d-Matrix Adopts NVIDIA NVLink Fusion to Deploy Raptor XPUs at Rack Scale
NVIDIA Blog · rss · 2026-09-10
Inference chipmaker d-Matrix will use NVIDIA NVLink Fusion to connect its next-gen Raptor XPUs to NVIDIA's AI factory platform, joining partners including AWS, Intel, Marvell, MediaTek, and Samsung.
- The problem: building custom silicon is only step one; rack-scale deployment requires networking, power, cooling, software, and supply chain. NVLink Fusion lets XPU makers skip building that infrastructure and plug into NVIDIA's proven platform.
- Performance: 6th-gen NVLink delivers 3x lower XPU-to-XPU latency than Ethernet, 10x higher packet rates, and 3 TB/s per-XPU all-to-all bandwidth.
- Deployment: d-Matrix will build a single scale-up domain with NVLink, interoperate with Vera Rubin NVL72 systems for disaggregated inference, and integrate Vera CPUs, ConnectX-9 SuperNICs, BlueField-4 DPUs, and Spectrum-X Ethernet.
- Ecosystem: the MGX rack architecture supports Arm, x86, and RISC-V, letting one data center design host GPUs, CPUs, and XPUs — a faster, lower-risk path for semi-custom AI factories.
More from Infra
- M.2-Oculink eGPU Link Silently Downgrades to PCIe Gen1 — Here's How to Check — El_90 · 2026-09-11
- DeepSeek launches V4.1-Flash with 1M-token context and 4x smaller KV-cache — matlabulous · 2026-09-11
- What Can You Still Run on 8GB VRAM? User Asks for Small Models With Tool Use — riceinmybelly · 2026-09-11
- Spain's hourly 80% renewable matching rules clash as France fast-tracks 700MW sites, UK cuts grid queues — eherrerosj · 2026-09-11
- AI could add 0.3-0.4 points to Europe's productivity growth, but the EU holds under 5% of global compute — rohanpaul_ai · 2026-09-11
- Qualcomm's next-gen Hexagon NPU runs 30B MoE models with 32K context on-device — lee_stott · 2026-09-11