[News] Apple Reportedly Eyes 2029 Enterprise Server Return With M8 Ultra, NVIDIA NVLink

Apple is reportedly considering a return to the enterprise server market as it seeks to extend its in-house silicon strategy into AI infrastructure. According to Reuters, citing The Information, the company is exploring an AI server that would combine its own chips with NVIDIA’s NVLink Fusion. The system could feature either two or four planned M8 Ultra chips, though it is not expected to launch before 2029 and could still be canceled or ultimately arrive without NVIDIA technology.

Unlike Apple’s existing server infrastructure, the proposed system would be sold to outside customers, potentially bringing the company back into a market it exited when it discontinued Xserve in 2011, MacRumors notes. Apple is reportedly targeting companies looking to run AI models on their own hardware.

The project would also extend Apple’s custom silicon strategy beyond consumer devices and into the rapidly growing AI inference market, Reuters notes. The effort reportedly began about a year ago with the backing of John Ternus, who was Apple’s hardware chief at the time and is now CEO.

The potential server push comes as Apple hardware gains traction among AI developers. TechPowerUp notes that Apple’s Mac lineup generated a record US$10.4 billion in sales last quarter, while some enterprises are building custom racks of multiple Mac Studios for local AI inference. Ars Technica, citing another report from The Information, adds that OpenAI has bought “tens of thousands” of Mac mini and Mac Studio systems for reinforcement learning and AI agent training, while Anthropic has rented Mac minis through AWS.

Apple is also developing dedicated AI server chips, codenamed Baltra, using Broadcom technology, with deployment targeted by 2027. The servers are intended to power cloud-based Apple Intelligence features, according to Bloomberg.

Why Apple Is Reportedly Considering NVIDIA NVLink for Its AI Servers

Apple is reportedly considering NVIDIA’s NVLink as it explores how to connect multiple chips in the proposed AI server. The company already builds server hardware for Private Cloud Compute, which handles Apple Intelligence workloads that require processing beyond users’ devices. However, scaling Apple’s existing chip-to-chip connections could create cost and speed constraints, leading the company to consider NVIDIA’s technology, Reuters notes.

An agreement would put NVIDIA technology inside an Apple product that could ultimately compete with NVIDIA’s own AI systems, MacRumors notes. Through NVLink Fusion, NVIDIA can provide networking technology to companies building systems around alternative processors, allowing it to participate in AI infrastructure even when customers are not using NVIDIA chips.

The discussions also point to closer cooperation between Apple and NVIDIA following years of disputes between the companies. Apple has already announced plans to extend Private Cloud Compute to Google Cloud using NVIDIA GPUs, according to MacRumors. The two companies have also reportedly explored other areas of AI collaboration, including the use of NVIDIA’s open-source models.

Read more

• [News] Inside iPhone 18 Pro and Duo: A20 Pro Debuts as Apple’s 1st 2nm Chip in iPhone, Duo Trades RAM for Storage
• [News] NVIDIA’s DLSS 5 Could Boost Graphics DRAM Demand, With GDDR6 Prices Reportedly Already Triple

(Photo credit: FREEPIK)