Apple may be late to the generative AI party, but don't rule it out just yet. According Bloomberg's Marc Gurman and MacRumors'Hartley Charlton, the company will use the M2Ultra on their own servers (in their own data centers) to fuel their growing GAI ambitions. Released in June 2023, the CPU, as used in the Mac Studio – remains the most complex piece of silicon ever released by Apple with 24 compute cores, up to 76 GPU cores and 32 AI accelerators.
The report does not mention whether Apple plans to revive its defunct Xserve range of rack servers or whether it will regain its Mac OS X Server OS. Both products have been discontinued for years as Apple shifted its focus away from the enterprise market early last decade. A separate article from WSJ He also adds that Apple is using the internal code name ACDC (Apple Chips in the Data Center).
In this article, authors Aaron Tilley and Yang Jie posit that Apple would use the formidable firepower of its data centers for training or more complex inference, while lighter workloads (or those that would require access to personal data ) would be handled locally on the device. itself, eliminating the need to run in the cloud.
This reflects what chipmakers like AMD and Intel, x86 stalwarts, have been advocating, in unison with Microsoft, with the PC with AI paradigm: large server chips (such as epic and Xeon) working in conjunction with smaller client processors (Ryzen or Core). The difference is, of course, that Apple uses an existing processor instead of a new one.
Another way to achieve AI hegemony?
Which raises another question; Did Apple plan this (to have a multipurpose CPU family) from the beginning? Considering that the M2 Ultra would probably be the only server processor in the world to have a GPU and an AI engine. Could it give way to a server-only version (the S1?) aimed at a data center environment, with many more cores, no GPU and much, much more memory?
However, there was never any doubt that sooner or later Apple would have started dabbling in server processors. It was a question of when, rather than if. Reports that Apple is building its own servers They date back to 2016 and is in line with Apple's doctrine of own the stack. In 2022The company also sought to hire an “upbeat and hard-working hardware validation engineer” to “develop, implement and complete hardware validation plans for its next-generation hyperscale and storage server platforms.”
Then, a year later, research carried out by the analyst firm Structure Research found that Apple was planning to triple the critical power capacity of its data centers to accommodate its two billion active devices (and nearly one billion iOS users) and provide more services.
Of course, hardware requires software and Apple has been increasingly vocal over the past 12 months, launching MLX, a machine learning framework designed specifically for Apple Silicon, a look at a AI-enhanced Siri and a new AI toolset called OpenELMs.
It will be immensely instructive to see how Apple manages to do generative AI at scale using more than just brute force GPUs (à H100). This could well have a direct impact on the fate of another trillion-dollar company called Nvidia.) WWDC, Apple's annual developer conference, will take place next month and it will have AI written all over it.
