d-Matrix adopts NVIDIA NVLink Fusion for rack-scale inference
d-Matrix plans to connect its next-generation Raptor XPUs to NVIDIA's AI infrastructure platform through NVLink Fusion for rack-scale inference.
#
Key takeaways
- d-Matrix plans to connect its next-generation Raptor XPUs through NVIDIA NVLink Fusion.
- NVIDIA says the setup combines custom silicon with NVLink, Spectrum-X networking, and MGX rack infrastructure.
- The focus is rack-scale inference, not training.
- The source gives a useful signal for readers watching inference infrastructure and energy-aware design.
What NVIDIA announced
NVIDIA announced on September 10, 2026 that AI inference chipmaker d-Matrix plans to connect its next-generation Raptor XPUs to NVIDIA's AI infrastructure platform through NVLink Fusion. The announcement frames the arrangement as a way to combine d-Matrix custom silicon with NVIDIA's broader infrastructure stack.
The source is specific about the components involved. It names NVLink Fusion, Spectrum-X networking, and MGX rack infrastructure. It also describes the effort as rack-scale inference.
What the arrangement means
Based on the source, the main idea is integration. d-Matrix is not described as replacing its own silicon approach. Instead, it plans to connect that silicon to NVIDIA's platform.
That matters because inference systems often depend on more than one layer. Chips, networking, and rack design all shape how the system works. The announcement points to a combined approach rather than a single-product story.
Why rack-scale inference matters
The source uses the phrase rack-scale inference. That suggests the system is being discussed at the rack level, not only at the chip level. In practical terms, that means the infrastructure around the chips is part of the story.
The announcement also highlights energy-aware design as relevant background for readers. The source does not provide technical measurements or performance claims. It only shows that infrastructure choices are part of the conversation.
Governance and operational considerations
The source does not mention policy, compliance, or deployment details. It also does not provide pricing, timelines, or customer commitments. Readers should treat the announcement as a product and infrastructure signal, not as proof of market adoption.
Operationally, the key consideration is coordination across layers. Custom silicon, networking, and rack infrastructure must work together. That makes integration a central issue in any similar setup.
Morocco relevance
The source reports no Morocco-specific deployment or local program. For readers in Morocco, the global lesson is simple: inference infrastructure is increasingly shaped by how chips, networking, and racks fit together.
Bottom line
This announcement is about infrastructure alignment. d-Matrix plans to connect its next-generation Raptor XPUs to NVIDIA's AI infrastructure platform through NVLink Fusion. The source presents it as a rack-scale inference arrangement built around custom silicon and NVIDIA's networking and rack stack.
For readers tracking AI infrastructure, the story is less about a single chip and more about system design. That is the main signal the source provides.
Follow us on Google
Add Intelligence Artificielle Maroc as a preferred source to see more of our relevant stories in Google Search.
What would you like to build?
We build custom AI platforms, SaaS products, intelligent business applications, and automation systems.
This form is for project inquiries, not general questions about artificial intelligence.
