SoftBank Corp. has launched its computing platform for the development of generative AI. SoftBank's subsidiary, SB Intuitions Corp, also began developing Japanese large language models (LLMs) using this computing platform. The company plans to finish developing 350 billion-parameter LLMs in 2024.
Developing Futuristic Social Infrastructure Using Generative AI
In May 2023, Softbank announced its long-term goal of creating next-generation social infrastructure for a digital society. The company is building a generative AI computing platform to advance this long-term objective. In July 2023, Japan's Ministry of Economy, Trade, and Industry (METI) categorized the computing platform as a "Cloud Program" under its Economic Security Promotion Act. This designation aims to define essential materials and ensure a secure supply.
Junichi Miyakawa, President & CEO of SoftBank Corp., commented on this development, saying, “Generative AI is a ground-breaking technology that has the potential to fundamentally transform society. However, significant investments in computing platforms and LLMs are now primarily being made by companies based outside of Japan. The progress in AI development using Japanese datasets is trailing behind advancements made in other languages. Given this situation, SoftBank is aiming to create its own advanced, large-scale, computing platform and develop homegrown LLMs that are tailored to Japanese culture and business practices. By doing this, we will support the implementation of generative AI solutions across various industries and work toward realizing the implementation of digitalization into society.”
The Pivotal Collaboration Between NVIDIA, ITOCHU, SoftBank and SB
The Japan top-level computing platform includes an NVIDIA DGX SuperPOD™ AI supercomputer with over 2,000 GPUs, NVIDIA Networking and NVIDIA AI Enterprise software. The computing platform was immediately operational thanks to ITOCHU Techno-Solutions Corporation's equipment installation and construction. SoftBank and SB Intuitions aims to complete investments and construction within the 2023 fiscal year, which ends on March 31, 2024, to provide the computing platform to universities, research institutions, companies, and others at an early date.
“Computing is undergoing a massive transition, driven by GPU-accelerated computing and generative AI. DGX SuperPOD is the ultimate AI supercomputer. We are delighted to partner with SoftBank to build a state-of-the-art AI infrastructure for the industry and researchers of Japan,” commented Jensen Huang, founder and CEO of NVIDIA.
Ichiro Tsuge, President and CEO of ITOCHU Techno-Solutions Corporation, also expressed his excitement regarding the collaboration. He elaborated by stating, “We, at ITOCHU Techno-Solutions, are excited to collaborate with SoftBank. Our joint effort involves operating a computing platform for developing generative AI. This partnership with SB Intuitions will enable us to initiate the fully-fledged development of homegrown LLMs specifically designed for the Japanese language. Our aim is to provide generative AI services that are tailored to Japanese business customs and culture. We also provide various services, including, services encompassing NVIDIA products, support for the deployment of generative AI environments and support for the development of customer-owned LLMs. We will continue to support SoftBank's business and contribute to the prosperity of society through the utilization of AI.”