Gartner just put out their 2026 AI spending forecast. $2.59 trillion, up 47% from last year.
Most of that money is going to infrastructure. AI Infrastructure alone is $1.43 trillion, which is 55% of all AI spending. The companies building GPUs, data centers, and cloud capacity are the ones writing the checks right now.
The enterprises that are actually supposed to use all this? They haven't really started spending yet.
John-David Lovelock at Gartner put it pretty clearly:
"Up to this point, AI spending has primarily been driven by technology companies and hyperscalers. Enterprises have yet to really flex their spending potential. That is coming and 2026 will be the inflection year."
Where the money is shifting
One thing worth noting: inference is overtaking training in infrastructure spend. 55% of AI-optimized IaaS spending now goes to running models rather than building them. That number is expected to hit 65%+ by 2029.
That makes sense. You only train a model once. You run it thousands of times a day.
How enterprises are approaching it
Most enterprises are starting tactical. Incremental efficiency gains. Practical use cases. Solving a specific problem before trying to transform the whole business.
That's probably the right move. The companies starting with real problems tend to learn faster and scale smarter than the ones chasing disruption for its own sake.
What's next
Gartner expects $3.49 trillion in AI spending by 2027. Enterprise budgets are supposed to start ramping this year.
It'll be interesting to see which approaches stick.