IN Brief:
- AMD’s Ryzen Embedded V3000 is being used in Cisco N9300 switching and 8000 routing platforms.
- The processor handles control-plane workloads including routing, telemetry, automation, and security policy functions.
- AI network buildouts are increasing demand not only for switching silicon, but also for embedded compute around the fabric.
AMD’s Ryzen Embedded V3000 processor family is powering control-plane functions in Cisco’s latest N9300 Series switches and 8000 Series service-provider routers, extending the reach of embedded x86 deeper into the networking hardware now being built around large-scale AI infrastructure. The announcement links AMD’s embedded platform to Cisco systems designed for dense data-centre fabrics, carrier routing, and AI-driven network automation, where switching silicon may take the headlines but surrounding compute still determines how the system behaves.
In Cisco’s N9300 switching line, the V3000 is being used alongside Silicon One-based ASIC platforms to manage control-plane workloads such as routing protocols, traffic policy, configuration, telemetry, automation, and security enforcement. Cisco’s broader AI networking push has centred on Silicon One, including the 102.4Tbps G300 architecture and newer 800G and 1.6T-ready systems, but the control plane still needs a processor that can deal with real-time analytics, orchestration, and management without blowing out thermal or space budgets. That is where the embedded CPU comes in.
The same applies on the routing side. Cisco’s 8000 Series routers are built for service-provider scale, with Silicon One routing ASICs handling the data plane and the embedded processor supporting management, automation, and larger routing-table and policy tasks around it. AMD says the V3000 family offers four, six, or eight x86 cores and up to 96GB of DDR5 memory, giving Cisco more headroom for control-plane compute while maintaining the sort of power efficiency and lifecycle support that operators expect from network infrastructure designed to stay in the field for years.
The pairing is a reminder that networking hardware is no longer a simple story of one big chip and some optics. AI clusters are placing heavier demands on every layer around the fabric: telemetry that can expose congestion before jobs stall, orchestration systems that can manage thousands of links and workloads, encryption that runs at line rate, and operating models that have to span cloud, enterprise, and service-provider environments. Cisco has already been arguing that the network is becoming part of the compute system itself, and the architecture backs that up. The more intelligent the fabric becomes, the more important the surrounding embedded compute becomes as well.
That is good news for embedded processor suppliers, but it also changes the design brief. In network equipment, long-term availability, software ecosystem breadth, thermal efficiency, and predictable behaviour still matter as much as peak throughput. The appeal of a mature x86 embedded platform is not difficult to understand in that context. It offers familiar software compatibility, solid control-plane performance, and enough flexibility to support multiple product families without forcing a fresh software migration each time the hardware is updated. For OEMs building around long service lives, that can be worth more than chasing novelty for its own sake.
The wider backdrop is AI infrastructure expansion beyond the hyperscaler tier. Cisco has been pitching newer Silicon One systems at enterprises, neoclouds, sovereign deployments, and service providers, with claims around faster job completion, denser optics, and sharply better energy efficiency in liquid-cooled systems. Whether all of that translates evenly across the market remains to be seen, but one point is already clear: AI networking requires more than raw packet-moving silicon. It also needs embedded compute that can support telemetry, management, policy, and automation at scale. AMD’s role in these Cisco platforms reflects that shift. In the AI era, the network box is becoming a more layered embedded system in its own right.
