AI spending will exceed traditional IT spending for the first time in 2026, according to the latest Gartner forecasts, with AI budgets reaching $2.67 trillion next year while non-AI IT spending shrinks to $3.69 trillion — a crossover that closes a gap that stood at $2 trillion just two years earlier.

Total IT spending rose 10.7 percent to $5.58 trillion in 2025 and is expected to grow 14.2 percent to $6.37 trillion in 2027, based on Gartner's numbers. The analyst firm has not yet published a 2027 total IT forecast, so industry observers estimate $6.98 trillion for that year, up 9.6 percent.

The split tells a sharper story. Aggregate AI spending grew 2.6X in 2025 to $1.79 trillion. Gartner anticipates a further 49.7 percent rise in 2026 to $2.67 trillion, and 36 percent growth in 2027 to $3.64 trillion.

Traditional IT tells the opposite story. Non-AI spending fell 12.5 percent in 2025 to $3.79 trillion. Gartner expects declines of 2.5 percent in 2026, to $3.69 trillion, and 9.5 percent in 2027, to $3.43 trillion. Traditional datacenter IT has been in recession since 2023, when the GenAI boom took off, and the data suggests it will stay there for the foreseeable future.

This forecast covers an $8.1 trillion investment in AI infrastructure and services across 2025 through 2027.

What is driving the upward revisions?

Gartner raised its 2025 AI infrastructure spending estimate — servers, storage, and switching — from $975.6 billion in its May forecast to $981.9 billion in the September update. For 2026, the firm added $52.9 billion to the forecast. For 2027, it added $87.4 billion.

The increase likely reflects two forces:

• Custom XPU designs from hyperscalers and cloud builders, used internally and sold to large AI model builders such as OpenAI and Anthropic
• Incremental Nvidia and AMD GPU-accelerated system sales

"The buildout of AI data center capacity is the largest infrastructure project humanity has even undertaken," said John-David Lovelock, distinguished vice president analyst and economist at Gartner, in the statement accompanying the updated forecast. "The capacity growth from hyperscalers and service providers purchasing AI-optimized servers will continue to be the largest single area of spending."

Gartner's latest forecast also breaks out GenAI model licenses and sales of AI agents and assistants from AI data science and machine learning platforms and generic AI development platforms — a category distinct from the broader AI software segment, meaning enterprise software with embedded AI functionality.

The distinction matters. Buyers replacing PCs and smartphones get local AI capabilities incidentally with the hardware upgrade; those devices do not count as AI infrastructure. Enterprise software follows a similar pattern. Customers may intentionally upgrade applications for AI features, but most upgrade for capability gains generally — as they always have in enterprise computing. AI is simply another set of functions in the application stack, and at some point this becomes what "software" does by default.

Where does the money come from?

Synergy Research Group's latest forecast of spending on cloud computing services, cloud software services, and leading consumer-oriented digital services quantifies the revenue pool. The major players are Google, Microsoft, Amazon Web Services, Meta Platforms, and ByteDance.

John Dinsdale, principal analyst at Synergy Research, expects revenue growth among these service providers to accelerate from 2026 onward, exceeding the pace recorded from 2020 through 2025. Adding up revenues across these providers for 2025, 2026, and 2027 — estimated from chart data rather than a published table — yields roughly $6.2 trillion from IaaS, PaaS, SaaS, search, and social media.

That revenue must also cover salaries, health insurance, and other operating costs. Assuming hyperscalers, cloud builders, and AI model builders account for 75 percent of AI spending, their share works out to $6.1 trillion. They would have to spend every penny of revenue to fulfill their AI infrastructure plans.

This arithmetic explains the turn to debt financing, vendor financing from Nvidia, AMD, and Broadcom, roundtripping deals, and other capital-raising mechanisms. Without them, the tech titans cannot invest in AI at the rate the Gartner forecast implies. Neoclouds, sovereign clouds, and commercial enterprises face the same constraint. AI will eventually have to start removing costs to support further AI investment.

Can the borrowing be repaid?

Synergy Research's projections for 2028 through 2031 show revenue streams more than doubling. If that holds, there is a large future to borrow against — and that borrowing is what makes the future happen. AI budgets keep absorbing every headwind: inflation, bubble worries, power constraints, supply chain problems, political instability, and economic uncertainty. Every forecast revision moves the numbers upward.

The only way to predict this future is to live it.