Rumor

Nvidia to utilize Intel Xeon 6 processors as host CPUs for Rubin NVL8 AI systems

Monday, March 16, 2026 at 06:35 PM

Nvidia is reportedly selecting Intel Xeon 6 processors to serve as the host CPUs for its next-generation DGX Rubin NVL8 AI systems. This hardware configuration choice is separate from the formal strategic partnership the two companies established in 2025, signaling a specific supply chain decision for the upcoming Rubin platform architecture.

Context

At the GTC 2026 conference, Intel and Nvidia announced that Intel Xeon 6 processors will serve as the host CPUs for the upcoming Nvidia DGX Rubin NVL8 systems. This decision integrates the Xeon 6776P processor to manage orchestration and data movement for the Rubin architecture, which was first unveiled at CES 2026. While Nvidia produces its own Vera ARM-based CPUs for proprietary superchips, it continues to utilize Intel's x86 architecture for the NVL8 platform to provide enterprises with balanced performance and a mature software ecosystem. The DGX Rubin NVL8 is a high-density AI system featuring eight Rubin GPUs and 2.3 TB of total memory, delivering up to 400 petaFLOPS of inference performance. The inclusion of Xeon 6 is critical for maintaining high throughput, offering 3X higher memory bandwidth than previous generations to prevent bottlenecks. Rubin-based products are expected to enter full production with availability through partners in the second half of 2026, targeting large-scale agentic AI and reasoning workloads.

Related Companies

Nvidia
Nvidia
NVDA
US
Intel
Intel
INTC
US