Cisco unveiled a sweeping set of AI infrastructure innovations with NVIDIA at GTC Washington, D.C., introducing the Cisco N9100 data center switch—the first NVIDIA Cloud Partner (NCP)-compliant architecture from a networking vendor. Built on NVIDIA Spectrum-X Ethernet silicon, the N9100 extends Cisco’s Nexus family to serve neocloud, sovereign cloud, enterprise, and telecom operators deploying AI at scale. Cisco also launched a new Cloud Reference Architecture for AI infrastructure based on Cisco Silicon One and NVIDIA technologies, along with updates to its Secure AI Factory and a role in the industry’s first AI-native wireless stack for 6G.
For data center operators, the N9100 series provides up to 51.2 Tbps of bandwidth and the flexibility to run either Cisco NX-OS or SONiC within a unified Nexus operating model. It integrates with Cisco Nexus Dashboard and supports NVIDIA ConnectX-9 SuperNICs and BlueField-4 DPUs, enabling high-performance AI clusters across thousands of GPUs. The new Cisco Cloud Reference Architecture extends NVIDIA’s Cloud Partner design principles to Cisco Silicon One-based switches, including the recently introduced 8223 model using the P200 ASIC. Cisco also expanded its Nexus Hyperfabric AI platform, designed to cut AI cluster deployment time from months to weeks through guided design, cabling automation, and on-site installer workflows.
For enterprise AI deployments, Cisco’s Secure AI Factory with NVIDIA adds tighter integration between Cisco AI Defense and NVIDIA NeMo Guardrails, alongside observability enhancements via Splunk. Cisco’s Isovalent platform now supports inference workloads on AI PODs, and new UCS servers with NVIDIA HGX B300 and RTX PRO 6000 Blackwell GPUs are available. Meanwhile, Cisco joined NVIDIA and telecom partners to debut the first AI-native wireless stack for 6G—an AI-RAN framework combining Cisco’s 5G core with NVIDIA AI Aerial, integrating sensing and communications for next-generation networks.
• Cisco N9100 series delivers 51.2 Tbps with NX-OS or SONiC support
• New NCP-compliant Cloud Reference Architecture unifies Silicon One and Spectrum-X networks
• Cisco Secure AI Factory integrates NVIDIA NeMo Guardrails for AI model protection
• Nexus Hyperfabric AI accelerates AI cluster rollout with automation and pre-validated design
• Cisco joins NVIDIA and partners in developing the first AI-native wireless stack for 6G
“We’re at the beginning of the largest data center build-out in history,” said Jeetu Patel, President and Chief Product Officer at Cisco. “Together, Cisco and NVIDIA are leading the way in defining the technologies that will power these AI-ready data centers in all their varieties—from emerging neoclouds to global service providers to enterprises and beyond.”
🌐 Analysis: Cisco’s latest moves deepen its alignment with NVIDIA across the AI stack, from Ethernet fabrics to AI-RAN architectures. The N9100 series brings Spectrum-X to Cisco’s Nexus ecosystem, offering hyperscalers and sovereign clouds an alternative path to high-performance Ethernet AI networking. With competitors such as Arista and Juniper also advancing AI-optimized platforms, Cisco’s dual-silicon strategy—combining Spectrum-X and Silicon One—positions it to address diverse AI infrastructure demands across data center, edge, and telecom domains.
🌐 We’re tracking the latest developments in AI infrastructure and networking silicon. Follow our ongoing coverage at: https://convergedigest.com/category/semiconductors/