d-Matrix Teams Up with Nvidia to Accelerate AI Inferencing

d-Matrix's CEO Sid Seth said that his company has spent seven years developing its platform of hardware and software designed to speed up AI inferencing workloads while driving down costs.

The startup, which has received over $500 million in funding, including from Microsoft's M12 venture arm, introduced its first inference accelerator platform, Corsair, in June. The platform takes a memory-centric approach that can help run inferencing workloads ten times faster than Nvidia GPUs alone.

Now, d-Matrix is pairing its Raptor XPU with Nvidia's Blackwell GPU accelerators in a rack. The XPU tightly couples the memory, such as SRAM or 3D-RAM, with GPUs and CPUs in the same rack to accelerate the generation of tokens and reduce costs.

The partnership between d-Matrix and Nvidia will make d-Matrix's Raptor XPU platform available through Nvidia's MGX reference architecture, deployed as part of Nvidia's AI factory offering. The partnership also includes future versions of d-Matrix's XPUs, including Lightning.