AI is transforming every industry—redefining how we innovate, compete, and create value. This shift demands a new class of data center infrastructure: one that’s high-performance, secure, and built to adapt to constantly changing workloads. The network is at the heart of this transformation—it must evolve to match AI’s pace, enabling organizations to unlock new efficiencies, insights, and opportunities.
The reality of AI networks
As organizations scale their AI environments, they’re discovering that traditional networks simply weren’t built for the job. Managing east-west GPU-to-GPU traffic for AI training—while also supporting frontend, storage, and management traffic—requires a fundamentally new approach.
Customers are asking for a unified operating model that spans on-premises and cloud, automation that works across open and integrated systems, and platforms tuned from silicon to software for AI-scale performance. The common Nexus operating model—combined with Nexus Dashboard and Nexus Hyperfabric—simplifies operations across diverse environments, giving network operators a single, consistent way to manage, automate, and gain visibility across all their networks.
What’s needed is a network designed to scale massively, adapt seamlessly to mixed workloads, and ensure consistent performance—exactly what our new architectures deliver.
Building future-ready AI network infrastructure requires more than performance—it requires a robust operating model. I’m thrilled to announce the expansion of our Cisco Nexus portfolio and our partnership with NVIDIA.
Introducing Cisco N9100 Series Switches
Building on its February 2025 announcement on deepening its NVIDIA partnership to accelerate AI adoption, Cisco is expanding its Nexus data center networking portfolio with the launch of the Cisco N9100 Series Switches.
The Cisco N9100 Series augments the broader Cisco Nexus portfolio, which includes switches powered by Cisco Silicon One and Cloud Scale silicon, by incorporating NVIDIA Spectrum-X Ethernet switch silicon—all unified under the Nexus operating model for seamless integration with Cisco Nexus Dashboard.
With the N9100 Series, customers also get the choice of operating systems—Cisco’s proven NX-OS for enterprise-grade reliability and operational consistency, or SONiC for organizations that prefer open networking standards with automation and operations with Nexus Dashboard.
Reference architectures for ease of deployment
To support the evolving needs of AI infrastructure, Cisco combines high-performance hardware and unified operations with proven reference architectures. This comprehensive approach ensures that customers can scale, adapt, and operate with confidence.
Cisco Cloud Reference Architecture with Silicon One
Our Cisco Silicon One platform remains the cornerstone for customers requiring maximum architectural flexibility. With industry-leading scale and programmability, Silicon One delivers unmatched performance across diverse workloads. Cisco Cloud Reference Architecture (CRA) with Nexus Dashboard and Nexus Hyperfabric AI showcases scalable networking for large-scale clusters. Nexus Hyperfabric AI, now generally available, simplifies and accelerates AI journey for customers with an easy start operational experience for deploying and managing full-stack AI clusters. It leverages Cisco Silicon One’s programmability to enable NVIDIA Spectrum-X Ethernet technology on Cisco switches that adhere to NCP design tenants for neocloud and sovereign cloud customers.
“We’re proud to partner with Cisco to bring next-generation networking innovations into our AI-native framework,” states Tarun Raisoni, CEO and Co-Founder of Gruve. “Gruve helps enterprises build the infrastructure and intelligence layer that powers enterprise AI adoption. Cisco’s 800G Silicon One and NVIDIA Spectrum-X silicon-based Nexus switches set a new standard for performance, scalability, and programmability, enabling us and the ecosystem to scale the foundation of modern AI infrastructure. Together, we’re building the intelligent, efficient, and future-ready connectivity that defines the era of AI.”
NCP reference architecture compliance with N9100
Cisco will offer an NVIDIA Cloud Partner (NCP) compliant reference architecture with the new Cisco N9100 Series Switches for neocloud and sovereign cloud customers. These switches combine the strengths of NVIDIA AI technology with Cisco enterprise-grade reliability and streamlined management. Orderable before the end of 2025, it is generally available early in 2026.
“BlueSky Compute is excited to be one of the first neoclouds to deploy Cisco’s N9100 Series Switches for the scale-out fabric in our AI training clusters—powering our vision to turn AI into ROI for the world’s enterprises by building larger, more interconnected B300 clusters, faster. Cisco’s NCP compliant reference architecture, based on N9100 Series Switches while maintaining the operational simplicity with Nexus Dashboard, is a game-changer.”—Ian Hartley, CEO, BlueSky Compute
Cisco and NVIDIA: Powering the next wave of AI
Cisco and NVIDIA’s partnership continues to accelerate as we work together to empower AI at scale—securely and at levels that can be observed, managed, and trusted. The advancements announced today with the Cisco N9100 Series Switches that are NCP RA compliant and the Cisco CRA with Nexus Dashboard and Nexus Hyperfabric AI are a testament to Cisco’s relentless pursuit of innovation to drive AI adoption across enterprises, neoclouds, and sovereign clouds. Cisco Secure AI Factory with NVIDIA encompasses these innovations to deliver comprehensive framework where compute, storage, security, and observability work in concert.
By building on a flexible, scalable, and secure network foundation, you unlock new opportunities for AI innovation—enabling your organization to realize its full potential of observable and future-ready AI success.
Discover how the Cisco N9100 Series Switches power AI-scale data centers.
Resources:
Cisco AI Networking in Data Centers
Cisco Nexus 9000 Cloud Partner Reference Architecture
Cisco Hyperfabric AI Cloud Partner Reference Architecture
Engineering Leadership Blog: Cisco drives AI networking innovation with NVIDIA