BIS warns: an AI investment bust could hit credit markets like 2008

The Bank for International Settlements (BIS), the body that coordinates the work of central banks in 63 countries, has warned in its annual report that a collapse in AI investment could shake credit markets with a force comparable to the 2008 mortgage crisis. The Basel-based institution placed AI risks alongside inflation and fiscal stress, calling them "pressure points requiring close attention."

Why the BIS is sounding the alarm

The regulator is concerned not about the technology itself but about how it is financed. BIS analysts describe a "circular financing" mechanism: AI companies borrow against future revenue that itself depends on a continuous inflow of AI investment. As long as the market believes in the technology's limitless potential, the system works. But if major investors begin to doubt the returns, the chain risks breaking.

Particularly worrying is that AI infrastructure spending, including building data centers and buying specialized chips, requires multi-year commitments. Monetizing this capacity may take longer than expected or fall short of forecasts, while the debt burden will not go away. Banks that have lent against AI assets risk being left with non-performing portfolios, and a credit squeeze would hit every sector of the economy, not just tech.

"Disappointment over returns could trigger a sharp reversal that would

ripple through the entire credit market," the BIS report warns.

Parallels with 2008

The comparison with the mortgage crisis is no accident. Back then, high demand for mortgage-backed securities created an illusion of safety shared by banks, rating agencies and regulators. When borrowers began defaulting en masse, it turned out the whole market had rested on a single assumption: that property prices could not fall.

Today the BIS sees similar logic in AI. Several structural risks add up to a familiar picture:

• AI company valuations are built on expected, not confirmed, revenue
• Lenders are eager to finance startups and infrastructure projects against future growth
• Risk is concentrated among a few key players: hyperscalers, chip suppliers and cloud providers
• The failure of one link could pull down a long chain of connected loans and guarantees
• Regulators do not yet have specialized tools to monitor AI-specific risks

The key difference from 2008: regulators today have a chance to spot the risks in advance. That is exactly why the BIS publishes warnings, to give the system time to respond.

The scale of the AI bet

Global investment in AI infrastructure over the past two years runs to hundreds of billions of dollars. The largest tech companies are competing to announce record capital expenditure on data centers, energy and chips. A significant share of this spending is financed with borrowed money or depends on revenue that has yet to materialize.

This gap between investment and actual returns, a "valley of disappointment," is the source of systemic risk. The question of "when will AI start making real money" is getting louder. The longer the answer is delayed, the greater the pressure on credit portfolios. The first signs of caution are already visible: some AI startups that raised large rounds in 2023-2024 are struggling to secure their next funding, a classic symptom of a cooling market.

What this means

The BIS warning is not a forecast of a crash but a call for preventive action. Regulators are urged to strengthen oversight of AI-related credit before the problem reaches systemic scale. For investors and businesses, it is a reminder that the AI industry's rapid growth rests on a fragile financial structure, and the risk of a "bubble" is more real than commonly assumed.