Amidst the backdrop of Amsterdam’s iconic canals and the town’s famend Science Park, Cisco Stay 2025 Amsterdam units the stage for the announcement of our new Cisco UCS C845A M8 AI Rack Server based mostly on the NVIDIA MGX reference structure.
We’ve engineered this groundbreaking addition to our AI server lineup to ship enterprise AI workloads, offering revolutionary NVIDIA accelerated computing options that equip companies to thrive in as we speak’s fast-paced digital surroundings. The UCS C845A M8 Rack Server is a perfect match for enterprise information heart environments as a result of it’s extremely scalable, versatile, and customizable—delivering highly effective AI capabilities to mainstream enterprise peripheral element interconnect specific (PCIe) servers.
Think about a healthcare analytics firm beginning its AI journey with the UCS C845A M8, initially utilizing two GPUs to research affected person information. As its fashions and information quantity develop, the corporate scales as much as eight GPUs, seamlessly enhancing capabilities to assist enhance affected person outcomes with out disrupting operations.
This may be completed as a result of the server relies on the NVIDIA MGX modular reference design, which helps two to eight NVIDIA PCIe GPUs (together with NVIDIA H200 NVL, H100 NVL, or L40S GPUs), NVIDIA BlueField-3 SuperNICs and DPUs, and NVIDIA AI Enterprise software program.
This platform is not only about uncooked energy; it’s about unlocking new potentialities and efficiencies that had been beforehand unimaginable. It builds on the capabilities of our UCS C885A M8 Rack Server—an eight-way accelerated computing system based mostly on the NVIDIA HGX platform—by delivering distinctive efficiency and suppleness throughout a variety of enterprise AI workloads.
Constructed for AI efficiency, effectivity, and scale
As we proceed our AI journey, this introduction marks a pivotal step. The UCS C845A M8 leverages the NVIDIA MGX modular reference design and is future-proofed to combine next-gen PCIe GPUs with out requiring a brand new platform. Cisco plans to supply configurations with further GPUs as they grow to be out there. It additionally includes a dense design, supporting as much as eight GPUs in a compact 4RU chassis.
We’ve additionally improved the reference design whereas preserving the intent of the design. This consists of enhanced energy supply, fewer printed circuit boards (PCBs), and higher cable routing for optimum airflow and thermal administration. Due to this, if a GPU had been to fail, we make it quicker and simpler to switch it after which get again up and operating. The system options E1.S strong state drives for native storage to extend storage density, enhance thermal administration, and supply excessive efficiency in a compact type issue.
From day 1, IT groups can handle the UCS C845A M8 by way of Cisco Intersight. This implies you possibly can have the identical operational capabilities for conventional and AI workloads. On the similar time, you additionally get the most recent {hardware} compatibility suggestions, safety alerts, and integrations along with your present instruments, like ServiceNow.
Along with delivering accelerated servers to handle compute-intensive AI workloads, we’re additionally providing AI PODs to assist shorten the time required to attain production-ready inferencing. Constructed on the muse of Cisco Validated Designs (CVDs), AI PODs embrace NVIDIA AI Enterprise and supply clients with a longtime place to begin, simply adaptable to fulfill their particular wants. These full-stack, pre-sized infrastructure bundles get rid of the guesswork from deploying AI inference options—from edge inferencing to large-scale clusters with NVIDIA accelerated computing. This implies quicker time to worth, constant efficiency, and lowered danger for AI tasks.
An AI revolution for information scientists, AI engineers, and CTOs
The UCS C845A M8 Rack Server’s superior structure is designed to allow AI innovation, providing the computational energy wanted for intensive duties and the effectivity required for fast deployment. By advancing past conventional capabilities, this new AI server allows quicker deployment of AI purposes, fostering innovation and serving to place companies as leaders within the AI area.
The server’s modular NVIDIA MGX design permits for versatile configurations and addresses a broad vary of AI use circumstances.
Listed here are just a few examples:
- Massive enterprises with vital information processing wants, similar to these in finance, healthcare, automotive, and manufacturing, can make the most of this server for his or her intensive computational necessities.
- Analysis establishments and universities, notably these specializing in AI and machine studying, can leverage such a server to research giant, complicated datasets throughout numerous fields to higher tune predictions, determine patterns, and acquire worthwhile insights that is likely to be close to unattainable with standard strategies.
- Cloud service suppliers can leverage AI for deep studying and inferencing to successfully ship IT companies and SaaS purposes.
- Authorities companies, notably these concerned in large-scale information evaluation, might have wants for tasks requiring superior computational capabilities.
Organizations with vital information processing wants can additional speed up their inferencing and RAG/CAG workloads by leveraging the SuperNIC or NIC card choices out there, like the most recent NVIDIA BlueField-3 SuperNIC or NVIDIA ConnectX-7 playing cards. This could additionally permit for extra flexibility to right-size workloads whereas saving energy and decreasing value.
As well as, relying on the mannequin and variety of GPUs configured, the UCS C845A M8 Rack Server is right for:
- GenAI coaching and fine-tuning
- Excessive-performance computing (HPC)
- Information analytics and visualization
- Hyperscale cloud purposes
- Design and simulation
- Language processing
- Conversational AI
- Graphics and rendering
- Digital AI workstations
As we mirror on the vitality and innovation showcased at Cisco Stay Amsterdam, as we speak’s unveiling stands as a testomony to our dedication to main the cost in AI know-how. The UCS C845A M8 Rack Server is designed to empower your online business to navigate the complexities of a data-driven world with confidence and precision.
We’d like to hitch you in your journey as you drive your enterprise ahead on this thrilling new period of AI. The Cisco UCS C845A M8 ships in early Q3CY25, with orders opening in Could 2025. For extra particulars, learn the Cisco UCS C845A M8 information sheet.
Share: