Cisco expands Nvidia AI factory stack with Supermicro servers

Cisco is adding Supermicro server systems to its Nvidia-backed Secure AI Factory portfolio, turning the offer from a networking-led architecture into a fuller rack-scale AI infrastructure stack for enterprises, neoclouds and sovereign cloud customers.

The company says the package combines Cisco AI networking, Nvidia Cloud Partner-compliant reference architecture and Supermicro liquid- and air-cooled compute. The aim is to make large AI clusters easier to buy, deploy and certify, including systems built for high-throughput inference and training on upcoming Nvidia Vera Rubin and HGX Rubin platforms.

This matters because the AI infrastructure market is moving from buying GPUs in isolation toward validated full-stack systems: compute, networking, cooling, observability and support sold together. For customers, that can reduce deployment risk and shorten the path from pilot projects to production workloads. For Nvidia, Cisco and Supermicro, it deepens the race to own the reference architecture around AI data centers, not just individual components.