AIartificial intelligence inference chipmaker d-Matrix has announced that its next-generation Raptor XPU1. accelerator processing unitXPU is used here for the compute package that hosts multiple processing elements.2. accelerated processing unitA collective term for a high-performance processor in the package architecture described.s will be connected to NVIDIA’s AIartificial intelligence infrastructure platform through NVLink Fusion. The move places d-Matrix inside a broader ecosystem of third-party silicon that can be integrated into NVIDIA-based racks and systems.
Technically, the setup combines NVLink scale-up with Spectrum-X scale-out networking, the NVIDIA MGX rack architecture and the wider NVIDIA AIartificial intelligence platform. According to d-Matrix, this gives the company a more direct route from custom silicon to large-scale deployment, while reducing the integration burden that normally comes with high-speed interfaces, networking validation and rack certification.
The company says its goal is to deploy Raptor in a broadly used, liquid-cooled architecture and to support ultralow-latency inference at scale. NVIDIA describes NVLink Fusion as an open extension of its platform that lets silicon companies focus on their processor design while relying on NVIDIA’s infrastructure for deployment. In this model, the rack, power, cooling, software stack and supply chain are treated as part of the platform rather than separate design problems.
d-Matrix also plans to integrate NVIDIA Vera CPUcentral processing unitMain processor that executes program instructions and performs data operations.s, ConnectX-9 SuperNICs, BlueField-4 DPUs and Spectrum-X Ethernet networking. Its racks are intended to work alongside NVIDIA GPUgraphics processing unitA GPU is a parallel graphics processor widely used for AI acceleration. systems such as Vera Rubin NVL72 for disaggregated inference. The result is a common rack and interconnect foundation that can support GPUgraphics processing unitA GPU is a parallel graphics processor widely used for AI acceleration.s, CPUcentral processing unitMain processor that executes program instructions and performs data operations.s and XPU1. accelerator processing unitXPU is used here for the compute package that hosts multiple processing elements.2. accelerated processing unitA collective term for a high-performance processor in the package architecture described.s within a unified AIartificial intelligence factory environment.
For d-Matrix, NVLink Fusion and MGX are meant to provide a validated base for bringing specialized inference hardware into large deployments with less risk and less integration effort than a fully custom rack design.