The number that unsettles the International Monetary Fund is not the headline total but how it was paid for. Of the $1 trillion or so now cycling between AI companies, roughly two-thirds comes from debt, the fund estimates. Managing Director Kristalina Georgieva called the sum modest in absolute terms; what caught her attention was how quickly it got there.
She made the remarks to CNBC's JP Ong on Monday, on the margins of the Association of Southeast Asian Nations summit in the Philippines. Her phrase for the world economy was a body "standing on one leg" — the AI leg — and a wobble there, she said, could turn into trouble for financial stability.
Key numbers
- $1 trillion in circular financing between AI companies, an IMF estimate, with about two-thirds funded by debt
- More than $1 trillion erased from Nvidia's market value over the past month, CNBC reported
- 3.1% the IMF's forecast for global growth in 2026, raised in October
- About 40% of the world's workforce the IMF estimates will be affected by AI
How the money goes round
The loop works like this. A large supplier takes a stake in a customer, the customer spends that money on the supplier's chips or services, and the revenue arrives back where it started. Nothing in the chain is hidden, but the same dollars can be counted as investment at one end and sales at the other.
Nvidia sits in the middle of several of these arrangements, CNBC reported, as one of the main financiers of the AI build-out. Its shares have fallen hard: more than $1 trillion of market value has gone in a month, in a sell-off CNBC linked to fallout from reports about the private conduct of former chief executive Jensen Huang, which has spread to AI-linked stocks worldwide.
In plain words: when a deal is funded with equity, a disappointment costs the investor some of the stake. When it is funded with debt, the borrowing has to be serviced whatever happens to the project, and corporate borrowing has grown more expensive as interest rates have risen. That is why the two-thirds share matters more than the headline $1 trillion.
What the fund has and has not said
I went through both reports on the interview, and neither has Georgieva calling a top or naming a trigger. She said investor enthusiasm has run ahead of what AI has so far delivered, that valuations climbing this way make "some corrections" necessary, and that the recent swings have so far looked "orderly" rather than destabilizing. She urged caution all the same.
The backdrop is that AI spending has been doing much of the work in the global economy. The IMF lifted its 2026 growth forecast to 3.1% in October on the strength of that investment, and Capital Economics calculated that as much as 80% of the recent expansion in world exports was in AI-related goods. The fund also published a worst case in October: a short-term disappointment setting off a disorderly correction across asset classes, with U.S. equity prices 20% lower in 2026 and global output 1.1% below baseline by 2028. That is a scenario the IMF drew up, not a base case.
The Defiant, summarizing the same interview, carried the two-thirds debt figure and the warning about dependence on AI investment. Neither outlet breaks the $1 trillion estimate down by company or by lender.
What to watch
- This week Nvidia's quarterly earnings, which CNBC described as a fresh test of investor appetite for the AI trade
- This week Anthropic's stock market listing, reported by CNBC alongside those earnings
