Apple is reportedly developing an AI server built around its high-performance M-series Ultra chips, a move that would push the company back into the enterprise server market for the first time in nearly two decades. The machine is expected to ship in 2029 and would be offered in two configurations, one containing two of Apple’s future M8 Ultra chips and another containing four.
People familiar with the project say the effort began roughly a year ago and carried backing from John Ternus, who supported the work while he led Apple’s hardware engineering group and is now Apple’s new chief executive. The proposed product draws on the same family of chips that power Apple’s Mac desktops, and it reflects the company’s effort to package that silicon for larger, rack-style deployments aimed at AI workloads.
The plan comes as Apple’s Mac mini and Mac Studio devices have seen growing demand from AI developers and companies that use the machines to run model training and inference tasks. That uptake of Mac hardware running M-series Ultra processors is an explicit factor reported to have encouraged Apple to pursue a dedicated enterprise server using its own chips.
If delivered on the current timeline, the server would represent a significant strategic shift for Apple, moving from desktop-focused deployments into a form factor designed for data centers and enterprise customers. The product’s two- and four-chip configurations signal an attempt to scale performance inside a familiar silicon architecture rather than adopt third-party accelerators.
Apple’s timetable and the machine’s target customers point to a multiyear effort: engineers must adapt desktop-class M-series Ultra designs for sustained, rack-mounted operation and for the management and networking expectations of enterprise buyers. The next visible milestone will be whether Apple holds to the 2029 horizon and how the company positions the server versus existing choices from established data center vendors.
