Update, October 1, 2026. On September 16 Gartner revised the 2026 figure to $2.7 trillion, up 49.5%. The May number below is the forecast this piece used when it ran. The revision is Gartner revised the AI bill.
In May, Gartner forecast worldwide AI spending of $2.59 trillion in 2026, up 47% from 2025. Infrastructure — servers, chips, networks, AI-optimized cloud — is more than 45% of that pile. John-David Lovelock, a Gartner analyst, put the awkward part in plain English: so far the spend is driven by tech vendors and hyperscalers. Enterprises “have yet to really flex their spending potential,” and CIOs still struggle to show a business outcome.
That is the split worth managing. The market is building capacity. Most companies are still buying pilots and calling them strategy.
What the surveys actually say
McKinsey’s 2026 State of AI survey, summarized in early September coverage of the report, found 37% of respondents attributing at least some EBIT impact to AI — roughly unchanged from the prior year. The share of “high performers,” defined as firms that both call the impact significant and attribute at least 5% of EBIT to AI, sat around 6%. Individual workers tell a different story: 80% said AI improved their own productivity; 50% said it helped them make better decisions.
Those two facts can both be true. A faster draft is not a cheaper close. Time saved at the desk does not automatically show up as margin if the workflow, the review step, and the headcount plan do not change.
About 20% of McKinsey respondents said AI operating costs, including tokens, were already constraining use. Most still planned to spend more. Gartner, in a separate September survey, said only 22% of organizations had successfully scaled AI across multiple business units.
A budget that grows 47% while the EBIT share stays flat is not a strategy. It is a bet that the spreadsheet will catch up.
Who is actually paying
The Ramp AI Index, reported by PYMNTS in August, showed how uneven the checkbook is inside the United States. In July the median company spent $11.95 per employee on AI. The top 1% spent a median of $7,400 per employee. Anthropic led paid business adoption in that dataset at 43.5% of U.S. businesses; OpenAI trailed on the latest monthly increment; xAI reached 4% of businesses.
Goldman Sachs, cited in the same PYMNTS piece, found that only 2% of S&P 500 companies quantified AI’s effects in second-quarter earnings. Of that tiny group, 11% cited measurable productivity in coding or support. Median earnings rose 17% at companies that quantified versus 14% at those that did not — a small gap, and not proof of causation. It is a reminder that the firms doing the counting are at least looking.
What to do with the numbers
If you run a mid-market shop, do not copy the top 1% per-employee spend. Copy their habit of attaching a cost to a workflow. Pick one process with a definition of done — close commentary, ticket macros, bid memos — and put a before-and-after number next to the license line. If you cannot show movement in 90 days, you have a subscription, not a deployment.
The Gartner forecast is not wrong because it is large. It is incomplete if you treat it as a mandate to match the industry’s check. Infrastructure companies will keep building. Your job is to make sure the next seat you buy retires an hour, a tool, or a rework loop — not just a slide.
Sources
- Gartner on X, AI spending up 49.5% in 2026, September 16, 2026
- Gartner on X, AI cost is becoming as important as deployment, September 24, 2026
- Gartner, “Forecasts Worldwide AI Spending to Grow 47% in 2026,” May 19, 2026
- Gartner, survey on scaling AI across business units, Sept. 1, 2026
- McKinsey, “The State of AI: Global Survey 2026” (figures as reported in HPCwire / AIwire, Sept. 2, 2026)
- PYMNTS, Ramp AI Index and Goldman Sachs earnings note, Aug. 17, 2026
Operator guides: How to tell AI spend from AI return. For one workflow, use token cost to gross margin, then Put an AI cost on one workflow. The September revision is Gartner revised the AI bill.