Intel's partnership with NVIDIA has finally begun to formalize with the recent announcement, as the company's Xeon 6 CPU series is being integrated into Rubin systems.
Intel's Xeon 6 Server CPU Gets Integrated Into NVIDIA's Baseline Rubin Offering; Larger Rack Adoption Remains Uncertain For Now
With today's agentic workloads, CPUs have become the next area of focus for hyperscalers and manufacturers like NVIDIA, as functions such as "orchestration, memory access, and model security" have become increasingly dominant. We did talk about Intel attending NVIDIA's GTC this year as a key partner, and with that, the company's recent announcement focuses on the Rubin DGX NVL8, which includes Xeon 6 CPUs, in a similar fashion to what we witnessed with the DGX B300 and the DGX H200.
Based on what Intel has disclosed, the Xeon 6 CPUs will complement inference with DGX NVL8, allowing NVIDIA to optimize TCO by essentially offloading some portion of the agentic workloads towards the host x86 CPU. Intel claims that with Xeon 6, they are bringing in several benefits to NVIDIA's offerings, and key ones include:
- Efficient performance per watt
- Optimized support throughout the ecosystem AI software stack, including new support for NVIDIA Dynamo to enable heterogeneous inference across CPU and forthcoming GPUs
- Proven reliability across mission critical environments
- Superior orchestration of GPU accelerated, heterogeneous systems
Intel hasn't specified the Xeon 6 CPU SKUs expected to be integrated into NVL8 Rubin systems, but based on what we have seen with the Blackwell series, the most likely option is the Xeon 6776P. This particular model features an all-P-core configuration with 64 cores and 128 threads, a base clock of 2.30 GHz, and support for "industry-leading" PCIe 5.0 lanes and MRDIMM. The integration of Granite Rapids also indicates that it is one of the more dominant choices for adoption, not just among hyperscalers but also among companies like NVIDIA.
We were hoping to see wider adoption, with Xeon CPUs integrated into NVL72 racks, but it appears NVIDIA has confined them to NVL8 for now. Since NVIDIA and Intel are also developing a joint x86 server CPU solution, it appears we could see that particular model gaining wider adoption within NVIDIA's rack-scale infrastructure, but for now, Xeon 6 is limited to the base Rubin offering.
About the author: Muhammad Zuhair is a hardware and technology reporter for Wccftech, specializing in the semiconductor industry and the complex interplay between technology, manufacturing, and geopolitics. His coverage focuses on the corporate strategies and technological roadmaps of industry giants like TSMC, NVIDIA, Samsung, and Intel. Zuhair's expertise lies in deconstructing complex topics such as fabrication nodes (e.g., 2nm process), the economic impact of policies like the CHIPS Act, and the strategic development of AI infrastructure from NVIDIA, AMD and Intel.
Follow Wccftech on Google to get more of our news coverage in your feeds.
