Supermicro 10U B200 (SYS-A21GE-NBRT-G1)
Power your most intensive workloads with this high-performance 10U single-node server, expertly engineered for High Performance Computing, Conversational AI, and Deep Learning Training. Business Intelligence & Analytics, and Industrial Automation. Driven by dual Intel® Xeon® Platinum 8597 processors, offering a combined 96 cores and bolstered by 32 high-speed 96GB DDR5-6400 RDIMM memory modules, this system ensures exceptional processing power and responsiveness across diverse projects and capacity at 3.84TB NVMe devices and 8x 3.84TB U.2 NVMe drives. Accelerate your AI initiatives with the cutting-edge NVIDIA H100 80GB GPUs. High-bandwidth connectivity is guaranteed through 4x 400GbE QSFP ports, a 200GbE DPU, and 2x 10GbE RJ45 LAN ports, facilitating seamless data transfer. Ready to deploy immediately, this robust server ships within 24 hours, minimizing downtime and maximizing productivity.
Accelerated Compute for Modern AI
Engineered around the NVIDIA HGX platform to deliver unprecedented GPU-to-GPU bandwidth via NVLink and NVSwitch. Train trillion-parameter models, accelerate scientific simulation, and serve high-throughput inference at scale.
Why It Stands Out
Up to 8 NVIDIA H200 SXM GPUs in a single chassis. 900 GB/s NVLink interconnect for distributed training. 400G InfiniBand for low-latency multi-node scale-out. Liquid-cooling ready for sustained peak performance.
Pre-validated with NVIDIA AI Enterprise, PyTorch, TensorFlow, and CUDA-X frameworks.
| Specification | Value |
|---|---|
| GPU Support | 8 × NVIDIA H200 / H100 SXM5 (HGX) |
| GPU Memory | Up to 1.1 TB HBM3e aggregate |
| CPU Sockets | 2 × Intel Xeon 6 / AMD EPYC 9005 |
| System Memory | 24 DIMMs, up to 3 TB DDR5-5600 |
| NVMe Storage | 8 × 3.84 TB Gen5 U.2 |
| Networking | 8 × 400G InfiniBand + 2 × 25GbE |
| PCIe | PCIe Gen5 with dedicated GPU lanes |
| Power | 6 × 5500W 80+ Titanium |
| Cooling | Liquid-cooled / DLC ready |
| Form Factor | 8U HGX Server |