D-Matrix adopts NVLink Fusion for rack-scale AI inference

d-Matrix Adopts NVIDIA NVLink Fusion for Rack-Scale XPU Deployment

d-Matrix adopts NVLink Fusion for Raptor XPU, achieving 3x lower latency and 3 TB/s bandwidth for inference scale-out. The move signals NVIDIA opening its AI factory platform to third-party chipmakers, enabling unified rack architecture for GPUs, CPUs, and custom XPUs without infrastructure duplication.