Bringing Rack-Scale AI to the Enterprise Together with Cisco

Our new collaboration with Cisco puts Supermicro liquid-cooled and air-cooled AI systems inside the Cisco Secure AI Factory with NVIDIA, delivered through the partners enterprises and AI clouds already work with.

Organizations have made their choice: run AI on infrastructure they control. Data sovereignty, security policy, and the economics of sustained inference all point in the same direction. The hard part is deployment. Bringing accelerated compute, three distinct network fabrics, high-performance storage, security, power delivery, and liquid cooling into a single operational system takes precision. Keeping that system running takes even more.

Supermicro and Cisco announced a strategic collaboration to overcome exactly this challenge. Supermicro liquid-cooled and air-cooled AI systems are now part of the Cisco Secure AI Factory with NVIDIA, validated and sold as part of the broader Cisco AI infrastructure portfolio. This joint solution is scheduled to available through Cisco and its authorized channel partners beginning October 2026.

"Supermicro has spent over three decades perfecting the building blocks of IT infrastructure, and today's announcement validates our focus on today's complete AI factories," said Charles Liang, president and CEO of Supermicro. "By working closely with Cisco, we are delivering our first-to-market AI systems and data center-scale deployment capability to global enterprises through Cisco's sales organization and partner ecosystem they already know and trust. Customers can focus on their AI applications and delivering results while we deliver the infrastructure, on time, and ready to run."

What the joint solution includes

The collaboration spans the full, NCP RA compliant infrastructure stack required to bring enterprise, neocloud and sovereign AI into production.

• Accelerated compute. Supermicro rack-scale systems built on NVIDIA GB300 NVL72 and NVIDIA Vera Rubin NVL72, with expanded support for liquid-cooled NVIDIA HGX B300 NVL8 and NVIDIA HGX Rubin NVL8 systems. High-density platforms for trillion-parameter training, and dense GPU systems for high-throughput inference.

• Integration and deployment: Rack-scale integration, testing, and validation performed by Supermicro, with onsite deployment services built on Data Center Building Block Solutions (DCBBS).

• Liquid cooling. Supermicro provides an end-to-end solution from cold plates through liquid-cooling-optimized rack enclosures to cooling distribution units, with in-rack, in-row, and liquid-to-air sidecar CDU options. Rack-to-fabric liquid cooling pairs Supermicro liquid-cooled servers with Cisco liquid-cooled AI networking systems.
  
  
• AI Networking. Cisco brings the familiarity and power of the Nexus operating model to AI by being the only NVIDIA technology partner to deliver NCP-RA compliant fabric on their own silicon. Cisco Silicon One powers the front-end fabric, and NVIDIA Spectrum-X Ethernet powers the backend, all on Cisco switches, with the AI backend fabric, frontend connectivity, and storage networking unified by Nexus One architecture.
  
  
• Cisco Validated Infrastructure Services (CVIS) for NCP-RA. Cisco-delivered validation services confirm the deployed system meets NCP-RA reference architecture requirements, giving you documented assurance that compute, fabric, storage and cooling operate as a certified whole.
  
  
• Security, observability and management. Cisco Cloud Control delivers unified management across the full stack, from rack-scale core to edge deployments with end-to-end observability, proactive hardware monitoring, automated fault detection, and security built into every layer.

Together, Supermicro and Cisco close the gap between AI ambition and AI in production. You get first-to-market rack-scale systems, NCP-RA compliant networking on Cisco switching and validated services that stand behind the entire stack, all deployed into your existing network fabric and operated from a console your teams already know. That is agentic AI infrastructure engineered to run at scale, from day one.

Learn more about our AI Solutions.