Intel Confirms Xeon 6 SoCs Powers NVIDIA DGX Rubin NVL8 Systems for AI Inference Push

At the NVIDIA GTC 2026, Intel officially announced that its Xeon 6 processors will power NVIDIA’s DGX Rubin NVL8 systems. The company is positioning Xeon as an integral part of modern AI infrastructure that can handle orchestration, scaling, and security for GPU-heavy systems. Earlier this year, at MWC 2026, Intel positioned the Xeon 6 SoCs as the foundation for AI-ready 5G and the path to 6G.

Intel Xeon 6 confirmed as host CPU for DGX Rubin NVL8 systems

In the announcement post, Intel confirmed that Intel Xeon 6 will help manage memory access, workload distribution, and overall system coordination. The company further highlighted that as AI moves toward real-time inference and agentic systems, the role of the CPU becomes increasingly important in keeping GPU performance efficient and stable.

Not to forget, Xeon 6 also brings support for up to 8TB of system memory, along with significantly higher memory bandwidth using MRDIMM technology. It also features PCIe 5.0 lanes, which allow for faster communication between GPUs and other components. Notably, the company is also pushing its efficiency claims, noting better performance per watt and balanced output across mixed workloads.

As far as platform capabilities are concerned, Xeon 6 is designed to support a wide AI software ecosystem, including new support for NVIDIA Dynamo for heterogeneous inference. The processor also focuses on reliability in enterprise environments, with improved orchestration of GPU-accelerated systems.

Intel has also focused on security, as it confirmed integration of features like Trust Domain Extensions and encrypted CPU-to-GPU data paths to protect sensitive AI models and workloads during operation. These additions may appeal to enterprises working with confidential data.

NVIDIA’s announcement from GTC 2026

It’s worth noting that NVIDIA has also unveiled DLSS 5, which it claims will make games look more cinematic. The company has also collaborated with Adobe to build next-generation Firefly AI models. Moreover, the company also announced that its latest computing platforms are bringing AI to orbital data centers and space operations. The AI chip giant unveiled these alongside the new Vera CPU and NemoClaw.

Article feature image source: Intel