- The Bank for International Settlements warned on June 28, 2026, that the five largest hyperscalers are on track to commit more than $1 trillion in AI capital expenditure across 2025 and 2026, outpacing their earnings and free cash flow.
- Much of that spending flows through opaque non-bank financing and “circular financing” arrangements, which the BIS says could trigger a “sharper, faster crash” than traditional banking crises if AI returns disappoint.
- The BIS also flags AI’s growing demand for electricity and semiconductors as a driver of “chipflation,” which could force central banks to tighten policy and accelerate an asset price correction.
The Bank for International Settlements is warning that the AI investment boom now resembles historical speculative manias, canal fever in the 1830s, railway mania in the 1840s, the dot-com collapse, and that the financial structures sustaining it are less visible, and potentially more fragile, than those that preceded past crashes. The BIS released its Annual Economic Report on June 28, 2026, and the central finding is blunt: the race for AI market share may have already produced overinvestment, leaving the sector exposed if commercial returns fall short of expectations.
Trillion-Dollar Capex Outpaces Cash Flow
The five largest hyperscalers are on track to commit more than $1 trillion to AI-related capital expenditure across 2025 and 2026, according to the BIS report. That pace is outrunning these firms’ earnings and free cash flow, pushing some toward heavy borrowing to sustain spending. Pablo Hernández de Cos, General Manager of the BIS, said the race for market share in AI may have led to overinvestment, making the sector vulnerable if AI under-delivers on expected returns.
The BIS draws the comparison to prior technology booms deliberately. Canal infrastructure, railways and the early internet all attracted capital well beyond what commercial returns could justify. Each ended in investment reversals and economy-wide recessions. The report positions AI spending within that lineage, a genuine technological advance accompanied by financing that has run ahead of demonstrated returns.
Opaque Financing and Non-Bank Risks
The BIS’s deeper concern is structural. A significant share of AI funding is flowing through hedge funds and private credit vehicles rather than regulated banking channels, which limits visibility into where risk is actually sitting. Zhang Tao, the BIS chief representative for Asia and the Pacific, cautioned that this reliance on non-bank finance could produce a “sharper, faster crash” than traditional banking crises, given how interconnected these financing channels are.
The report specifically calls out “circular financing” arrangements: chipmakers and cloud giants take equity stakes in AI labs, which then commit to buying chips and computing capacity from those same investors. The circular structure means assets can be pledged multiple times, and the true financial health of the sector becomes difficult to assess from the outside. If stress hits one part of that circle, the report suggests it could spread quickly into the broader economy rather than being contained within the AI sector.
From Investment Bust to Macroeconomic Turmoil
If AI returns disappoint, the BIS sees a clear transmission mechanism from market correction to macroeconomic damage. Investor disappointment triggers a pullback in financing; the capital expenditure boom becomes what the report calls a “protracted investment bust”; and the resulting feedback loops push toward economy-wide recessions. The US market’s dominance in AI valuations amplifies this risk: a sharp correction there, the BIS argues, could produce more pronounced wealth effects and a steeper consumption pullback than past market corrections have delivered.
AI’s Inflationary Pressure Points
The financial stability concerns sit alongside a separate but related inflation problem. Surging demand for advanced semiconductors, electricity and grid equipment is already straining supply chains and pushing up input costs. The BIS describes this as potential “chipflation”, a price dynamic specific to AI infrastructure that could spill into broader inflation measures, compounded by existing geopolitical tensions and energy supply pressures.
That puts central banks in an uncomfortable position. Tightening monetary policy to contain AI-driven inflation could itself precipitate a sharp correction in AI asset prices, after a prolonged period of risk-taking. The BIS framing here is not speculative, it is a described mechanism, with the conditions for it already partially in place. For more coverage of AI policy and regulation, visit our AI Policy & Regulation section.
