The company’s AI Data Center GPU Cloud is intended for customers facing restrictions on transferring data overseas and those seeking infrastructure optimized specifically for the Japanese market
In sum – what to know:
Sovereign AI focus – Softbank said the platform is designed for customers facing restrictions on overseas data transfers, offering AI infrastructure and operations hosted within Japan.
Enterprise AI demand – The company is targeting large enterprises and technology firms seeking inference environments for internally developed AI models.
Nvidia infrastructure – The service combines Nvidia GB200 NVL72 systems with Softbank’s “Infrinia AI Cloud OS” to support AI training, inference and Kubernetes-based operations.
Japanese company Softbank said its new AI cloud platform will focus on enterprises requiring sovereign AI infrastructure within Japan, positioning the service as a domestic alternative to global hyperscalers for organizations with data residency and security requirements.
In responses to questions from RCR Wireless News, a Softbank spokesperson said the company’s “AI Data Center GPU Cloud” is intended for customers facing restrictions on transferring data overseas and those seeking infrastructure optimized specifically for the Japanese market.
“Our AI Data Center GPU Cloud is particularly well suited for customers facing restrictions on transferring data overseas, as it enables the provision of a secure environment that ensures both data sovereignty and a high level of security,” the spokesperson said.
The company added that, unlike global hyperscalers operating largely standardized global platforms, Softbank plans to provide services tailored to local enterprise requirements in Japan.
Softbank officially announced the new cloud platform as part of its “Activate AI for Society” growth strategy. The commercial launch is scheduled for October 2026, while a beta version became available this week for internal use across Softbank group companies.
The platform combines AI infrastructure deployed in Softbank’s Japan-based data centers with the company’s “Infrinia AI Cloud OS,” a software stack designed for AI data center operations.
The service will support Kubernetes as a Service (KaaS) for multi-tenant environments and Inference as a Service (Inf-aaS) for large language model inference through APIs.
Softbank said the infrastructure uses Nvidia GB200 NVL72 systems and Nvidia Blackwell GPUs interconnected through Nvidia NVLink technology. According to the company, the platform is intended to support AI workloads ranging from model training and inference to data processing.
The company said centralized and automated GPU resource management, Kubernetes-based operations and workload orchestration are designed to reduce infrastructure management complexity and lower operational costs.
Softbank also said the service is intended to support broader efforts around “Telco AI Cloud,” its initiative to build AI infrastructure using telecommunications network assets. The company plans to integrate the GPU cloud platform with AI-RAN infrastructure to optimize AI processing from training through inference while supporting distributed AI deployments with lower latency.
The company said it expects early demand primarily from large enterprises, technology firms, and organizations seeking inference environments for internally developed AI models.
“We are primarily targeting large enterprises, tech companies and other companies seeking inference environments for the AI models they have developed,” the spokesperson said.
Softbank said the beta rollout is currently focused on internal testing and deployment across its own group companies ahead of broader commercial availability later this year.
“As we have only just launched the service, we will continue preparing for the full-fledged rollout based on feedback from internal users and group companies,” the spokesperson added.
