Cisco Unveils Silicon One G300 Chip to Power Large-Scale Enterprise AI

Silicon One G300 chip was the centerpiece of Cisco’s AI infrastructure announcements at its Cisco Live conference in Amsterdam, where the company outlined updates aimed at supporting large-scale enterprise AI deployments.

The launch comes as enterprises move from AI experimentation to building full-scale data center infrastructure capable of handling intensive AI workloads. Networking performance, security and operational visibility are increasingly seen as critical layers beneath AI models.

The move positions Cisco to compete more directly with established AI infrastructure players including Nvidia, which dominates AI chip markets, and Broadcom, which has also been expanding its networking solutions for AI workloads. Unlike pure-play chip makers, Cisco is emphasizing its integrated approach that combines silicon, networking hardware, security and management software.

Cisco said the G300 chip is designed to scale AI clusters more efficiently. The company claims the chip enables what it calls Intelligent Collective Networking, delivering a 33% increase in network utilization and a 28% improvement in job completion time compared with non-optimized traffic.

Cisco is introducing G300-powered N9100 and 8000 systems targeted at hyperscalers, cloud providers, sovereign private deployments, service providers and large enterprises.

Beyond hardware, Cisco introduced Nexus One, a unified management platform intended to simplify operations across on-premises and cloud-based data centers. The platform is positioned as a way to manage increasingly complex hybrid AI environments through a single control layer.

Speaking to more than 21,000 IT professionals at the event, Jeetu Patel, President and Chief Product Officer at Cisco, said the company is focusing on integrating networking, security, observability and sovereignty features within a unified platform. Patel said enterprises need infrastructure that allows them to deploy AI securely as innovation accelerates.

Cisco also introduced AgenticOps, a set of tools designed to automate IT operations in AI-driven environments. The platform draws telemetry data from across Cisco’s networking, security and observability portfolio — including Splunk — to provide system-wide visibility.

In security, Cisco announced updates to Cisco AI Defense, adding AI supply chain governance and runtime protections designed to reduce risks such as system compromise or manipulation. Enhancements to Cisco Secure Access Service Edge (SASE) include what the company describes as “intent-aware inspection,” which evaluates AI-generated traffic patterns to detect threats.

Recognizing regulatory requirements in certain markets, Cisco expanded support for sovereign deployments. Cisco Critical National Services Centers (CNSCs) have been established in the UK, France and Spain, with another under development in Italy. These facilities operate under strict controls with dedicated infrastructure and personnel. Cisco said it has maintained a similar center in Germany for 15 years.

The announcements reflect a broader shift in the AI market, where infrastructure providers are focusing less on standalone AI models and more on the networking, security and governance layers required to run AI systems reliably at scale.