Forget the AI bubble. The IMF says the real threat is the mountain of debt behind it — and 60% of planned data centers haven't even broken ground

“What is quite worrisome from a financial stability perspective is that the major tech firms are starting to leverage up themselves,” Adrian told Bloomberg’s Francine Lacqua, adding there was a “potential maturity mismatch in between the duration of the physical assets and the duration of the debt.”

A maturity mismatch occurs when firms rely on short-term debt to finance long-term assets. In this case, the concern stems from hyperscalers pouring cash into AI equipment like data centers and chips that could lose value before the debt is repaid due to unexpected circumstances in the rapidly evolving tech sector. In the worst-case scenario, financing dries up before tech firms generate meaningful profits to justify the original spending.

At the moment, investors are happy to fork over cash to AI companies. Still, there are signs of a slowdown in physical infrastructure. A J.P. Morgan analysis from May indicated that 60% of data center capacity intended to be completed by 2027 still hasn’t reached the construction stage. Another 7% is delayed.

The AI borrowing binge

Demand for corporate bonds in the AI sector is red-hot with a borrowing binge underway.

Last month, Alphabet announced it was raising $85 billion in equity to finance its AI buildout — and it’s far from alone in doing so. AI hyperscalers such as Amazon, Alphabet, Meta, Microsoft and Oracle have issued $159 billion in corporate bonds in the first five months of 2026. The staggering sum surpasses the companies’ total borrowing in the past five years, according to financial services firm Dealogic.

Nvidia also issued $25 billion in corporate bonds last month, its first round since 2021. The chipmaker had no trouble finding willing customers among investors frantic to expose their portfolios to the AI craze.

Adrian cited strong earnings from tech companies as a reason to downplay the possibility of an AI credit boom going bust.

Other AI concerns from the IMF

IMF officials recently warned about the possibility of frontier AI models wreaking havoc on global finance. In April, Anthropic announced a new model known as Mythos capable of exploiting cracks in the software running power grids, major banks and other institutions. Anthropic held off on releasing it to the public, and only shared it with a dozen trusted partners in the corporate sector based in the U.S.

The IMF, though, believes the frontier model is only the tip of the iceberg.

“What we recognize is that Mythos is just the beginning, there will be more like it,” IMF Managing Director Kristalina Georgieva said last month, according to Politico, adding that advanced AI models can be weaponized by bad actors to “destroy the financial system” without safeguards in place.