What the IMF and central banks actually measured on AI valuations

'Is AI a bubble?' is usually argued with charts and opinions. Financial stability reports are different: they are written by institutions with access to supervisory data, and they have to show their measurements. In 2026, three of them made AI a central theme.

Bank of England, July 7: valuations and leverage

Governor Andrew Bailey's remarks on the July 2026 Financial Stability Report cite the cyclically adjusted price-to-earnings (CAPE) yield for the S&P 500, an inverse valuation measure, which 'has fallen to levels not seen since the dot-com bubble.' The detail that matters: even excluding the 30 most AI-exposed S&P 500 stocks, as defined by a JPMorgan basket, the level is 'around the lowest seen since 2007.' In other words, the Bank sees stretched valuations beyond AI itself.

The Bank's bigger worry was leverage. It reported record prime brokerage balances, rising hedge fund leverage in equities, and rapid growth in levered ETFs, 'particularly in those holding AI-related stocks,' mostly held by retail investors.

Reserve Bank of Australia, October: what happened in July

The Reserve Bank of Australia's October 2026 review describes a stress episode. AI-related equities 'experienced significant volatility in July,' reflecting concerns about revenue growth, debt-financed spending and competition from China. It was most pronounced in South Korea and Taiwan, where retail investors had piled into single-stock leveraged ETFs on AI names; Korean regulators restricted those products in July. The review also states that the hedge fund Situational Awareness suffered major losses on leveraged AI positions that month and was forced to sell to meet margin calls, and that these fire sales were reported to add to volatility and to losses elsewhere.

IMF, October 13: circular financing

The International Monetary Fund's October 2026 Global Financial Stability Report names an AI-driven market correction as one of three rising risks. Its most concrete number comes from a review of 14 'circular financing' transactions since 2023, deals in which AI firms invest in, lend to and buy from one another.

The concern, in the IMF's words: stress at a company receiving investment 'could lower revenue forecasts and investment valuations, which in turn would weaken the value of the accumulated debt in those deals.' The Fund's policy recommendation includes limiting leverage in AI and requiring banks to hold capital against concentrated exposures to the sector.

Where the three agree

All three point to the same chain: high valuations concentrated in a few AI firms, more leverage on top of them (from hedge funds, retail levered ETFs and corporate debt), and financing links that tie stronger and weaker AI companies together. None of them predicts a crash. Each describes how a disappointment in AI earnings could spread further than it would have in 2024.

How to read these reports as a trader

These are risk assessments, not forecasts or price targets. Central banks publish them to push regulators and lenders to build buffers. The July episode is the useful case study: the trigger was a change in sentiment about AI revenue and debt, and the damage was amplified by leverage. The reports point to where that leverage sits, which is the part a trader can actually monitor.