NTT Docomo Business launches GPUaaS for AI development

NTT Docomo Business has announced the launch of ‘GPUaaS’ (GPU as a Service), a cloud-based AI development platform scheduled to begin service on September 30. The service allows users to rent high-performance NVIDIA GPUs, such as the B300 and A100, starting from a single unit. This model is designed to support various stages of development, from small-scale Proof of Concept (PoC) testing to full-scale production.

A key feature of the service is the integrated AI/GPU controller, which enables users to manage GPU resource allocation, visualize usage, and handle tasks such as training job execution and dataset management through web browsers using tools like JupyterLab and Visual Studio Code. The platform also supports hybrid configurations, allowing users to combine cloud-based GPU resources with dedicated, private GPU environments.

To address data security and sovereignty, the service supports ‘Sovereign AI’ by allowing companies to manage sensitive data and AI models within Japan. By utilizing NTT Docomo Business’s domestic data centers and secure closed networks, the service ensures that AI processing, data storage, and integration with existing business systems can be completed entirely within the country.