Analysts agree: PCs with dedicated AI hardware will take over the enterprise market. They differ, however, on how quickly the transition will take place.
Consulting firm Gartner recently projected that AI PCs—which it defines as any PC with a neural processing unit (NPU), a type of chip designed to run machine learning and large language models efficiently—will become ubiquitous in just a couple of years. Almost one-quarter (22%) of total PCs shipped in 2024 will have an NPU, according to Gartner, and it will become a universal feature of enterprise PC purchases by the end of 2026.
With companies like Qualcomm, AMD, and Intel all entering the market, NPUs will soon cease to be a feature only found in high-end PCs, Gartner Senior Research Director and Analyst Ranjit Atwal told IT Brew. NPU-equipped PCs are the result of a “supply-side push,” according to Atwal, with manufacturers waiting on developers like Microsoft to create AI features that will justify the investment.
As NPUs “become more pervasive, it doesn’t become a choice,” Atwal said.
“You have the capability, you have the install base, and then you maximize those capabilities over time...It’s an introduction to the capabilities as much as anything, [but] won’t be the finished article,” he added.
Morgan Stanley analysts were less effusive in their own recent projection, forecasting AI PCs will comprise around 28% of the total PC market by 2026, reaching 64% by 2028. Forrester analyst Andrew Hewitt advised buyers to skip this year entirely, but begin buying enterprise AI PCs in 2025.
Hewitt noted that AI PCs still lack a killer app to justify the additional expense of an NPU, telling IT Brew that most features powered by generative AI, like picture enhancement on video calls, fall into the user experience category. That’s not what drives enterprise purchasing, according to Hewitt.
Top insights for IT pros
From cybersecurity and big data to cloud computing, IT Brew covers the latest trends shaping business tech in our 4x weekly newsletter, virtual events with industry experts, and digital guides.
“Much of the story around the benefit from AI PCs is focused on end user experience,” Hewitt told IT Brew. “And so what’s missing from that conversation is what benefit is there for IT.”
Michael Cherry, the Windows analyst at IT information and advisory service Directions on Microsoft, cautioned it was also unclear whether NPUs made by different manufacturers will be cross-compatible with all services due to the lack of a common framework.
“It’s a really bad time to buy a PC,” Cherry said. “And not because there aren’t deals: because you could be buying something that’s not optimal or that is going to become obsolete on a faster cycle than normal.”
One of the key purported benefits of NPUs is reduced reliance on costly cloud services. The Wall Street Journal reported some CIOs are wary, however, of investing before the technology is demonstrated to benefit their bottom line, or are limiting NPU upgrades in the short term to specific roles like data scientists.
According to Atwal, enterprises should consider investing now as a method of future-proofing, especially with the looming end-of-life date for Windows 10 in October 2025.
“Whether they’re going to use them now or not isn’t really the question, because they’re going to have those devices for three or four years,” Atwal said.
“You don’t want to be in a place where you’re not able to take advantage of those features, unless you’re so certain that the next iteration of all this in three years’ time is where you want to be,” he added.