Cisco Systems has launched a networking chip, the P200, and its new 8223 router, specifically designed to meet the massive connectivity and power demands of modern AI data centers. Major cloud computing players, including Microsoft and Alibaba, have already signed on as early customers, signaling strong industry adoption.
Cisco’s P200 chip sits at the core of the 8223 router, enabling data centers spread across hundreds of miles to operate as a single, unified computing system. This innovation addresses one of AI’s biggest challenges: connecting sprawling data centers efficiently while minimizing power consumption. According to Cisco, the P200 replaces what previously required 92 separate chips with just one, and the resulting router uses 65 percent less power than comparable devices.
Martin Lund – Executive Vice President, Cisco’s Common Hardware Group, highlighted that AI compute is outgrowing the capacity of even the largest data centers. Cisco 8223, powered by the P200 chip, provides the bandwidth, scale, and security needed for connecting multiple data centers over long distances.
Revolutionizing AI Data Center Connectivity
As AI workloads grow exponentially, data centers must distribute processing across multiple locations. This “scale-across” approach demands ultra-fast, reliable, and secure interconnects. The Cisco 8223, powered by the P200, delivers 51.2 terabits per second (Tbps) of Ethernet bandwidth, making it the first fixed router designed for AI-scale networking. Its deep buffering capabilities prevent network slowdowns and data loss, ensuring seamless AI training across multi-site infrastructures.
Power Efficient, Scalable, and Secure
The Cisco 8223 is optimized for power efficiency, handling surging AI workloads with minimal energy consumption. As a 3RU system, it is also the most space-efficient fixed router of its kind. With support for 64 ports of 800G and scalability up to 3 Exabits per second, the 8223 enables high-density, high-performance data center interconnects. Its programmable P200 silicon allows networks to adapt to new protocols and AI workloads without costly hardware upgrades.
Future-Ready AI Infrastructure
With AI compute outgrowing the capacity of even the largest data centers, the Cisco 8223 and P200 chip are poised to become essential components of next-generation AI networking. The system combines scalability, flexibility, programmability, and security, allowing organizations to build future-ready data centers that meet the demands of the AI era.
About Cisco Silicon One
The P200 chip is part of Cisco’s Silicon One portfolio, which provides a unified, programmable networking architecture across AI, hyperscaler, enterprise, and service provider environments. Introduced in 2019, Silicon One has become a critical backbone for global networks, and the P200 continues this legacy by delivering unmatched bandwidth, efficiency, and reliability for AI workloads.
Industry Leaders Back Cisco’s Innovation
Dave Maltz – Technical Fellow and Corporate Vice President, Azure Networking, Microsoft, praised the P200’s innovation, noting its ability to absorb bursts of data and expand AI networking capabilities.
Dennis Cai – Vice President, Head of Network Infrastructure, Alibaba Cloud, highlighted the chip’s high bandwidth, low power consumption, and programmability as a key enabler for scaling their cloud infrastructure.
Other major partners, such as Lumen, emphasized the technology’s role in supporting AI-driven services and multi-cloud connectivity.
Dave Ward – Chief Technology Officer and Product Officer, Lumen, mentioned that Lumen is advancing its network infrastructure to support the AI-driven economy. As AI adoption accelerates, Cisco’s P200 chip and 8223 routing system offer a power-efficient, scalable, and secure solution for connecting multi-site data centers. By enabling massive AI workloads to run seamlessly across distances, Cisco is setting a new standard for the infrastructure needed in the AI-driven future.
Rajani Baburajan
