Apple Prepares Enterprise Server for AI Inference, Targeting 2029

Apple is developing an enterprise-grade server designed for AI applications, according to The Information. The device is expected to feature Apple's in-house M8 Ultra chip, and the company has held talks with Nvidia about using its NVLink Fusion interconnect technology. The server could launch as early as 2029, with a focus on AI inference workloads—marking Apple's first serious evaluation of re-entering the enterprise server market since it discontinued the Xserve in 2011.

Newly appointed CEO John Ternus has been a key driver behind the project, which started about a year ago when he was still leading Apple's hardware engineering division. The potential partnership also reflects a thawing in the long-strained relationship between Apple and Nvidia, which have already worked indirectly on Siri improvements and deepened discussions around Nvidia technology and open-source AI models over the past year.

Apple's Mac product line has become its fastest-growing business, with latest-quarter sales rising nearly 29% year over year to $10.4 billion, but supply constraints have highlighted a structural gap in meeting enterprise demand with consumer products. If the server initiative materializes, it could position Apple as a competitor to Nvidia's dominance in the AI server market.