Apple weighs Nvidia's NVLink Fusion for planned AI server with M8 Ultra chips

Apple has held discussions about using Nvidia's NVLink Fusion networking technology to connect chips inside a planned artificial intelligence inference server built around its own silicon, a move that would mark the company's re-entry into the server hardware market.

The proposed machine is being developed as an enterprise AI server aimed at AI developers, businesses and governments, and could be offered in configurations carrying either two or four of Apple's planned M8 Ultra processors, described as its highest-performance chips.

NVLink Fusion has been considered as a way to link those M8 chips within the server. The product is not expected before 2029, and the company could still scrap the plan or bring it to market without Nvidia's technology.

Apple's shares, which had been trading lower earlier, reversed direction and edged up modestly in premarket trading following the report.

A return to servers would revive a category Apple exited more than a decade ago. The company last sold dedicated server hardware under its Xserve line, which was discontinued in 2011 as it shifted its enterprise focus toward standard Mac hardware and consumer devices.

Any collaboration with Nvidia would come against a long backdrop of friction between the two companies. Ties have been strained for nearly two decades, after an alleged defect in Nvidia graphics chips caused widespread MacBook failures and led Apple to cut off the relationship.

Neither Apple nor Nvidia responded immediately to requests for comment, and the reported plans could not be independently verified.