In February of this 12 months, Cisco introduced an expanded partnership with NVIDIA to mix two of probably the most extensively used networking portfolios: Cisco Silicon One and NVIDIA Spectrum-X Ethernet. This collaboration aimed to offer enterprises with unparalleled flexibility and choices for his or her AI information heart investments.
I’m delighted to announce three vital milestones in our partnership:
- The Cisco Cloud Reference Structure, that includes switches primarily based on Cisco Silicon One with the add-on license for deeper integration with NVIDIA Spectrum-X Ethernet know-how.
- A Cisco NVIDIA Cloud Accomplice(NCP)-compliant Reference Structure, constructed on the NVIDIA Spectrum-X Ethernet silicon–powered Cisco N9100 Collection switches, with early orderability in November 2025.
- The orderability in November 2025 of Cisco Nexus Hyperfabric AI.
These milestones spotlight the power of our partnership and show our dedication to innovation and excellence in delivering cutting-edge options to our clients.
Cisco Cloud Reference Structure
The brand new Cisco Cloud Reference Structure is predicated on NCP design tenets for Cisco Silicon One primarily based switches with an add-on license for deeper integration with NVIDIA Spectrum-X Ethernet know-how.
This add-on license permits the NVIDIA Spectrum-X adaptive routing functionality to work seamlessly with Cisco Nexus 9300 Collection switches and NVIDIA Spectrum-X Ethernet SuperNICs. The mixture of low latency and congestion-aware per-packet load balancing on Cisco 9300 switches, together with out-of-order packet dealing with and end-to-end congestion administration on NVIDIA SuperNICs, considerably enhances community efficiency. These enhancements are important for AI networks, optimizing crucial metrics corresponding to job completion time.
Cisco NVIDIA Cloud Accomplice Compliant Reference Structure
With the Cisco N9100 Collection, we now introduce to the market a NVIDIA Cloud Accomplice(NCP)-compliant reference structure. This growth is especially vital for neocloud and sovereign cloud clients constructing information facilities with capacities starting from hundreds to probably a whole bunch of hundreds of GPUs, because it permits them to diversify their provide chains successfully.
The Cisco N9100 Collection delivers a formidable 51.2 Tbps of bandwidth, guaranteeing distinctive scalability, efficiency, and safety. With the addition of NVIDIA Spectrum-X Ethernet silicon alongside Cisco Silicon One and Cisco Cloud Scale ASICs, Cisco affords a wider vary of swap silicon choices. This enables clients to standardize on a unified AI networking platform that helps each Cisco Silicon One and NVIDIA Spectrum-X Ethernet silicon, streamlining administration via the Cisco Nexus Dashboard. Moreover, customers have the pliability to decide on between totally different Community Working Techniques, together with Cisco NX-OS or the open-source SONiC.
The Cisco N9100 Collection empowers organizations to effectively deploy each AI backend and frontend networks, successfully assembly the demanding necessities of recent AI workloads
Cisco Nexus Hyperfabric AI
Our clients have shared that they really feel stress to deploy bigger AI clusters, and they’re looking for help from infrastructure companions like Cisco. We developed Cisco Nexus Hyperfabric AI resolution particularly to assist IT groups shorten their time to worth realization and simplify the supply and operation of advanced AI infrastructure.
Since its preliminary announcement in June 2024, many purchasers and companions have expressed curiosity in its capabilities, amongst them: Wake Forest College, Sharon AI and Gruve.
“We’re thrilled to take part within the Cisco Nexus Hyperfabric AI pilot, made attainable via our collaborative partnerships with Cisco, VAST Knowledge, and MCNC. This cutting-edge know-how permits us to discover new methods for Wake Forest College school and college students to entry important instruments that may considerably speed up their analysis, advance AI initiatives, and result in groundbreaking discoveries.”
— Adam Carlson, Assistant Director of Analysis Computing”Sharon AI is worked up to check Cisco Nexus Hyperfabric AI as a result of it affords a turnkey, cloud-managed, full-stack AI infrastructure resolution that simplifies deployment and accelerates AI innovation, completely aligning with our mission to energy sovereign AI options and AI factories.”
— Dan Mons, Chief Expertise Officer at Sharon AI“As a Cisco associate, Gruve views Cisco Nexus HyperFabric AI as a serious step ahead in simplifying AI infrastructure—integrating compute nodes, material interconnects, and VAST storage right into a cohesive, scalable framework. Our partnership with Cisco permits us to speed up buyer deployments, improve throughput, and guarantee safe, high-performance AI operations.”
— Rajeev Khanolkar, Chief Technique Officer of Gruve, Inc
As clients construct their very own AI clusters on premises, they take care of an rising quantity of complexity as they should handle a number of programs with specialised know-how stacks for which they don’t essentially have the suitable sources with the suitable expertise.
To simplify and speed up their AI journey, Cisco is reworking the operational expertise for deploying and managing AI clusters. Many purchasers put money into massive GPU clusters however usually spend months getting them absolutely operational, which may result in a poor return on funding.
Cisco Nexus Hyperfabric AI goals to scale back deployment time from a number of months to only a few weeks and even much less. By planning the material format prematurely (determine 1), creating detailed cabling plans (determine 2), and providing on-line guided help for installers (determine 3), we will considerably save time through the deployment course of.
Figure1: Design an NVIDIA Enterprise Reference Structure in one-click from a template (no guesswork)
Determine 2: Complicated interconnect cabling format is robotically generated and auto-documented by the Cisco Nexus Hyperfabric controller (no extra spreadsheets)
Determine 3: Day-1 Deployment is automated utilizing Cisco Nexus Hyperfabric’s on-site interface, which guides installers via all of the steps of deployment.
This elevated effectivity instantly interprets to a shorter time to first token and the simplified processes for deploying advanced AI clusters encourage adoption amongst clients who could not have the required experience or sources.
These three milestones mark an thrilling new chapter within the Cisco-NVIDIA partnership and supply a basis for patrons to fulfill the demanding community necessities of AI reasoning and different inference workloads.
