Central bankers have grown unusually blunt. The debt that powers today’s artificial intelligence surge now threatens to drag the global financial system into crisis, they say. Hyperscalers and their suppliers have poured hundreds of billions into data centers, chips and power infrastructure. Much of it comes financed through loans and opaque structures that echo vulnerabilities last seen before the 2008 meltdown.
The Bank for International Settlements delivered the clearest warning yet in its latest annual report. Public debt sits at record highs. Asset prices look stretched. And the AI investment wave, while lifting confidence and growth forecasts, carries familiar risks of overbuilding and sudden reversal. “Policy actions must reinforce each other,” BIS General Manager Pablo Hernandez de Cos said. “Success depends on sound fiscal and financial foundations.”
But the foundations look shaky. Amazon, Microsoft, Meta, Google and Oracle together plan more than $1 trillion in AI-related capital spending between 2025 and 2026. Hyperscaler outlays for 2026 alone track near $750 billion, a 77 percent jump from the year before. Morgan Stanley projects that hyperscalers and AI joint ventures could issue $250 billion to $300 billion in new debt by the end of next year. Already in 2025, AI-focused firms raised at least $200 billion through borrowing.
Yet earnings and free cash flow have not kept pace. Companies spend beyond what they generate internally. They turn to debt markets and, increasingly, to private credit funds that operate with lighter oversight than traditional banks. Direct lending vehicles have quadrupled their exposure to AI and technology over the past five years. The chain of financing runs from Big Tech to semiconductor makers, energy providers, real estate developers and specialized suppliers. A slowdown in one link can cascade quickly.
Fast. That single word appears often in the BIS analysis. Zhang Tao, a senior BIS official, told the South China Morning Post that any market correction could move with greater speed than previous banking crises. “If the market has any sort of correction, the interconnectedness of the financial system and interplay of vulnerabilities could mean the speed of a correction could be much faster than previous banking crisis episodes,” he said.
The structure itself amplifies danger. Private credit channels sit outside conventional regulatory perimeters. Complex funding arrangements spread risk across hedge funds, non-bank intermediaries and leveraged players. Core bond markets already show fragility from elevated valuations and investor complacency. Add a sovereign-financial stability nexus created by record public debt and heavy participation from leveraged hedge funds, and the stage is set for sharper swings in government bond prices that tighten conditions across the economy.
Frank Smets, acting head of the BIS monetary and economic department, captured the dynamic. “The new fiscal-financial stability nexus may mean more frequent and sharper drops in sovereign bond values… could rapidly tighten financial conditions.”
History offers uncomfortable parallels. The dot-com boom of the late 1990s ended in a painful bust when investment in internet infrastructure outran actual demand and profitability. Central bankers see similar signs now. Intense competition, supply bottlenecks in chips and energy, and uncertainty over how quickly AI will deliver broad productivity gains all point toward potential overinvestment. Expectations of transformative returns have already lifted stock prices and corporate confidence. A reversal could hit borrowers throughout the supply chain who counted on continued spending.
But the warning does not come in isolation. The International Monetary Fund flagged related concerns earlier this year. An abrupt drop in AI-related stocks could spill over and damage household wealth globally, the IMF noted, given the geographic concentration of the boom, rising use of debt financing and growing foreign investor holdings of U.S. technology shares. The Bank of England and other supervisors have voiced parallel worries about stretched valuations in AI-focused companies and the trillions in debt expected to fund sector growth over the next several years.
Even within financial institutions the risks compound. Earlier analysis from the European Central Bank highlighted how heavy reliance on a small number of AI providers could create single points of failure, amplify operational weaknesses and distort market signals if models produce similar biases across participants. Herding behavior, correlated trading and asset bubbles become more likely. The Financial Stability Board, in its June meeting, discussed exactly these emerging vulnerabilities alongside sound practices for responsible AI adoption in banks.
So far markets have shrugged off the cautions. Technology stocks continue to drive indexes higher on optimism about future profits. Yet the BIS report strikes a different tone from past assessments. It treats the AI surge not as abstract technological progress but as a concrete macro-financial event with measurable transmission channels to credit, liquidity and confidence. Hernandez de Cos described the message as “urgent.” Policymakers should prioritize price stability, bring down debt, strengthen oversight of non-bank activity and pursue structural reforms that improve productivity without adding fresh fragilities.
The irony sits plain. AI itself could help detect and manage some risks through better data analysis and stress testing. At the same time it accelerates the very capital spending that creates those risks. Central bankers do not prescribe halting investment. They urge awareness that current exuberance may not prove sustainable and that the unwind, when it arrives, could prove abrupt.
Recent coverage reinforces the point. A Fortune analysis from earlier this month noted that 2026 increasingly resembles 1999 in sentiment, with earnings expectations detached from realistic outcomes. Goldman Sachs equity research head James Covello has questioned whether the productivity payoff will arrive fast enough to justify current valuations. Senator Elizabeth Warren has raised alarms about circular spending arrangements, private credit exposure and the potential for damage larger than 2008 if the sector falters.
Investors face a difficult calculus. Continued spending may deliver genuine breakthroughs in efficiency and new products. Or it may prove another case of capital misallocation on a grand scale. The difference will show up first in corporate balance sheets, then in credit markets, and finally, if unchecked, in broader economic activity.
Central bankers rarely issue such stark alerts without cause. Their message this time carries extra weight because it comes not from one institution but as a coordinated view from the body that coordinates them all. The AI boom has delivered impressive stock gains and breathless headlines. Whether it also delivers lasting economic value without a painful correction remains the open, and increasingly uncomfortable, question.
WebProNews is an iEntry Publication