Dive Brief:
- Global IT spending will grow 15%, reaching $7.35 trillion in 2027 as organizations funnel money into AI infrastructure and software in a bid to “industrialize AI deployment,” according to Gartner.
- Spending on AI-optimized servers and infrastructure as a service, as well as AI software platforms, will continue to strengthen into next year. Hyperscaler, technology and service providers’ continued investments increasingly position infrastructure as “the foundational layer of the AI economy” and a strategic asset, the analyst firm said.
- “Right now we have more capacity than we have demand,” John-David Lovelock, distinguished VP analyst at Gartner, told CIO Dive. “We have a lot of premium model giveaways, subsidized model and token pricing, in order to get usage coming. As much as CIOs are worried about the cost of AI today, it’s not going to be this low again.”
Dive Insight:
Heightened AI infrastructure spend is putting the hardware market in a bind, as the push to build enough data centers to support the compute-hungry technology causes a surge in memory costs and supply constraints.
In the last year, demand from data center systems led memory prices to more than double, Lovelock told CIO Dive. Cost increases rippled throughout devices, storage, server and networking supply chains.
The “memflation” effect — a term Gartner coined to describe the surge in memory pricing — will persist long through 2027, likely into 2030, Lovelock said. The trend means suppressed end-user demand as increased costs hit PCs, laptops, tablets and mobile phones — any device that needs a memory or computer chip.
“This is the first place where we’re seeing the massive investment by the hyperscalers, the neoclouds, the folks building out the AI infrastructure, it’s now leaking over and starting to impinge on CIOs,” Lovelock said. “Everything that has a memory chip or a computer chip in it is going up in price and there’s no negotiating.”
Even with hyperscaler, technology and service providers’ significant AI infrastructure investments, enterprises are behind on where they could be with AI, Lovelock said. Google, Microsoft and AWS, along with other technology and service providers, have launched forward deployed engineering programs to speed enterprise AI deployment.
“All the service providers, whether they’re hyperscalers or regular, are working with enterprises to find where’s the blockage internally and help them get past it and get to this sort of more productive, more prosperous future with AI in it,” Lovelock said.
