SoftBank Makes Domestic AI Data Centre Move

Why It Matters

The offering gives Japan a domestic high‑performance AI compute platform, reducing reliance on foreign cloud providers and bolstering local AI innovation.

Key Takeaways

• SoftBank beta tests AI Data Centre GPU Cloud for October launch.
• Rack includes 72 Nvidia Blackwell GPUs and 36 Grace CPUs.
• Infrinia AI Cloud OS delivers LLM and multi‑tenant capabilities.
• Service offers Kubernetes and inference as separate as‑a‑service layers.
• Nvidia VP praises platform’s security and scalability for Japanese firms.

Pulse Analysis

The rapid expansion of large‑language models and generative AI has turned compute capacity into a strategic asset for enterprises worldwide. In Japan, concerns over data sovereignty and latency have spurred demand for locally hosted AI infrastructure. Responding to that pressure, SoftBank Corp unveiled a beta version of its AI Data Centre GPU Cloud, slated for commercial rollout in October. By bundling high‑end hardware with a proprietary operating system, the telecom giant aims to give Japanese firms a home‑grown alternative to foreign cloud providers such as AWS, Azure, and Google Cloud.

The core of SoftBank’s offering is a single‑rack chassis equipped with 72 Nvidia Blackwell GPUs and 36 Grace CPU modules, a configuration that rivals the most powerful AI clusters on the market. The hardware is paired with Infrinia’s AI Cloud OS, which abstracts the underlying resources into ready‑to‑use services for large‑model training, multi‑tenant workloads, Kubernetes orchestration, and inference‑as‑a‑service. According to SoftBank president Junichi Miyakawa, the combined compute and software stack is designed to meet the “computing power and operational software” requirements that modern AI projects demand.

Strategically, the move positions SoftBank as a key enabler of Japan’s AI ambitions, offering a secure, low‑latency platform that complies with domestic data‑privacy regulations. Nvidia’s endorsement through VP Charlie Boyle underscores the technical credibility of the solution and hints at deeper collaboration between the two firms. If the service gains traction, it could pressure global cloud giants to establish more localized data centres or partner with Japanese providers, accelerating the country’s AI talent pipeline and potentially attracting foreign AI startups seeking a compliant hosting environment.