Gartner has nudged down its expected growth in worldwide IT spending for 2024 from 8 percent to 7.5 percent, with the total figure now expected to reach $5.26 trillion.

The global IT research biz and consultancy had previously upped its growth forecast from 6.8 percent to 8 percent based largely on repricing of communications services.

Among the drivers of spending growth are datacenter systems, spending on which is forecast to increase 24 percent in 2024, up from the previous quarter’s estimate of 10 percent growth, largely down to investment in the development of GenAI models.

“The compute power needs of GenAI are being felt across the datacenter, and spending in that segment reflects this ravenous demand,” said John-David Lovelock, Distinguished VP Analyst at Gartner told The Register.

The downward revision of the growth figure for the year was made in part because the analyst has changed what it is tracking, but also because CIOs are holding off starting big projects for longer than expected.

IT services spending is now projected to rise 7.1 percent in 2024, down from the previous forecast of 9.7 percent, for example. The slowdown was a result of so-called “change fatigue” among CIOs lasting longer than Gartner had expected.

Lovelock explained that, following the pandemic, large scale digital transformation initiatives became the responsibility of CIOs, who had big projects and contracts in the offing from 2021 to 2023. But since then they have been reluctant to take on such large projects, focusing instead on helping departments with incremental cost savings.

“It wasn't a money thing. Nobody was seeing budgets seriously cut or impinged. It was just a slowing down in committing to new things. That continued to slow down into April, which means coming up it's going to be a little bit bumpier and longer than we projected,” he said.

While GenAI spending is driving growth in datacenter investment, software vendors are expected to invest in the technology - most likely via third parties - while end users are yet to be convinced of the value GenAI may bring to the enterprise. At the moment, GenAI is akin to a tax on software vendors, he said.

“For most software companies, it's not like they've had the time or the attention to develop their own GenAI models - yet. They are going to go to a third party for that model… and hook into that via an API. The problem is the cost of the generative AI model inferencing is not necessarily tied to the value derived. So we have the vendor having to charge the cost of the Gen AI and the client paying for the value of the AI. Hopefully those two align but they don't by necessity,” he said.

Lovelock said most software companies did not have the resources to train their own AI models, and business users were still uncertain about the value of GenAI. As such, software companies will be forced to make some “bets” and hedging to give themselves as many possibilities in terms of the GenAI they integrate in their software.

In January, Garnter said that most large user organizations were still in the planning stages of how they would deploy GenAI.

https://www.theregister.com//2024/07/16/gartner_nudges_down_global_it/