For BlackRock's Fink, AI's biggest risk is a capital shortage, not a bubble

BlackRock CEO Larry Fink has pinpointed a different threat in the AI surge: a shortage of capital. He warns that government budget deficits and heavy technology-sector investment are chasing the same pool of funds, and the result could be higher borrowing costs for homebuyers and companies.

In a Wall Street Journal op-ed, Fink described the present contest for funding as the most acute he has witnessed in his 50 years in finance.

Deficits and AI Compete for the Same Capital

In Fink's telling, budget deficits in developed countries are now rivaling strong private demand, a big portion of which comes from technology companies. Electricity, housing, manufacturing capacity and other infrastructure will require trillions of additional dollars. He pointed out that US federal debt on its own has climbed to a record $40 trillion.

“When all that investment demand collides with a limited supply of capital, borrowing becomes more expensive for everyone, from families buying homes to businesses looking to invest,” he wrote.

The effect is already visible in borrowing conditions. The yield on the 10-year Treasury, a reference point for home loans and corporate credit, has advanced by roughly 1.2 percentage points over the current year. Earlier this month it reached 5.34%, a level not seen since 2002.

Hyperscalers, in Fink's assessment, have an advantage because their resources are the largest. He argued that their expansion ought to be funded without siphoning capital away from the broader economy.

Fink's Plan for Broadening the Capital Pool

Simply reallocating the same dollars from one asset class to another will not finance the need, in Fink's view. He instead proposed 3 steps.

• Expand AI financing: Fink said current AI infrastructure funding is concentrated among a small set of companies and lenders. That arrangement is not large enough for the scale of the need. A former mortgage trader, he proposed securitizing computing infrastructure, the way home loans once were, so that a broader group of investors could supply the money.
• Develop the power behind AI: He cited the expansion of electricity generation in China and argued that abundant, low-cost power will count for as much as the technology. He pressed for swifter permitting and more training programs for electricians, welders and technicians.
• Attract more investors: This measure mattered most to Fink. He noted that roughly 4 in 10 Americans own no stock. Additional long-term investors would bring new capital and give people a direct interest in growth.

Bubble Predictions Suggest a Steeper Fall

Fink said he remains bullish on the technology itself and on America's capacity to finance it. He acknowledged that a number of companies might collapse, just as railroad and internet firms did in earlier eras. Even so, he said it is not difficult to picture a handful being worth $10 trillion each.

Other recent warnings have centered on the AI trade itself. Ray Dalio, founder of Bridgewater, described AI as a textbook bubble near its breaking point, pointing to debt and rising interest rates.

Joachim Klement, a strategist at Panmure Liberum, predicts the bubble will burst in 2027 or 2028. The unwinding, he cautioned, could trigger the most severe collapse since the global financial crisis.

Klement also stated that hyperscalers have mostly exhausted their free cash flow as the cost of debt rises. Fink, in turn, depicts those same businesses as possessing the largest financial reserves.

Fink's scenario requires fresh sources of financing, considerably quicker permitting and added infrastructure. If those conditions are met, he wrote, growth could be fast enough to keep the debt load manageable.