Apple is reportedly developing an enterprise AI server featuring its M-series Ultra chips, with an expected launch in 2029. This marks Apple's return to the server market for the first time in nearly two decades, capitalising on surging demand from AI developers who have been purchasing Mac mini and Mac Studio computers in large quantities to power AI workloads.

The server would be available in two configurations, with either two or four of Apple's forthcoming M8 Ultra chips. Major AI companies including OpenAI have purchased "tens of thousands" of Mac devices for training AI agents, whilst Anthropic has rented Mac hardware through Amazon Web Services. Apple is reportedly exploring the use of Nvidia's data centre networking technology to connect the chips, though the project remains subject to cancellation. The initiative has support from new CEO John Ternus, who assumed leadership on 1 September; however, the industry faces ongoing memory chip shortages that could impact production timelines.