Fitch Ratings has gathered a number of the market’s quieter anxieties into a single sentence, warning that a possible AI market correction is now emerging as one of the biggest credit risks facing the global economy.

The judgement comes in the agency’s third-quarter Global Risk Outlook, published this week, and it lands at a moment when the money flowing into artificial intelligence has grown large enough that a stumble would be felt well beyond the technology sector.

The core of the argument is about entanglement rather than any single company’s balance sheet. Equity markets, corporate bond issuance, and even headline economic growth have all leaned heavily on AI over the past year, so a reassessment of long-run returns would not stay contained.

“The scale of AI investment is such that the exposure of the economy and overall capital market to such a correction is significant,” the agency wrote, adding that the way capital markets and economies have become intertwined with AI has “created a vulnerability for credit.”

The worry has been circulating among policymakers for months, and it echoes the BIS warning that an AI bust could hit credit markets as hard as 2008.

Some of the figures Fitch cites make the concentration concrete. It notes that the cyclically adjusted price-to-earnings ratio on the S&P 500 is now close to levels last seen during the late-1990s dotcom boom, and that US corporate bond issuance jumped 26% in the first half of 2026.

Much of that borrowing traces back to a handful of names, since Amazon, Alphabet, Nvidia, Meta, Oracle, and SpaceX together sold roughly $182bn of investment-grade bonds, a spree that has already pushed Big Tech’s AI debt into European markets.

Because the spending shows no sign of slowing, the exposure keeps building. Capital expenditure at Alphabet, Amazon, Meta, and Microsoft is projected to rise about 75% to some $700bn a year, a pace that is now outrunning the cash those businesses generate.

IT investment alone added 1.4 percentage points to US GDP growth in the first quarter, which is the same intertwining seen from the other direction.

When outlays on that scale are funded increasingly through debt, the question of whether AI revenues arrive on time stops being a purely equity-market problem and becomes a credit one, which is precisely the point.

The strain is not hypothetical. Earlier this month S&P cut Oracle to BBB-, one notch above junk, as its data-centre build burned through cash, and Fitch’s outlook reads as a broader signal that the Oracle downgrade may be less an outlier than an early example. Even so, the agency is careful about scale.

A modest pullback would be absorbed, it argues, though a larger, prolonged correction could carry “wider market, macro and credit effects depending on its scale, duration and contagion.”

Fitch’s caution extends beyond markets to the real economy. It has flagged that AI’s efficiency gains could weaken employment and erode tax bases in developed economies even as the technology lifts productivity, and it has separately pointed to private credit as an area worth watching, while judging that segment unlikely to pose a systemic risk on its own.

Across its recent research the through-line stays consistent, because the upside of AI for corporate credit remains largely incremental and hard to quantify, whereas the downside is increasingly easy to model.

None of this is a forecast that the correction will happen, and Fitch does not claim it is imminent. What the agency is describing is a build-up of exposure, the sense that so much of the market’s recent strength now rests on a single bet paying off.

Uncertain revenues, historically stretched valuations, and record borrowing all sit together, and whether that turns out to be prudence or an underpriced risk depends, as Fitch itself concedes, on future returns that nobody can yet see.

Get the TNW newsletter

Get the most important tech news in your inbox each week.