AI's share of global IT spending jumped from 13.7 percent in 2024 to 31.7 percent in 2025 — but Gartner's latest forecast shows the era of repeated large upward revisions is ending, because DRAM and HBM memory makers are running at full capacity and cannot expand much further.

That supply constraint is now the gating factor for the AI accelerator market. Price increases passed through to customers will nudge forecasts up modestly, but the leaps in revenue forecasts seen every few months over the past several years will not repeat.

Gartner's latest numbers bear this out. The research firm raised its AI infrastructure sales forecast by 4.8 points for 2026 and 8.1 points for 2027 — and much of that reflects opportunistic pricing rather than capacity expansion. AI software forecasts moved up only a few tenths of a point since the January forecast. AI services forecasts for this year and next were actually tweaked downward by a few tenths.

The revisions still describe enormous incremental money flowing into the IT sector. That marks a stark contrast with the years before the GenAI boom started in late 2022, when IT spending growth of 5 to 8 percent drew constant complaints.

What changed in the 2024–2025 numbers?

Gartner characterized the market somewhat differently back in 2024, but comparable AI segments show a four-year horizon of AI budgets worldwide in US dollars. Growth from 2024 to 2025 was large across every AI spending category.

Two forces drove AI's jump to 31.7 percent of the IT pie:

• AI spending itself grew sharply across hardware, software, and services.
• The rest of the IT spending pie shrunk by 12.6 percent.

Anyone claiming AI is not eating into other parts of the IT budget is wrong, and that trend looks set to continue indefinitely — even amid a spending boom to modernize and consolidate traditional servers to free room and budget for AI projects.

Recent financial results from Dell and Hewlett Packard Enterprise reflect that consolidation wave. It will not last forever, though. Companies will eventually return to incremental installations for traditional servers, and spending on traditional back-office and analytics systems will resume its slide.

Will AI spending overtake non-AI spending in 2027?

Gartner has not yet published a 2027 IT spending forecast, so the author of the analysis produced his own projection. If that guess holds, aggregate AI spending on IT hardware, software, and services will exceed non-AI spending by a smidgen for the first time in history — and the gap should widen from there.

That crossover carries a large assumption: AI proponents must prove the investment pays off through increased efficiency and productivity, reduced costs, and higher revenues from AI projects. This is not a foregone conclusion, and the outcome depends on your assessment of the world economy and AI itself.

What could break the trend?

A single recession could produce three very different outcomes:

• It could stop the AI spending spree outright.
• It could maintain spending at current levels.
• It could accelerate AI spending at the expense of legacy IT systems.

There is no way to predict which path materializes short of living through it.