Cisco has launched its latest Silicon One chip, the P200, powering the new Cisco 8223 fixed system aimed at high-scale networking for Artificial Intelligence workloads. Positioned as the company’s most optimized routing platform for Artificial Intelligence networking, the 8223 is designed to enable interconnect bandwidth scale of more than 3 exabits per second, according to Cisco. The company says the system is the only 51.2-terabit-per-second Ethernet fixed router currently on the market oriented to the intense traffic between data centers supporting Artificial Intelligence.
The rollout follows rising pressure on data centers from Artificial Intelligence adoption. Jeetu Patel, Cisco’s president and chief product officer, said the industry is facing an “infrastructure constraint,” citing insufficient power, compute capacity and network bandwidth to satisfy Artificial Intelligence growth. Martin Lund, executive vice president of Cisco’s Common Hardware Group, added that organizations are relocating data centers to more power-abundant locations, which in turn demands unprecedented bandwidth to connect massive training clusters over large distances. He cautioned that long-running training jobs can last months and any network downtime risks idling billion-dollar facilities.
At the core of the 8223 is Cisco Silicon One P200 routing silicon, the latest entry in a portfolio the company introduced six years ago and says is deployed by five of the top six hyperscalers. Cisco highlights power efficiency, scalability and security as key design goals for the platform, describing it as purpose-built for scale-across networking demands tied to Artificial Intelligence. The 8223 is available now as a fixed system, while the P200 chip will also be deployable in modular platforms and disaggregated chassis. Cisco said its Nexus portfolio will support systems running NX-OS based on the P200 in the near future.
Cisco frames the 8223 and P200 as tools to reduce the networking bottlenecks that can stall large-scale Artificial Intelligence efforts. By pairing high-throughput Ethernet routing with a unified Silicon One architecture, the company is targeting the dual challenges of moving enormous datasets across distributed training sites and sustaining reliability over long-duration jobs. With Artificial Intelligence pushing data center operators to rethink locations, capacity planning and interconnect design, Cisco’s latest silicon and system are pitched as a way to expand bandwidth headroom while tightening operational resilience.
