The B200 is NVIDIA's first-generation Blackwell data center GPU: two reticle-limited dies joined by a 10 TB/s NV-HBI link and packaged as one SXM6 module with 180 GB of HBM3e. The system is built on eight NVIDIA B200 Tensor Core GPUs. The NVIDIA DGX B200 (1,440 GB) system includes the following components. The system includes six power supply units (PSU). However, in real-world deployments, GH200-based AI servers and clusters require a large number of high-speed optical modules to enable communication between servers, switches, and the broader data center network infrastructure. It sells as a bare module only inside an 8-GPU HGX B200 baseboard, never standalone, and that baseboard is the building. NVIDIA DGX B200 delivers a unified platform for training, fine-tuning and inferencing in a single solution optimized for enterprise AI workloads, powered by the NVIDIA Blackwell GPU. Analysts and technical sources have estimated the number of optical modules per GH200 chip based on cluster network configurations: In a cluster with 256 GH200 chips, each chip contributes to the total interconnect load. Assuming 32 ports at 800 G per NVLink switch and that only L2 uses optical.
[PDF Version]Contact us for competitive quotes and expert installation services
Get a Quote