AI chip startup D-Matrix received $110 million in Series B investment from Microsoft

D-Matrix, an AI chip startup founded in 2019 by Sid Sheth, a former director at Broadcom, and Sudeep Bhoja, recently secured an investment of $110 million in a Series B round, supported by notable investors such as Singapore’s Temasek, California’s Playground Global, as well as Ericsson, Marvell, SK Hynix, Samsung, and Microsoft’s venture arm, M12.

Currently, D-Matrix boasts a workforce of around 100, with Sid Sheth serving as CEO and Sudeep Bhoja as CTO. Their product, leveraging in-memory computing technology combined with chip-level horizontal scaling, is designed for data center-scale AI computations. This facilitates a computational speed surpassing traditional CPUs and GPUs, and it is optimized for large-scale AI models like the Transformer.

In a recent interview with Reuters, Sid Sheth elucidated that their AI technology focuses primarily on enhancing the inference stage, rather than vying with competitors in training large-scale AI models. He emphasized that while extensive AI models can expedite their training process through GPU acceleration, pure hardware acceleration isn’t the sole answer to enhance inference efficiency. Their product can elevate the efficiency of the inference process by 10-30 times.

Having secured investment from major players including Microsoft, Sheth interprets this as a sign of the growing market demand for AI computation and suggests the company’s potential for deeper collaboration with these industry giants. Depending on D-Matrix’s product performance anticipated for the following year, Microsoft might evaluate further investment opportunities.

D-Matrix projects a revenue nearing $10 million this year and anticipates an increase to a range of $70-75 million within two years, aiming for a balanced profit and loss.