Cisco Unveils Silicon One G300 for Gigawatt-Scale AI Networking, Boosts Efficiency and Security

Cisco's Gigawatt-AI Networking Offensive

Cisco unveils a new AI networking stack built around the Silicon One G300, a 102.4 Tbps switch silicon the company says is designed to power gigawatt‑scale AI clusters for training, inference and real‑time agentic workloads. Announced at Cisco Live EMEA in Amsterdam, the G300 underpins new N9000 and 8000 systems and is offered in 100% liquid‑cooled configurations paired with new high‑density optics to meet the thermal and power density of modern AI racks.

Cisco says the G300 maximizes GPU utilization and cuts job completion times by about 28%, while system-level innovations deliver up to nearly 70% improvement in energy efficiency versus prior architectures. The combination of higher throughput, liquid cooling and denser optics is positioned to lower operating and cooling costs, enable denser fabrics and allow customers from hyperscalers to enterprises and sovereign clouds to expand AI deployments without breaching thermal or power budgets.

To simplify deployment and operations, Cisco enhances its Nexus One management plane to provide a unified control and operational fabric that accelerates fabric stand‑up, predictable scaling and secure operations across on‑premises and cloud environments. Jeetu Patel, Cisco’s president and chief product officer, frames the release as a full‑stack approach — from silicon to systems and software — aimed at enabling broader adoption of hyperscale performance for a wider range of AI network builders.

Bolstering AI Defenses and AgenticOps

Alongside the networking launch, Cisco expands its AI security portfolio to address risks from agentic systems. The company introduces an AI Bill of Materials for centralized visibility, an MCP (model context protocol) Catalog for inventorying model servers and registries, advanced algorithmic red‑teaming, and real‑time agentic guardrails. Cisco also upgrades its SASE offering with AI‑aware traffic detection and routing enhancements, adding post‑quantum cryptography and operational resilience for encrypted communications at scale.

Cisco also advances AgenticOps, its agent‑first IT operating model, extending autonomous execution with oversight across networking, security and observability. The company says these tools combine unified network data, purpose‑built models and governance to automate repetitive tasks, accelerate incident resolution and maintain human control, targeting faster, more secure operations as enterprises scale AI‑driven workflows.

Strong Quarter Underscores AI Momentum

Cisco reports record Q2 fiscal 2026 results, with revenue of $15.3 billion, double‑digit year‑over‑year growth and expanded margins. Management highlights $2.1 billion of AI infrastructure orders from hyperscalers and increased product order momentum, while reaffirming fiscal guidance and modestly raising shareholder returns. The results and product announcements together signal Cisco leaning into AI networking, security and operations as a platform play for large‑scale enterprise and cloud customers.