Concerns are mounting over the sustainability of the AI investment boom as the market for AI providers is becoming increasingly competitive and the ability of frontier AI developers to charge premium prices is weakening, says Fitch Ratings.
In a report Tuesday, the rating agency said the gap between the capabilities of the leading-edge models being developed by the frontier AI labs and much cheaper, open models is narrowing — a development that has corporate buyers looking at cheaper alternatives to the premium AI providers, which undermines their pricing power.
This decline in pricing power for frontier firms is increasingly “at odds” with accelerating AI investment, the report said, adding that consensus estimates point to annual spending by so-called “hyper-scalers” topping US$1 trillion in the 2027–2029 period.
That level of capital expenditure is “substantially above” the level that Fitch views as “economically justifiable,” the report said.
“Frontier labs continue to generate significant operating losses, and their paths to profitability are increasingly uncertain as pricing power erodes ahead of planned IPOs,” it noted.
Against that backdrop, Fitch said it believes that “a period of reassessment is increasingly likely in the next one to three years.”
From a credit perspective, the implications of a potential shift in AI investment trends “vary significantly” across the sector, the report added.
“Compute providers face the most direct counterparty and re-contracting risk, though capex flexibility provides a meaningful offset,” it said.
Memory chipmakers would “face significant volume risk from an AI capex pullback,” it said — whereas data centres and structured finance transactions are less exposed to aggregate demand for computing power.