Apple steps up as AI infrastructure diversifies

AI is changing the tech stack—and for Apple, that’s an opportunity to enter markets traditionally dominated by PCs.

Enterprises rushing to deploy AI solutions are hitting the reality that building infrastructure takes time and money—and hardware isn’t always available. Delays can impact project timelines and even an organization’s health.

Fixit. Many IT teams are turning to Apple hardware to pick up the slack when it comes to the on-prem training and use of AI models. Matt Vlasach, SVP of product and solutions engineering at Jamf, told IT Brew that a number of factors including token cost, data sovereignty, and the capabilities of Apple’s chipset are at play in moving to the company’s hardware.

“It’s created a lot of interest in leveraging Apple devices as a great tool for proving out local device inference and exploring on-device models to figure out what sort of workloads can be handled locally for the sake of cost reduction, as well as for the sake of data sovereignty,” Vlasach said.

Kush Patel, CIO at IT services company Tusker, told IT Brew that some software can’t be run on Mac, Patel said, which can present challenges in a Windows-heavy environment. Ultimately, however, the decision will depend on what works for the individual organization—cost, efficiency, or stack needs—and how you manage compliance, identity, and the usual IT intricacies.

“If you’re going as an end user device, it’s the whole consulting path we have to go down, and how do we make a Mac work in your world and work in your compliancies?” Patel said. “If it’s an AI play with Apple, it’s easier.”

Cashing in. The main motivation is cost, which has gone up exponentially for AI solutions. Tusker found that sometimes this can result in lower costs—one enterprise built an AI cluster with Apple laptops for $15,000, a price cut from the $60,000 to $70,000 price tag on other AI hardware. Another company used three Apple laptops instead of spending $75,000 on a Nvidia solution.

That has made Apple’s hardware, despite its reputation for premium price tags, appealing for enterprises.

“The power draw of a Mac Studio or Mac laptop is much less than a full server, a rack full of servers—it’s small enough to fit under my desk,” Patel said, adding: “The power is much less, the cost of acquiring the hardware is much less.”

Vlasach sounded a similar note, explaining that overall pricing is changing, and with it, the motives for using different tech stacks.

“There were periods of time where Macs were considered just too expensive—they just wouldn’t even consider it,” Vlasach said. “We got to a point here earlier in the year where if Macs weren’t the same price, they were actually cheaper than some of their PC counterparts.”