AI data center owners face a $4.9 trillion shortfall through 2030, according to Gartner. That is the gap between the $8.5 trillion in revenue they need to cover announced capital spending, financing costs and expected returns, and the $3.6 trillion in enterprise and consumer AI revenue Gartner actually forecasts for the period.

The verdict comes from Will Sommer, Senior Director Analyst at Gartner. He said: "Enterprise and consumer AI revenue will be insufficient to support profitable returns on all planned AI capital expenditures during this period." In plain terms: money in will not cover money out. Gartner did not publish a breakdown of which workloads generate the $3.6 trillion figure, so buyers should treat it as a forecast, not a measurement.

How large is the gap?

Gartner's arithmetic is blunt. Data center owners need $8.5 trillion in cumulative revenue by 2030. Sommer's forecast caps attributable AI revenue at $3.6 trillion. Put differently, projected revenue covers roughly 42 cents of every dollar the industry says it must recover. The remaining 58 cents depends on revenue streams Gartner does not currently see. To put the number in enterprise terms, $8.5 trillion over roughly six years is more than $1.4 trillion per year in required AI revenue.

The gap is not a Gartner-only claim. Bain and Company ran its own numbers and reached a similar conclusion, estimating annual spending on AI infrastructure could hit $1.5 trillion by 2031. Bain also calculated the AI services market would need to exceed $6 trillion annually to pay for that buildout. The two firms surveyed different ground with different methods. Their conclusions converge: a substantial share of AI infrastructure investment will not return a profit.

What happens to neocloud providers?

The pressure lands hardest on smaller players. Gartner states that hyperscalers are large enough to absorb the financial strain. Neocloud providers are not. Gartner projects that cost pressure will force more than 80% of neocloud providers to cease independent operations.

That figure carries consequences for enterprise buyers. If four out of five independent GPU-cloud vendors exit the market, workloads migrate — often on short notice. CIOs contracting with neoclouds today should read the 80% number as a concentration-risk signal and plan exit clauses accordingly. Gartner did not name which providers it expects to fold, nor did it give a timeline beyond 2030.

What should enterprises take from it?

The two analyst houses agree on the core point: some AI infrastructure investment will simply lose money. Gartner's framing is that companies need to prepare for a degree of financial loss. Bain's framing matches it at a different scale — $1.5 trillion in annual spend by 2031 against a required $6 trillion services market.

For IT leaders, three practical implications follow from the numbers:

• Vendor durability. With 80% of neoclouds forecast to exit, contract terms and data-portability clauses matter more than headline GPU prices.
• Capacity timing. If the $3.6 trillion revenue ceiling holds, some planned capacity will chase demand that never arrives, and pricing could swing in buyers' favor.
• Investment scrutiny. Internal AI business cases should be stress-tested against the possibility that aggregate industry revenue stays below the level needed to fund the buildout.

Every promise of AI-driven returns now carries an analyst-grade caveat attached: the industry as a whole has committed capital that its own revenue forecasts cannot repay.