Apple is reportedly developing a dedicated AI server that would leverage the high-end M-series Ultra silicon currently found in its desktop computers. If the rumored 2029 launch window holds, it would represent the company's first foray into the enterprise server market in nearly 20 years, potentially leveraging the current trend of AI researchers adopting Apple hardware.
Technical Specifications and Leadership
According to The Information, this server project could feature configurations with either two or four upcoming M8 Ultra chips. The initiative reportedly began a year ago with backing from John Ternus, who was then the head of hardware engineering. Ternus has since succeeded Tim Cook as Apple's CEO, taking the reins on September 1 after Cook's 15-year tenure.
Market Demand from AI Giants
The news follows a surge in Mac Studio and Mac mini sales to AI firms looking to run specific workloads. OpenAI has reportedly acquired "tens of thousands" of these machines for reinforcement learning tasks, while Anthropic is utilizing Mac minis via Amazon Web Services.
Nvidia Collaboration and Supply Chain Risks
Apple is also exploring the use of Nvidia's NVLink Fusion technology to interconnect these M8 chips, though the project remains subject to change or cancellation. A major hurdle remains the global memory chip shortage, driven by the rush to build AI data centers. As Ars has documented, this shortage has already driven up costs for various electronics, including flagship smartphones, gaming devices, and Apple's own product line.
