Apple Eyes Enterprise Server Return to Fuel AI Ambitions

Apple is developing an enterprise AI server that would utilize the company's high-performance M-series Ultra chips, according to reporting from Ars Technica. The project, which reportedly began a year ago, is eyeing a 2029 debut and would mark Apple's first foray into the server market since the retirement of the Xserve in January 2011.

As reported by The Information via Ars Technica, the proposed server would be available in two configurations featuring either two or four future M8 Ultra chips. The initiative reportedly has the backing of Apple CEO John Ternus, who led the company's hardware engineering efforts when the project started. Ternus succeeded Tim Cook as CEO on September 1.

The push into enterprise hardware follows a surge in demand for Apple's Mac mini and Mac Studio among AI developers. The Information reports that OpenAI has purchased "tens of thousands" of these devices to train AI agents using trial-and-error reinforcement learning, while Anthropic has rented Mac minis through Amazon Web Services.

While Apple is exploring the use of Nvidia's NVLink Fusion technology to connect its M8 chips, Ars Technica notes that The Information cautioned the project could still be canceled or proceed without Nvidia's involvement. Additionally, the effort faces industry-wide headwinds; Ars Technica reports that a shortage of memory chips driven by the AI data center boom has already led to price increases for various flagship smartphones and Apple products.