Three risks, and AI is one of them
The International Monetary Fund has named an AI-driven market correction as one of three risks that could sharply tighten global financial conditions and set off market turmoil. The other two are a sharp rise in sovereign bond yields and cyber threats. The assessment appears in Chapter 1 of the October 2026 Global Financial Stability Report, titled “Beneath the Resilience: Rising Risks, Unresolved Vulnerabilities” and released at the Fund’s annual meetings.
The Fund’s trend marker for the AI risk is “about the same” since its April report, while both bond yields and cyber threats are marked “worse”. What has changed is the detail: the chapter puts numbers on the leverage sitting underneath the AI build-out, and the numbers are the news.
$950bn in, $1.4tn promised
The chapter counts fourteen publicly disclosed transactions since 2023 in which hyperscalers led roughly $300bn of equity investment into what the Fund calls “financially weaker AI firms”, supplemented by $650bn in debt, in exchange for $1.4tn in future revenue commitments through services or chip orders. About $320bn of that debt is carried by the investees themselves, including debt backed by investor guarantees, contingent financing and other off-balance-sheet obligations.
The Fund’s concern is not the solvency of the investors. It says firms on the investing side are “generally financially sound”. The problem is the shape of the arrangement: distress at a key investee would mean that “equity and debt investments in the firms depreciate in value precisely when future revenue streams tied to them are impaired”. The report gives that failure mode a name — “wrong-way risk” — and notes the financing and the revenue promises are concentrated in a handful of investees.
Three places the leverage sits
The chapter identifies three growing sources of leverage, all measured against 2024. Hyperscalers’ share of global investment-grade corporate bond issuance has risen to 10% from less than 1%, with free cash flows “rapidly declining” as those companies try to fund their AI commitments. A footnote spells out why that matters beyond the sector: investment-grade issuance accounted for 40% of all bond issuance, sovereign bonds included, across the euro area, the United Kingdom and the United States on a duration-weighted basis in the first quarter of 2026, so AI borrowing now competes directly with governments for the same investors.
Second, hedge funds’ gross equity positions plus equity margin loans have grown from 10% to about 15% of advanced-economy GDP. Third, the share of single stocks in leveraged exchange-traded funds has climbed to 28% from 16%, which the Fund attributes to retail investors buying levered exposure to AI names.
What a correction would cost
The Fund’s companion World Economic Outlook models an unwinding of AI investment reducing global GDP by as much as 0.4%, investment by 1.3% and consumption by 0.6% by 2028. The equity damage would be uneven: advanced Asian markets show the highest beta to a global AI index, while the euro area and the United Kingdom come out low. The chapter also cites Bank of America’s August fund manager survey, in which respondents named the “AI boom” their most important risk.
What the Fund is asking for
The policy line is narrower than the diagnosis. Supervisors should “contain leverage used by investors to gain exposures to AI where excessive, including limits on derivatives, exchange-traded products, and margin loans”, and strengthen resilience with sector capital requirements, concentration limits and capital buffers. Nothing in the chapter asks anyone to slow the build-out itself.
The next test of the assessment is the April 2027 report, where the same three risks get a fresh trend marker — and where the circular-financing map will either have more lines on it or fewer.
