The World’s Proven Choice for Enterprise AI
Expand the frontiers of business innovation and optimization with NVIDIA DGX™ B200 and NVIDIA DGX™ H200. Thesepowerful systems form the backbone of cutting-edge AI infrastructures, enabling enterprises to take innovation and efficiency to a new level – with maximum computing power, scalability, and reliability.
NVIDIA DGX B200 Datasheet
Download
NVIDIA DGX H200 Datasheet
Download
DGX B200
- 8 NVIDIA B200-GPUs with 1,440 GB of Total GPU Memory
- 2 NVIDIA NVSwitchesTM
- 14.4 TB/s aggregate bandwith
- 14.4 TB/s aggregate bandwith
- 10 NVIDIA CONNECTX®-7 400 GB/s Network Interface
- 1 TB/s of peak bidirectional network bandwith
- 1 TB/s of peak bidirectional network bandwith
- Dual Intel Xeon Platinum 8570 processors, 112 cores total, and 2 TB System Memory
- Powerful CPUs for the most intensive AI jobs
- Powerful CPUs for the most intensive AI jobs
- 30 TB NVMe SSD
- High speed storage for maximum performance
The Proven Standard for Enterprise AI
DGX H200
- 8 NVIDIA H100-GPUs with 1,128 GB of Total GPU Memory
- 18 NVIDIA® NVLink® connections per GPU, 900 GB/s of bidirectional GPU-to-GPU bandwith
- 18 NVIDIA® NVLink® connections per GPU, 900 GB/s of bidirectional GPU-to-GPU bandwith
- 4 NVIDIA NVSwitchesTM
- 7.2 TB/s of bidirectional GPU-to-GPU bandwith, 1.5x more than previous generation
- 7.2 TB/s of bidirectional GPU-to-GPU bandwith, 1.5x more than previous generation
- 10 NVIDIA CONNECTX®-7 400 GB/s Network Interface
- 1 TB/s of peak bidirectional network bandwith
- 1 TB/s of peak bidirectional network bandwith
- Dual Intel Xeon Platinum 8480C processors, 112 cores total, and 2 TB System Memory
- Powerful CPUs for the most intensive AI jobs
- Powerful CPUs for the most intensive AI jobs
- 30 TB NVMe SSD
- High speed storage for maximum performance
