Fitch Ratings warns weaker AI pricing power undermines capex sustainability

Fitch Ratings is sounding an alarm that the AI investment frenzy might be running ahead of the revenue to justify it. The credit agency’s core concern: if AI service providers can’t maintain pricing power in an increasingly crowded market, the hundreds of billions being poured into data centers and infrastructure may not pay off.

The spending spree meets economic gravity

Hyperscalers like Microsoft, Amazon, and Alphabet have been committing staggering sums to AI infrastructure, with annual capital expenditures expected to remain elevated through the late 2020s. The investments span data centers, specialized chips, cooling systems, and the sprawling physical footprint that generative AI demands.

Fitch projects that AI service revenue could land somewhere between $800 billion and $1.4 trillion by 2030. That’s a wide range, and the gap between those two numbers, roughly $600 billion, is itself larger than the GDP of most countries. Where revenue actually falls within that band depends heavily on one variable: pricing power.

Fitch has documented a capital expenditure boom driven by AI infrastructure and data center demands dating back to at least 2024. The shift from asset-light to asset-heavy business models among major tech firms introduces a kind of financial risk that Silicon Valley hasn’t had to grapple with at this scale before.

Customer concentration and counterparty risk

Beyond pricing pressure, Fitch highlights a structural vulnerability that doesn’t get enough attention: customer concentration. Companies across technology, utilities, and real estate sectors that have tied their fortunes to AI revenue are increasingly dependent on a handful of massive clients. Fitch notes that rising counterparty exposure tied to large customers like Oracle and Microsoft threatens firms reliant on AI compute services.

The utilities angle is equally important. AI data centers are extraordinarily power-hungry, and utility companies have been racing to sign long-term power purchase agreements.

An AI correction as a global credit risk

Perhaps the most striking element of Fitch’s analysis is the agency’s identification of an AI market correction as a top global credit risk, flagged in its July 2026 Global Risk Outlook report. The reasoning combines three factors: stretched valuations across AI-related equities and debt, uncertain revenue trajectories, and record levels of borrowing within the sector.

Fitch suggests this could lead to meaningful differentiation in credit ratings, with companies that demonstrate strong monetization capabilities separating from those running on ambition and borrowed money. The agency’s framework essentially divides the AI landscape into winners and losers based on a simple question: can you actually charge enough for your AI services to justify what you spent building them?