NVIDIA and SoftBank are partnering to build Japan's most ambitious AI infrastructure yet, marking a significant step in the country's push for AI sovereignty. The partnership aims to position Japan to become a major player in the global AI race, with a unique focus on combining AI with the country's strengths in robotics and telecommunications.
Why it Matters
The collaboration comes at a pivotal moment, as countries worldwide push to develop their own sovereign AI initiatives. Japan’s efforts with NVIDIA and SoftBank mirror similar moves in Europe and North America, all working towards a future where AI is a key pillar of national and economic strategy. The partnership also shows how AI and 5G convergence is pushing us into a new telecommunications era—one where networks themselves are powered by and optimized for AI.
What's Happening
The collaboration centers on three key initiatives:
- AI Factory: SoftBank will receive the world's first NVIDIA DGX B200 system, which will be the foundation of the new NVIDIA DGX SuperPOD supercomputer. This supercomputer will harness 25 exaflops of AI power to accelerate everything from generative AI projects to complex research for universities and enterprises in Japan.
- AI-RAN: 5G Meets AI: NVIDIA and SoftBank are introducing the world’s first live AI and 5G hybrid network, known as AI-RAN (Artificial Intelligence Radio Access Network). By integrating the NVIDIA AI Aerial engine into SoftBank’s 5G radio, this new system will redefine telecommunications infrastructure—not just transmitting data but evolving into an "AI network" capable of carrying and processing AI models in real time.
- AI Marketplace: SoftBank is also setting up an AI store to allow local and third-party developers to share AI applications. The marketplace will cater to Japan's need for secure, locally-developed AI solutions, allowing businesses, students, and consumers to innovate with access to cutting-edge technology.
The Bigger Picture
This partnership isn't just about building supercomputers or upgrading network infrastructure. It’s about creating an integrated AI ecosystem in Japan—one that spans everything from AI research and development to its commercialization in telecommunications and beyond. By leveraging NVIDIA's Blackwell GPUs and Grace CPUs, SoftBank aims to turn Japan into an AI hub where the entire spectrum of AI activities, from model training to deployment, can be efficiently executed.
Between the Lines
The creation of AI-RAN, in particular, promises to unlock billions in revenue for telecom operators, according to SoftBank. The partnership transforms traditional telecom infrastructure from a cost center into a potential revenue generator. SoftBank's 5G network will be able to run AI workloads during periods of low network usage – typically about two-thirds of its capacity.
By The Numbers
- The supercomputer will deliver 25 AI exaflops of performance
- SoftBank estimates a 219% return on investment for each AI-RAN server added to its infrastructure
- Telecom operators could earn $5 in AI inference revenue for every $1 invested in new AI-RAN infrastructure
What they're Saying
"Japan has a long history of pioneering technological innovations with global impact," says NVIDIA CEO Jensen Huang. "With SoftBank's significant investment in NVIDIA's full-stack AI, Omniverse and 5G AI-RAN platforms, Japan is leaping into the AI industrial revolution."
What's Next
SoftBank also plans to build additional acceleratored supercomputers to run extremely computationally high workloads using the NVIDIA Grace Blackwell platform. The company's AI marketplace will facilitate AI development and deployment across Japanese industries, from transportation to healthcare.
The Bottom Line
NVIDIA and SoftBank's collaboration marks a major leap towards transforming Japan into an AI-centric economy. This infrastructure buildout is part of a broader initiative supported by Japan's Ministry of Economy, Trade and Industry to establish sovereign AI capabilities.