TOKYO — SoftBank is spearheading a strategic effort to build a domestic artificial intelligence ecosystem in Japan, establishing a new unit designed to develop homegrown AI capabilities. The initiative, reported by Nikkei, involves a consortium of eight corporate investors, including electronics giant NEC and automaker Honda Motor.
The move signals a pivot toward “sovereign AI,” a growing global trend where nations seek to develop their own large language models (LLMs) and computing infrastructure to reduce dependence on American technology providers like OpenAI, Google, and Microsoft. For Japan, the goal is to ensure that AI systems are deeply integrated with the Japanese language and cultural nuances whereas keeping sensitive industrial data within national borders.
By uniting a diverse group of industrial leaders, SoftBank is attempting to bridge the gap between raw computing power and practical, industry-specific applications. The partnership with Honda suggests a focus on integrating AI into robotics and autonomous mobility, while NEC’s involvement brings deep expertise in network infrastructure and biometric security.
The push for sovereign AI in Japan
The drive for SoftBank homegrown AI is not happening in a vacuum. Japan has historically struggled to keep pace with the rapid advancements in generative AI seen in the United States and China. While Japanese firms have excelled in hardware and robotics, the software layer—specifically the massive datasets required to train frontier models—has largely been dominated by Silicon Valley.
Developing a domestic model allows Japan to mitigate “cultural drift,” where AI models trained primarily on English-language data may fail to capture the social hierarchies, linguistic subtleties, and business etiquette essential for the Japanese market. The Japanese government has been increasingly vocal about the need for digital sovereignty to protect national security and intellectual property.
SoftBank, led by Masayoshi Son, has spent the last several years repositioning itself as the primary financier and infrastructure provider for the AI era. This new unit is the logical next step in a broader strategy to move beyond simply investing in other companies’ startups and instead owning the foundational layers of the AI stack.
Infrastructure as the foundation
Software is only one half of the equation. To run sophisticated AI, a country needs immense amounts of compute power, typically provided by high-complete GPUs (Graphics Processing Units) and the data centers to house them. SoftBank has been aggressively expanding its physical footprint in Japan to support this ambition.
The company is investing heavily in the construction of massive data centers across the country, aiming to create a “GPU cluster” that can rival the capabilities of the world’s largest AI labs. This infrastructure is critical due to the fact that the bottleneck for AI development is no longer just talent, but the availability of hardware and the electricity required to power it.
By controlling the data centers, SoftBank can offer its consortium partners—like Honda and NEC—the ability to train specialized models on their own proprietary data without that information ever leaving the domestic network. This “closed-loop” system is particularly attractive to the automotive and defense sectors, where data leakage is a primary concern.
Strategic roles within the consortium
While the full list of eight investors has not been detailed in every report, the roles of the known participants highlight the multidisciplinary nature of the project:
- SoftBank: Providing the primary capital, strategic vision, and the underlying cloud and GPU infrastructure.
- NEC: Contributing expertise in system integration, high-performance computing, and enterprise AI deployment.
- Honda Motor: Focusing on the application of AI in “embodied intelligence,” including autonomous driving and humanoid robotics.
| Dimension | Objective | Key Driver |
|---|---|---|
| Linguistic | Japanese-native LLMs | Cultural nuance and accuracy |
| Hardware | Domestic GPU clusters | Reduced reliance on US clouds |
| Industrial | Robotics & Automation | Labor shortage in Japan |
| Security | On-shore data residency | National security & IP protection |
The broader vision: Toward Artificial Super Intelligence
This initiative aligns with Masayoshi Son’s public obsession with what he calls “Artificial Super Intelligence” (ASI)—an AI that surpasses human intelligence in every field. Son has frequently stated that the transition to ASI is inevitable and that those who control the infrastructure and the foundational models will lead the next century of economic growth.
For SoftBank, this unit is a hedge. While the company maintains a massive stake in Arm, which designs the architecture for the chips that power AI, owning the actual AI models and the data centers creates a vertically integrated powerhouse. If the world moves toward a fragmented “splinternet” where different regions apply different AI standards, SoftBank ensures Japan is not left behind.
However, the path forward is not without challenges. The cost of training a frontier model is astronomical, often requiring billions of dollars in electricity and hardware. SoftBank must compete for a limited pool of global AI researchers who are often lured by the massive salaries offered by Big Tech firms in the U.S.
Disclaimer: This article discusses corporate investments and financial strategies. It is intended for informational purposes and does not constitute financial or investment advice.
The next major milestone for the unit will be the unveiling of its first prototype model or the announcement of its first commercial partnership. Market analysts expect further details on the remaining five investors and the specific funding targets for the unit in upcoming quarterly filings or corporate briefings.
What do you consider about Japan’s move toward sovereign AI? Share your thoughts in the comments or share this story with your network.
