The NVIDIA 2-Way NVLink Bridge for H200 NVL delivers a 900 GB/s bidirectional GPU-to-GPU interconnect using 4th Generation NVLink technology, enabling two H200 NVL GPUs to operate as a unified memory and compute resource within a single server. Running 7x faster than PCIe Gen5 and offering 50% higher bandwidth than the H100 NVL equivalent, it eliminates PCIe bottlenecks for demanding AI training, large language model inference, HPC, and data-intensive analytics workloads.

Compatible with HPE ProLiant DL380a Gen12 and DL385 Gen11 servers, one bridge is required per pair of H200 NVL GPUs. The low-latency direct GPU-to-GPU fabric reduces cost of service while maximising parallel compute throughput in tightly coupled multi-GPU configurations. Get it now at authentic hardware, reliable service.