• has announced the Silicon One G300 , an innovative 102.4 Tbits/s switching chip designed to address the growing demands of AI data centers in the agentic era. • This new silicon powers the next-generation Cisco N9000 and Cisco 8000 switch systems, setting new benchmarks in performance, efficiency, and scalability. • Engineered for hyperscale, enterprise, and sovereign cloud environments, the G300 is specifically designed to support massive AI clusters for training, inference, and real-time workloads, achieving a 28% improvement in job completion times and maximizing GPU utilization. • The Silicon One G300 introduces Intelligent Collective Networking, combining advanced packet buffering, path-based load balancing, and proactive telemetry. • This innovation ensures high network utilization, reduces packet loss, and accelerates response to link failures, enabling a 33% increase in network efficiency, according to Cisco. • The platform’s programmability also allows for seamless updates, ensuring long-term adaptability to evolving network demands.
Sources: