Japanese telco SoftBank enters the sovereign AI race with planned AI data center cloud in what is the latest move into enterprise AI by Asian telcos.
SoftBank is set to enter the sovereign AI business, unveiling new GPU cloud infrastructure targeted to deploy in October. It says its new AI data center cloud, built on NVIDIA GB200 and other advanced chips, will enable customers to carry out AI workloads while ensuring secure data operations within Japan.
The new hardware combines with SoftBank’s Infrinia AI Cloud OS, which supports Kubernetes-as-a-service (KaaS)*3 and inference-as-a-service (Inf-aaS) for LLMs. CEO Junichi Miyakawa said the new cloud platform would “provide integrated computing infrastructure and software that can be securely used within Japan as a neocloud provider.”
He said that as AI becomes increasingly important as a source of competitiveness, customers need additional capabilities such as compute power and the operational software to support it.
SoftBank Corp, the telecom arm of the SoftBank Group, says the next goal for its AI infrastructure will be to integrate the GPU cloud with AI-RAN while building out its distributed and sovereign AI capability.
The closely related sovereign AI and GPU cloud markets are still nascent services for Asian telcos. They’re not competing directly with hyperscalers but are serving their own customers who seek a trusted local provider.
Military-specific AI
SoftBank’s domestic rival NTT Data has built out a sovereign AI cloud platform targeted at enterprise as well as GPU-based infrastructure offerings. It has just introduced enterprise AI factories to provide domain-specific AI solutions and also offers Tsuzumi, a language and enterprise-focused LLM.
South Korea’s SK Telecom has been running its GPU-as-a-service since January 2025. At MWC in March it unveiled a sovereign AI package that integrates its AI data center, sovereign AI model, and agentic and other AI services. It has just partnered with the Defense Ministry to develop and verify military-specific AI capabilities based on SKT’s AI foundation model.
The other regional telco active in sovereign AI is Singtel. It debuted RE:AI, its AI cloud and sovereign AI offering, in October 2024 with the stated aim of making AI capabilities more widely available. In its latest move RE:AI has teamed up with France’s Mistral AI to develop AI use cases in verticals including defense, financial services and healthcare.
Singtel CEO Yuen Kuan Moon said last week the company had allocated 1.2 billion Singapore dollars (US$940 million) for GPU as a service and sovereign AI.
https://www.lightreading.com/ai-machine-learning/softbank-joins-sovereign-ai-race?shem=rimspwouoe,
