AI Market Faces Valuation Reset as Trillion-Dollar Boom Shows Cracks

The Nasdaq 100 dropped 5.5% last week, and suddenly the question everyone had been avoiding became impossible to ignore: Is a trillion dollars in annual artificial intelligence investment sitting on something solid, or are we watching a structure built on air?

The wobble started small. An Australian datacentre company called Firmus had planned to list on the ASX with a valuation that would have made it worth more than Telstra—this before it had actually delivered the earnings it promised. The IPO never happened. That single failure rippled outward, forcing investors across Australia and beyond to reconsider whether similar businesses were worth anywhere near what they'd been valued at. It was, for many, a moment of reckoning.

Yet the market bounced back by Monday night, and the Nasdaq 100 remains nearly three times larger than it was at the start of 2023. The so-called magnificent seven—Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia, and Tesla—now represent 35% of the S&P 500, up from 12% a decade ago. Their combined market value has grown more than 50% in just two years. Gerard Minack, an independent market analyst, put it plainly: investors are fundamentally overestimating what AI companies will actually earn. But he also admitted he has no idea when the market will finally correct itself.

Raisah Rasid, a strategist at JP Morgan Asset Management, offered a different reading. What she saw in last week's sell-off was not panic about whether AI works, but rather a recalibration of what companies should be worth. This is a structural shift in the economy, she argued—nothing like the dotcom bubble. Yet she identified two genuine threats. One is that an open-source AI model, possibly from China, could undermine the entire economics of the industry and the hundreds of billions already committed to it. The other is public backlash: anger over datacentre construction, or a broader conviction that AI will harm workers and society more than it helps.

The real financial risk may lie elsewhere. Amazon, Microsoft, Google, and Apple have borrowed heavily to fund datacentres and AI infrastructure. These companies spent $425 billion on AI-related capital expenditure in 2025 and are on track to spend $800 billion this year and over $1.1 trillion next year. Their corporate bonds now account for nearly 10% of the global investment-grade credit market, up from 0.5% at the end of 2024. The International Monetary Fund last week listed the bursting of the AI investment bubble as one of three top risks to the global economy, alongside rising borrowing costs and cyber-attacks.

There is also a structural vulnerability in how the money moves. Companies have begun arranging circular financing deals—a chip manufacturer helps fund a datacentre in exchange for the datacentre buying its chips. These arrangements are only as strong as their weakest link. If demand for AI computing doesn't materialize, the datacentre doesn't get built, and the chips don't need to be made. Sally Auld, chief economist at NAB, noted this is far more of a concern for the United States than for Australia.

Australia is expecting $150 billion in datacentre investment by 2030 and $260 billion by the mid-2030s—echoing the mining boom of the 2010s. But Auld emphasized this is meaningful without being central to the economy. Australia would not be at the epicentre of any AI bust. The real question—whether this week's market pullback is a pause or the beginning of something larger—remains unanswered. The earnings season ahead will offer the next verdict on whether these companies are worth what the market has decided they are.