HGX H200
NVIDIA's flagship multi-GPU interconnect platform for data centers, featuring an 8-GPU configuration connected via NVLink and NVSwitch high-bandwidth interconnects, with a total memory capacity of 1.1 TB (HBM3e). Compared to the previous-generation H100, the platform delivers substantially increased memory capacity and bandwidth, and is optimized for large-model training and inference workloads.
Timeline (last 90 days)
The purchase price of an 8-GPU HGX H200 server is approximately in the range of $320,000 to $420,000, with a median of about $370,000.
The 8-GPU configuration of the HGX H200 interconnects 8 H200 GPUs with high bandwidth via NVLink and NVSwitch, with a total VRAM of up to 1.1 TB (HBM3e).
Comparing a purchase price of $370,000 against a rental cost of $35.20 per hour, considering hardware alone, the break-even point is approximately 14.4 months at 100% utilization, approximately 24 months at 60% utilization, and approximately 36 months at 40% utilization.
All Facts (2)
The 8-GPU configuration of the HGX H200 interconnects 8 H200 GPUs with high bandwidth via NVLink and NVSwitch, with a total VRAM of up to 1.1 TB (HBM3e).
50%UnverifiedComparing a purchase price of $370,000 against a rental cost of $35.20 per hour, considering hardware alone, the break-even point is approximately 14.4 months at 100% utilization, approximately 24 months at 60% utilization, and approximately 36 months at 40% utilization.
50%