The International Monetary Fund warned in its October Global Financial Stability Report, as reported by the Guardian and Reuters, that a sharp fall in AI-related share prices could spread across global markets. The report was presented on Tuesday, October 13, in Bangkok, which was late Monday night in the United States, at the IMF and World Bank annual meetings there.
The figures below are as relayed by Reuters and the Guardian, each attributed to the outlet that carried it; Plainly Now had not seen the report’s own text when this article was written.
What the IMF warned
According to the Guardian’s report from Bangkok, the IMF said valuations of AI companies, meaning how high their share prices are relative to their earnings, are very high and the debt those companies carry is growing, which keeps risks to the financial system elevated. It named three risks it called plausible near-term catalysts for market turmoil: a sharp fall in AI share prices, rising government borrowing costs and a major cyber-attack.
Reuters described the report as a warning against complacency in markets that have so far held up well. It also reported that the IMF said these risks by themselves do not necessarily threaten financial stability, but that they could become systemic if they are amplified by weaknesses already in the system. In Reuters’s framing, then, the warning is conditional.
Reuters quoted Athanasios Vamvakidis, a deputy director in the IMF’s monetary and capital markets department, as saying: “An ensuing AI market correction would feed back through the real economy.”
Why AI valuations and debt matter
The Guardian reported that a correction could be set off if AI investment or profits turn out weaker or less durable than expected, and that it could be made worse by borrowing, concentrated exposures and what it called circular financing, which it listed among the amplifiers. It described the deals involved as tying stronger AI companies to weaker partners in the same sector.
The Guardian cited several figures. In its own account, the Magnificent Seven, a nickname for seven large technology companies that it listed as Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia and Tesla, make up more than 35% of the S&P 500, the main US share index, up from 12% a decade ago, and their combined market value is up more than 50% in two years. In its account of the IMF’s findings, it reported $300 billion of transactions over three years between large companies that build AI infrastructure and weaker AI firms, supported by $650 billion of debt, with $1.4 trillion of future financial commitments in exchange. In the Guardian copy read, it is not shown whether the IMF or the newspaper computed these three deal figures, and one rewrite of the report credited them to the IMF. The Guardian’s account names no company as responsible for the deals; the Magnificent Seven list is a share-of-index figure only.
The Guardian gave Amazon Web Services, Google, Meta and Oracle as examples of the companies building AI infrastructure. Separately, Reuters reported that hyperscalers, the largest cloud and AI infrastructure companies as a group, now account for about 10% of global investment-grade bond issuance, meaning bonds from borrowers rated as lower-risk, up from under 1% in 2024.
The Guardian listed where the borrowing sits: corporate debt, hedge fund investment in the sector and retail investors borrowing to buy AI stocks. In a downturn, it said, these could force selling and push volatility, meaning the size of price swings, higher. It reported the US and Asian markets as especially exposed, with the UK and euro area possibly more insulated, although investors there who hold US assets could still lose.
Borrowing costs and cyber risk
Reuters reported that government bond yields, the interest rates governments pay to borrow, have risen since the IMF’s April report, helped by higher energy prices, and that heavy public debt and unresolved budget deficits in major economies add to the strain. The IMF urged governments to put public debt on a sustainable path and policymakers to curb excessive leverage, which means using borrowed money or financial contracts such as derivatives to increase exposure. Bond markets have so far absorbed the stress, Reuters said, and hedge funds unwound some borrowing in the repo market, where investors borrow cash for short periods against securities they hold, without causing turmoil.
On cyber risk, Reuters reported that in over 70% of cases so far in 2026, software providers had almost no time to patch vulnerabilities before they were exploited, up 20 percentage points from a year earlier. According to Reuters, AI can help address some of these risks but also makes attacks more sophisticated, and a major attack on a critical service could ripple through markets quickly.
Reuters also reported on a stress test, a hypothetical exercise that checks how banks would fare in a downturn, built around an AI slowdown and bond-market stress. Bank capital cushions held up overall, but under the test banks holding 22% of global banking assets had weak capital, up from 18% in the 2025 test. The coverage labelled this weak capital and did not define the term further, so the 22% should not be read as a failure rate.
How it fits earlier IMF reports
The IMF’s April 2026 report, published on April 14 and framed mainly around the war in the Middle East, had already flagged stretched valuations and concentration, particularly in AI-related firms. In that report, AI-related capital spending through 2029 was estimated at about $3.4 trillion. Its stylized scenario, an illustrative what-if rather than a forecast, said hyperscaler debt of $800 billion could exceed $1 trillion under assumptions of a three-year useful life for the assets and debt-financed investment. That $800 billion scenario figure and the Guardian’s $650 billion of debt supporting the deals are different measures and should not be lined up as a trend.
The IMF’s October 2025 report, titled Shifting Ground beneath the Calm and published on October 14, 2025, said the forward price-to-earnings ratio of the S&P 500, the share price divided by expected earnings over the next 12 months, sat near the 96th percentile of readings since 1990, meaning higher than about 96% of those readings, against about the 81st percentile implied by its fair-value model. It said the Magnificent Seven made up about 33% of the index. The Guardian’s more than 35% figure for 2026 and this 33% for October 2025 may not be the same measure, so they should not be read as an exact change.
What the other chapters say
Two chapters were already posted on the IMF’s website. The chapter on hedge funds says their gross assets have tripled since 2013, to about $13 trillion, with gross notional exposures of about $40 trillion, according to the chapter summary and an IMF blog. It says hedge funds help markets function in calm periods but that leverage and crowded positions can force synchronized selling under stress. The IMF blog from October 6 added that hedge funds held about 9% of US Treasurys, government debt securities of all maturities, in 2025. As far as the blog opened for this article goes, it does not discuss AI-specific exposure.
The IMF’s release page summarizes the chapter on tokenization, which means representing assets on shared digital ledgers that can run programmed rules, as finding that tokenized markets are small and fragmented, with legal certainty, regulatory clarity, the ability of systems to work with each other and the assets used for settlement named as constraints.
What markets did
An investingLive market wrap published Monday evening, which does not state a time zone, put the S&P 500 at 7,807.24, down 0.06%, and the Nasdaq Composite, another main US share index, at 27,245.90, down 0.44%. Monday was a holiday session in the US: stocks traded, while the cash market for US Treasurys was closed. Those moves came before the report and were not a reaction to it.
On Tuesday, according to an Asia-Pacific market wrap from investingLive stamped 3:38, with the time zone not stated, Asian stocks fell sharply, with the Kospi, South Korea’s main share index, down about 7% and Japan’s Nikkei index down about 2.6% at points in the session. That wrap attributed the declines to a chip-stock selloff, doubts about how long the AI rally can last and high oil prices, and it did not link them to the IMF report. US stocks had not yet opened for Tuesday as of the writing, about 12:45 a.m. EDT, so no US market reaction to the report is available.
What comes next
The IMF and World Bank annual meetings in Bangkok run through October 18. The IMF’s release page lists Chapter 1 for the Oct. 13 launch, so the chapter text and executive summary may be posted later that day; the page is the place to check the IMF’s own wording against the figures above. The IMF has published the Global Financial Stability Report twice a year, in April and October in recent issues, though its page states no formal schedule.