An Institution That Speaks Carefully, On Purpose
The Bank for International Settlements does not issue casual warnings. It is, in its own description, the central bank for central banks, the Basel-based body that coordinates monetary policy across the world’s major economies. In its Annual Economic Report 2026, the BIS named the sustainability of the AI boom as one of four pressure points threatening the global economy, alongside returning inflation, strained public finances, and growing financial vulnerabilities. The institution chose its language with evident deliberation. The BIS stops short of calling the AI boom a bubble outright. Its prescription is for “robustness,” a word it uses carefully and repeatedly to describe what it wants policymakers to build beyond the fragile “resilience” the global economy has demonstrated so far.
That distinction between robustness and resilience is not bureaucratic hedging. It signals an institution trying to name a risk without triggering the very panic it hopes to prevent. Even with that caution, the historical comparison the BIS chose to draw was unambiguous. The report compared the current AI investment surge to canal and railway mania in the 1800s, electrification exuberance of the 1920s, and the dotcom boom of the 1990s, noting “all shared one common trait: a genuine technological breakthrough that attracted capital in excess of what commercial returns could ultimately justify.” The report continued with a warning that should give any AI investor pause: “These episodes ended with an eventual reversal in investment, inducing economy-wide recessions.”
What the BIS Is Actually Worried About
The headline framing of “AI bubble” obscures the BIS’s more precise concern. An “AI bubble” does not mean AI stops being useful. It means the money funding AI infrastructure grows faster than the cash flow that infrastructure generates, and investors keep paying anyway. That is a financing structure problem, not a verdict on whether large language models work. The scale driving this concern is substantial by any measure. The report says the five largest hyperscalers are set to spend more than $1 trillion on AI-related capital expenditure between 2025 and 2026 alone. These commitments now exceed both earnings and free cash flow, pushing some firms to issue debt to continue funding the buildout.
What makes this BIS warning distinct from generic skepticism is the specificity of the mechanism it names. The BIS warns that AI infrastructure is increasingly financed through private credit and hedge funds, channels with less oversight than banks. The institution drew a comparison to past financial crises and what made this one different in its timing. Every infrastructure boom eventually outgrows the balance sheets that started it, from railroads to telecom fiber to housing. What separates this one is the BIS naming the exact channel, non-bank private credit, that previous crises took years to identify after the damage was done. The numbers behind that channel are not trivial. Private credit funds originated over $40 billion in loans to AI-related companies in 2025, compared with about $3 billion in 2010, and AI-related direct loans now account for approximately 4% of total private credit holdings.
The Circularity Problem the BIS Flagged
Beyond debt structure, the BIS pointed to a financing pattern that should concern anyone trying to assess genuine demand for AI infrastructure. A related concern is circular financing, where chip and cloud giants take equity stakes in AI startups that then spend that same capital buying chips or compute from the investor. The risk is that these deals can inflate the appearance of demand. Revenue looks real on paper, but a chunk of it is the investor’s own money moving through one company and back into another. This circularity matters precisely because it complicates the most basic question any infrastructure investor should ask: is this demand organic, or is it manufactured by the financing structure itself? The BIS does not claim to know the answer with certainty. It claims, instead, that the opacity of these structures makes the answer unknowable from the outside, which is itself the risk.
Why the Skepticism Resonates Beyond Finance
The BIS report focuses on financing mechanics. A separate, quieter pattern inside the AI infrastructure build itself lends weight to the BIS’s underlying skepticism, without needing a single dollar figure to make the point. Operators across the industry have been reserving electrical capacity and expanding data center footprints on the assumption that more megawatts secured translates directly into more computational value produced. That assumption has not held as cleanly as planning models expected. Capacity, once deployed, does not automatically convert into productive output. Training pipelines pause. Orchestration systems queue jobs that compete for the same resources. Reserved infrastructure can sit available while researchers simultaneously describe a shortage of usable compute. Both observations can be true at once, and the discrepancy exists because productive value depends on scheduling quality and workload orchestration, not on installed capacity alone.
This pattern is instructive for the BIS’s broader thesis, even though it sits entirely outside the financing question the BIS examines. It shows that infrastructure scale and infrastructure productivity are not the same variable, and that an industry capable of confusing the two at the operational level is also an industry where outside observers, including central bankers, may reasonably struggle to verify whether capital is chasing genuine commercial return or chasing the appearance of demand. The financing risk and the operational risk are different problems. They share a common root: both depend on infrastructure converting into productive output at a pace that current evidence does not yet fully confirm.
The Counterargument the BIS Itself Acknowledges
Fairness requires noting that the BIS does not present this as a settled conclusion. The institution’s own report leaves room for AI to validate the investment thesis. The BIS itself notes that the build-out’s hunger for electricity is already pressuring prices and input costs, with potential spillovers to inflation, though it stresses, as do many economists, that AI could yet prove disinflationary if its promised productivity gains eventually arrive. That caveat matters. The BIS is not predicting failure. It is naming a structural vulnerability and urging preparation before any reversal occurs, not after. BIS general manager Pablo Hernández de Cos said the message was one of “urgency,” with policymakers urged to act before any reversal makes the eventual adjustment more painful.
Should You Believe the Central Bankers?
The honest answer is that the BIS has earned a hearing, not blind deference. Central banking institutions have been wrong before, and they have also been early and right when markets preferred not to listen. What distinguishes this warning is its specificity: a named financing channel, a documented shift from cash flow to debt, and a circular financing pattern the institution describes with unusual directness for a body known for diplomatic restraint. The BIS is not asking investors to abandon AI. It is asking regulators to extend oversight to the parts of the financing structure currently operating with the least transparency, and asking market participants to distinguish between capital that reflects validated commercial return and capital that reflects competitive momentum.
Practitioners should watch whether regulators use this as backing to push disclosure requirements on off-balance-sheet AI financing arrangements before the next cycle of capex commitments. That is a reasonable, measured ask from an institution whose job is to see systemic risk before it becomes a systemic crisis. Whether the industry listens, or treats this as one more headline in a cycle of warnings it has weathered before, will determine which side of the historical pattern the BIS just described canal mania, railway mania, electrification, dot-com — this current boom ultimately resembles.
