Cisco
Cisco designs UCS AI servers with its own boards and management plane, and sells them as part of a networking-led stack. The rational choice when the cluster fabric, security posture and support contract already run through Cisco.
- UCS C-Series dense GPU servers (C885A M8, C880A M8) built on NVIDIA HGX reference architecture with Cisco-designed board and NIC integration and Cisco BMC
- UCS X-Series: own modular blade chassis architecture (X9508, X440p PCIe GPU node, X-Fabric)
- Cisco Intersight: own cloud-based lifecycle management across UCS
- Networking-first engineering base: Silicon One ASICs, Nexus switching and N9100 Spectrum-X switches anchor its AI play
- In-house DLC demonstrated on a liquid-cooled Silicon One switch; no shipping DLC GPU-server SKU confirmed as of mid-2026
| Model | Platform | Form factor | Cooling |
|---|---|---|---|
| UCS C880A M8 Dual Xeon 6, 8x board-integrated ConnectX-8 800G NICs | HGX B300 NVL8 | 10RU | air |
| UCS C885A M8 8-GPU dense AI server, Intersight-managed | HGX H100/H200 | 8RU | air |
| UCS C845A M8 RTX PRO 6000 Blackwell Server Edition, enterprise AI PODs | PCIe | rack | air |
| UCS X9508 + X440p Up to 16 GPUs per chassis via X-Fabric | PCIe | modular blade chassis | air |
- Cisco Secure AI Factory with NVIDIA: joint initiative since GTC March 2025
- N9100: first NVIDIA-partner-developed datacenter switch on Spectrum-X silicon; first partner with an NVIDIA Cloud Partner-compliant reference architecture
- UCS C240 M8 and C845A M8 are NVIDIA-Certified Systems
- Named among expected Rubin system makers at CES 2026, but not in NVIDIA's June 2026 Vera Rubin full-production builder list; no Cisco-branded NVL72 rack found as of July 2026
- Global contract-manufacturer network rather than owned factories: Foxconn, Jabil, Celestica, Flex, Sanmina, Wistron named as Cisco CM partners
- Networking and security company entering AI infrastructure fabric-first; UCS compute is the attach to its switching business
- Surpassed its $1B annual AI infrastructure order target from webscale customers a quarter ahead of schedule
- Smaller GPU-server entrant vs Dell, HPE, Supermicro, Lenovo; not in the top AI-server revenue tier in 2024-2025 market reports
Is Cisco an OEM, ODM or integrator for NVIDIA AI servers?
Cisco sits in the "Global OEMs" tier of our index. UCS C-Series dense GPU servers (C885A M8, C880A M8) built on NVIDIA HGX reference architecture with Cisco-designed board and NIC integration and Cisco BMC
Does Cisco design its NVIDIA servers in-house?
Cisco designs UCS AI servers with its own boards and management plane, and sells them as part of a networking-led stack. The rational choice when the cluster fabric, security posture and support contract already run through Cisco.
Which NVIDIA platforms does Cisco ship?
Cisco ships systems on HGX B300 NVL8, HGX H100/H200, PCIe. The systems table on this page lists the specific models.
Where does Cisco manufacture its AI servers?
Global contract-manufacturer network rather than owned factories: Foxconn, Jabil, Celestica, Flex, Sanmina, Wistron named as Cisco CM partners