Description
Accelerate your most demanding AI and HPC workloads with the Gigabyte G894-SD3-AAX7, a flagship 8U GPU server built on the NVIDIA HGX B300 platform. Powered by 8 NVIDIA Blackwell Ultra SXM GPUs delivering 2.3TB of HBM3e memory, dual Intel Xeon 6767P processors, and 3TB of DDR5 ECC memory, this system is engineered for large-scale AI training, LLM reasoning, generative AI, and complex scientific computing.
The G894-SD3-AAX7 delivers exceptional GPU-to-GPU communication with 1.8 TB/s of bandwidth per GPU via fifth-generation NVLink and NVSwitch, and 14.4 TB/s of total system interconnect bandwidth — enabling seamless scaling across all eight GPUs for the largest model training runs. Onboard NVIDIA ConnectX-8 SuperNICs provide 8x 800 Gb/s OSFP ports supporting InfiniBand XDR or dual 400 Gb/s Ethernet, ensuring this server integrates seamlessly into high-performance AI fabrics and multi-node clusters. An additional Mellanox ConnectX-6 dual-port 100GbE NIC is included for dedicated management and storage networking.
With 15.36TB of high-speed NVMe storage, redundant 80 PLUS Titanium power supplies, 4x PCIe Gen5 expansion slots, and support for NVIDIA BlueField-3 DPUs, the G894-SD3-AAX7 provides a dense, air-cooled, and scalable foundation for next-generation AI infrastructure. Backed by a 3-year limited warranty with return-to-depot service, this system is ready to deploy in production AI factories, neoclouds, and enterprise data centers.
