Rumor
Intel and NVIDIA to collaborate on custom Xeon with NVLink for AI host nodes
Friday, January 23, 2026 at 10:50 AM
Intel is reportedly collaborating closely with NVIDIA to develop a custom Xeon processor that integrates fully with NVIDIA's NVLink technology. This collaboration aims to provide top-tier x86 performance for AI host nodes. The announcement suggests a strategic partnership to enhance AI infrastructure.
Context
Intel and Nvidia are collaborating to develop a custom Xeon processor that natively integrates Nvidia’s proprietary NVLink interconnect. This marks a pivotal strategic shift, as NVLink—the high-speed fabric essential for modern AI clusters—was previously exclusive to Nvidia’s own ARM-based Grace CPUs. By porting this technology to the x86 architecture, the two companies aim to capture the massive enterprise market requiring Intel’s legacy compatibility alongside Nvidia’s industry-leading AI acceleration.
This partnership secures Intel’s relevance in the high-growth AI server market while significantly expanding Nvidia’s hardware footprint. The custom silicon is expected to support bandwidth speeds of 1.8 TB/s, neutralizing the data bottlenecks that often plague massive training clusters. Shipments are projected to begin in late 2025, targeting the 2026 global data center refresh cycle. This alliance reinforces the dominance of both firms against the rising threat of in-house silicon developed by major hyperscale cloud providers.
Related Companies
Nvidia
NVDA
Intel
INTC