ECB counts $4.5 trillion in AI commitments kept off tech balance sheets

ECB counts $4.5 trillion in AI commitments kept off tech balance sheets

ECB finds $4.5 trillion in AI debt that never reaches company balance sheets

The European Central Bank (ECB) estimates that eight large US tech companies have $4.5 trillion in AI-related commitments that do not appear on their balance sheets. That is 287% higher than the off-balance sheet commitments the ECB counted in 2025. The figure came out in the ECB's Financial Stability Review, published the day before Anthropic's $1 trillion IPO.

These commitments are years of promised payments for data center rentals, cloud computing, and chips. The ECB says they are debt-like in economic terms, yet they do not show up as liabilities under US accounting rules.

Key numbers from the review

  • The ECB put off-balance sheet AI commitments at $4.5 trillion, up 287% from its 2025 count.
  • The figure covers Alphabet, Amazon, Meta, Microsoft, Nvidia, Oracle, Broadcom, and Anthropic.
  • The US Federal Reserve, in its own report, counted more than $3 trillion in similar obligations at seven of the same firms.
  • The ECB published its review one day before Anthropic's IPO.

What the ECB report says

The ECB's tally focuses on hidden, off-balance sheet commitments. It says these dwarf what the companies report as formal liabilities like debt.

Under US accounting rules, a lease only appears on a balance sheet once the asset is available for use. Purchase commitments that have not started are listed in footnotes, and the ECB says those footnotes have grown sharply over the past year.

The report also describes how large cloud providers turn clients' long-term compute contracts into collateral for loans. It points to Oracle's $300 billion deal with OpenAI as an example. The ECB warned: "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."

The report repeatedly flags circular financing among AI companies. ECB Vice-President Boris Vujcic explained the pattern: chipmakers and cloud providers may invest in AI labs that, in turn, commit to buying their hardware and services.

The ECB also warned that the rapid shift toward debt finance invokes historical parallels with the dotcom and subprime bubbles.

The Federal Reserve reached a similar finding

The US central bank reached its own conclusions separately. In a speech, Fed Governor Michael Barr counted over $100 billion of stakes that hyperscalers and GPU makers hold in AI labs and cloud startups, and called the pattern "circular financing."

The Fed's financial stability report, published the Friday before the ECB's review, tallied more than $1 trillion of signed but unstarted leases. It added more than $2 trillion of take-or-pay contracts and guarantees at seven of the same firms. By the Fed's calculation, the nominal value of these off-balance sheet obligations now exceeds $3 trillion.

According to the source report, moving cash around without true revenue from consumers or outside industries can misleadingly inflate top-line numbers and forecasts for AI companies.

What is confirmed and what is an estimate

The $4.5 trillion figure is the ECB's own estimate, not a confirmed total of money owed. The Fed's figure of more than $3 trillion is also its own calculation, and it covers seven firms rather than eight. The two central banks used different methods, so their totals differ.

The ECB's warning about parallels with past bubbles is its stated concern, not a confirmed prediction of a crash.

Why this matters

If the central banks are right, the true financial commitments behind the AI build-out are far larger than company balance sheets show. That matters for anyone assessing the health of these firms, because the obligations are debt-like but sit in footnotes instead of liability lines. The ECB's concern about circular financing also points to a risk: if the money flows mainly between AI companies themselves, reported revenue may overstate real demand from outside customers.

Sources

Newisty Editorial Team
Written by

Newisty Editorial Team

Technology · Crypto · Digital Economy
View all posts

Newisty Editorial Team covers technology, cryptocurrency, digital products, online platforms, developer tools and the wider digital economy. Our content is researched from official sources, company announcements, public documentation, market data and other primary or reputable sources. Articles are reviewed and edited before publication for clarity, accuracy and useful context.

Comments (0)

Leave a comment
Your comment will appear publicly after submission.
No comments yet. Be the first to comment!