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

AI inference chipmaker d-Matrix today announced it will use NVIDIA NVLink Fusion to connect its next-generation Raptor XPUs to NVIDIA’s AI infrastructure platform — joining a growing roster of ecosystem partners. By connecting Raptor to NVIDIA NVLink scale-up and Spectrum-X scale-out networking, the NVIDIA MGX rack architecture and the broader NVIDIA AI platform, NVLink Fusion gives d-Matrix an accelerated, lower-risk path from custom silicon to large-scale deployment. “Demand for inference is soaring, but capital, time and energy remain finite,” said Sid Sheth, cofounder and CEO of d-Matrix during a press briefing yesterday. “With NVLink Fusion and MGX, we can integrate our Raptor XPUs into a broadly deployed, liquid-cooled architecture, giving customers a faster, lower-risk path to deploy and scale ultralow-latency inference.” The NVIDIA AI platform is vertically integrated and horizontally open. NVLink Fusion extends this openness to XPUs and CPUs, allowing silicon companies to focus on their processor innovations while using NVIDIA infrastructure to deploy them at AI factory scale. From Custom Silicon to Rack-Scale Deployment NVLink Fusion — Quick Reference What is NVIDIA NVLink Fusion? NVIDIA’s platform for integrating third-party custom XPUs and CPUs with NVLink scale-up networking, MGX rack architecture, and the full AI factory platform — spanning compute, networking, storage, security, power, cooling and software. What problem does it solve? Building a custom XPU is hard. Deploying one at scale is harder. NVLink Fusion lets XPU makers skip building rack-scale infrastructure from scratch and plug directly into NVIDIA’s proven, globally deployed AI factory platform. How fast is it? 3x lower XPU-to-XPU latency than off-the-shelf Ethernet, 10x higher packet rates, and 3 TB/s per XPU of all-to-all bandwidth via sixth-generation NVLink. Who are the partners? d-Matrix, AWS, Arm, Intel, Fujitsu, SiFive, Alchip, Astera Labs, GUC, Marvell, MediaTek, Samsung, Cadence, Synopsys, Ayar Labs and Lightmatter. What CPU architectures does it support? All major CPU architectures — Arm, x86 and RISC-V — alongside NVIDIA GPUs. What is a semi-custom AI factory? A semi-custom AI factory pairs NVIDIA’s proven infrastructure with a partner’s custom XPU. NVLink Fusion provides the interconnect, rack architecture, software and supply chain; the partner focuses on differentiated silicon. Building an XPU is only the first step. Deploying it at AI factory scale requires a complete platform spanning networking, rack architecture, power, cooling, software and a proven supply chain. Each of the steps — sourcing chips, integrating high-speed interfaces, validating a scale-up networking solution, designing and certifying rack architecture — adds time, costs and risks. NVLink Fusion lets silicon innovators connect directly into NVIDIA’s proven platform . It’s the high-bandwidth, low-latency technology that connects custom XPUs and CPUs to the NVIDIA stack. By adopting NVLink Fusion, d-Matrix can tap into the NVIDIA MGX ecosystem’s mature, validated rack designs, supply chain, power and cooling infrastructure. By standardizing on a common rack, data centers can be built once and support GPUs, CPUs and XPUs — without requiring a separate rack architecture for each processor type. NVLink Fusion Integrates d-Matrix XPUs Into AI Factories Using NVIDIA NVLink, d-Matrix plans to connect its XPUs in a single high-bandwidth, low-latency scale-up domain. Its racks can also work alongside NVIDIA GPU-based systems like NVIDIA Vera Rubin NVL72 for disaggregated inference. d-Matrix also plans to integrate NVIDIA Vera CPUs, NVIDIA ConnectX-9 SuperNICs, NVIDIA BlueField-4 DPUs and NVIDIA Spectrum-X Ethernet networking. Together with NVLink and MGX, these technologies give d-Matrix a proven foundation to deploy specialized inference alongside NVIDIA systems within flexible, unified AI factories. NVLink Fusion Opens NVIDIA AI Factories to Specialized XPU Architectures The NVIDIA full-stack AI factory platform includes NVIDIA Vera Rubin NVL72, Groq 3 LPX, the Vera CPU rack, Vera BlueField-4 STX storage and Spectrum-6 SPX Ethernet networking. It’s designed to be completely fungible — running every AI workload, model and model architecture — with the best performance per watt and lowest cost per token. NVLink Fusion gives customers the flexibility to match the right compute to each workload within a common AI factory platform, opening access to NVIDIA networking, systems, software and global supply chain. Silicon innovators like d-Matrix can use NVLink Fusion to increase performance, accelerate time to market and reduce the risk of deploying semi-custom AI factories.