Cisco Live Amsterdam 2026: Entering the AI era of networking

Recently I had the privilege of hosting over 25 Softcat customers at Cisco Live in Amsterdam, with the opportunity to hear feedback directly from customers and learn more about Cisco’s ambition to lead the industry into an AI‑defined future.

This year’s announcements centred on building high‑performance, energy‑efficient and operationally autonomous networks capable of supporting the scale, speed and resilience required by organisations.

For those planning major refresh cycles and AI‑ready architectures, the message was clear: The network is now the platform, and AI is changing the game.

As not everyone in the industry could attend the event (although that would have been fun!), I have provided a summary of the key updates and announcements.

AI-driven operations and governance

Cisco agentic ops and the deep network model

A major theme at Cisco Live was the NetOps transformation required for AI‑era networks. Cisco announced significant expansions to AgenticOps, its automation framework designed for cross‑domain, AI‑driven operations.

AgenticOps integrates with telemetry sources such as Nexus One, Meraki Dashboard, ThousandEyes and Splunk to power its Deep Network Models. This is used by NetOps agents to provide autonomous troubleshooting, optimisation and validation of network state.

This shift moves IT teams from reactive to proactive operations, reducing noise, improving accuracy and ensuring AI workloads run on stable, predictable infrastructures The main benefit this technology will bring for organisations is radical operational efficiency, however the challenge many will need to overcome is trusting Agents to undertake tasks autonomously. From my perspective, just like self-driving cars, this will happen over time.

Observability for AI agents

During the conference, Cisco showcased advances in observability for AI, including AI agent behavioural monitoring, performance tracing and dependency correlation across network, application and security domains. These capabilities reinforce the need for deep visibility when implementing large‑scale, autonomous systems.

Cisco secure access with MCP

Cisco has expanded its Secure Access architecture to protect AI‑driven interactions using the Model Context Protocol (MCP), a standards‑based framework that governs how AI agents securely request, receive and act on contextual information. MCP allows Cisco Secure Access to enforce identity, policy and authorisation controls, not just for human users, but also for machine‑to‑machine and agent‑driven workflows.

By integrating MCP into Secure Access, Cisco ensures that AI agents operate within defined trust boundaries, accessing only the resources they are explicitly permitted to use. As MCP and Agent adoption increases, so does complexity and attack surface. Ensuring security is baked into adoption of these technologies is key to managing risk.

AI Defense: protecting AI agents and workflows

Cisco has extended its ‘AI Defense’ platform with new capabilities designed to protect not just data and applications, but the AI agents themselves. Enhancements include runtime protections against prompt injection, model manipulation and secure behavioural boundaries for agentic workflows. These protections are increasingly critical as semi-autonomous AI systems interact with sensitive infrastructure. 

Silicon, switching and sustainable scale

Cisco smart switches

Cisco also expanded its smart switching portfolio with systems designed to bring intelligent telemetry, simplified automation and security controls leveraging in-switch DPUs. Cisco reinforced its commitment to bringing security controls closer to the network layer. Through intelligent switching platforms, AI‑driven operational insight now extends deeper into access and aggregation layers.

Cisco Silicon One: the foundation for networking for AI

Cisco expanded its Silicon One portfolio with the launch of the G300 series, a 102.4 Tbps programmable switching ASIC designed specifically to power high‑density AI clusters and reduce job completion times by 28%. The chip introduces Intelligent Collective Networking, combining shared packet buffers, path‑based load balancing and proactive telemetry to increase network utilisation and stability under AI workloads.

These silicon advances directly underpin Cisco’s newest data centre switching systems, enabling customers to scale up, scale out and scale across distributed AI data centres spanning multiple geographies.

Liquid‑cooled switches and high‑density optics

To support the heat and bandwidth demands of AI clusters, Cisco introduced fully liquid‑cooled variants of its Nexus 9000 and 8000 platforms, driving up to 70% improvements in energy efficiency for dense GPU environments.

Alongside this, Cisco unveiled next‑generation high‑density optics, including 1.6T OSFP and 800G low‑power pluggable modules, designed to reduce power per bit and support ultra‑high east‑west traffic volumes. These optics are essential for organisations building AI fabrics where data movement, not compute, is the limiting factor.

Fabric, edge and unified control

Nexus HyperFabric

One of the event’s most anticipated updates was the broadened vision for Nexus HyperFabric. This is Cisco’s cloud-managed, fabric-as-a-service approach to data centre networking, designing, deploying and managing multiple scalable fabrics from a single cloud controller.

HyperFabric integrates silicon, systems, optics, software and security into a tightly aligned AI fabric, enabling organisations to build predictable, high‑performance networks for training, inference and real‑time agentic workloads.

Cisco Nexus One

At the operational layer, Cisco Nexus One has now evolved into a unified management plane spanning on‑premises and cloud environments. It consolidates configuration, lifecycle management, telemetry, and compliance into a single operational interface, reducing friction and enabling consistent operations across distributed network topologies.

Cisco Unified Edge

Cisco Unified Edge is Cisco’s AI-ready edge platform that brings data centre grade compute, networking and storage to the edge so you can deploy and run small language models (SLMs) locally for low-latency, real-time inference. This unlocks the ability for organisations to gain fast, secure, real-time insights and automations locally, without relying on the cloud.

Cisco Data Fabric

Finally, Cisco advanced its Data Fabric initiative in partnership with Splunk, delivering a cross‑domain, federated data plane designed to unify telemetry from networking, security, observability and AI systems. This fabric will become essential for organisations seeking to correlate events, detect anomalies and streamline AI‑powered operations at enterprise scale. Unified, high‑quality data enables organisations to accurately correlate events, spot anomalies sooner and power reliable AI‑driven operations at scale.

A new chapter for networking

At Cisco Live 2026, one thing was clear to me: modern networking is no longer just about connectivity. It’s about building a resilient, intelligent, secure foundation for AI‑driven enterprises. With breakthroughs across silicon, optics, cooling, network operations, security and data platforms, Cisco is rewriting the blueprint for what tomorrow’s networks must deliver.

By adopting these technologies, organisations can unlock radical operational efficiency gains, enable business innovations and gain data driven insights to help steer future strategy.

At Softcat, we’re excited to help customers understand these innovations and how they can apply them to real‑world use cases.

If you’d like to explore what these announcements mean for your organisation, contact your Softcat Account Manager or Network & Security Specialist to continue the conversation.