The canal madness of the 1830s. The British railway bubble of the 1840s. The dot-com crash of 2000. Both began with a real technological breakthrough that attracted more capital than the commercial returns could ultimately justify. Each ended in recession.
The Bank for International Settlements – the Basel-based institution that coordinates the world’s central banks and serves as the global financial system’s most authoritative watchdog – sees the $1 trillion AI investment boom along the same lines. And the comparison is not subtle.
“The scale and speed of the current AI investment boom, accompanied by expectations of large productivity gains, bear similarities to these precedents,” the BIS writes in its 2026 Annual Economic Report published on Sunday. “These episodes ultimately ended with a reversal in investment that triggered an economy-wide recession.”
Table of Contents
A bet that already exceeds the balance
The top five hyperscalers are on track to spend more than $1 trillion combined on AI-related investments in 2025 and 2026 – a sum that the BIS says is already outpacing their profits and free cash flow, forcing some to issue debt to fill the gap.
The BIS’s concern is not that AI is a scam. The technology is real, and the report acknowledges that task-level studies consistently show productivity gains of 20% to 50% through time savings. However, there is concern that all major hyperscalers are making the same big bet at the same time, driven by the assumption that only a handful of players will ultimately dominate the market. This logic, warns the BIS, is a recipe for collective over-commitment.
“Intense competition increases the risk that companies will allocate too many resources to investment projects with still uncertain returns,” the report says, “leaving all companies vulnerable to disappointment in AI returns.”
Using competition theory models, BIS economists find that the net economic surplus for the sector as a whole – total payouts less investment costs – is falling and could become negative in adverse scenarios if competitive pressures drive up capital spending. A disappointment in returns, the report warns, “could trigger a sudden pullback in funding, turning the investment boom into a protracted investment downturn.”
The hidden wiring underneath
What makes an AI bust particularly dangerous, the BIS argues, is not just the size of the spending, but also the way it is financed.
Hyperscalers, chipmakers and AI labs are linked by what the report calls “a complex web of private agreements.” The most well-known is circular financing: hyperscalers participate in AI laboratories, which in turn commit to buying chips or computing power from the same hyperscalers over several years. Data centers are outsourced to third-party providers who lease the facilities back under long-term contracts with embedded exit clauses.
“The terms of such deals are typically poorly disclosed,” writes the BIS, “with the risk that the same asset will be pledged multiple times.”
If the hyperscalers slow or stop their aggressive investment deployment, the entire supply chain – infrastructure contractors, chipmakers, AI labs and the private lenders behind them – would simultaneously face a loss of revenue. The engineering and construction companies at the bottom of this chain are particularly vulnerable because they have “comparatively weak” balance sheets and little cushion against a sudden turnaround.
BIS Asia-Pacific representative Zhang Tao told the newspaper South China Morning Post that a correction could subside “much more quickly than in previous banking crises” – precisely because much of the financing flows through hedge funds and private credit vehicles, which are subject to less regulatory oversight than traditional banks.
Such warnings are commonplace on Wall Street. For example, Torsten Slok, chief economist at Apollo Global Management, argued in mid-May that AI was “penetrating every corner of the financial markets” and that a stock market phenomenon was mutating into a capital market-wide transformation. AI now accounts for nearly half of all investment-grade debt issuance, 87% of venture capital funding, and a growing share of high-yield debt.
The problem of the wealth effect
The financial consequences would not be contained in Silicon Valley or in the balance sheets of hyperscalers. U.S. stocks now make up about 64% of the MSCI Global Index, and household equity exposure as a share of income has more than doubled since 2010.
The BIS warns that a major repricing of AI-related stocks “could have stronger wealth effects and a sharper decline in consumption than in the past.” And given the U.S. market’s global footprint, wealth destruction would spread internationally.
Direct lending funds – already a $1 trillion-plus ecosystem – have quadrupled their lending to the AI and IT sectors in the last five years and now make up about 15% of their portfolios. Signs of stress are already visible: some retail-focused direct lending funds have faced increasing redemption requests, forcing the liquidation of assets.
“A major shock,” writes the BIS, “whether from a renewed rise in inflation or a strong AI-driven repricing, could trigger a broader credit crunch.”
The Hormuz Complication
AI risk does not exist in a vacuum. The report’s opening chapter documents a second major shock that occurred in early 2026: the closure of the Strait of Hormuz following the start of the Iran conflict in late February, eliminating more than 10 million barrels per day of crude oil from global supplies – a greater disruption than the 1973 oil embargo or the 1979 Iranian Revolution.
Oil prices rose 67% in two weeks to an intraday high of $120 a barrel. Fertilizer and plastic prices each rose by 50%. Overall global inflation has risen by half a percentage point since the conflict began.
The energy shock and AI risk interact in unpleasant ways. Financial markets remain buoyant – equity valuations are high, credit spreads are tight – and expect the disruption in Hormuz to be temporary and the AI boom to continue. However, if inflation proves more stubborn than expected and central banks are forced to raise interest rates, the same tightening needed to curb energy-related inflation could burst the AI-funded debt bubble.
“The current tension between excessive risk-taking and increased macroeconomic risks,” writes the BIS, “could be resolved abruptly.”
The BIS is afraid to directly describe the AI boom as a bubble. Her prescription is “robustness” – a word she uses carefully and repeatedly to describe what she wants policymakers to build beyond the fragile “resilience” the global economy has demonstrated so far.
That means central banks remain vigilant about inflation even when things are politically uncomfortable, governments restore fiscal space rather than stimulus, and regulators extend prudential standards to the nonbank financial institutions now at the heart of AI finance.
https://fortune.com/2026/06/29/bis-central-bank-warning-hyperscaler-data-center-1-trillion-gamble-recession/
