애플이 M8 울트라 기반 AI 머신과 엔비디아 네트워킹 관련 논의를 통해 기업용 서버 시장으로의 복귀를 검토 중인 것으로 알려졌다.
Apple is exploring a return to dedicated server hardware for the first time since discontinuing the Xserve line in 2011. According to a new report from The Information, the company is developing enterprise AI inference systems centered on forthcoming M8 Ultra chips and has been in discussions with Nvidia regarding its NVLink Fusion technology to interconnect those processors.
The planned machines would target AI developers, businesses, and governments seeking to run models on-premises rather than relying on public cloud infrastructure. Two configurations are under consideration: a two-chip system and a four-chip version built around Apple’s highest-performance M-series silicon. A commercial launch is not anticipated before 2029, and the initiative remains subject to potential cancellation or could ultimately ship without Nvidia networking components.
Development began approximately a year ago with backing from John Ternus, then Apple’s hardware chief and now CEO. Surging demand from AI teams purchasing Mac mini and Mac Studio units in volume appears to have accelerated Apple’s interest in a rack-scale offering. While Apple already manufactures servers for its own Private Cloud Compute stack, it has previously declined partner requests to commercialize that hardware. Internal interconnect options were viewed as too slow and expensive for larger deployments, prompting the exploration of NVLink Fusion.
NVLink Fusion is Nvidia’s suite of switches, chiplets, and software designed to deliver high-bandwidth, low-latency connections between processors, including non-GPU silicon. By opening this architecture to third-party manufacturers, Nvidia can supply networking infrastructure even when customers opt for alternative compute architectures. An agreement with Apple would mark a notable shift given the historically strained relationship between the two companies, which traces back to GPU reliability problems in MacBooks nearly two decades ago. More recently, Apple has integrated Nvidia hardware into cloud infrastructure supporting a revamped Siri.
Neither Apple nor Nvidia commented on the report, which has not been independently verified. Following the story’s circulation, Apple’s shares, which had traded lower earlier, reversed modestly in premarket trading. If brought to market, the product would extend Apple Silicon beyond consumer devices and Macs into enterprise AI infrastructure. It would also serve as a real-world test of whether efficient Arm-based M chips can compete at rack-scale inference workloads, while simultaneously evaluating whether Nvidia can maintain its position as the foundational connectivity layer in systems built around competing processors.
Regardless of the project’s outcome, the leadership transition represents a meaningful development for the company. As noted in recent commentary, having “a product person, per se” back in charge is noteworthy. To paraphrase Steve Jobs, when product-focused leaders are driven out of the decision-making process, organizations forget what it means to make great products. That original product sensibility and genius, which drove past success, is often eroded by executives who lack a clear conception of what constitutes a good product versus a bad one.