How Cisco Boosts AI Data Centres with New Infrastructure

Speculation of an AI bubble is in the air and executives are rushing to get return on investment (ROI) on booming AI investments. 

ROI measures profit generated from money spent. As AI projects expand, organisations rely on data centre strategy to ensure AI expenditure translates into revenue rather than risk.

At the Cisco Live Conference in Amsterdam, Cisco stepped up its AI game with new infrastructure innovations. 

The company positioned its portfolio to help organisations build profitable data centres free from complexity and maximise AI returns. 

Complexity refers to the operational burden of managing hybrid environments, multiple vendors and expanding AI workloads.

Cisco presented software and hardware innovations alongside AI defence capabilities. These enable customers to raise ambitions for secure and trusted agentic AI. 

Agentic AI describes AI systems acting autonomously, making decisions and taking actions across tools and environments. Security and governance are central to implementing these systems safely.

Jeetu Patel, President and CPO at Cisco says: “AI innovation is moving faster than ever before and we’re delivering the critical infrastructure our customers need to move fast and adopt AI safely and securely.

“Today’s announcements highlight the power of Cisco as a unified platform, showcasing how our innovations in silicon and systems, AgenticOps, security and observability come together to unlock value for our customers from the data centre to the workplace and beyond.”

Hardware and software innovations for AI

At the core of Cisco’s announcement is Silicon One G300, a 102.4Tbps switching silicon. Tbps measures how much data moves through a network per second. 

Switching silicon is the chip in network switches directing traffic between servers and storage. In AI clusters, switching performance directly affects training and inference speed.

The G300 supports Intelligent Collective Networking, delivering a 33% increase in network utilisation and a 28% improvement in job completion time compared to non-optimised traffic. Network utilisation shows how effectively bandwidth is used, and job completion time measures AI task duration.

Cisco’s G300-powered N91000 and 8000 systems target hyperscalers, neoclouds, sovereign private deployments, service providers and enterprises.

These systems are 100% liquid cooled, provide higher bandwidth and improve energy efficiency by 70%. 

Liquid cooling removes heat more effectively than air cooling, reducing operational costs in high-density AI data centres.

Alongside hardware, Cisco introduced Nexus One, a unified management plane for operations across on-premises and cloud-based deployments. 

Unified management simplifies monitoring and control across environments, reducing silos and manual work.

Martin Lund, Executive Vice President of Cisco’s Common Hardware Group says: “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.

“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 utilise their compute and scale AI securely and reliably in production.”

Deterministic networking delivers predictable performance, critical for AI workloads dependent on consistent latency and throughput.

AgenticOps and AI defence

Cisco extends focus to operations and security through AgenticOps, which “automates, scales and simplifies IT operations in the AI era”. 

Automation reduces manual tasks, scaling supports growth, and simplification addresses operational complexity.

AgenticOps collects telemetry across Cisco Networking, Security Cloud Control, Nexus One, Splunk and more. 

Telemetry gathers monitoring data from systems. This enriches networking, security and observability – the ability to understand a system’s internal state through metrics, logs and traces.

Cisco also updates AI Defence for supply chain governance and runtime protection. AI supply chain governance ensures integrity of models, data and third-party components. Runtime protection secures AI systems while they operate, defending against attacks.

Enhancements to Secure Access Service Edge, or SASE, include “intent-aware inspection of agentic AI interactions and tool requests, evaluating the ‘why’ and ‘how’ of agentic traffic to ward off novel threats”. 

SASE combines networking and security functions in a cloud model, while intent-aware inspection adds context to traditional traffic monitoring.

As AI adoption accelerates, Cisco announces rollout of Critical National Services Centres in UK, France and Spain. 

These centres support organisations building AI confidently with Cisco, offering in-country expertise and services aligned with national requirements.

Cisco positions its unified platform to help organisations turn AI ambition into measurable performance, energy efficiency and trusted deployment across the data centre and beyond.