THE CRUNCH

Apple is reportedly planning a return to the server market with custom chips, potentially using Nvidia's interconnect technology, according to reports. The company, which retired its Xserve line in 2011, is said to be developing an enterprise server powered by its own M8 Ultra chips. The device is intended for AI inference and could launch as soon as 2029, though the project remains subject to change.

The server would ship in two configurations with either two or four M8 Ultra chips, according to sources familiar with the project. Apple is considering Nvidia's NVLink Fusion technology to connect the chips, enabling fast communication inside data centres. This move comes as AI labs increasingly bulk-buy Mac Minis and Mac Studios for workloads, driving up Apple's Mac revenue.

The project is reportedly being led by Apple's new CEO, John Ternus, who backed the initiative about a year ago. While the timeline suggests a debut around 2029, the effort could still be scrapped or proceed without Nvidia's technology. The move highlights Apple's attempt to capture a share of the booming AI compute market, which has seen high demand for powerful hardware.

WHAT HAPPENS NEXT

The project could still be scrapped or proceed without Nvidia's technology, with a potential launch around 2029.