Story
Sep 16, 2026
AI’s Safety Pause Turns a Market Darling Into a Test of Whether the Boom Has Gone Too Far
A global sell-off in AI hardware stocks exposed investors’ fear that data-center spending depends on ever-faster model development. Analysts remain split between a temporary overreaction and evidence of an unsustainable bubble.
The AI trade faced its sharpest reality check yet after leading executives urged the industry to slow development, sending chipmakers, data-center suppliers and other infrastructure stocks lower worldwide. The sell-off raised a question markets had largely avoided: can the enormous spending on computing capacity still be justified if the race itself is deliberately restrained?
The bearish case is straightforward. Anthropic chief executive Dario Amodei warned that “building [AI] too fast is reckless,” while investors began asking who would pay for leases, debt and power commitments if demand weakened.1 South Korea’s KOSPI fell 3.7%, SK Hynix dropped 5.75% and SoftBank slid as much as 13%, illustrating how quickly anxiety spread beyond AI developers to the companies financing and supplying them.
That interpretation gained support from the broader market. Nvidia fell 3.4% and Micron 5.3%, while the Nasdaq lost 0.6%. Analysts said the concentration of expected earnings growth in a handful of technology companies leaves portfolios vulnerable to any sustained retreat. Capital Economics sees further near-term upside for the S&P 500 but expects an AI-driven bubble to burst next year; Vital Knowledge’s Adam Crisafulli offered a less absolute warning: “This pace of spending is absolutely not sustainable. But that doesn't mean everything that is associated with AI has to suffer.”2
The more optimistic view is that the market has confused a proposed training slowdown with a collapse in AI demand. CNBC’s analysis argued that inference—the use of existing models—would continue, alongside new applications built on them. Its conclusion was blunt: “nothing has fundamentally changed” in the underlying buildout, though investors should avoid aggressive purchases while rates and political risks add uncertainty.3
Both sides agree the market has become excessively concentrated and vulnerable to headlines. They differ over what Monday’s rout means: an overdue repricing of infrastructure spending, or another short-lived overshoot by investors chasing a new narrative.