Central bankers warn debt-fuelled AI boom risks a global financial crash

The Bank for International Settlements warned on Sunday that substantial debt-fuelled spending on artificial intelligence is driving up the risk of a global financial crisis. The BIS, known as the bank for central banks, said “excessive” spending on new AI data centres and opaque, debt-fuelled transactions risked a meltdown similar to the global credit crunch nearly two decades ago.

The warning is one of the strongest yet from a mainstream financial regulator on the risks lurking in the AI boom. It lands at a moment when global tech stocks have already begun to wobble. The Nasdaq dropped as much as 1.3 percent on Friday after Apple said it would increase prices because of higher microchip costs. South Korea’s blue chip index, heavily tied to the fortunes of the country’s AI chip businesses, has seen swings of more than 10 percent in a single day.

What is new here is not the observation that AI stocks are overvalued. The Bank of England warned in December that share prices were the “most stretched” since the 2008 crisis. The International Monetary Fund has compared AI valuations to the excesses of the dotcom bubble. What is new is the BIS’s focus on the structure of the financing behind the AI buildout — and its conclusion that the system is dangerously opaque.

Pablo Hernández de Cos, the BIS general manager, said there were major questions about whether the boom would benefit the wider economy. “One risk is that large-scale investment in AI infrastructure becomes excessive, as each firm tries to outcompete rivals and dominate market share,” he said. “This could leave the sector more vulnerable if AI underdelivers, possibly bringing the current investment boom to an abrupt end, with large macroeconomic consequences.”

The BIS report identifies a specific mechanism for a crash. Big tech groups such as OpenAI and Nvidia have turned to complex financial transactions to fund AI’s development. Bot developers often receive loans from chipmakers to buy the chipmaker’s microchips. Shadow banks — private credit funds that operate outside traditional banking regulation — have piled money into AI data centres. The shadow banking industry has expanded rapidly as regulators around the world tightened rules governing mainstream banking after 2008.

The BIS says this web of financial ties between AI giants, shadow banks and data centre builders could unravel. “Should hyperscalers slow or halt the aggressive pace of capex deployment, many borrowers across the supply chain could struggle to replace lost revenue and service their debt,” the report states. “The opacity of AI-sector financing compounds these vulnerabilities.”

The report adds that “signs of stress are already visible” in private credit funds, with many inundated with redemption requests and in some cases forced to block withdrawals. This is the kind of liquidity crunch that can cascade through a system where nobody fully knows who owes what to whom.

“Should hyperscalers slow or halt the aggressive pace of capex deployment, many borrowers across the supply chain could struggle to replace lost revenue and service their debt.”

The BIS draws explicit historical parallels. It notes similarities between the AI infrastructure surge and the dotcom boom, the British railway mania of the 1840s, and the “roaring 20s” before the Great Depression. “The scale and pace of the current AI investment boom accompanied by expectations of large productivity payoffs bear resemblance to these precedents, highlighting potential downside risks in the near term,” the report said.

For AI builders and the broader industry, the BIS warning carries immediate practical implications. If the financing spigot tightens — because private credit funds freeze redemptions, or because chipmakers stop extending vendor loans, or because hyperscalers cut capex — the downstream effects will be felt in data centre construction, GPU availability, and startup runway. Many AI startups depend on access to compute that is effectively subsidised by the same debt-fuelled infrastructure boom the BIS is flagging.

The BIS also warns that “bottlenecks” in data centre construction or a shortage of chip supplies could threaten the AI boom. AI giants have already taken to rationing access to their most powerful tools at peak times because of a lack of capacity. The Financial Times reported on Sunday that Google had capped Meta’s use of the search giant’s Gemini AI because it was struggling to meet the Facebook owner’s demand. These are not hypothetical stresses. They are happening now.

The BIS’s warning is not a prediction of imminent collapse. It is a risk assessment from the institution whose job is to worry about systemic financial stability. But the risk it identifies is structural, not cyclical. The AI industry has built its expansion on a financing model that combines record levels of corporate debt, unregulated shadow banking, and opaque vendor-financing arrangements. That model works as long as AI revenue grows fast enough to service the debt. If AI underdelivers — or if the market simply decides it has overpaid — the unwind could be brutal.

The question the BIS leaves unanswered is whether any regulator has the tools to intervene. Shadow banks are, by design, outside the perimeter of traditional banking regulation. The cross-border nature of AI financing — chip loans from US companies, data centre construction in multiple continents, private credit funds domiciled in offshore centres — makes national-level oversight nearly impossible. The BIS can warn. It cannot stop.

For the AI industry, the lesson is uncomfortable. The boom has been financed on the assumption that the technology will deliver transformative productivity gains. The BIS is saying that assumption is itself a risk factor. If the boom is built on debt, the bust will be built on the same foundation.