Cisco, continuing to reinforce its vision of the network as a fundamental innovation platform for AI innovation, has announced the Silicon One G300, its latest 102.4 Tbps switching silicon designed from the ground up to power the next generation of large-scale AI data center and cluster environments. The Cisco Silicon One G300 is designed to accelerate the next wave of AI infrastructure with the Cisco N9000 and Cisco 8000 series, continuing Cisco’s leadership in high-end, scalable, and secure AI networking.
Similarly, the new systems have been optimized to address the operational requirements of latest AI solutions with improved liquid cooling solutions and support for high-density optical interconnect solutions to provide greater efficiency in terms of power consumption as well as system throughput. In addition to this, Cisco has improved its Nexus One to provide greater ease of management for its AI networking infrastructure for on-premises as well as cloud deployments.
“We are spearheading performance, manageability, and security in AI networking by innovating across the full stack – from silicon to systems and software,” said Jeetu Patel, President and Chief Product Officer, Cisco. “We’re building the foundation for the future of infrastructure, supporting every type of customer—from hyperscalers to enterprises—as they shift to AI-powered workloads.”
Also Read: Expereo and Cato Networks Partner for Global Enterprise Connectivity
“As AI training and inference continues to scale, data movement is the key to efficient AI compute; the network becomes part of the compute itself. It’s not just about faster GPUs – the network must deliver scalable bandwidth and reliable, congestion-free data movement,” said Martin Lund, Executive Vice President of Cisco’s Common Hardware Group. “Cisco Silicon One G300, powering our new Cisco N9000 and Cisco 8000 systems, delivers high-performance, programmable, and deterministic networking – enabling every customer to fully utilize their compute and scale AI securely and reliably in production.”
Silicon One G300: The Networking Foundation for the Agentic Era
Cisco Silicon One G300 establishes a new performance benchmark for AI backend networking with 102.4 Tbps of switching capacity, enabling the construction of massive, distributed AI clusters that demand consistent performance, resilience, and operational predictability. The platform reflects Cisco’s continued investment in purpose-built silicon to support emerging agentic and large-scale AI architectures.
A key differentiator of the G300 is Cisco’s Intelligent Collective Networking, which integrates a fully shared, high-capacity packet buffer, advanced path-based load balancing, and proactive, real-time network telemetry. Together, these capabilities enable large data centers to better handle bursty AI traffic patterns, react more quickly to link and path failures, and minimize packet loss that can disrupt distributed training and inference jobs.
Intelligent Collective Networking enables up to 33% higher network utilization and delivers a 28% reduction in job completion time compared with simulated environments using non-optimized path selection. These gains translate directly into higher efficiency and profitability for AI data centers by improving overall throughput and increasing the number of tokens generated per GPU-hour.
The Cisco Silicon One G300 is also designed for long-term adaptability. This is achieved due to its high programmability, which helps in the upgrading of network systems after the product has been implemented, allowing new workloads and network roles to be added. Security features, which can be implemented at speeds, help in the safeguarding of AI clusters.
This platform remains the most scalable and programmable unified networking architecture in the industry. Silicon One offers a broad portfolio of products that cover AI infrastructure, hyperscale networks, enterprise networks, data centers, and service provider networks. Since its launch in 2019, Cisco Silicon One has become an integral part of the most demanding networks globally.