IMF Warns $1.4tn in Circular AI Deals Could Deepen Market Correction
The International Monetary Fund (IMF) has cautioned that a correction in AI-related stocks and debt could trigger significant market turbulence. This risk was identified as one of three potential scenarios that could sharply tighten financial conditions, alongside rising government bond yields and cyber threats.
The Circular AI Deal Risk
According to the IMF's Global Financial Stability Report, which was presented at the annual meetings in Bangkok, about $300 billion in equity investments and $650 billion in debt have been funneled into financially weaker AI companies since 2023. In return, investors led by hyperscalers secured $1.4 trillion in future revenue commitments.
A concerning aspect is that approximately $320 billion of this debt remains with the receiving firms, with a handful of them holding most of the financing and revenue promises. The IMF describes this as "wrong-way risk": equity and debt investments depreciate when future revenue streams are impaired.
Rising Borrowing by Hyperscalers and AI Firms
The report highlights a dramatic increase in borrowing by hyperscaler data center operators, with their investment-grade bond issuance rising from under 1% in 2024 to 10% of the world's total. These companies' free cash flow is declining rapidly as they continue to invest heavily in AI.
Other AI firms, such as neoclassical cloud operators, carry higher debt levels and are less profitable, raising concerns about potential weak links in the financial system. A recent legal analysis mapped out how AI data center debt could unravel.
Investors' Borrowing and Geared Investments
Investors, including hedge funds and retail investors using geared exchange-traded funds (ETFs), have also increased their exposure to AI stocks. Geared ETFs magnify returns, leading to a significant 28% of these funds now tracking single AI stocks, up from 16%.
Advanced Asian stock markets are expected to be the most vulnerable in the event of an AI-driven correction, while Europe and the UK may still face stress through their higher holdings of US shares.
Cyber Threats and AI Model Access
The IMF report also addresses cyber threats, noting that state-sponsored groups, ransomware gangs, and insiders now have access to frontier AI models capable of powerful network attack simulations. The rapid evolution of AI technology is outpacing the ability of software providers to patch vulnerabilities, with less than 80% of cases addressed before exploitation in 2026, compared to over 90% in 2021.
Bank Stress Test Findings
The IMF conducted a 2026 bank stress test that included an energy price shock, higher inflation and bond yields, and an AI-led market correction. Despite these challenges, the Fund anticipates that banks would remain broadly resilient.