The AI Boom May Be Sowing Seeds of Another Financial Crisis

AI infrastructure investment could exceed $10 trillion if current plans proceed, funded through corporate cash, public and private borrowing, leases, and guarantees. The scale and structure of the boom create risks of overinvestment: at least seven or eight companies are competing to become major AI providers, but economies of scale and network effects may leave only two or three winners. It is also uncertain whether the industry can earn enough to justify its spending. Rapid depreciation of data centers means the investment may require more than $3 trillion in annual revenue to produce an adequate return. If the boom collapses, equities could fall sharply and lenders could face losses.

A crisis on the scale of 2008 appears less likely because AI-related risks are already widely discussed and much of the financing comes from long-term investors, reducing the danger of forced sales caused by short-term funding withdrawals. But hidden and complex obligations - especially leases, guarantees, and unclear counterparty exposures - could make losses difficult to trace. Some guarantees also create “wrong-way” risk: they may be triggered just as the guarantor’s own business weakens. The proposed safeguards are to give securities regulators an explicit financial-stability mandate, require clearer disclosure of who owes what and under which contingencies, and review accounting rules so they capture actual and potential liabilities. Better information could help investors price risks and curb excessive borrowing before trouble spreads.