Santa Clara, Calif. – On Thursday, d‑Matrix, a fast‑growing artificial‑intelligence chip maker, revealed a strategic partnership with Nvidia that will allow its Raptor inference processors to plug directly into Nvidia’s data‑center servers. The collaboration uses Nvidia’s NVLink Fusion technology, which provides high‑speed connectors and dedicated memory so custom AI chips can operate alongside Nvidia’s larger GPU platforms.
Why the partnership matters for AI inference
AI workloads are shifting from the expensive, power‑hungry training phase to the everyday inference phase, where speed and low latency are critical. While Nvidia’s graphics processors continue to dominate the training market, d‑Matrix specializes in inference, offering chips designed to deliver rapid responses for applications such as coding assistants, chatbots and voice agents. By integrating Raptor chips into Nvidia’s server racks, the combined system promises to accelerate these services without sacrificing efficiency.
Timeline and rollout
The Nvidia‑compatible server racks are slated for release in 2027. d‑Matrix expects the Raptor chips to complete their final design stage by the end of 2026, positioning the company to meet growing demand for high‑performance inference solutions as AI becomes a staple of everyday software.
Financial backing and valuation
The partnership follows a $110 million financing round in 2023 that was led by Microsoft, underscoring strong investor confidence in d‑Matrix’s technology. The startup shipped its first AI chip in November 2024 and was valued at $2 billion after raising $450 million in a 2025 round. Although the financial terms of the Nvidia collaboration were not disclosed, the deal signals a significant vote of confidence from one of the world’s leading semiconductor firms.
Additional collaborations
d‑Matrix is also working with connectivity specialist Astera Labs to develop custom solutions that ensure fast data flow across the combined system. This joint effort aims to eliminate bottlenecks and maintain the low‑latency performance that AI inference customers demand.
Implications for the tech ecosystem
By enabling third‑party inference chips to operate within Nvidia’s established data‑center infrastructure, the partnership could broaden the market for specialized AI hardware and give customers more flexibility in building AI‑powered services. Industry analysts note that such collaborations may accelerate the transition from monolithic GPU‑only solutions to more diverse, best‑of‑breed architectures.
For the Santa Clara tech community, the announcement highlights the region’s continued role as a hub for cutting‑edge semiconductor innovation. Local talent pipelines and venture‑capital networks are likely to benefit as d‑Matrix scales its operations and hires additional engineers to meet the upcoming production schedule.
Original reporting: Appleton, WI News Feed (HLL/CB) — read the source article.
