The International Monetary Fund warned on October 13, 2026 that elevated valuations of artificial intelligence companies and their growing debt burden rank among the highest risks to global financial stability, in its semiannual Global Financial Stability Report, presented in Bangkok.
The Guardian reports that the body identified three standout risks: a correction in AI markets, rising government borrowing costs and the threat of a large-scale cyberattack. The report quantifies the concentration behind the first: the seven largest U.S. technology companies account for more than 35% of the S&P 500 index, versus 12% a decade ago. The diagnosis lands as the AI boom sustains global growth and sector earnings, according to the report itself. Athanasios Vamvakidis, deputy director of the IMF's Monetary and Capital Markets Department, noted that the rest of the world's exposure to U.S. assets is at its highest level on record. For Mexico and Latin America the channel is indirect: a wealth effect and global pension fund portfolios.
The leverage figures complete the picture. The report counts 300 billion dollars in transactions between the large AI infrastructure companies and weaker firms in the sector over the past three years, accompanied by 650 billion in debt, in exchange for 1.4 trillion in future revenue commitments. The IMF also lists three sources of leverage: debt taken on by the tech companies themselves, hedge fund investment in the sector and retail investors who borrow to speculate. Infobae, which covered the presentation in Bangkok, details that the IMF warns about circular financing schemes and concentrated exposure, and that bank balance sheets remain broadly sound.
The IMF recommends limiting excessive exposure to AI and placing public debt on a sustainable path. On Tuesday, The Guardian reported, semiconductor makers' shares fell after Broadcom's chief compared many AI investors to casino gamblers. The body's next read on these risks comes with the April 2027 edition.
This article was written with the assistance of artificial intelligence from verified sources and reviewed by a human editor before publication.