SoftBank installs some 4,000 Nvidia Hopper GPUs in Japan top-level AI computing platform

• SoftBank making progress toward goal of achieving 25.7 exaflops of compute capability

SoftBank Corp. (“SoftBank”) announced that on October 31, 2024, it finished installing approximately 4,000 NVIDIA Hopper GPUs in its Japan top-level^*1 AI computing platform^*2 now under expansion.

SoftBank began operating its 0.7 exaflops^*3 AI computing platform equipped with approximately 2,000 NVIDIA Ampere GPUs in September 2023. With this installation of approximately 4,000 NVIDIA Hopper GPUs, the AI platform's total computing performance has been boosted to approximately 4.7 exaflops with approximately 6,000 GPUs installed. From the current fiscal year that ends on March 31, 2025 (FY2024), to FY2025, SoftBank plans to further strengthen the AI computing platform's performance by deploying an NVIDIA DGX SuperPOD with NVIDIA DGX B200 systems by increasing the total number of GPUs installed to approximately 10,000, and, by subsequently installing additional GPUs, aims to achieve a Japan top-level^*1 total compute capability of 25.7 exaflops.

This AI computing platform will initially be utilized by SB Intuitions Corp. (“SB Intuitions”), SoftBank's subsidiary dedicated to developing homegrown large language models (LLMs) specialized for the Japanese language. SB Intuitions is aiming to build LLMs with approximately 390 billion parameters that support multiple modalities within FY2024. With this AI computing platform enhancement, development of LLMs will be further accelerated.

SoftBank is committed to building different types of social infrastructure to enable a society that co-exists with AI. The construction of SoftBank's AI computing platform was certified for supply assurance by Japan's Ministry of Economy, Trade and Industry (METI) as a “Cloud Program,” which designates important materials based on its Economic Security Promotion Act^*2. By providing computational resources that can be utilized by a wide range of generative AI developers, SoftBank will contribute to the establishment of development infrastructure and strengthen the service provision framework in Japan. In addition to utilizing the AI computing platform among its own group companies, SoftBank plans to provide the infrastructure as a service to companies and research institutions based in Japan.

*1 One of the largest computing platforms in Japan for LLMs learning. Based on publicly available information as of October 31, 2024, and SoftBank Corp. research.

*2 More details are available in SoftBank Corp.'s press release dated May 10, 2024 (in Japanese).

*3 An exaflop is a measure of performance for a supercomputer that can calculate at least 1018 floating point operations per second.

SoftBank, the SoftBank name and logo are registered trademarks or trademarks of SoftBank Group Corp. in Japan and other countries.

Other company, product and service names in this press release are registered trademarks or trademarks of the respective companies.