October 28, 2025
In a significant move to accelerate the development of AI infrastructure, Cisco Systems and Nvidia have announced a suite of new products and integrations, headlined by the first data center switch built on Nvidia's Spectrum-X Ethernet networking platform.
The announcements, made at the Nvidia GTC conference in Washington, D.C., encompass advancements in data center switching, AI security, and future mobile networks. The centerpiece is the new Cisco Nexus 9100 (N9100) series switch, the first of its kind to utilize Nvidia Spectrum-X Ethernet silicon. Designed for AI data centers, the switch is scheduled for availability before the end of the year and will offer customers a choice between Cisco's NX-OS or the open-source SONiC operating systems, providing greater flexibility for neocloud and sovereign cloud deployments.
Built upon the N9100, Cisco is delivering a reference architecture compliant with the Nvidia Cloud Partner program. This integration creates a uniform operating model for the entire Nexus data center switch portfolio, managed via the Cisco Nexus Dashboard and incorporating Silicon One, cloud-scale ASICs, and the new Spectrum-X technology.
Jeetu Patel, Executive Vice President and General Manager of Security and Collaboration at Cisco, emphasized the need for new architectural approaches. "The infrastructure that will power the agentic AI applications and innovation of the future requires new architectures designed to overcome today’s constraints in power, computing, and network performance," Patel stated.
Enhanced Security for AI Factories
The collaboration also expands on the Cisco Secure AI Factory, initially announced in March. This initiative is gaining new security and observability capabilities. A key integration now links Cisco AI Defense with Nvidia NeMo Guardrails to bolster the cybersecurity of AI applications. This solution is available for on-premises deployment, allowing security and AI teams to protect models and applications directly. Furthermore, the Splunk Observability Cloud will provide teams with tools to monitor the performance, quality, security, and cost of their AI application stacks.
On the infrastructure front, Cisco's Isovalent technology has been validated for inference workloads on AI PODs, enabling high-performance Kubernetes networking. The new cloud-managed Cisco G200 Silicon One switch, which offers high-density 800G Ethernet, is now available as a deployment option within AI PODs. These PODs also feature the Cisco UCS 880A M8 rack servers with Nvidia HGX B300 and Cisco UCS X-Series modular servers with Nvidia RTX PRO 6000 Blackwell Server Edition GPUs, supporting high-performance workloads like generative AI fine-tuning and inference.
Pioneering AI-Native Wireless for 6G
In a forward-looking move for telecommunications, Cisco, Nvidia, and telecom partners have developed what they term the first U.S. AI-RAN stack for mobile networks. This stack integrates sensing and communication functions, with multiple pre-6G applications being demonstrated at the conference. It is designed to allow telecom providers to integrate AI into their mobile networks, beginning with 5G Advanced services, while simultaneously laying the groundwork for 6G. The stack combines Cisco’s user plane function and 5G core software with the Nvidia AI Aerial platform, creating a foundation for "physical AI" and integrated sensing with high efficiency and security.
This development is critical as wireless networks face increasing pressure from the proliferation of connected devices, including AR glasses, autonomous vehicles, and robotics, which require support for billions of simultaneous connections.
Ecosystem Expansion with New Partners
The ecosystem around the Cisco Secure AI Factory is also expanding. Nvidia Run:ai software is now available through Cisco and its partners for intelligent AI workload and GPU orchestration. The Nutanix Kubernetes Platform has been added as a supported platform, with Nutanix Unified Storage and Nutanix Enterprise AI serving as validated storage and software components to simplify containerized inference services.
Additionally, the two companies are collaborating on the Nvidia AI Factory for Government, an end-to-end reference design tailored for AI workloads in highly regulated environments. A separate reference architecture for neocloud and sovereign cloud customers includes the Cisco 8223 switch based on Silicon One P200, Nvidia BlueField-4 DPUs, and Nvidia ConnectX-9 SuperNICs.
SOURCE techzine
