Debt-driven financing has come to underpin much of the artificial intelligence expansion, the International Monetary Fund has noted, adding that an abrupt unwinding of those positions carries a projected drag on worldwide output of 0.4% through to 2028.

The IMF's latest Global Financial Stability Report found that markets have held up, buoyed by earnings tied to artificial intelligence, though debt levels and financial vulnerabilities across the sector are climbing. Bond markets are registering the scale of demand from a small group of dominant technology platforms: roughly ten cents of every dollar raised through investment-grade debt globally now flows to that cluster of companies, up from around one cent per dollar the prior year.

Debt levels and appetite for speculative positions have been climbing throughout financial markets. Borrowings tied to hedge fund stock holdings, alongside margin-based lending facilities, have grown to a combined level representing 15 percent of the total economic output across developed nations. Single-stock products have grown to account for 28% of all leveraged ETF assets, nearly doubling their previous share of 16%.

Among the vulnerabilities the document outlined was a pattern in which established AI corporations take ownership stakes in smaller businesses that subsequently become customers of those same corporations. Under the arrangement, each side provides financial backing to the other, with pending contract commitments offered as the stated justification for that support.

Across the 14 publicly reported transactions logged from 2023 onward, equity stakes totalling $300 billion and debt arrangements reaching $650 billion have been structured around pledges for $1.4 trillion in anticipated revenue. The IMF noted that investment valuations face downward pressure if financially weaker firms cannot sustain operations, with projected revenue streams potentially falling short of earlier forecasts.

Modelling by analysts points to a potential contraction in worldwide investment of around 1.3% through to 2028 should the current wave of AI-driven growth lose momentum, with developed economies across Asia identified as particularly vulnerable to that outcome.

The IMF put forward a call for tighter oversight of margin lending and leveraged ETFs — products whose price movements are amplified in both directions — positioning stricter supervisory standards as a tool for keeping those risk exposures in check. Among its proposals were tighter capital buffers and guardrails on concentrated exposure to parts of the AI sector where vulnerabilities had been identified.

According to the Fund's assessment, banking institutions worldwide would retain adequate capital buffers even if an AI-driven market downturn coincided with a notable uptick in bond yields. Elevated risks to financial stability, the notice added, had yet to recede.

Analysts note that the tools available to policymakers — both on the spending side and through interest-rate levers — have narrowed in ways that leave less buffer for absorbing a large economic disruption than was typically the case over much of the preceding two decades.