日本語 
SoftBank has unveiled ‘Infrinia AI Cloud OS,’ a software stack built for AI data centers. It is designed to help operators run Kubernetes and AI workloads at scale without the complexity of managing every layer themselves. The system supports Kubernetes as a Service and Inference as a Service, letting users deploy Large Language Models via APIs and run GPU cloud services more efficiently.
The goal is to reduce operational burden and cut costs compared with building custom solutions in-house. Operators can automate everything from BIOS and GPU drivers to networking and storage, and scale clusters dynamically based on workload. The software also handles secure multi-tenancy and automated maintenance, including monitoring and failover.
Also Read: TGES Joins CBRE & La Clé du Joie for Data Centers
SoftBank will first deploy Infrinia AI Cloud OS in its own GPU cloud. The company plans to expand it to overseas data centers over time. The shift takes place at the same time as the increasing need for GPU-assisted AI at a quick rate in the areas of generative AI, robotics, simulations, drug discovery, and materials development. SoftBank’s strategy of easing deployment and optimizing GPU power aims to bring advanced AI infrastructure within the reach of cloud service providers and enterprise users, and to make it more adaptable.
English