Japan's SoftBank to Invest Additional 1.3 Trillion Won in Generative AI... "Developing ChatGPT-Level Technology"

Japanese telecommunications company SoftBank will invest an additional 150 billion yen (approximately 1.337 trillion KRW) by next year to develop high-performance generative artificial intelligence (AI), local media reported on the 22nd.

According to the Nihon Keizai Shimbun (Nikkei) on the same day, SoftBank plans to use this additional investment to develop a world-class model comparable to ChatGPT-4, created by the US-based OpenAI. Last year, SoftBank invested 20 billion yen (approximately 178 billion KRW) in the computational infrastructure necessary for generative AI development.

Masayoshi Son, Chairman of SoftBank

Nikkei stated, "This is the largest scale of investment among Japanese companies (in generative AI)," and expressed expectations that SoftBank will purchase semiconductors from US-based Nvidia to enhance its computational power by several tens of times compared to the current level.

SoftBank is developing a large language model (LLM), the core technology behind generative AI. By March next year, it plans to complete a 3.9 trillion parameter model and then begin developing a model specialized in the Japanese language with approximately 10 trillion parameters.

Nikkei also reported, "SoftBank is promoting the establishment of AI data centers tailored to domestic generative AI development," adding, "Recently, it decided to invest 65 billion yen (approximately 580 billion KRW) to build Japan's largest AI data center in Hokkaido." It further noted, "Domestic generative AI has become important from the perspective of economic security, and both the government and companies have started to emphasize 'data sovereignty,' managing their own data within the country."

Earlier, Masayoshi Son, chairman of SoftBank Group, announced in October last year that he wanted to make SoftBank Group the group that utilizes AI the most in the world, signaling an expansion of AI business.