The inaugural analysis of over 1,000 companies suggests the generative AI sector is finally generating enough revenue to justify its enormous infrastructure bets.
The generative AI economy pulled in more than $110 billion in sales over the past year. That is not a projection, a TAM fantasy, or a McKinsey slide deck estimate. It is the central finding of Exponential View’s inaugural State of the AI Economy report, published June 25, 2026.
Here’s the thing: the sector’s annualized revenue run rate now exceeds $175 billion.
What the numbers actually say #
The report, led by Exponential View founder Azeem Azhar, is built on AI spending data from over 1,000 companies. The figures were adjusted to avoid double counting, a detail that matters more than it sounds. In AI revenue discussions, the same dollar often gets claimed by the cloud provider, the model maker, and the enterprise software layer simultaneously. Stripping that out gives you a cleaner picture.
In Q1 2026, global AI sales outside of China hit approximately $25 billion. That single-quarter figure surpasses the estimated $21 billion in quarterly depreciation costs on data centers and chip infrastructure that companies have been racking up. In English: the money coming in is now larger than the cost of the buildings and hardware wearing out.
Why this report exists now #
Azeem Azhar has been publishing Exponential View as a newsletter and research platform for years, tracking the intersection of technology and economics. This report represents the organization’s first attempt to produce a comprehensive, ground-up estimate of the generative AI economy’s actual revenue, not forecasted revenue, not addressable market, not vibes.
What this means for investors #
The most important number in this report is not $110 billion or $175 billion. It is the $25 billion versus $21 billion quarterly comparison.
Revenue exceeding depreciation at the infrastructure level is the kind of inflection point that changes how institutional capital thinks about a sector. That said, depreciation is not the same as total cost. It does not account for operating expenses, energy costs, talent acquisition, or the billions being spent on model training that may or may not produce commercially viable products. The infrastructure math working does not mean the entire value chain is profitable. It means one critical piece of the puzzle has clicked into place.
The report’s China exclusion is notable. Global AI sales of $25 billion per quarter outside of China implies that Chinese AI revenue sits on top of that figure, unmeasured here.
One risk worth monitoring: the $175 billion run rate is an annualization of current trends. Run rates are famously flattering. They assume the most recent quarter’s momentum holds steady, which in a fast-moving sector driven by enterprise adoption cycles and model capability leaps, is far from guaranteed.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our