The AI economy has raised $110 billion: real revenue and hidden growth drivers
The scale of artificial intelligence is no longer an abstract concept. Over the past 12 months, the real revenue of the AI industry, after eliminating double counting, reached $110 billion. Meanwhile, the current annualized rate already stands at $175 billion. These are not just numbers—they are a clear signal that the technology has moved from the experimental stage to a phase of sustainable commercial application.
A key nuance of the methodology: each dollar is counted only once. For example, $1 spent on Claude is recorded once, even if part of that amount goes to Amazon or another infrastructure provider. The metric is measured by end-customer spending, not by revenue along the entire supply chain. The calculation excludes China, internal AI economies, advertising effects, consulting, and system integration. This makes the figure as close as possible to actual consumption.
Growth Rate: Faster Than Mobile Technology and the Internet
The AI industry is growing roughly three times faster than the adoption of mobile technology or the internet in their respective eras. The pace of revenue generation has accelerated sharply: each new $1 billion in revenue now appears in less than two days, whereas in 2023 it took 180 days. This is unprecedented momentum, even by tech sector standards.
Enterprise AI has moved beyond pilot projects, but deep, company-wide implementation is still in its early stages. Mentions of AI on earnings calls have reached 31% of tracked companies in the S&P 500 index. However, only 20% of them have made quantitative statements about the technology's impact on their business. This means that a concrete, measurable effect is currently confirmed by a minority of firms. The rest are merely studying or testing it.
Infrastructure Economics and Price Elasticity
Revenue from cloud giants currently roughly covers the depreciation of AI infrastructure, but the economics of GPUs heavily depend on the assumption of a six-year lifespan. Meanwhile, other AI infrastructure is modeled over 14 years. This creates a certain imbalance in valuations that investors should take into account.
Particular attention should be paid to the conclusion about token prices. A decrease in cost does not automatically reduce revenue: every 10% drop in token price leads to a 12–18% increase in its consumption. This means that demand for AI appears elastic—cheaper prices expand usage faster than the cost declines. This effect could become a key driver for further scaling.
The main constraints on growth are identified as electricity availability and the cost of data centers. These factors will limit the development of the AI economy in the future. The team worked on the calculations for several months, underscoring the depth of the analysis.
My opinion: The numbers confirm that AI has entered a mature market stage, but infrastructure constraints and uneven adoption among S&P 500 companies indicate that the real explosion is still ahead. Investors should keep an eye not only on revenue but also on energy capacity—this is the new limiting factor for the entire ecosystem.