Accelerate AI Factories with Supermicro and NVIDIA-Certified Servers
What is a Supermicro AI Factory?
AI factories from Supermicro and NVIDIA are complete, turnkey solutions simplifying the deployment of enterprise AI at scale for faster time-to-online and time-to-revenue, with full-stack solutions including compute, software, networking, and storage.
Supermicro delivers AI infrastructure optimized for performance and efficiency, with fully-integrated solutions based on NVIDIA Enterprise Reference Architectures and NVIDIA-Certified Systems™ for guaranteed full-stack performance and compatibility.
Supermicro’s industry-leading rack-level testing, validation, and deployment services ensure quality and seamless plug-and-play deployment for complete AI confidence.
Supermicro
First-to-Market NVIDIA-Certified Systems
Rack-Scale Integration, Testing, and Validation before Shipping
Cluster-scale Deployment, Services, and Support
Storage and Networking Integration
NVIDIA
NVIDIA Accelerated Compute
NVIDIA Spectrum™-X Ethernet Networking Platform
NVIDIA Software Stack
NVIDIA AI Data Platform
Supermicro and NVIDIA Deliver Everything You Need to Reduce Complexity and Deploy AI Faster
Supermicro and NVIDIA Deliver Everything You Need to Reduce Complexity and Deploy AI Faster
- Proven first-to-market track record for new NVIDIA acceleration technologies to market
- Flexible building block approach enables faster adoption cycles
- Production capacity in the USA of over 5,000 racks per month
- Supermicro Data Center Building Block Solutions® (DCBBS) provides everything needed to facilitate the deployment of AI factories
Flexible, End-to-End AI Solutions Tailored to Your Enterprise
- Industry-leading broad portfolio of accelerated AI systems
- Flexible, modular architectures fine-tuned to maximize performance and efficiency in enterprise environments
- Cluster-level integration expertise including networking, testing, and validation
- Storage solutions for all stages of the AI data pipeline
Proven Quality. Unmatched Performance. Complete AI Confidence
- Close cooperation between Supermicro and NVIDIA ensures performance-optimized AI hardware can be easily integrated into full-stack AI solutions
- Full portfolio of NVIDIA-Certified Systems for guaranteed performance
- Single-vendor solutions with complete quality, integrity, and compatibility control throughout the entire supply chainh
- Complete L11 testing and validation beyond industry standards for seamless plug-and-play deployment
Supermicro NVIDIA-Certified Systems™ – The Foundation of AI Factories
Supermicro’s flexible, modular architectures mean configurations and form factors have been fine-tuned to maximize performance and efficiency in enterprise environments, resulting in a simplified process of integrating AI-optimized hardware into existing enterprise environments where thermal, power, and space constraints may limit the use of one-size-fits-all or ready-made solutions. Supermicro’s industry-leading portfolio of NVIDIA-Certified Systems have been fully tested and validated for performance, reliability, and compatibility with NVIDIA Enterprise software and NVIDIA Spectrum-X networking, and form the building blocks for seamlessly scaling AI factories.
NVIDIA RTX PRO™ Servers
Available in a range of form factors and densities to optimized for enterprise environments. From industry-standard 2U systems designed to replace CPU-based servers to thermally-optimized high density systems designed for maximum performance.
- 2U, 4U, and 5U form factors
- Up to 8 NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs per system
- Multi-workload support including AI, HPC, and visual computing
- Optimized for air cooled environments with support for ambient temperatures up to 35°C
- AMD EPYC™ or Intel® Xeon® CPU options
SYS-522GA-NRT
Thermally-optimized 5U system supporting up to 8 GPUs and Intel Xeon 6 CPUs
AS -5126GS-TNRT2
Thermally-optimized 5U system supporting up to 8 GPUs and AMD EPYC 9005 series CPUs
SYS-422GL-NR
4U RTX PRO Server supporting up to 8 GPUs and Intel Xeon 6 GPUs
SYS-521GE-TNRT
5U RTX PRO Server supporting up to 8 GPUs and 5th Gen Intel Xeon GPUs
NVIDIA HGX™ Servers
Specialized architectures designed for maximum AI performance. NVIDIA HGX systems offer unprecedented computational performance, density, and efficiency with next-generation air-cooled architectures as well as multiple CPU options.
