News · 2026-06-28
The world's central-bank watchdog warns an AI bust could spill into the wider economy
The Bank for International Settlements -- the institution central banks use as their own central bank -- issued a warning this week that the AI investment boom carries risks reaching well beyond the technology sector. As reported here, the BIS cautioned that if the flood of money into AI reverses, the fallout could ripple from economic growth into the credit system, where it would touch businesses and households with no direct stake in the technology at all.
A little context on who is talking, because the messenger is the message. The BIS is a deliberately boring, deeply conservative body whose job is watching for the slow build-up of risk in the global financial plumbing. When it uses words like ripple effects and credit, it is not chasing headlines -- it is doing the thing it exists to do, which is to flag a fragility before it breaks. So the notable fact is not that some commentator called AI a bubble; people have done that for two years. It is that the central banks' watchdog now considers the scenario serious enough to put in print.
The mechanism it is worried about is straightforward once you lay it out. An enormous amount of capital has poured into AI -- chips, data centers, the power to run them, and the companies building on top. A lot of that money is borrowed or staked on the assumption that the revenue will eventually show up to justify it. If that assumption wobbles -- if the returns arrive slower or smaller than the spending implied -- then the financing can pull back sharply. Venture funding dries up, credit lines for AI-heavy firms tighten, and because modern finance is interconnected, the stress does not stay politely inside the tech bucket. Banks exposed to AI borrowers, suppliers selling into AI build-outs, regions banking on data-center jobs: the BIS is pointing at all the wires running out of the AI sector into everything else, and asking whether anyone has stress-tested them.
The so-what is that this reframes the AI conversation from a purely technological one into a macro-financial one. For most of the boom the debate has been about capability -- can the models do the thing. The BIS is asking a different and colder question: what happens to the rest of the economy if the spending got ahead of the substance. That does not require AI to fail. It only requires the gap between what was invested and what was earned to close the wrong way, fast.
The reaction was split along the usual fault line. One camp called the warning well-timed and overdue, pointing at valuations that only make sense if near-everything goes right. The other argued the fundamentals are genuinely different this time -- that AI is already generating real productivity and real revenue, and that financing cycles come and go without negating the underlying technology. The honest caveat cuts both ways: the BIS is flagging a risk, not forecasting a crash, and a warning from a cautious institution is a smoke detector, not a fire. But it is the kind of detector that is worth taking seriously precisely because it so rarely goes off.