Gartner's latest worldwide IT spending forecast exposes a trend every CIO has already felt in budget negotiations. Total spending will climb 14.2 percent in 2026, reaching $6.37 trillion.
It seems that the IT infrastructure market is simply having a strong year.
However, the more useful insight is how unevenly that growth is distributed.
The Headline Number vs. The Real Story
Data center systems and infrastructure as a service (IaaS) are absorbing capital at a pace several multiples faster than devices, communications services, and traditional IT services.
This is not incremental growth spread across a healthy portfolio.
It is a wholesale reallocation of enterprise technology budgets toward AI infrastructure, made on the expectation that demand for AI workloads will justify the outlay before that demand has been fully proven.
Where the Money is Going
Data center systems are forecast to grow 62.5 percent in 2026, that's up from 51.6 percent growth in 2025, reaching $822 billion. The acceleration, not just the size of the number, is the signal worth reading closely: capital is not simply following AI demand, it is chasing an expectation that demand will keep compounding.
IaaS follows a similar trajectory, growing from 25.3 percent in 2025 to a projected 29.3 percent in 2026 and reaching $287 billion, as enterprises rent AI compute capacity rather than commit fully to owned infrastructure.
Where Investment is Restrained
Software spending is forecast to grow 15.5 percent, to $1.47 trillion, as AI-ready platforms absorb budget that once funded broader software refresh cycles.
By contrast, device spending is expected to grow just 9.8 percent, communications services 4.4 percent, and IT services 5.3 percent. These were once the steady core of enterprise technology budgets. They are now, in relative terms, somewhat of an afterthought.
John-David Lovelock, Distinguished VP Analyst at Gartner, put a name to the scale of what is being built, describing the AI compute buildout as "the largest infrastructure project ever attempted by humanity."
That framing captures the moment. It also hints at the risk inside it: no one has run this experiment before, so no one can forecast its return on investment with precision.
The Strategic Question for the C-Suite
For technology leaders, the challenge is not deciding whether to fund AI infrastructure. That decision has effectively been made across the industry, and the reallocation already underway will accelerate as hyperscalers and enterprises compete for the same constrained pool of compute, power, and talent.
The harder decision, and the one boards are not yet asking clearly enough, is how to distinguish AI investment tied to workloads with demonstrated, repeatable economic value from investment made against a general belief that AI demand will eventually arrive.
Lovelock's own caution, that a growing budget does not mean every category benefits equally, is a useful discipline for any CFO reviewing a 2027 capital investment plan.
Proven Value vs. Speculative Hypothesis
A small number of AI applications, agentic workflows in specific back-office functions, coding assistance, and targeted customer service automation among them, have moved past the pilot stage and into measurable, repeatable value creation.
A much larger number remain speculative: broad language model deployment with no clear owner of the business outcome, or infrastructure sized for projected workload growth rather than validated demand.
Enterprises funding the second category at the same pace as the first are effectively underwriting the industry's forecast rather than their own business case.
The organizations that look disciplined in 2027 will be the ones that can answer, category by category, which AI infrastructure dollars are chasing proven usage and which are chasing a hypothesis.
That being said, I believe the distinction, more than the topline growth number, should be the starting point for every enterprise IT infrastructure budget conversation this year. This thoughtful approach is a risk reduction checkpoint that every executive team must debate and resolve.