The Cisco UCS C880A M8 Rack Server accelerates advanced AI and High-Performance Computing (HPC) workloads in every data center with next-generation NVIDIA HGX B300 NVL8 GPUs.

Based on the NVIDIA HGX platform, the Cisco UCS C880A M8 Rack Server is a high-density, air-cooled rack server designed to power the most demanding Artificial Intelligence (AI) and High-Performance Computing (HPC) workloads. It integrates the NVIDIA HGX platform with eight NVIDIA HGX B300 (SXM) GPUs and is powered by two Intel^® Xeon^® 6^th Gen Processors, making it ideal for real-time Large Language Model (LLM) inference, next-level training performance, and large-volume data processing. The C880A M8 supports customers across the entire AI stack, from large-scale model training and fine-tuning to real-time inferencing and large-volume data processing. It integrates seamlessly into Cisco’s AI strategy, connecting and protecting the AI era by providing robust compute infrastructure. This server expands the Cisco UCS^® dense AI server portfolio, offering a powerful solution for enterprises across various industries, including service providers, financial services, manufacturing, healthcare, life sciences, and automotive. With its advanced architecture, the C880A M8 ensures unparalleled performance, scalability, and enterprise manageability, making it ideal for compute-intensive AI use cases such as large-scale AI model training, fine tuning, and inferencing.

The Cisco UCS C880A M8 Rack Server stands out by integrating the cutting-edge NVIDIA HGX platform with eight NVIDIA B300 (SXM) GPUs. This powerful GPU configuration is at the heart of its capability to deliver next-level performance for the most demanding AI workloads, including large-scale AI model training, fine tuning, and real-time inferencing. The B300 GPUs provide immense parallel processing capabilities and high-speed GPU interconnects, which are critical for accelerating complex deep learning models and large language models. This integration ensures that enterprises can achieve higher token throughput and improve the economics of their AI operations, enabling profitable scaling of LLM and agentic workloads.

Beyond raw power, the Cisco UCS C880A M8 Rack Server is architected specifically to meet the unique demands of AI and HPC. Its design supports real-time large language model Inference, enabling rapid deployment and responsiveness for AI-driven applications. It also excels in next-level training performance, significantly reducing the time required to train complex AI models. Furthermore, its capacity for large-volume data processing makes it an ideal platform for data-science and big-data analytics, including GPU-accelerated ETL processes. This specialized design ensures that organizations can build, optimize, and utilize AI models efficiently, accelerating business growth with scalable and high-performance solutions.

The Cisco UCS C880A M8 Rack Server is a dedicated rack server platform designed to host and accelerate AI and HPC workloads. It supports various operating systems and virtualization platforms typically used in data center environments for AI/HPC deployments. Specific software stack compatibility includes NVIDIA AI Enterprise and NVIDIA NIM (NVIDIA Inference Microservices) for AI application deployment and optimization. For more information, see Cisco UCS C880A M8 Rack Server Data Sheet.

The Cisco UCS C-Series rack server supports operating systems such as Ubuntu, Red Hat Linux, and so on. For more information on supported operating systems, see the UCS Hardware and Software Compatibility. You can use Cisco Baseboard Management Controller 4.0 (Cisco BMC 4.0) to install an OS on the server using the KVM console and vMedia.

The Cisco Baseboard Management Controller 4.0 (Cisco BMC 4.0) is the management service for the Cisco UCS C880A M8 Rack Server servers. Cisco BMC 4.0 runs within the server.

You can use a web-based GUI, an SSH-based CLI, or a REST API to access, configure, administer, and monitor the server. Each interface offers different capabilities, and the tasks supported by each interface are described in their respective configuration guides.