It is reported that Gartner estimates that by 2030, over $3.5 trillion in AI infrastructure spending may face the risk of impairment. The institution is concerned that the expansion of investments has exceeded the revenue growth required to support its cost and return requirements.
This calculation takes into account financing costs and capital returns as part of the threshold requirements. A previously released Global X study predicted that the revenue related to AI could exceed infrastructure depreciation around 2027.
According to the analysis of Gartner, owners of AI data centers will need to generate a cumulative revenue of 8.5 trillion US dollars by 2030 in order to cover capital expenditures and financing costs and achieve a return on invested capital of 12%. The institution estimates that during the same period, the revenue attributable to the AI business of enterprises and consumers will not exceed 3.6 trillion US dollars, which is less than half of the aforementioned requirement.
Gartner is also concerned that the rapid depreciation of computing assets may render them obsolete before they are able to generate sufficient returns.
Global X adopted a different measurement approach. Its model estimates the AI related revenues of 15 listed and unlisted AI platforms and service providers, covering areas such as model development, cloud computing, digital advertising, and subscription services; for asset depreciation, a straight-line depreciation method over a five-year useful life is used, which is only included in the capital cost.
The institution expects that the AI revenue of these companies may increase from approximately $146 billion in 2025 to $317 billion in 2026, $518 billion in 2027, and approach $1.5 trillion by 2030. Comparing annual revenue with annual depreciation, Gartner lists the cumulative revenue targets up to 2030.
The forecast from Global X also indicates that by 2030, the annual depreciation cost for each $1 of AI related assets may correspond to $2.36 in AI revenue.
There are still cost boundaries to this set of forecasts. Global X clearly reminds us that depreciation is just a part of the costs; the 12% return threshold mentioned in Gartner also includes financing costs. The two calculations measure separately the ability of income to cover depreciation and to meet capital return requirements; exceeding depreciation cannot be directly equated with achieving the latter.
Gartner Senior Research Director Will Somer ( Will Sommer ) pointed out that only a portion of each dollar spent on computing power goes to the owners of that computing power. The institution believes that service providers that specialize in selling AI computing capacity are more directly affected by the demand for computing power than large cloud vendors with diversified businesses.
Gartner It is estimated that due to cost pressures and limited ability to differentiate, more than 80% of new cloud service providers will no longer be able to operate independently by 2030. Somer stated that these companies are attempting to develop software and services with higher profit margins, but few may be successful.
CoreWeave ( CRWV.O ) tumbled on Wednesday. According to reports, CEO Mike Introtto ( Mike Intrator ) stated in an employee memo that the company's backlogged orders are sufficient to support its debt obligations.
InterraT also stated that the difficulties and delays in competitors' listings and debt financing are affecting the market narrative surrounding CoreWeave. The company previously disclosed that it had $104.2 billion in backlogged orders for the second quarter, higher than $99.4 billion in the first quarter; the company said that it also secured over $25 billion in new customer commitments for the third quarter.
