Apple Revives Server Ambitions with Nvidia Partnership

Apple's plans to re-enter the server market have been in the works since at least 2026, when John Ternus, Apple's new CEO, expressed support for the project. The proposed servers will use Apple's M-series Ultra chips and pair them with Nvidia networking equipment, targeting AI developers, enterprises, and government clients.

The product is expected to launch in 2029 and will come in two versions: a lower-spec version integrating two M8 Ultra chips under development, and a higher-spec version carrying four M8 Ultra chips to form a compute cluster. To solve the problem of high-speed interconnect among multiple chips, Apple has held talks with Nvidia about adopting its NVLink Fusion interconnect solution.

This move marks Apple's return to the server market, which it previously left in 2011 due to insufficient emphasis on the enterprise market and weak customer support. The new product will not be available until 2029, and the project may still be canceled, but its potential impact is significant: if realized, it could help Apple sell differentiated private AI compute at a premium.