Japanese media such as the Nihon Keizai Shimbun and Tokyo Broadcasting System reported on the 12th that, softbank Group, NEC, Honda and Sony Group co-founded“Japan AI basic model development company” in April 2026, start the research and development of domestic AI basic model. This is a joint layout of Japan’s major technology and manufacturing companies around the next generation of AI, with the goal of accelerating domestic AI technology research and development and industrial application as US and Chinese companies continue to lead. According to experts interviewed by the global times, this move not only involves the restructuring of technology research and development, but also profoundly reflects Japan’s coping strategies in the face of changes in the global industrial landscape, however, it is difficult to touch the core weakness of Japan’s AI development.
Differentiated competitive strategy
Japanese media said the new company is headquartered in Shibuya, Tokyo, softbank Group, NEC, Honda and Sony Group as the core shareholders, each holding more than 10 percent of the shares. The new company will also absorb 100 senior Japanese AI R & D technicians to improve development efficiency and speed of industrial landing.
In the specific division of labor, each enterprise will play its own advantages, form a complete chain from the underlying model research and development to practical application. Softbank and NEC will be mainly responsible for the development of AI basic model, undertake the construction of the“Brain” part, including large-scale model training and basic technical architecture. Honda and Sony will be responsible for applying the development to their respective areas of focus. Honda will focus on autonomous driving systems and robotics, while Sony plans to use them in games, entertainment and semiconductors.
The new company plans to launch AI models with 1 trillion parameters by 2030, the Nihon Keizai Shimbun reported. TBS reports that the AI model developed by the new company will not only be used internally by the sponsoring companies, but is also planned to be opened up to other Japanese companies to facilitate wider industrial applications.
Fu Borui, a researcher at the World Think Tank Research Center at Shanghai International Studies University, told a reporter from the Global Times on the 13th that the division of Labor at the new company reflects the idea of combining software and hardware, the communication and software companies will be responsible for model development, while the automotive and electronics companies will provide hardware carriers. The strategy is to use japanese-made AI models as a central hub of control, embedded in physical devices where it already has an advantage. Japan has shifted its strategic focus to the field of physics AI, which is a differentiated choice based on its comparative advantage. The fundamental reason is that there is a substantial gap between Japan and Chinese and n Enterprises on the traditional general large model track. Fu further analyzed that Japan did not inject a large amount of capital in the early stage of AI development, resulting in its lack of computing power and data reserves needed to train models with very large parameters and computations, it has yet to come up with a competitive base model. At the same time, Japanese enterprises tend to be conservative in technology application, and the penetration rate is significantly lagging behind, which directly leads to the inability to effectively accumulate the large-scale feedback data required for algorithm iteration. This forces Japan to turn to the physical application segmentation track where it still has an advantage.
‘growing sense of crisis’
The establishment of the new company is in line with the Japanese government’s ongoing support policy for domestic AI. On December 23,2025, the Japanese government officially adopted the“Artificial Intelligence Basic Plan”. The document proposes to make Japan“The easiest country in the world to develop and apply AI”.
According to the AI Basic Plan, Japan has fallen significantly behind other major countries in AI development and investment, and the sense of crisis continues to deepen. Only 49.7 per cent of Japanese companies have a policy on using generative AI, compared with 84.8 per cent in the US, according to the data. According to the Asahi Shimbun, the low penetration rate of generative artificial intelligence in the Japanese corporate sector has become a structural problem that hinders the development of corporate innovation in Japan. If this problem is not seriously addressed, japan’s industrial competitiveness will“Continue to be left behind”.
In this context, the Japanese government proposed to provide about 1 trillion yen (100 yen) of support funds in the next five years to promote the construction of domestic AI basic models and related infrastructure. New companies are seen as one of the main applicants for the funds. In the policy direction, Japan puts forward four pillars: “Using AI”, “Creating AI”, “Enhancing credibility” and“Collaborating with AI”. In the“Creating AI” section, the government proposes to build a complete AI ecosystem in Japan, from data centers, semiconductors, basic models to applications.
In an effort to“Reshape the competitiveness of Japan’s semiconductor industry,” eight Japanese companies, including Sony and Toyota, set up Rapidus in 2022 with massive government subsidies. The Japanese government will provide an additional 631.5 billion yen in subsidies to Rapidus, Kyodo News reported Monday. But the Nihon Keizai Shimbun had previously reported that the company was facing difficulties in winning over high-end global customers and an imperfect manufacturing chain.
‘core weaknesses’ hard to reach
Japanese media believe that under the background of the continuous increase in investment in AI by American and Chinese Enterprises, Japanese enterprises have made joint investments to set up new companies, integrate technology and capital resources, promote the development and industrial application of local AI basic models, and hope to achieve breakthroughs in the next generation of AI competition.
Fu Borui told reporters that the newly established AI company has provided a set of technical governance solutions, for example, the pressure of computing power cost can be alleviated by special government subsidies, and the organizational barriers can be broken by centralizing the core R & D personnel of each enterprise. These measures can indeed improve the research and development efficiency of local basic algorithms to a certain extent, and alleviate the backwardness of technology. However, these measures can not address the core shortcoming of Japan’s AI development, that is, the current social mode of production in Japan has been unable to provide the innovation vitality needed for the development of disruptive technologies. The prosperity of AI industry needs sensitive trial-and-error mechanism, highly mobile talent ecology and the ability to deal with systemic risks. The conservative management habits of Japanese enterprises and their risk aversion to disruptive innovation have fundamentally suppressed the innovation potential of the society. This rigid mode of production leads to a lack of interest in developing new technologies in the home market.
Fu Borui believes that simply setting up a new enterprise with the nature of an industrial alliance to repair the industrial technology level can not fundamentally solve the deep-seated problem of the lack of vitality of social mechanisms, japan’s long-term development in the AI sector is still subject to severe underlying structural constraints.