NVIDIA DGX H200 – 8× 141GB GPUs | 2TB DDR5 RAM | 400Gb/s Networking | Hyperscale AI Supercomputing
NVIDIA DGX H200 AI Supercomputing Server – 8× H200 GPUs | Dual Xeon Platinum 8480C | 2TB RAM | Ultra-High-Speed Networking
Specifications
CPU: 2 × Intel Xeon Platinum 8480C (56 Cores each, 2.0GHz)
RAM: 2TB DDR5 System Memory
Storage: 2 × 1.9TB NVMe M.2 SSDs + 8 × 3.84TB NVMe U.2 SSDs
GPU: 8 × NVIDIA H200 141GB Tensor Core GPUs (Total 1128GB HBM3e Memory)
Networking: 4 × OSFP (8 × Single-Port 400Gb/s InfiniBand/Ethernet)
2 × Dual-Port NVIDIA ConnectX-7 VPI
1 × 400Gb/s InfiniBand/Ethernet Port
1 × 200Gb/s InfiniBand/Ethernet Port
Support: Includes 3 Years NVIDIA Business-Standard Support
The NVIDIA DGX H200 represents the new frontier of AI supercomputing. Featuring 8 NVIDIA H200 Tensor Core GPUs with a combined 1128GB of high-bandwidth HBM3e memory, dual 56-core Intel Xeon Platinum CPUs, 2TB system memory, and massive NVMe storage, the DGX H200 is built for training the largest language models, generative AI, AI simulation, and scientific discovery workloads. With advanced 400Gb/s networking and the NVIDIA AI Enterprise software stack, it provides seamless scalability for hyperscale AI environments.
