d-Matrix is a semiconductor and computing platform company founded in 2019 and focused exclusively on generative AI inference. Rather than adapting existing GPU architectures, the company built its platform from scratch around a core technical bet: that the principal constraint in running large AI models at scale is not raw compute power, but the movement of data between memory and processors. Its answer is digital in-memory compute, a design approach that brings computation closer to where data is stored.
The company's flagship product, Corsair, is a computing platform engineered specifically for inference workloads - the process of running a trained AI model to generate outputs - rather than for training. Corsair is designed to deliver low latency and high throughput while keeping energy consumption and costs at levels that make large-scale generative AI commercially viable for a broader range of organisations, not only the largest technology companies.
d-Matrix has grown from a small founding team to more than 200 people. Its engineering work spans silicon design, software, and systems engineering, with a stated emphasis on first-principles thinking applied to hard problems at the intersection of all three disciplines. The company operates in the AI infrastructure space, targeting cloud and enterprise inference at scale.