d-Matrix launches Squadron clusters for data center AI inference

d-Matrix announced Squadron, a fleet of multi-rack compute clusters built to run AI models at data center scale with high speed and energy efficiency. The systems use the company’s memory-centric compute architecture and are aimed at AI labs, enterprises, and sovereign organizations running latency-sensitive workloads.

Target applications include coding, personal agents, cybersecurity, and high-frequency trading. Squadron was developed with Arista, Broadcom, and Supermicro, and has been deployed in d-Matrix’s Arizona data center for internal research and development. The clusters are now shipping to priority customers, while the underlying Corsair servers and Corsair SquadRacks are generally available.