d-Matrix, which designs chips optimized for the “inference” part of AI processing, raised a $110M Series B led by Temasek, taking its total funding to $154M
Context & Ripple Effects
D-Matrix had already raised a $44M Series A for its chiplet-based Nighthawk architecture, positioning the company around faster matrix-math processing. This Series B extends that financing path with Temasek as lead investor and gives the company a larger capital base to pursue its inference-focused design.
The raise also foreshadows the company’s later $275M inference-chip financing at a $2B valuation, showing that investors continued to fund D-Matrix’s specialization rather than treating the Series B as a one-off hardware bet.
First-order effects
- D-Matrix adds $110M and reaches $154M in total funding, increasing its capacity to advance and commercialize chips aimed at AI inference workloads.
- Temasek becomes the lead backer in this round, tying its capital to a specialized AI-chip company rather than a broader software or consumer-internet bet.
Second-order effects
- The new funding raises the competitive bar for other inference-chip startups: competing designs will need capital not only for chip development but also to sustain the path toward deployment.
- A larger privately funded D-Matrix gives prospective buyers and partners another specialized supplier to evaluate for inference, alongside more general-purpose AI compute options.
Third-order effects
- If follow-on funding continues—as the later $275M round indicates—AI hardware financing may increasingly favor companies with a narrowly defined workload advantage, such as inference, over undifferentiated accelerator claims.
- The pattern points to a more capital-intensive AI infrastructure market in which investor conviction must persist across multiple rounds before a chip architecture can prove commercial traction.
The trend: This is one data point in the financing of specialized AI infrastructure companies built around distinct stages of the compute workload, particularly inference.