SoftBank Group Corp. will be the first to develop a supercomputer with chips based on Nvidia Corp.'s new Blackwell design, demonstrating the Japanese company's aim to catch up in AI.

SoftBank's telecom unit intends to develop Japan's most powerful AI supercomputer to support a variety of local services, the two companies said. That computer will be built on Nvidia's DGX B200 platform, which combines computer processors and so-called AI accelerator chips. Grace Blackwell will return in a more advanced version.

SoftBank Corp., Japan's third-largest wireless provider, saw its stock price rise 1.5 percent on Wednesday. SoftBank Group, which includes the conglomerate's investment teams, dropped as much as 3.3 percent.

Nvidia chips have become a valuable commodity for the world's largest technology businesses, which use them to build and run AI algorithms. The process requires software to be blasted with data, which accelerator chips excel at.

The announcement suggests that SoftBank, which owned 4.9% of Nvidia until early 2019, has secured a favorable position in line for the chips. SoftBank founder Masayoshi Son has stated that he is planning to'swing for the fences' in AI ventures. On Tuesday, the company declared a return to profitability, citing higher technology valuations.

Nvidia Chief Executive Officer Jensen Huang announced the new Blackwell lineup early last year, but production issues hampered the launch.

Despite Huang's assurance that supply will be sufficient once manufacturing begins, buyers have been keen to get their hands on the first new chips.

Son joined Huang on stage in Tokyo on Wednesday for Nvidia's AI Summit. When reminded of SoftBank's previous Nvidia stake, which would be worth approximately $178 billion today, Son shuddered and chuckled before wrapping Huang in a bear hug.

Nvidia has been traveling the world to conduct such events, promoting what it terms the "new industrial revolution." Events in India and now Japan are intended to widen the deployment of AI systems to national-level endeavors and reduce Nvidia's dependency on a few large US customers.

In addition to the new machine and plans for a second, telecom subsidiary SoftBank Corp. will leverage Nvidia technology to provide AI services across cellular networks. Traditional hardware, which is built on bespoke chips designed to maximize mobile data flow, is not ideal for new AI services.

"What will result is an AI grid that runs across Japan," Huang explained, noting that this will transform the communications network into an AI network.

According to Huang, new AI-RANs, or AI radio access networks, will be more suitable for remote robotics, autonomous car support, and powering other applications. They will also use less electricity.

The telecom unit, which controls the country's third largest wireless carrier, will start testing the network with partners Fujitsu Ltd. and International Business Machines Corp.'s Red Hat.

Japan is on the verge of change, according to Son, who has frequently chastised both the government and companies for their tardy adoption of new technologies. This time, the Japanese government is not impeding AI development, he claimed.

"But they should encourage more," he advised Huang. "As you mentioned, this is the reset, the catch-up period for this revolution. We cannot afford to miss this opportunity."