Apple Reportedly Planning M8 Ultra AI Servers with Nvidia NVLink Fusion for 2029
9to5mac.com

Apple Reportedly Planning M8 Ultra AI Servers with Nvidia NVLink Fusion for 2029

Tech News
2 min read

Published by AINave Editorial • Reviewed by Ramit

TL;DRApple is reportedly developing enterprise AI servers using M8 Ultra chips for sale to AI developers, enterprises, and governments. Two configurations (two and four chips) are under consideration, with Nvidia NVLink Fusion for inter-chip communication. A 2029 launch window is cited, but plans remain uncertain and could be canceled.

Apple is reportedly building enterprise AI servers based on its future M8 Ultra chips and plans to sell them, not just use them internally. According to The Information, two configurations are under consideration: a smaller server with two M8 Ultra chips and a larger one with four M8 Ultra chips. Apple has also explored using Nvidia NVLink Fusion networking to connect the chips, which would mark a notable collaboration between the two longtime rivals.

The servers would target AI developers, other businesses, and governments, according to Bloomberg's coverage of the report. Development has been underway for about a year, and CEO John Ternus is reportedly in favor. However, the project has not been finalized, and a launch is not expected before 2029.

What this means for AI builders

The direct sale of Apple silicon servers would open a new option for teams running inference or fine-tuning workloads. Today, some builders already cluster Mac Studios to run models locally, but a formal server product would offer better integration, support, and scale. The two- and four-chip configurations suggest Apple is targeting both smaller deployments and higher-capacity setups, potentially competing with GPU-based servers from Nvidia and AMD.

Cautious optimism

While the prospect is intriguing, the timeline is distant and uncertain. A 2029 launch means this product is still several chip generations away. Apple has not confirmed the project, and the report notes it could be canceled before then. The inclusion of Nvidia NVLink Fusion is also under consideration, not finalized, so the final interconnect approach may differ.

The bigger picture

If Apple moves forward, this would represent a significant shift: the company leaving the server market in 2011 with the Xserve cancellation and now potentially re-entering with a focus on AI. For builders, the key takeaway is that Apple is seriously exploring data-center hardware. That alone is worth tracking, even if concrete buying decisions are years away. Keep an eye on whether Apple prioritizes this internally or lets it slip.

FAQs

The M8 Ultra is described in reporting as Apple's upcoming high-performance chip intended for AI server workloads. It would serve as the core compute component in Apple's proposed enterprise servers, similar to how M-series chips power Macs but scaled for data center use. More details on the chip's architecture are not yet available.

Sources

Latest Tech News