The artificial intelligence boom has transformed data centers from a specialist infrastructure business into a central pillar of the global technology economy, but the financial returns may not keep pace with the capital flowing into the sector. Gartner warns that data center owners face a substantial gap between the revenue their investments must generate and the income that enterprise and consumer AI markets can realistically deliver by the end of the decade. Its analysis puts the cumulative revenue required to justify data center investments at $8.5 trillion through 2030, against a forecast maximum of $3.6 trillion in enterprise and consumer AI revenue attributable to that infrastructure. The resulting $4.9 trillion shortfall raises a fundamental question for investors, operators and technology companies: who will ultimately pay for the computing capacity now under construction?

Gartner’s warning arrives as data centers command an expanding share of construction investment and corporate capital allocation. The Building Owners and Managers Association’s 2026 market study examines the economic impact of commercial real estate in the United States, although the publicly accessible study page does not independently establish the specific share of private office construction attributable to data centers in 2025. That shift illustrates how rapidly digital infrastructure has changed the physical investment landscape, with AI adding another layer of urgency to decisions about land, power, cooling and computing equipment. However, the scale of construction does not establish that the resulting capacity will generate adequate returns throughout its useful life. Operators must convert infrastructure into paying workloads, while customers must find sufficient economic value in AI applications to sustain spending. The distinction matters because a market can expand operationally while delivering disappointing financial results to the companies financing its foundations.

Gartner identifies mounting AI infrastructure impairment risks

Gartner estimates that more than $3.5 trillion of AI infrastructure spending could face impairment by the end of 2030 if providers continue spending at their current pace. An impairment occurs when an asset’s carrying value exceeds the amount recoverable under the applicable accounting rules, potentially requiring the company to recognize a write-down. Gartner also warns that more than $4 trillion in asset value could face write-down risk through 2032, extending the potential consequences beyond the immediate investment cycle. These figures describe exposure rather than a certainty that companies will lose the entire amount, and their realization will depend on demand, pricing, utilization, asset lifetimes and the financial performance of individual providers.

William Sommer, senior director analyst at Gartner, said: “In 2026, AI providers are expected to spend nearly $1 trillion on AI infrastructure, and are forecast to invest an additional $6 trillion between 2027 and 2030. However, our research indicates that enterprise and consumer AI revenue will be insufficient to support profitable returns on all planned AI capital expenditures during this period.” The forecast places the investment cycle under pressure because spending commitments extend well beyond the near-term revenue available to support them. Gartner’s analysis, as reported by Fortune, projects cumulative global AI capital spending of $7.5 trillion between 2025 and 2030, with annual spending approaching $2 trillion by 2030. Such commitments create a difficult balancing act for boards and investors: capacity must arrive ahead of demand to support growth, but excess capacity can weaken pricing and extend the period required to recover investment.

The revenue model leaves infrastructure owners exposed

Gartner’s $8.5 trillion revenue requirement reflects more than the cost of constructing facilities and purchasing computing equipment. Its calculation assumes data center owners must cover announced capital spending and financing costs while earning a 12% return on invested capital, the hurdle rate used in the analysis. Gartner does not expect the industry to reach that return threshold before the end of 2032 at the earliest. The distinction between revenue and return is important: generating substantial sales does not guarantee that a capital-intensive business will compensate investors for its financing costs, risks and long investment horizon. Facilities may attract customers and still fail to deliver the returns needed to justify their original valuations. For infrastructure leaders, therefore, occupancy and contracted capacity must sit alongside a broader assessment of asset productivity, customer creditworthiness and the economics of each deployment.

The revenue distribution across the AI ecosystem creates another complication. Fortune’s reporting on Gartner’s analysis puts total revenue from AI software, services and consumer products at $4.4 trillion by 2030 and estimates that no more than 30% will be booked against the underlying capital investments. These figures require attribution to that reporting unless the underlying Gartner research confirms the precise calculation. Gartner’s October 14 announcement separately forecasts no more than $3.6 trillion in enterprise and consumer AI revenue attributable to data centers, highlighting the distinction between overall AI market revenue and the income available to support infrastructure investments. Revenue can accrue to companies selling applications, models and services without providing an equivalent return to the organizations financing the underlying computing capacity. Investors must therefore distinguish the total revenue generated across the AI economy from the share that supports infrastructure ownership and capital recovery.

Neocloud providers face a tougher test of financial resilience

The financial pressure will not affect every infrastructure provider equally. Large cloud companies such as Amazon, Microsoft and Alphabet can draw on diversified businesses and established cash flow sources to support AI investments, even when returns take longer to materialize. Their broader operations may provide financial flexibility, although diversification does not eliminate the risk of overinvestment or disappointing returns. Specialized cloud infrastructure providers, commonly known as neoclouds, face a different challenge because their business models often depend more directly on demand for AI-optimized computing capacity. These companies must secure customers, maintain access to financing and manage expensive hardware through a market in which prices and technology requirements can change quickly. Their exposure makes the relationship between utilization, contractual commitments and debt obligations particularly important.

Gartner’s press release states: “Neoclouds must soon earn returns on AI-optimized IaaS and associated platforms to sustain operations. By 2030, Gartner predicts that cost pressure and a lack of a differentiating business model will force over 80% of neocloud providers to cease independent operations.​” Gartner’s forecast concerns the loss of independent operations, not necessarily insolvency or the closure of every affected business. A provider could cease operating independently through a merger, acquisition or another corporate restructuring while its services continue. The forecast does not establish which companies will follow these paths or how many will ultimately close. For customers, the possibility of industry consolidation raises questions about supplier choice, contractual continuity and bargaining power, although the effects will depend on how the market develops.

AI investment risks extend beyond infrastructure owners

Gartner also warns that product leaders cannot treat potential AI capital write-downs as a problem confined to infrastructure companies. Gartner’s broader research on AI market consolidation identifies provider concentration, pricing volatility, model discontinuations, service rationalization and sovereign AI requirements as risks that can affect product economics, development plans and customer commitments. These risks extend beyond companies that own AI infrastructure because changes in suppliers, service availability and commercial terms can affect the economics of products built on external AI platforms. “Product leaders should therefore design AI-dependent products assuming that today’s AI supply landscape will become materially more concentrated through the end of the decade.”

The broader implication is that the AI investment cycle will be judged by economic value creation, not construction volume alone. Gartner’s projections do not establish that AI demand will collapse or that every infrastructure investment will fail, but they highlight the distance between the capital committed and the revenue available to justify it. Investors will need to scrutinize the quality of demand, the durability of customer contracts and the ability of operators to adapt as hardware and service markets evolve. Infrastructure leaders, meanwhile, must connect expansion plans to credible pathways for utilization, pricing power and returns across the asset lifecycle. The next phase of AI growth may depend less on who can build the most capacity and more on who can turn that capacity into a defensible business. For the sector, the $4.9 trillion gap is a warning that technological momentum and financial sustainability are not the same thing.