AI borrowing binge has ‘historic parallels’ with subprime bubble, says ECB
Reported by Financial Times, Fortune and Tech Times3 sources ↓
The ECB says eight U.S. companies had more than $4 trillion in AI-related off-balance-sheet commitments in the second quarter of 2026, and warns that the sector’s shift toward debt financing echoes past bubbles.
Key points
- The ECB’s estimate covers Alphabet, Amazon, Meta, Microsoft, Nvidia, Oracle, Broadcom and Anthropic
- The commitments include obligations not recorded as balance-sheet liabilities under current accounting rules
- The ECB says debt financing and circular deals distinguish the current boom from the equity-funded dotcom bubble
- The Federal Reserve reported more than $3 trillion in related commitments at seven companies, using a different sample and method
Sources
- Financial TimesRead ↗AI borrowing binge has ‘historic parallels’ with subprime bubble, says ECB
- FortuneRead ↗European Central Bank sees ‘historical parallels with the dotcom and subprime bubbles’ because of AI’s ‘rapid shift’ into debt finance
- Tech TimesRead ↗ECB Draws Subprime Parallels as Over $4 Trillion in AI Debt-Like Commitments Stay Off Balance Sheets
Details
The ECB’s November 2026 Financial Stability Review says the eight companies had more than $4 trillion in AI-related commitments as of the second quarter of 2026, exceeding their outstanding on-balance-sheet liabilities. The estimate draws on company filings and includes Anthropic, based on its public IPO filing.
The ECB says signed leases and purchase obligations can remain outside reported liabilities until accounting rules require them to be recorded. It warns that the shift toward debt financing, circular financing and off-balance-sheet commitments resembles aspects of the subprime episode; it contrasts this with the dotcom bust, which it says was largely equity-financed and led to a shallow recession. The Federal Reserve’s separate November report counted more than $3 trillion in commitments at seven companies, using a different company list and methodology.
Written from coverage as of
Images
- Fortune
- Tech Times
Stocks in this story
Change since open: 20 Nov to 21 Nov 2026
Last refresh