Apple is working on an enterprise AI server that would be powered by its in-house M8 Ultra chips and aims to bring the product to market in 2029, according to the report. Company engineers have engaged in conversations with Nvidia about the possibility of using the chipmaker's NVLink Fusion interconnect to enable high-speed communication between chips.
The server under development would be offered in two configurations - one that pairs two M8 Ultra chips and another that houses four. Sources familiar with the project describe the target audience as AI developers, businesses and government entities. Some engineers involved in the effort view Nvidia's NVLink Fusion - an integrated approach that includes switches, chiplets and supporting software to let chips from different vendors communicate rapidly - as the most suitable connectivity solution available.
People familiar with the work say the project began about a year ago while the executive who recently became Apple CEO led the companys hardware engineering group. That executive supported the server concept at its inception and is reported to have continued backing the effort after taking on the chief executive role, a factor cited as important to the project's continued momentum. Neither Apple nor Nvidia has publicly confirmed the discussions, and the initiative could still be cancelled.
The plan to develop a server reflects a clear commercial motive. In Apples most recently reported quarter, the Mac product line was the company's fastest-growing segment, with Mac revenue rising by nearly 29 percent to $10.4 billion. That growth has been driven in part by unanticipated demand from AI developers buying Mac mini and Mac Studio systems, which has led to acute supply shortages and exposed the limitations of consumer-oriented hardware for enterprise AI workloads.
For Nvidia, which currently dominates the AI server market, the potential of Apple adopting NVLink Fusion is relevant to its business model. Nvidia derives roughly one-fifth of its data-center revenue from networking components. Securing Apple as a NVLink Fusion customer would add a significant new buyer for that technology and would reinforce Nvidia's strategy of monetising interconnect solutions even as competitor chips seek to challenge the companys GPU position.
While the reported plans outline a multi-chip server architecture and identify potential customers, the report emphasizes that the product remains a work in progress. Discussions over interconnect technology are ongoing and the company-level plans have not been publicly announced.
Key Takeaways
- Apple is developing an enterprise AI server built around M8 Ultra chips with a target launch in 2029.
- The server would come in two-chip and four-chip configurations and is aimed at AI developers, businesses and government users.
- Apple has held talks with Nvidia about using NVLink Fusion to connect M8 Ultra chips; such a deal would add a large-volume customer for Nvidia's interconnect technology.
Risks and Uncertainties
- The project could be cancelled and neither Apple nor Nvidia has publicly confirmed the reported talks - regulatory or strategic developments could alter the plan.
- Supply constraints that have affected Mac systems suggest execution and manufacturing challenges for scaling from consumer hardware to enterprise-grade servers.