Morning Edition · Sunday, June 28, 2026

AI Revenue Reaches $110 Billion and, for a Second Quarter, Outpaces the Cost of the Buildout

Exponential View's first State of the AI Economy report offers the most rigorous deduplicated revenue figure to date, and a tentative answer to whether the capital spending is justified.

Save

AI Revenue Reaches $110 Billion and, for a Second Quarter, Outpaces the Cost of the Buildout

For three years, the central question for the AI sector has been whether revenue could ever justify the capital invested in chips and data centers. A new State of the AI Economy report from Azeem Azhar's Exponential View, summarized by AI Post, gives the first deduplicated answer: roughly $110 billion in real end-customer revenue over the trailing twelve months, with an annualized run rate now above $175 billion.

The figure matters less for its size than for how it was built. It draws on spending data from more than 1,000 companies and removes the supply-chain double counting that inflates most market estimates, in which the same dollar is booked separately by a chip vendor, a cloud provider, and a model lab. The result is a bottom-up measure of what customers actually pay. On that basis, the report finds AI revenue growing roughly three times faster than the mobile and internet adoption cycles did at comparable stages, according to Crypto Briefing's account of the findings.

The more consequential claim is that revenue has crossed above the cost of the buildout. In the first quarter of 2026, AI sales outside China reached approximately $25 billion, exceeding an estimated $21 billion in quarterly depreciation on data-center and chip infrastructure, the second consecutive quarter revenue has cleared that bar. Depreciation is the accounting measure of how fast the physical buildout is being used up, so revenue above it is the first arithmetic sign that the industry is generating more value than it consumes in hardware each quarter.

Skepticism is warranted for two reasons. The depreciation figure depends on assumptions about useful life that vendors and skeptics dispute, and shortening the assumed life of a graphics processing unit (GPU) would remove the crossover. The estimate also excludes China, where pricing and accounting are opaque, so the global picture is incomplete. What is verified is the revenue methodology and the direction of the trend. What remains asserted is that the trend is durable rather than the result of a one-time surge in enterprise pilots.

Veracity: Plausible

71/100

If true, who benefits: Hyperscalers, chip vendors and model labs raising capital, since a "revenue now exceeds the cost of the buildout" narrative justifies continued capital spending and supports AI-sector valuations.
The nuance: The $110 billion revenue figure and methodology are corroborated, but the crossover above depreciation rests on disputed GPU useful-life assumptions and excludes China, so "self-funding" is an accounting interpretation, not a settled fact.

An open-source-intelligence read of how likely this story is true with its real nuance, not a judgment of any outlet. It assesses the claim, weighing independent and adversarial reporting. How we label confidence.

What this means

For investors weighing the AI capital-expenditure cycle, a revenue line that grows faster than depreciation is the measure that separates a productive buildout from a bubble. It does not validate any individual company's spending, but it shifts the burden of proof. The question moves from whether demand will ever materialize to which companies capture the margin. A reversal in the depreciation-coverage ratio would be the earliest warning that the cycle has grown faster than its customers.

What to watch

Whether the next quarterly reading keeps AI revenue above data-center depreciation, since a single quarter back below it would suggest the crossover was a matter of timing rather than a trend.
How hyperscalers set GPU depreciation schedules in upcoming filings, because lengthening useful-life assumptions improves the same ratio this report relies on.

Observations to monitor, not financial advice.

3 sources

Synthesized from: Polylog editors · Exponential View · Crypto Briefing

Part of a tracked trend

AI Revenue Versus Buildout Economics

As AI capital spending compounds, deduplicated end-customer revenue and its margin over depreciation become the decisive gauge of whether the compute buildout is self-funding or speculative, and each new data point moves capital toward or away from the trade.