Apple develops AI server using proprietary M8-Ultra chips

Apple is reportedly developing a new AI server for data centers powered by its proprietary M8-Ultra chips. According to reports from The Information, the company is considering two configurations: one utilizing two M8-Ultra chips and another utilizing four.

To connect these chips, Apple is exploring the use of NVIDIA’s NVLink Fusion technology, which allows multiple chips to function as a unified computing system. This potential move marks Apple’s first attempt to re-enter the server market since discontinuing its Xserve line in 2011.

The project is currently in its early stages, and a market launch is not expected before 2029. The initiative could still be modified or canceled. The development follows increased demand for Apple hardware from AI companies; OpenAI has reportedly purchased tens of thousands of Mac mini and Mac Studio units for reinforcement learning tasks, while Anthropic has utilized Mac mini systems via Amazon Web Services.