SoftBank is collaborating with key technology partners such as Fujitsu and IBM’s Red Hat who will test and develop the AI network using Nvidia’s GPUs and AI accelerator.
SoftBank Group Corp is taking a bold step in advancing its position in AI technology with plans to build a supercomputer powered by Nvidia’s new Blackwell chips. This move signals SoftBank’s intention to close the rapidly developing AI field gap. By leveraging Nvidia’s next-generation architecture, SoftBank can tap into higher-performance computing, which is crucial for AI workloads like training large models and running complex simulations.
Its decision to build a supercomputer based on Nvidia’s DGX B200, which integrates AI accelerator chips with high-performance processors, is a bold and strategic move that could significantly bolster the company’s ambitions in artificial intelligence. The DGX B200 is a high-performance AI supercomputer platform designed for large-scale AI workloads, enabling advanced capabilities in training and running complex AI models. Combining Nvidia’s cutting-edge AI accelerators (like GPUs and other specialized chips) with traditional processors, this system is optimized to handle the massive parallel computing required by AI applications—particularly for large-scale machine learning and deep learning tasks.
SoftBank Group’s return to profitability, spurred by rising tech valuations, signals a shift in the company’s financial health, especially after the tumultuous years following its high-risk bets in various sectors. The Nvidia stake, which would be worth US$178 billion today, is a powerful reminder of the potential opportunities SoftBank missed.
The company’s plan to integrate Nvidia’s hardware into its telecom unit to power AI services over cellular networks is an exciting and transformative development for SoftBank and the broader telecommunications and AI landscape. The key innovation here is that Nvidia’s AI chips and AI radio access networks (RAN) will enable SoftBank to re-engineer its cellular networks to handle traditional data traffic and support AI-driven services. Instead of using custom chips that focus on optimizing the movement of data, the network would be enhanced with AI accelerators capable of running AI models in real-time, both at the network’s edge and in the cloud.
To bring this vision to life, SoftBank is collaborating with several key technology partners such as Fujitsu and IBM’s Red Hat. Together, these companies will test and develop the AI network using Nvidia’s GPUs and AI accelerators, marking the beginning of a collaborative effort to transform telecom into a new type of AI infrastructure.
To bring this vision to life, SoftBank is collaborating with several key technology partners such as Fujitsu and IBM’s Red Hat. Together, these companies will test and develop the AI network using Nvidia’s GPUs and AI accelerators, marking the beginning of a collaborative effort to transform telecom into a new type of AI infrastructure. It allocated ￥4 trillion (about US$26 billion) for the domestic semiconductor sector is a massive financial commitment to strengthen Japan’s ability to produce cutting-edge chips. One of the most ambitious elements of Japan’s semiconductor push is the Rapidus Corp initiative. This moonshot project is designed to build a state-of-the-art semiconductor foundry from scratch in Japan, to challenge Taiwan Semiconductor Manufacturing Co. (TSMC), the dominant player in the advanced chip manufacturing space.
The government’s pledge of more than US$65 billion to support the semiconductor and AI sectors highlights the growing recognition of their symbiotic relationship. Prime Minister Shigeru Ishiba’s strong commitment to backing these initiatives signals that Japan is taking a long-term approach to revitalizing its technology sector.
