Description
Accelerate your most demanding AI and HPC workloads with the Supermicro SYS-A22GA-NBRT-G1, a 10U GPU server built on the NVIDIA HGX B200 platform. Powered by 8 NVIDIA Blackwell SXM GPUs delivering 1.44TB of aggregate HBM3e memory, dual Intel Xeon 6960P processors with 144 total cores, and 2TB of DDR5-6400 ECC memory, this system is engineered for large-scale AI training, LLM reasoning, generative AI, and complex scientific computing. As a Gold Series quick-ship configuration, it is optimized for rapid deployment with no lead time on custom builds.
The SYS-A22GA-NBRT-G1 delivers exceptional GPU-to-GPU communication with 1.8 TB/s of bandwidth per GPU via fifth-generation NVLink and NVSwitch, providing 14.4 TB/s of total bisection bandwidth across all eight GPUs for the largest distributed training runs. Eight NVIDIA ConnectX-7 NICs provide 8x 400 Gb/s OSFP ports supporting InfiniBand NDR or 400GbE, ensuring seamless integration into high-performance AI fabrics and multi-node clusters. An NVIDIA BlueField-3 DPU adds a dedicated 200 Gb/s network path for infrastructure offload, while dual 10GbE RJ-45 ports and a dedicated 1GbE BMC/IPMI port handle management and storage networking.
With 11.4TB of raw NVMe storage across M.2 and U.2 drives, six redundant 5250W power supplies in a 3+3 configuration, 10x PCIe 5.0 x16 low-profile and 2x PCIe 5.0 x16 full-height expansion slots, the SYS-A22GA-NBRT-G1 provides a dense, air-cooled, and scalable foundation for next-generation AI infrastructure. Backed by a 3-year onsite next-business-day service warranty and NVIDIA Enterprise Standard Support for BlueField software, this system is ready to deploy in production AI factories, neoclouds, and enterprise data centers.
