d-Matrix Joins Nvidia's NVLink Fusion Ecosystem

California-based AI inference chip maker d-Matrix announced it is joining Nvidia's NVLink Fusion ecosystem. Its rack-level XPU systems will use Nvidia's MGX universal rack reference architecture.

Starting with its next-gen "Raptor" inference XPU, d-Matrix products will integrate with Nvidia's Vera CPU, NVLink switches, BlueField-4 DPU, ConnectX-9 SuperNIC, and Spectrum-X Ethernet networking to enable both scale-up and scale-out expansion.

This partnership lets d-Matrix gain rack-level scalability without complex design work, allowing it to focus more on developing its own AI XPU.