Cisco UCS C885A M8 Rack Server

The Cisco UCS C885A M8 Rack Server (part number UCS-C885A-M8) is a GPU-dense server for AI workloads, supporting 8x NVIDIA H200 GPUs for LLM training and inferencing. Part of Cisco’s Secure AI Factory with NVIDIA, it integrates with Nexus 6000 Series Switches for high-performance AI clusters. Announced in 2024 and orderable now, it delivers scalable compute for AI data centers. No GPUs are included; NVIDIA GPUs are supported.

get a quote

No items found.

Technical Specifications

Form Factor

8RU rack-mount serverProcessor2x AMD EPYC CPUs (no GPU included, supports 8x NVIDIA H200 SXM GPUs)

Network Interfaces

8x NVIDIA BlueField-3 B3140H SuperNICs for East-West traffic (400G)
1x NVIDIA BlueField-3 B3220 DPU NIC (2x200G mode) for North-South traffic

Performance Features

Designed for AI workloads: large language model (LLM) training, fine-tuning, inferencing, and retrieval-augmented generation (RAG)
Supports NVIDIA HGX H200 reference architecture with 8x H200 GPUs interconnected via NVLink
Up to 3 TB DDR DRAM, 30 TB NVMe local storage
High-throughput, low-latency connectivity for AI GPU clusters

Management

Cisco Intersight for cloud-based management and automation
Integrates with Cisco Nexus Hyperfabric for AI cluster management

Power Supply

Hot-swappable fan trays and redundant power supplies
Optimized for data center energy efficiency

Target Market

Hyperscalers and enterprises deploying GPU-intensive AI/ML workloads

Speak to an engineer

Ready?

Our team of experts are available and ready to assist you with any inquiries and offer customized 3-5 year financing options that align with your budget and goals.