Chip Startup d-Matrix to Use Nvidia Chip-Linking Tech in AI Servers

Chip startup d-Matrix said on Thursday it will use Nvidia’s chip-linking technology to connect its artificial intelligence processors directly to Nvidia’s data-center systems, marking a significant step for the young company as demand for AI computing shifts increasingly toward inference.

The Santa Clara, California-based startup said its new Raptor chips will use Nvidia’s NVLink Fusion technology, which allows custom AI processors to connect to Nvidia’s larger data-center platforms. The technology combines high-speed connections and specialized memory to allow third-party chips to work within Nvidia-based server systems.

The partnership focuses on AI inference, the process of running trained AI models to generate responses and perform tasks for users. While Nvidia’s graphics processors have become dominant in AI model training, d-Matrix specializes in inference workloads, where speed and low latency are particularly important.

The companies are targeting applications such as coding assistants, chatbots and voice agents, where users expect AI systems to respond almost immediately. D-Matrix says integrating its processors with Nvidia’s infrastructure could help deliver these services more efficiently while giving data-center operators another option for handling increasingly demanding AI workloads.

D-Matrix’s Raptor chips are expected to complete their final design stage by the end of this year, while Nvidia-compatible server racks are projected to become available in 2027. The startup is also working with connectivity company Astera Labs to develop customized solutions for moving data rapidly across the combined systems.

Financial terms of the collaboration were not disclosed. Still, the agreement gives d-Matrix access to an important ecosystem at a time when AI companies and data-center operators are looking beyond conventional graphics processors to meet rapidly growing demand.

The startup has attracted significant backing from major technology investors. Microsoft has supported d-Matrix since its $110 million financing round in 2023. The company shipped its first AI chip in November 2024 and was valued at about $2 billion after raising $450 million last year.

The move also illustrates how Nvidia is increasingly opening its data-center architecture to specialized processors. For d-Matrix, compatibility with Nvidia’s systems could provide a path to reach customers without requiring data centers to replace their existing infrastructure.

As AI applications move from experimental systems into everyday services, the competition to provide faster and more cost-efficient inference hardware is intensifying. D-Matrix is betting that its specialized processors, combined with Nvidia’s connectivity technology, can capture a share of that expanding market.