Artificial intelligence (AI) startup d-Matrix announced on Wednesday that it has successfully secured $110 million in a Series B funding round.
Artificial intelligence (AI) startup d-Matrix announced on Wednesday that it has successfully secured $110 million in a Series B funding round. Specializing in generative AI compute solutions for data centers, the company intends to utilize the funds to advance its digital-in-memory compute platform, Corsair, based on chiplet-based inference compute technology.
Last month AI Startup, d-Matrix introduced Jayhawk II, the latest iteration of its silicon. Despite being a newcomer in the competitive AI computing and inference market, d-Matrix faces off with industry behemoth Nvidia.
According to Cambrian AI analyst Karl Freund, while Nvidia dominates the AI chip market, d-Matrix offers an in-memory compute platform that promises to be faster and more cost-effective due to its utilization of static random access memory instead of high-bandwidth memory.
Freund expressed confidence in d-Matrix’s potential to be a formidable contender in the market, distinguishing it from other startups like Cerebras. This conviction attracted early-stage investment firm Playground Global to d-Matrix, initially serving as an advisor in 2019 before becoming an investor during the vendor’s inaugural funding round.
Sasha Ostojic, venture partner at Playground Global, highlighted d-Matrix’s exceptional attributes, including its proficient team, cutting-edge technology, and its strategic alignment with the burgeoning AI market. The company’s early bet on generative AI and LLM (Large Language Models) proved prescient, positioning it well for the evolving demands of the industry.
Freund emphasized that the market is gravitating towards d-Matrix’s cost-effective approach to inference, particularly as current methods are deemed excessively expensive. He cited the resource-intensive nature of inferencing processes, requiring a substantial number of GPUs, and stressed the need for more sustainable solutions.
Looking ahead, d-Matrix’s challenge lies in streamlining the deployment of its generative AI models with its Jayhawk II development platform. Ostojic emphasized the importance of execution, with a focus on delivering an enterprise-class software stack for seamless deployment in data centers and hyperscalers.
The availability of d-Matrix’s chiplets is projected for 2024. The Series B funding round was led by investment firm Temasek and included participation from Microsoft’s venture capital fund M12, as well as Playground Global.