- 8U and 10U form factors allow optimal system performance in air cooled environments
- NVIDIA B300, B200 and H200 8-GPU with NVIDIA NVLink® for maximum GPU-GPU communication
- AMD EPYC 9005 or Intel Xeon 6 CPU options
SYS-822GS-NB3RT
Air cooled 8U system with NVIDIA HGX B300 8-GPU and Intel Xeon 6 CPUs
SYS-A22GA-NBRT
Air-cooled 10U system with NVIDIA HGX B200 8-GPU and Intel Xeon 6 CPUs
SYS-821GE-TNHR
Air-cooled 8U system with NVIDIA HGX H200 8-GPU and 5th Gen Intel Xeon CPUs
AS -8125GS-TNHR
Air-cooled 8U system with NVIDIA HGX H200 8-GPU and AMD EPYC 9004 CPUs
AI at Scale with Supermicro AI Factory SuperClusters
Supermicro’s AI Factory SuperClusters are based on NVIDIA Enterprise Reference Architectures and provide enterprise customers with complete, rack-scale and cluster-scale solutions that ensure full-stack performance and compatibility, simplifying the deployment of complete AI factories. Supermicro’s testing and validation goes beyond industry standards, with complete testing of all nodes and cluster-level (L12) testing before shipment to ensure seamless plug-and-play deployment for customers of any size. Supermicro AI Factory solutions are endorsed by NVIDIA for Infrastructure Configuration, Spectrum-X networking, and Software Reference Stack and based on the NVIDIA Enterprise Reference Architecture for RTX PRO 6000 Blackwell Server Edition and HGX B200.
| GPU | NVIDIA RTX PRO 6000 Blackwell Server Edition GPU | NVIDIA HGX B200 | NVIDIA HGX B300 |
|---|---|---|---|
| Maximum Cluster Size | Up to 32 nodes, 256 GPUs per scalable unit | Up to 32 nodes, 256 GPUs per scalable unit | Up to 32 nodes, 256 GPUs |
| Nodes per Rack (Typical) | 4–8 per rack | 4 per rack | 4 per rack |
| GPU System Node SKU(s) | |||
| GPU Configuration per Node | 8x NVIDIA RTX PRO 6000 Blackwell Server Edition (96GB GDDR7 per GPU) | 8x NVIDIA HGX B200 (192GB HBM3e per GPU) | 8x NVIDIA HGX B300 (288GB HBM3e per GPU) |
| Rack Power (4 Nodes) | 33.3–36.6kW | 53.6kW | 60kW |
| Networking | NVIDIA Spectrum-X | NVIDIA Spectrum-X | NVIDIA Spectrum-X |
| NVIDIA Software Stack | NVIDIA AI Enterprise/ | NVIDIA AI Enterprise/ | NVIDIA AI Enterprise/ |
| Target Deployment Use Case | AI inference / Retrieval Augmented Generation (RAG), HPC, and visual computing | Foundational AI model training, large-scale AI inference, and FP64 HPC workloads | Foundational AI model training, large-scale AI inference, and HPC workloads |
| Links | Datasheet | Datasheet | Datasheet |
NVIDIA AI Software Platforms
Supermicro’s NVIDIA-Certified Systems™ have been fully tested and validated for performance, reliability, and compatibility with the NVIDIA AI software stack including NVIDIA AI Enterprise, NVIDIA Omniverse, and NVIDIA Run:ai, enabling the building and deployment of production-ready agentic AI and physical AI systems anywhere—across clouds, data centers, or at the edge.
NVIDIA AI Enterprise
NVIDIA AI Enterprise is a cloud-native suite of software tools, libraries, and frameworks, including NVIDIA NIM and NeMo microservices, that accelerate and simplify the development, deployment, and scaling of AI applications.
NVIDIA Omniverse
NVIDIA Omniverse is a platform of APIs, SDKs, and services that enable developers to integrate OpenUSD, NVIDIA RTX™ rendering technologies, and generative physical AI into existing software tools and simulation workflows for industrial and robotic use cases.
NVIDIA Run:ai
NVIDIA Run:ai accelerates AI operations with dynamic orchestration across the AI life cycle, maximizing GPU efficiency, scaling workloads, and integrating seamlessly into hybrid AI infrastructure with zero manual effort.