Picking Apart An Ebullient GenAI Spending Forecast

April 2, 2025 Timothy Prickett Morgan

In one of the big ironies of our time, companies are embedding AI acceleration functions into PCs that most consumers do not actively want while at the same time datacenter operators (be they captive or independent) are actively trying to get AI compute engines (mostly Nvidia and sometimes AMD GPUs) so they can train their GenAI and other machine learning models – and they can’t get enough of these accelerated servers.

Because a lot of PCs are already enhanced with AI functions, more and more PCs are falling under the label of an “AI PC,” and therefore the market for AI PCs seems to be exploding, if the latest data from Gartner is any test. And because of the pseudo-popularity of these AI PCs, which drove $199.6 billion in sales in 2024, up by a factor of 9.5X compared to 2023 and which is expected to nearly double to $398.3 billion in 2025, the GenAI spending forecast put together this week by the prognosticators based in Stamford, Connecticut, is perhaps quite a bit rosy.

So don’t feel bad if you are shelling out the millions or billions on GenAI quite yet.

The fun bit is that an enormous amount of money is being spent even though Gartner finds trials of GenAI models embedded in applications have been disappointing.

“Expectations for GenAI’s capabilities are declining due to high failure rates in initial proof-of-concept work and dissatisfaction with current GenAI results,” John-David Lovelock, distinguished vice president analyst at Gartner, who created the GenAI model above, explained in a statement accompanying the numbers. “Despite this, foundational model providers are investing billions annually to enhance GenAI models’ size, performance, and reliability. This paradox will persist through 2025 and 2026.”

Lovelock says that companies will be looking for third party software providers to add AI features rather than try to do all of this GenAI creation and integration themselves, which reflects the attitude that most IBM i shops have had thus far. Straw polling we have seen shows that companies are waiting for either IBM or application software makers to give them the tools to have AI functions in their applications.

If you take the AI PCs and AI-enhanced tablets and smartphones out of the data and just look at servers, software, and services, you get a more reasonable growth rate for GenAI spending. If you do the math on the growth to calculate this datacenter GenAI spending for 2023, you get $62.5 billion, which grew by 2.65X to $165.4 billion in 2024 and which is only expected to grow by 48.5 percent to $245.5 billion in 2025.

Now, when you look at this data, you must remember that GenAI spending is still only about a third of total AI spending, and it may only grow to be half of spending in the next few years. There is a lot of traditional, statistical machine learning that is done for recommendation engines, search engines, and other AI workloads that do not generate tokens of some fashion. So just don’t think that this is all of the AI market, because it is not.

This Issue Sponsored By

Table of Contents

Content archive

Recent Posts

Subscribe

Pages

Search