AI Revenue Crosses Key Threshold, Covering Data Center Costs

Artificial intelligence revenue has reached a critical milestone that suggests the industry's massive infrastructure spending may be economically viable, according to a new report from research firm Exponential View.

Global AI sales outside China totaled $25 billion in the first quarter of 2026, surpassing the estimated $21 billion in depreciation costs tied to data centers and chip investments. This marks the second consecutive quarter that AI revenue has exceeded these capital expenses, Bloomberg first reported.

Why it matters

The findings address a fundamental question that has shadowed the AI boom: whether customer demand can justify the unprecedented capital expenditures flowing into the sector. Major U.S. technology companies including Meta, Alphabet, Microsoft, and Amazon plan to spend up to $725 billion on capital expenditures in 2026, with much of that directed toward AI infrastructure. This represents one of the largest corporate spending sprees in history, and evidence that the economics can work—even marginally—provides crucial validation for investors and executives alike.

Thin margins leave little room for error

While the revenue milestone is significant, the report emphasizes that margins remain extremely tight. Depreciation charges still consume more than two-thirds of AI revenue, leaving limited cushion to cover operational expenses such as electricity, labor, and financing costs.

"For now, the economics are holding," the Exponential View report states. "But the margin for error is narrow," particularly as more financing risk shifts into capital markets through leases, debt, and equity arrangements among newer cloud providers.

Azeem Azhar, founder of Exponential View and an investor in numerous startups, told Bloomberg that the current economics are appropriate for this stage of capital-intensive investment. "It just about clears the depreciation hurdle, and roughly speaking, it's improving over time," Azhar said. He noted that dramatically exceeding the hurdle at this early stage would suggest companies had underinvested.

Demand side comes into focus

Much of the AI investment cycle has been tracked through supply-side indicators—public semiconductor companies like Nvidia and hyperscale cloud providers have disclosed their infrastructure spending. Quantifying actual customer demand has proven more difficult because major AI laboratories including OpenAI and Anthropic remain private companies.

The Exponential View analysis provides a rare window into whether end-user spending is keeping pace with the buildout. The fact that revenue has exceeded depreciation costs for two quarters suggests genuine commercial traction, though the narrow margins indicate the industry has not yet achieved the kind of profitability that would make the investment cycle self-sustaining without continued capital infusions.

The findings were published Thursday by Exponential View and detailed by Bloomberg News.