Analysts warn AI infrastructure spending could amplify a market drop

In short: Some analysts are warning that today’s big, debt-backed spending on AI data centers could fall short of expectations and make any market downturn worse.

What's going on

Warnings about an “AI bubble” are getting louder because the spending is massive and the payoff is uncertain. One estimate from Gartner suggests data-center owners would need about $8.5 trillion in revenue through 2030 to cover what they are investing, plus financing costs and a target return. Another analysis suggests AI revenue flowing to data centers could be closer to $3.6 trillion, which would leave a gap.

A second concern is where the borrowing sits. Some of it may be outside the balance sheets of the biggest tech companies, for example with data-center operators and private-credit funds (lenders that are not regular banks). Commentators also point to “circular” financing, where an AI company helps fund a customer or supplier, and that partner then uses the money to buy the AI company’s products. It is a bit like a store giving someone a loan so they can shop at that same store.

Markets also look more exposed than before. A J.P. Morgan analysis cited by NPR said 42 AI-related companies drove about two-thirds of S&P 500 gains since early 2024. If those stocks drop sharply, many retirement and investment accounts could fall with them.

What to watch

Comparisons to 2008 are mainly a warning, not a certain forecast. The 2008 crisis was centered on housing loans and a freeze in bank lending, and some writers argue bank safeguards are stronger now. Still, if AI spending is heavily debt-financed and revenues do not catch up, the knock-on effects could include layoffs, lower stock values, and weaker business investment.

Source: NYTimes