Technology and Science17 September, 20263 min read
Apple Develops AI Enterprise Servers With M8 Ultra Chips And Nvidia Networking For 2029
This article is an automated aggregation of coverage from news outlets. Our editorials offer in-depth analysis curated by the NewsCord team.
Apple developing two configurations of enterprise AI servers powered by M8 Ultra chips. Plans include integrating Nvidia networking technology, potentially NVLink Fusion, to connect processors.
8 outlets told this the same way.
Apple eyes AI servers
Apple is reportedly developing enterprise servers for AI demand that would use Apple’s future M8 Ultra chips and Nvidia networking equipment, with The Information saying the effort targets a 2029 launch. The reported server designs under review include a smaller model combining two M8 Ultra chips and a larger model equipped with four, and Apple has been considering using Nvidia’s NVLink Fusion to connect the chips. The Information’s reporting also says Apple could cancel the project or proceed without using Nvidia technology, even as Apple and Nvidia began cooperating in 2001 and later saw ties cool under Steve Jobs.
Ars Technica adds that an expected release in 2029 would mark the first Apple server to hit the market in nearly two decades, and that the enterprise server would come in two configurations with either two or four of Apple’s future M8 Ultra chips. Ars Technica further links the push to surging popularity of Apple hardware among AI developers, describing Mac mini and Mac Studio demand as AI developers and companies snap up the computers to run AI workloads.
NVLink Fusion debate
The Information’s reporting, as relayed by 디지털투데이, says some people involved in the Apple project see NVLink as the most suitable connection method, and it notes NVLink Fusion was originally designed to link Nvidia chips but can now connect chips from other companies. TechPowerUp frames the same development as Apple weighing a return to the server market with Nvidia networking technology, describing a plan to interconnect two or four Apple M8 Ultra chips using NVIDIA NVLink for socket-to-socket communication. TechPowerUp also says the effort would delegate every interconnect question to NVIDIA NVLink Fusion while Apple focuses on the actual system-on-chip, and it describes the product as targeting enterprises and governments.
Ars Technica reports that the project received support from new Apple CEO John Ternus when it began a year ago, back when Ternus led Apple’s hardware engineering efforts. Ars Technica’s account also ties the server work to AI hardware buying patterns, saying The Information reported that AI companies such as OpenAI have bought "tens of thousands" of Mac minis and Mac Studios to train AI agents.
Launch, market, and risk
Multiple outlets say the project is not expected to reach the market before 2029, with The Information reporting that it could be cancelled or proceed without Nvidia technology. 9to5Mac says the report specifically says Apple plans to offer the AI-focused servers for sale, not just use them internally, and it repeats that the servers are not expected to go on sale until 2029. The Next Web adds that the machines would be aimed at AI developers, companies and governments rather than ordinary Mac buyers, and it describes two versions with one carrying two M8 Ultra chips and a larger one carrying four.
디지털투데이 reports that Apple sold its own server product, Xserve, from 2002 to 2011, but did not achieve major results due to a lack of support for corporate customers, leaving attention on whether Apple can secure a meaningful presence this time. Ars Technica notes the stakes for Apple’s positioning in enterprise AI infrastructure, saying the 2029 server could capitalize on the surging popularity of Apple hardware among AI developers and mark Apple’s first server in nearly two decades.