AI sales start to justify data-center spending boom, report says

Revenue from artificial intelligence has reached a tipping point, showing that the hundreds of billions of dollars tech companies are spending on it may be economically sustainable, according to a report from research firm Exponential View.

Global AI sales, excluding China, reached $25 billion in the first quarter of 2026, exceeding the industry’s estimated $21 billion in depreciation costs tied to investments in data centers and chips for the second consecutive quarter. While the milestone suggests that AI companies are beginning to cover the cost of their capital spending, the margins are thin. Depreciation charges still consume more than two thirds of revenue, leaving a small buffer to cover other costs such as power, labor and financing.

“For now, the economics are holding,” the report, published Thursday, states. “But the margin for error is narrow,” it adds, with more financing risk shifting into capital markets through leases, debt and equity, especially among the so-called neoclouds.

The findings speak to one of the central questions hanging over the AI boom: Whether customer demand is large enough to justify the hundreds of billions of dollars being poured into chips and data centers. The biggest US tech companies, including Meta Platforms Inc., Alphabet Inc., Microsoft Corp. and Amazon.com Inc. plan to spend as much as $725 billion this year on capital expenditures, much of it on AI infrastructure, in one of history’s largest corporate spending sprees.

“It just about clears the depreciation hurdle, and roughly speaking, it’s improving over time,” Azeem Azhar, founder of Exponential View and investor in dozens of startups, told Bloomberg News. “At this stage of an investment in any kind of capital expenditure, you wouldn’t expect to have dramatically jumped over that hurdle because if you had, you were probably leaving something on the table.”

Much of the AI boom has been measured from the supply side, through disclosures from public semiconductor companies like Nvidia Corp. and hyperscalers like Alphabet. Demand has been harder to quantify because many of the most important AI labs, including OpenAI and Anthropic, remain private.

Generative AI revenue, excluding China, reached $110 billion over the past 12 months and is scaling three times faster than any previous information technology wave including the internet, mobile applications and the cloud, according to the report.

The figures are based on a dataset that Exponential View built tracking AI spending across more than 1,000 companies. They used sources including company filings, executive statements, press reporting and cloud-provider disclosures, and then adjusted the figures to avoid double-counting between layers of the AI supply chain.

The analysis assumes a six-year depreciation life for IT equipment including graphics processing units, or GPUs, the chips used to train and run advanced AI models. Some investors argue this is optimistic given the rapid pace of chip innovation, which can render older hardware less valuable within a few years.

If GPUs lose economic value faster than assumed, companies could face higher depreciation charges, asset writedowns or earlier replacement costs. Michael Burry, the investor known for betting against the US housing market before the 2008 financial crisis, has described understated depreciation as “one of the most common frauds of the modern era.”

However, data in the report suggests older chip models are not collapsing in value. The rental price for an hour of access to Nvidia’s H100 chip remains almost 80% of its launch level. “Even into its fourth year, it is completely in demand,” Azhar said, noting it’s become more expensive over the last year, as demand for AI compute outstripped supply of Nvidia’s new Blackwell chips.

That chimes with comments from Matt Garman, the chief executive officer of Amazon Web Services, who said in February the company had not retired six-year-old Nvidia A100 servers due to continuing demand.

The report also shows more users are moving toward open-weight and Chinese AI models such as DeepSeek. Data from OpenRouter, a platform that gives developers access to multiple AI models, shows the share of tokens requested from Google, OpenAI and Anthropic models fell to 33% in June 2026 from 72% a year earlier.

Azhar said that reflects power users moving toward cheaper and faster models for simpler tasks. “You don’t always need a Nobel laureate to extract a number from your receipt to put into an expense spreadsheet,” he said.

That does not necessarily spell trouble for leading foundation-model companies, he added, but it raises the bar for charging higher prices. They will need to compete with “additional services, with more lock-in, and with all of the things that allow you to charge a premium,” he said.