Apple is reportedly targeting NVIDIA's NVLink Fusion technology for its M8-based Baltra servers, marking a strategic return to enterprise hardware after 2011.

Aligning with the industry-wide trend of leveraging existing infrastructure nodes, Apple is reportedly considering pairing its upcoming “Baltra” ASIC with NVIDIA’s bespoke low-latency connectivity solution. This strategic move would significantly expand options for Apple’s Private Compute inference architecture. Reports indicate the company is committing fully to NVIDIA’s NVLink connectivity framework to ensure seamless interoperability with its M8 chip-based servers.

Apple has been collaborating with Broadcom on its first AI server chip, internally codenamed “Baltra.” According to prior reporting by The Information, the ASIC may utilize TSMC’s 3nm ‘N3E’ process and feature a chiplet-based design. Broadcom is expected to assist in designing individual chiplets, while Apple will manage their integration into a single package. This siloed development strategy would allow Apple to keep the overall AI ASIC architecture confidential, even from partners like Broadcom.

Regarding server production, Foxconn has been tasked with manufacturing the units. Apple’s primary assembly partner will also receive design support from Lenovo and its subsidiary. The Information has now published new reporting indicating that Apple is evaluating the production of dedicated AI servers built around its M8 chip. Crucially, the company is exploring support for NVIDIA’s NVLink Fusion connectivity solution, which enables XPUs to integrate into NVIDIA’s rack-scale AI systems. Adopting this architecture would allow Apple to market these servers to enterprise clients, marking a return to the data center hardware segment it exited in 2011 following the discontinuation of the Xserve line.

Under Apple’s updated AI architecture, a central orchestrator gathers necessary data and routes each Siri query either to on-device models or to the cloud. In cloud environments, encrypted NVIDIA GPUs and Apple’s Private Compute protocols enforce strict user privacy standards. For a detailed breakdown of how Apple’s Private Compute protocols now leverage NVIDIA GPUs, Intel CPUs with TDX, and Google’s Titan chip, readers can refer to our dedicated analysis.

Writing is his one incontrovertible passion. A graduate of the University of Toronto’s Rotman Commerce Program, he has spent the past six years writing extensively on financial and technology sectors, producing over 2,200 articles totaling nearly one million words. He has served on the Wccftech mobile team since 2025. Outside of journalism, he travels globally, seeking out independent confectioneries and restaurants as an aspiring food connoisseur. Readers can follow Wccftech on Google News to stay updated on further coverage.