Why You Should Expect an AI Crash

Commenters who expect an AI crash point to circular financing, deals booked ahead of delivery, and enormous infrastructure spending that they say may not make economic sense. Some compare these practices to past accounting scandals and wonder whether Nvidia’s CFO succession signals concern about liability or a change in how such deals are handled. Others say a correction is inevitable but impossible to time; a failed Anthropic IPO could expose fragile valuations. One commenter also reports cloud capacity shortages and argues that competing for AI infrastructure could make software more expensive to build and undermine SaaS economics.

Others question whether crash predictions have become too conventional to be useful, or whether investors have already priced in the risk. Some think a downturn could unwind without major damage if data centers are cancelled before construction, while chips and completed facilities could still find uses. Commenters also distinguish a financial bubble from the technology’s usefulness: a market fall would not necessarily erase demand for AI, and the internet’s dot-com bust is offered as an example of a failed boom whose broader vision still had merit. The divide is between seeing today’s spending and valuations as unsustainable, and seeing a volatile but valuable technology whose infrastructure may remain productive even after a correction.