The company is designing an enterprise system powered by two or four M8 Ultra chips targeting AI developers and government agencies. Apple is considering the use of Nvidia's NVLink Fusion networking technology to connect the chips for AI inference on private infrastructure, The Information reported. A launch is aimed for 2029, though internal plans could still be modified or canceled.

This move follows a 29% increase in Mac revenue to $10.4 billion last quarter, driven by demand for Mac mini and Mac Studio systems from AI professionals. If the project launches, it will be the first Apple server product since the 2011 discontinuation of Xserve.