Gerard Minack of Minack Advisors said investors were fundamentally overestimating the returns AI providers will deliver, while conceding he had no confidence in calling the precise top of the market. JP Morgan Asset Management's Raisah Rasid called the sell-off "a valuation reset, rather than the market questioning the efficacy of AI itself", and NAB chief economist Sally Auld likewise saw investors taking a breath rather than the trade running out of puff.

The debt angle is where the story gets sharper. American hyperscalers spent $US425bn on AI-related capital expenditure in 2025, Minack said, and are on track for $US800bn this year and over $US1.1tn next. Their corporate bonds now make up nearly 10% of the global investment-grade credit market, up from 0.5% at the end of 2024, according to the IMF, which last week ranked a bursting of the AI investment bubble as the top of three risks to the global economy.

Rasid flagged two deeper threats: an open source AI model, potentially from China, undercutting the economics of today's spending, and popular backlash over datacentres or AI's effect on workers. She also pointed to "circular" financing deals, where a chipmaker helps fund a datacentre in return for chip orders, which Auld noted are only as strong as their weakest link if AI demand fails to materialise.

For all the angst, the Nasdaq 100 is still worth nearly three times more than at the start of 2023, and the seven biggest tech names now account for 35% of the S&P 500, up from 12% a decade ago. The imminent US earnings season will offer the next test of whether those valuations hold up.