Global networking leader Cisco Systems has just unveiled and begun rolling out its Silicon One G300 networking silicon and systems — marking a significant commercial deployment aimed at scaling artificial intelligence (AI) data centers worldwide.
At Cisco Live EMEA 2026, the company introduced the G300-powered Cisco N9000 and Cisco 8000 systems, built around the new 102.4 Tbps Silicon One G300 switching ASIC. These systems are now commercially available and designed for hyperscale AI environments, including cloud service providers, enterprise datacenters, sovereign clouds, and next-generation network operators.
The deployment represents a strategic shift in how networking infrastructure supports AI compute workloads. Unlike traditional IP networking silicon, the G300 platform is optimized to handle gigawatt-scale AI clusters, accelerating training, inference, and real-time “agentic” workloads. According to Cisco, the silicon delivers up to a 28 % improvement in job completion time while maintaining secure, predictable scalability.
Beyond raw throughput, Cisco is integrating advanced optics and unified management capabilities into these systems, enabling customers to deploy and operate high-performance fabrics more efficiently. New liquid-cooled* configurations further boost energy efficiency by nearly 70 %, a critical factor for organizations facing both power constraints and sustainability goals.
“Cisco Silicon One G300 is engineered for the next phase of AI networking,” said a senior Cisco executive at the launch event, highlighting the necessity for predictable, secure, and energy-efficient AI infrastructure as global compute demand surges.
Industry analysts see this deployment as a pivotal moment — networking silicon is no longer a supporting actor but a core enabler of AI performance. Early adopters of the Cisco G300 platform are expected in large cloud and enterprise environments throughout 2026, with broader availability to follow as part of Cisco’s global sales and support ecosystem.
