Apple is reportedly building an enterprise AI server with M8 Ultra chips
Apple is reportedly developing an enterprise server powered by its own silicon and aimed at AI developers, businesses and governments, according to The Information, which cited people familiar with the project. The server is intended for AI inference, meaning it would run models that have already been trained rather than train new ones. It would ship in two versions, one with two M8 Ultra chips and another with four. For connecting the chips inside data centers, Apple is considering Nvidia's NVLink Fusion technology, which was originally built for Nvidia's own chips and has since been opened up to third-party hardware; the report notes the project could still proceed without it, or be scrapped entirely. A launch would not happen before 2029. Apple's new CEO, John Ternus, backed the effort about a year ago, while he was still heading the company's hardware division, seeing an opening in the growing AI compute market. The project may have picked up momentum recently: AI labs including OpenAI and Anthropic are already buying Mac Minis and Mac Studios in bulk for AI workloads, and Apple's Mac revenue rose nearly 29 percent last quarter to $10.4 billion.
Key facts
- Apple is reportedly building an enterprise AI inference server using its own M8 Ultra chips, per The Information, citing people familiar with the project.
- Two configurations are planned: one with two M8 Ultra chips, another with four.
- Apple is weighing Nvidia's NVLink Fusion to link the chips, but the project could go ahead without it or be scrapped; launch would not come before 2029.
- CEO John Ternus backed the effort about a year ago while still heading Apple's hardware unit.
- OpenAI and Anthropic are already buying Mac Minis and Mac Studios in bulk for AI workloads, and Apple's Mac revenue jumped nearly 29 percent last quarter to $10.4 billion.
Why it matters
Apple has stayed on the sidelines of the AI data center buildout that Nvidia, AMD and the hyperscalers have driven for years. An enterprise server built on its own M8 Ultra silicon would be Apple's first direct move into that market, targeting the inference side of AI workloads rather than the training clusters that dominate the industry's current spending.
Who it affects
The report names AI developers, businesses and governments as the intended customers. It also points to a precedent already underway: OpenAI and Anthropic are buying Mac Minis and Mac Studios in bulk for AI workloads, which the report frames as a possible reason the project has gained momentum. Nvidia would be affected too, either as a technology partner if Apple licenses NVLink Fusion, or as a competitor if Apple builds its own interconnect instead.
How to use it
No pricing, technical specifications beyond chip count, or product name have been reported. What is known is the chip configuration: versions with two or four M8 Ultra chips, built for running already-trained models rather than training them.
How solid is it
The account comes from The Information, citing people familiar with the project, not an Apple announcement or confirmation. Key details remain unsettled by the report's own account: whether Apple will use Nvidia's NVLink Fusion is described only as under consideration, and the project itself could still be scrapped.
Risks and caveats
A launch is not expected before 2029, leaving years for the plan to change or be dropped entirely, as the report itself acknowledges. The interconnect technology, product name, pricing and specific customers are all unconfirmed. As an unannounced project reported through unnamed sources, it should be read as a reported plan, not a committed roadmap.