Nvidia to Invest in AI Inference Chip Startup d-Matrix

Nvidia is planning an equity investment in AI inference chip startup d-Matrix to expand its hardware ecosystem, according to a report by The Information. The planned funding builds on an existing technical partnership and follows Nvidia shares closing down 3.16% at $221.30, compared to a previous close of $228.52, as of October 14, 2026, according to Finnhub market data.

Hardware Integration and NVLink Fusion

The upcoming investment follows an agreement by d-Matrix to integrate its next-generation Raptor XPUs with Nvidia's NVLink Fusion and MGX rack architecture for speculative decoding workloads. This hardware integration allows d-Matrix accelerators to connect directly into Nvidia-based server configurations, combining alternative chip designs with Nvidia's high-speed interconnect infrastructure.

Speculative decoding is a technique used in large language model inference where a smaller, faster model generates candidate tokens that are then verified in parallel by a larger model. By pairing d-Matrix hardware with Nvidia systems via NVLink Fusion, data centers can run specialized inference workloads with lower latency and higher throughput compared to standard homogeneous clusters.

Valuation and Funding History

For d-Matrix, backing from the dominant market leader in artificial intelligence hardware arrives as the startup scales its private capital raising. According to industry reports, d-Matrix closed a $275 million Series C funding round in November 2025 at a $2 billion valuation, and has subsequently targeted a valuation of at least $5 billion in ongoing fundraising discussions.

The startup's technology focuses specifically on artificial intelligence inference rather than model training. While training requires massive computational horsepower to build foundational models, inference handles the execution and deployment of models during user queries, a segment of the data center market experiencing rapid growth as enterprise adoption expands.

Strategic Shifts in the Accelerator Market

Nvidia's decision to fund a company developing alternative chip designs highlights a shift toward interoperability within specialized data center infrastructure. Rather than maintaining a closed hardware ecosystem, the chipmaker is working to ensure that competing architectures can integrate cleanly with its proprietary networking standards like NVLink.

For enterprise buyers deploying artificial intelligence clusters, the convergence of multiple accelerator types within a single rack architecture reduces deployment friction. As data center operators look for ways to optimize energy consumption and computational efficiency for inference workloads, collaborations between dominant market incumbents and specialized startups are reshaping hardware deployment strategies across the industry.