The IMF says the eye-watering money pouring into AI hasn't yet produced matching profits, and that mismatch is now a real risk to the global economy.
The fund's new stability report lays out three specific failure modes: a sharp correction in AI company valuations, a jump in government bond yields, or a major cyberattack hitting financial infrastructure. It singles out the web of circular deals between AI companies and their partners as the real vulnerability. Reviewing 14 such transactions since 2023, the IMF found about $300 billion in equity investments and $650 billion in debt exchanged for $1.4 trillion in promised future revenue. That's a lot of IOUs riding on AI living up to the hype.
This matters because AI spending has quietly become one of the few things propping up global growth while geopolitical shocks drag on everything else. The IMF's point is that these companies are no longer separate risks. They're financially entangled, so a stumble at one doesn't stay contained, and a single cyberattack on a bank or exchange can now ripple through a dozen interlinked balance sheets at once.
In other words, the AI trade has gotten so big and so interconnected that it's started to look less like a sector and more like a systemic risk, the kind regulators usually reserve for banks.