SoftBank is planning a supercomputer with Nvidia Blackwell chips: A breakthrough for the Japanese AI market

SoftBank Group Corp. has announced an ambitious project - to create Japan's most powerful AI supercomputer based on the next-generation Blackwell chips from Nvidia Corp. The move is a testament to SoftBank's determination to catch up with global leaders in AI and become a key player in the AI technology field.

SoftBank, Japan's third-largest mobile operator, has joined forces with US-based Nvidia $NVDA to carry out a project that promises to push the country's technology level. The supercomputer will be built on Nvidia's DGX B200 system, a solution combining processors with accelerators specialised in processing large data volumes, which is essential for the efficient operation of AI models. The next phase of the project is to develop an even more powerful system based on Grace Blackwell technology, the latest and most advanced version of Nvidia's chips to date.

Shares of SoftBank Corp. reacted to the news by rising 1%, while parent company SoftBank Group lost 1.7%. The announcement shows that SoftBank has managed to secure priority in acquiring these sought-after chips, which is evidenced by the close relationship between SoftBank and Nvidia, which was still a shareholder in the company in 2019.

Masayoshi Son and Jensen Huang: Shared visions and opportunities for Japan

During the recent AI Summit in Tokyo, SoftBank founder Masayoshi Son spoke alongside Nvidia CEO Jensen Huang, who introduced Blackwell chips earlier this year. In a friendly atmosphere, Son wryly lamented the sale of his original 4.9% stake in Nvidia, which today would be worth approximately $178 billion.

Son and Huang talked about the vision they share: to introduce AI at the national project level and reduce dependence on a few large U.S. clients. With this goal in mind, Nvidia held events not only in Japan but also in India to introduce AI technology to a wider audience. These efforts are intended to encourage innovation and open the door for new opportunities in the AI field in various countries.

Next-generation AI networks and the transformation of telecommunications

SoftBank plans to connect its mobile networks with Nvidia technologies and build an "AI grid" - a network that will provide a platform for AI-based services across Japan. The current hardware solution, optimised for mobile data, is not sufficient for AI applications, so it is moving to so-called AI-RANs (Radio Access Networks), which are AI-powered radio access networks. These networks will be beneficial for autonomous vehicles, remote robotics and other low-power applications.

SoftBank will begin testing these networks in collaboration with IBM's Fujitsu of Japan and Red Hat, who will provide know-how and support. Son has also pledged extensive investment in Nvidia chips to ensure sufficient capacity for the development of AI services in Japan.

Support from the Japanese government and the national chip strategy

SoftBank recognizes that the rapid adoption of AI technologies requires support not only from the private sector but also from the government. Therefore, the Japanese government has allocated ¥4 trillion (approximately $26 billion) to increase domestic semiconductor manufacturing capacity. This investment includes a project by Rapidus Corp. to build a state-of-the-art factory to compete with the world's leading chipmaker, Taiwan Semiconductor Manufacturing Co. (TSMC). In addition, Prime Minister Shigeru Ishiba has pledged an additional $65 billion to support the semiconductor and AI industries in Japan.

Masayoshi Son stressed that the government should further support these steps as Japan has the opportunity to catch up with the world in AI and semiconductor technology. This moment is a "moment of renewal and catch-up" for the country, when it has a chance to actively participate in global AI development.

Disclaimer: There is a lot of inspiration to be found on Bulios, but stock selection and portfolio construction is up to you, so always conduct thorough self-analysis.

Source: Reuters, Yahoo Finance.