Apple is considering re-entering the enterprise server market with rack-mounted systems built around its upcoming “M8” series Ultra chips, according to multiple reports. Apple last sold the Xserve, discontinued in 2011, and the new effort is described as targeting organizations that want to run AI models on their own hardware.
Across outlets, the reported plans are still tentative. The discussions cite two possible server configurations: one using two M8 Ultra chips and another using four M8 Ultra processors (at least some described as early in development). Several sources say Apple is evaluating Nvidia networking or chip-interconnect technology—specifically NVLink Fusion—to enable faster, lower-latency data exchange between chips, addressing performance and scaling concerns.
Reporting also varies in emphasis. Some outlets focus on whether the servers would be sold to outside customers and when they might launch, with a potential 2029 timeframe frequently mentioned. Others note that Apple already builds server infrastructure for Private Cloud Compute, but that scaling it for wider external use has drawbacks. Sources also describe the project as ongoing for about a year, with potential challenges including hardware design, customer support, and software resources for AI developers. The effort could be changed or canceled.
