Cisco partners with NVIDIA for AI data centre networking

Cisco has released a suite of AI-ready data centre technology, aiming to be the backbone provider for organisations with artificial intelligence deployments across hyperscale and enterprise environments.

With NVIDIA, the company showcased their first technical integration at the recent Cisco Live. event.

This featured a demonstration of NVIDIA Spectrum-X Ethernet networking based on Cisco Silicon One architecture.

The integration supports NX-OS, Nexus Hyperfabric AI and Software for Open Networking in the Cloud (SONiC) deployments on Cisco G200-based switches with NVIDIA network interface cards (NICs).

Expanded AI PODs now offer configurable solutions for diverse workloads including training and fine-tuning, with the NVIDIA RTX PRO 6000 Blackwell Server Edition GPU available for Cisco UCS C845A M8 servers.

Cisco's new Unified Nexus Dashboard consolidates network operations across LAN, SAN, IPFM, and AI/ML fabrics into a single management interface.

The platform converges Application Centric Infrastructure (ACI) and NX-OS VXLAN Evolved Virtual Private Network (EVPN) fabrics with unified data, control, and policy enforcement.

Enterprises can now optimise AI workload operations through Cisco Intelligent Packet Flow, which dynamically steers traffic using real-time telemetry and congestion awareness.

The solution provides end-to-end visibility across networks, graphics processing units (GPUs), and distributed AI jobs for proactive issue detection.

"The world is moving from chatbots intelligently answering our questions to agents conducting tasks and jobs fully autonomously," Jeetu Patel, CIsco's chief product officer said.

Patel referred to this as the agentic era of AI, where billions of autonomous agents will demand high-bandwidth, low-latency networking infrastructure.

Meanwhile, Cisco AI Defense and Cisco Hypershield provide end-to-end enterprise AI workflow protection, now included in NVIDIA Enterprise AI Factory validated designs.

Cisco has secured strategic partnerships with three neocloud providers as demand grows for GPU-as-a-service and Infrastructure-as-a-service solutions.

Saudi Arabia's HUMAIN is collaborating with Cisco to build what it calls the world's most open, scalable AI infrastructure using Nexus, UCS, Hypershield and Splunk technologies.

United Arab Emirates-based G42 announced a collaboration to advance AI innovation and infrastructure development across public and private sectors.

Cisco has also joined the Stargate UAE consortium as a preferred technology partner, providing networking, security and observability solutions for next-generation AI compute clusters.

The company's Agile Services Networking architecture enables service providers to modernise infrastructure and monetise AI-driven services through fundamental network architecture shifts.

New converged access and edge router devices powered by Cisco Silicon One expand the 8000 series portfolio for enhanced network efficiency.

A multi-agentic framework for Cisco Crosswork Network Automation introduces AI capabilities to accelerate operations and decision-making through collaborative AI agents.

The company also unveiled seamless integration between terrestrial and satellite networks, supporting service assurance and dynamic resource allocation for maritime, aviation, Internet of Things (IoT), and disaster response sectors

Cisco's new 400 gigabit/s bidirectional optics enable cost-efficient transitions to 400G networks while preserving existing duplex multi-mode fibre infrastructure.

The networking giant said it exceeded its annual US$1 billion AI infrastructure target from hyperscalers a full quarter early in Q3 FY25.