Cisco is broadening its Secure AI Factory with NVIDIA initiative by adding Supermicro as a compute partner, incorporating Supermicro’s liquid- and air-cooled systems into Cisco’s full-stack AI infrastructure architecture. The expansion targets enterprise, neocloud, and sovereign cloud customers building out large-scale AI deployments.
“We are at the beginning of one of the largest datacenter buildouts in history,” said Jeetu Patel, President and Chief Product Officer at Cisco, arguing that organizations scaling AI need compute and networking delivered as an integrated, secure offering rather than assembled piecemeal. “AI factories are revenue-generating infrastructure, where compute produces intelligence, and intelligence drives revenue,” said Justin Boitano, vice president of Enterprise AI at NVIDIA, adding that the expanded stack is meant to get customers into production faster and generate more value from every watt.
Infrastructure and Architecture
Through the partnership, Cisco will offer Supermicro’s liquid- and air-cooled systems as part of its broader AI infrastructure portfolio, giving customers access to rack-scale and dense GPU configurations that are validated for use alongside Cisco’s existing stack. The arrangement is designed to let customers manage high-density AI clusters and conventional workloads within the same operational framework.
A key element of the expansion is rack-to-fabric liquid cooling, combining Cisco’s liquid-cooled AI networking systems with Supermicro’s liquid-cooled servers. This configuration supports trillion-parameter training and high-throughput inference workloads, with compatibility extending to NVIDIA Vera Rubin NVL72 and NVIDIA HGX Rubin NVL8 platforms.
On the networking side, the expanded architecture offers NVIDIA Cloud Partner (NCP)- compliant solutions for neocloud and sovereign cloud customers. Cisco Silicon One-based switches handle the front-end network, while Cisco’s NVIDIA Spectrum-X-based switches manage the back-end, with both unified under Cisco Nexus One. Cisco notes that it is the only NVIDIA technology partner using its own switches and network operating system in an NCP-compliant solution. The Secure AI Factory with NVIDIA is also extending its Enterprise Reference Architectures to incorporate the latest generation of NVIDIA AI infrastructure for enterprise deployments.
Supply Chain, Validation, and Operations
Cisco positions the partnership as a way to reduce deployment risk, pointing to the combined global supply chain expertise of Cisco and Supermicro in navigating GPU and memory availability constraints. Infrastructure built on NVIDIA reference architectures is intended to be pressure-tested for current workloads, with security and resiliency addressed from the silicon layer through software agents, backed by Cisco’s partner ecosystem and support organization.
To accelerate certification after delivery, Cisco is introducing Cisco Validated Infrastructure Services (CVIS), aligned with NVIDIA Infrastructure Services (NVIS). CVIS is meant to confirm that deployed infrastructure matches its intended design and conforms to reference architectures. Cisco is also establishing a dedicated, large-scale AI lab to develop tools and test software that support CVIS.
On the operations side, the architecture integrates NVIDIA AI Enterprise software with AgenticOps through Cisco Cloud Control, allowing teams to use familiar tooling rather than adopting an entirely new operational model. The combination is intended to give customers end-to-end observability, correlating job health with compute, NIC, optics, and network performance metrics.
Availability
Cisco will begin offering Supermicro compute solutions as part of the Secure AI Factory with NVIDIA starting in October 2026.
Amazon