Gartner: Only 1% of a $2.67T AI Bill Goes to the Models

Worldwide spending on AI is forecast to reach $2.67 trillion in 2026, a 49.5% increase from 2025, according to Gartner research reported by Techstrong on 21 September 2026. Gartner expects spending on generative AI models to reach $28.3 billion, and Techstrong reports that models account for only about 1% of total AI spending.

Investment firm Social Capital posted the split on X on 24 October 2026: “Worldwide spending on AI is projected to reach ~$2.7T this year. Yet, only ~1% of it is expected to go to the models themselves.” Its chart cites Gartner’s “Worldwide AI Spending by Market” of 16 September 2026.

This Week In The AI Supercycle · VIDEO

The Business Engineer’s explainer for this week’s issue, 18 to 24 October 2026: the companies selling AI’s parts say they are booked for years, while the companies paying for the build found money dearer. It does not discuss Gartner’s forecast.

The Numbers Gartner Gives

Gartner expects AI infrastructure spending to reach $1.48 trillion this year, about 56% of the worldwide total, Techstrong reports. In plain words, AI infrastructure is the hardware and cloud capacity that AI runs on: servers, chips, networking gear and rented computing.

Beyond infrastructure, Gartner forecasts AI services spending of $576.5 billion and AI software spending of $461.6 billion in 2026. It forecasts $29.2 billion for AI agents and assistants, and $28.3 billion for generative AI models, up 117% from $13 billion last year.

By our arithmetic, $28.3 billion is about 1.06% of $2.67 trillion, in line with Gartner’s “about 1%” as Techstrong reports it.

Source: Gartner, “Worldwide AI Spending by Market” (16 September 2026), as charted by Social Capital on 24 October 2026. Techstrong’s report gives the infrastructure, services, software, agents and models figures; the cybersecurity, data science, application development and data rows appear in Social Capital’s chart.

What Was Posted on X

Social Capital says worldwide spending on AI is projected to reach about $2.7 trillion this year, with only about 1% expected to go to the models themselves.

Who Buys, and What Grew

Gartner says the spending is driven mainly by hyperscalers and technology service providers buying AI-optimized servers, semiconductors, networking equipment and cloud capacity, and that technology providers account for about 35% of worldwide AI spending, Techstrong reports.

Gartner forecasts AI-optimized IaaS spending of $42.3 billion in 2026, up 96%, with $23.3 billion for inference and $19 billion for training. IaaS, or infrastructure as a service, is computing power rented from cloud providers rather than owned.

Gartner has raised the forecast several times this year: in January it projected $2.53 trillion for 2026, including $1.37 trillion for infrastructure, and the latest estimate adds about $143 billion, with about $118 billion of that attributed to infrastructure. “The buildout of AI data center capacity is the largest infrastructure project humanity has ever undertaken,” Gartner analyst John-David Lovelock said.

Connecting the Coverage

Our Gartner IT piece of 23 October reported Gartner’s 21 October forecast that data center systems spending rises 55.5% to $1,366 billion in 2027, about half of that year’s IT spending increase by our arithmetic. Our Gartner impairment piece of 15 October reported Gartner’s warning that more than $3.5 trillion of AI infrastructure spending could be exposed to impairment by the end of 2030.

Our collision piece of 24 October set an H100 rent 44% higher than in December, by our arithmetic from Silicon Data, beside token prices about 50% lower year on year in YipitData’s panels. Our Apollo piece reported Torsten Slok’s view that mainstream companies are starting to treat token spending as a cost center that weighs on margins.

The Structural Read

As we read it, the money in the forecast goes to the layers below the model: Gartner says hyperscalers and technology service providers buying servers, semiconductors, networking and cloud capacity drive the spending, Techstrong reports.

As we read it, this sits beside our collision piece: H100 rent rose 44% by our arithmetic from Silicon Data while YipitData measured token prices about 50% lower year on year.

As we read it, Gartner’s own upward revisions point the same way: about $118 billion of the $143 billion added since January went to infrastructure, Techstrong reports.

Three Implications

FOR MODEL MAKERS Gartner puts generative AI models at $28.3 billion of $2.67 trillion in 2026, about 1% as Techstrong reports it.

FOR INFRASTRUCTURE SUPPLIERS Gartner expects AI infrastructure spending of $1.48 trillion this year, about 56% of the total.

FOR SERVICES AND SOFTWARE Gartner forecasts $576.5 billion for AI services and $461.6 billion for AI software in 2026.

The Business Engineer Lens

This story maps onto the Business Engineer framework The Nine Layers of AI.

The framework puts it this way: “That is why the AI economy is not a list of companies racing each other. It is a layered industrial stack — and at any given moment, the binding constraint sits at a different layer than it did six months ago.”

As we read it, Gartner’s forecast is a spending map of that stack: most of the 2026 budget sits in the infrastructure layers and about 1% in models. The framework was published in May 2026 and does not discuss Gartner’s forecast.

What Is Not Established

These are Gartner forecasts for 2026, not reported spending. We could not open Gartner’s own release, which blocked our reader; the figures come from Techstrong’s report of it and from Social Capital’s chart, which cites Gartner’s table. Four of the nine market values appear only in that chart. Gartner’s market definitions decide what counts as “models”: spending on cloud capacity used to run models sits in other markets. We did not contact Gartner or Social Capital.

The Bottom Line

Gartner forecasts $2.67 trillion of worldwide AI spending in 2026, with $1.48 trillion, about 56%, on AI infrastructure and $28.3 billion, about 1%, on generative AI models, according to Techstrong’s report of the forecast.