Field report NE-56 / M9 · sheet /macro

Boardroom MacroFiled from the territory

ECB Warns AI Debt Buildout Echoes Dotcom and Subprime Bubbles

The ECB counts over $4 trillion in off-balance-sheet commitments at eight AI-linked firms and warns the shift to debt finance echoes the dotcom and subprime bubbles.

Grid
NE-56 / M9
Surveyed
November 19, 2026
Reading
4 min read
Correspondent
Elena Vasquez
European Central Bank sees ‘historical parallels with the dotcom and subprime bubbles’ because of AI’s ‘rapid shift’ int
Fig. NE-56 / M9European Central Bank sees ‘historical parallels with the dotcom and subprime bubbles’ because of AI’s ‘rapid shift’ int — AI-generated

Field notes

  1. ECB estimates more than $4 trillion in off-balance-sheet commitments at eight tech firms as of Q2 2026
  2. ECB report says the rapid shift to debt finance 'invokes historical parallels with the dotcom and subprime bubbles'
  3. Fed counted more than $3 trillion in similar off-balance-sheet tech obligations a day earlier
  4. IMF puts GPU economic life at about two years versus seven for standard equipment; hyperscalers stretched server accounting life from three years to six
  5. ECB staff used an agentic AI model on ChatGPT 5.6-Sol to extract the figures from SEC filings

The European Central Bank has put the AI infrastructure boom's hidden debt load at more than $4 trillion and warned that "the rapid shift towards debt finance invokes historical parallels with the dotcom and subprime bubbles."

The estimate appears in the ECB's twice-yearly Financial Stability Review, published Wednesday. The central bank counted "total off-balance-sheet commitments of more than USD 4 trillion as of the second quarter of 2026" at eight companies — Alphabet, Amazon, Meta, Microsoft, Nvidia, Oracle, Broadcom and Anthropic — a figure it called "well in excess of their outstanding on-balance-sheet liabilities." Anthropic's numbers come from its IPO filing.

The ECB's count lands a day after the Federal Reserve's own November Financial Stability Report, which tallied more than $3 trillion in signed leases not yet started, take-or-pay contracts and residual value guarantees at seven big tech companies, none of it booked as debt. The two tallies cover different companies and use different methods.

What does the ECB say mirrors 2008?

In a dedicated box on AI debt financing, six ECB authors list five parallels to the pre-crisis years:

  • Debt-like obligations off the books. Hyperscalers sign contracts to use data centers and chips owned by others; the owners then borrow against those promised payments from "banks, bond investors or private credit funds." The ECB wrote that "such cash flow commitments are debt-like in economic terms, but they do not appear as liabilities on the balance sheets of hyperscalers under US accounting standards." A lease is not booked until the asset goes into use.
  • Circular demand. Cloud providers and AI labs buy and lease chips from chipmakers that often finance the purchases, inflating "the expected revenues of both the chipmaker and the service provider." A footnote notes "other income" at Alphabet and Amazon stems from revaluing their Anthropic stakes, and cites Nvidia's positions in CoreWeave, OpenAI and Anthropic.
  • Distrusted ratings. Spreads on hyperscaler bonds have widened versus same-rated corporate bonds, especially at longer maturities, which "could be an indication that markets currently view rating agencies' risk assessments as too benign." Credit default swap spreads for several big tech firms have climbed "to record levels."
  • Debt outlasting the assets. Debt maturities exceed "the economic lifetimes of a substantial share of the physical capital funded by this debt, especially in the form of advanced chips." The ECB cites market intelligence putting AI equipment life at one to three years; the IMF puts GPUs at about two years against seven for standard equipment. Most hyperscalers have stretched server accounting life from three years in 2020 to six in 2026.
  • Leverage outside the banks. Euro area funds and other non-bank investors have significantly increased AI exposures and begun buying AI firms' debt. The ECB points to "the near collapse of Situational Awareness, a US hedge fund, in July 2026" as concentration meeting leverage.

The historical anchors are explicit. The Financial Crisis Inquiry Commission concluded that "a combination of excessive borrowing, risky investments, and lack of transparency put the financial system on a collision course with crisis," and that rating-agency failures were "essential cogs in the wheel of financial destruction." In December 2007, Citigroup took $49 billion of structured investment vehicle assets onto its balance sheet to prevent downgrades and fire sales — vehicles built to sit off its books.

Where does the AI boom differ so far?

The ECB flags the differences too. "So far, however, hyperscalers' earnings have broadly kept up with expectations," the authors wrote. ECB Vice-President Boris Vujčić said "circular financing is making intrinsic demand hard to gauge," and that even if banks are not exposed to an AI correction "in material terms," they "could face spillovers via strong links with non-banks."

The central bank also separates the two named bubbles. The dotcom bust produced "only a shallow recession thanks to a large share of equity finance." A footnote states that credit booms that turned into bust were associated with "a significant and prolonged contraction in real GDP," and that a U.S. shock "would likely result in a significant contraction in euro area GDP as well." The authors' bottom line: "the rapid acceleration of debt financing raises concerns that the AI buildout might become debt-driven and unsustainable."

How big is the model transaction?

One concrete case shows the mechanics. Meta's $27 billion Hyperion data center in Louisiana was built through a joint venture in which Blue Owl Capital owned 80%, keeping the full debt off Meta's balance sheet. The bonds were rated A+ and yielded 6.58%.

Even the ECB needed AI to count the debt

The scale of the obligations made manual tracking impractical. A chart note says ECB staff extracted the off-balance-sheet figures from SEC filings using "an agentic AI model developed by ECB staff," running on OpenAI's ChatGPT 5.6-Sol. Staff trained it against hand-compiled first-quarter totals, then checked its output manually.

On the ECB's own tracker, earnings are the one item still holding up — and every other item depends on them. The central bank's next check is due in May, if it keeps to this year's schedule.

via fortune.com/feed/ (Source)

Share this article:

More from Elena Vasquez

Elena Vasquez

Show full bio

Staff writer covering companies, deals and strategy in corporate strategy and chief-executive leadership at CEO Frontier.

472 articles

Filed under

Adjacent sheets

« Previous articleNext article »

End of sheetNE-56 / M9 · November 19, 2026