Cisco and NVIDIA announced an expanded alliance for strategic development. The goal of this partnership is to give businesses unmatched options and flexibility for their investments in AI data centers.
The Cisco Cloud Reference Architecture, which includes Cisco Silicon One-based switches with an add-on license for more advanced NVIDIA Spectrum-X Ethernet technology integration.
Based on the NVIDIA Spectrum-X Ethernet silicon-powered Cisco N9100 Series switches, this Cisco NVIDIA Cloud Partner (NCP)-compliant Reference Architecture will be available for early order in November 2025.
Cisco Nexus Hyperfabric AI’s order will be available in November 2025. These developments show the strength of collaboration and dedication to quality and creativity in providing clients with state-of-the-art solutions.
With an add-on license for further integration with NVIDIA Spectrum-X Ethernet technology, the new Cisco Cloud Reference Architecture is built on the principles of NCP design for Cisco Silicon One-based switches.
The NVIDIA Spectrum-X adaptive routing capabilities may be smoothly integrated with NVIDIA Spectrum-X Ethernet SuperNICs and Cisco Nexus 9300 Series switches due to this add-on license.
Network performance is greatly improved by the combination of NVIDIA SuperNICs’ end-to-end congestion control and out-of-order packet handling with Cisco 9300 switches’ low latency and congestion-aware per-packet load balancing. These enhancements, which optimize vital parameters like job completion time, are crucial for AI networks.
We now release a reference architecture that complies with NVIDIA Cloud Partner (NCP) standards with the Cisco N9100 Series. Because it enables companies to efficiently diversify their supply chains, this breakthrough is especially important for neocloud and sovereign cloud clients constructing data centers with capacities ranging from thousands to possibly hundreds of thousands of GPUs.
