper oversight.