AI Startup d-Matrix to Integrate Nvidia NVLink Fusion and MGX Designs
2 sources · United Kingdom
Who reported this
- Reuters
- The Register
What the colours mean
- Left
- Centre-left
- Centre
- Centre-right
- Right
- A hatched block means the outlet is affiliated with, or controlled by, a state.
Political lean describes where an outlet sits within the politics of its own country. It is never a position on a single global scale.
Political lean is comparable inside one country and not across them, which is why the bar groups by country first. Publicly funded broadcasters are not marked as state-linked.
The owner of each outlet is listed as a matter of record, not as a judgement about the outlet.
Every source for this story reports from United Kingdom.
Every outlet covering this story shares the same political lean; read with that in mind.
AI infrastructure startup d-Matrix announced on Thursday that it will license Nvidia's NVLink Fusion interconnect technology and MGX rack scale reference designs. This agreement allows d-Matrix to integrate high speed chip to chip interconnects into its future chip designs, including its upcoming Raptor accelerators. By adopting these standards, d-Matrix enables its customers to deploy its chips using the same racks and NVSwitch fabrics as Nvidia.
d-Matrix expects to offer systems by the end of next year featuring up to 144 Raptor accelerators connected by a single all to all NVLink fabric. The company's Raptor cards are designed to address memory bandwidth bottlenecks in AI inference. According to reports, the Raptor cards feature 3D stacked DRAM to provide high memory bandwidth, which may allow the company to use fewer chips per model compared to competitors like Groq.
Beyond the interconnect technology, d-Matrix plans to pair its accelerators with other Nvidia products, including Vera CPUs, NVSwitch appliances, BlueField and ConnectX NICs, and SpectrumX Ethernet products. This move follows a trend of other chip designers, such as MediaTek, Marvell, and Qualcomm, licensing Nvidia's IP to avoid the challenges of designing their own scale up networks or rack systems.
How each side framed it
- Centre
- The coverage focused on the technical specifications of the hardware and the strategic business implications of Nvidia's IP licensing model.
Sources
100% of the statements in this article were traced back to the source articles listed above.