Japan Bets Big on Physical AI With SoftBank, Honda, Sony and NEC

• Physical AI is Japan’s sovereign bet, with SoftBank, NEC, Honda, and Sony building a trillion-parameter model.
• Japanese government pledges ¥1 trillion, steelmakers and banks hold stakes.

Japan has stopped waiting for Silicon Valley to solve its industrial problems. On April 12, SoftBank, NEC, Honda, and Sony jointly established a company called Japan AI Foundation Model Development, according to reports from Nikkei Asia and Kyodo News.

The venture has the objective to build a physical AI foundation model that Japan owns, trains on Japanese data, and deploys in Japanese factories and machines, without routing through foreign cloud platforms.

The model is targeting roughly one trillion parameters, to be tuned for real-world control tasks. As cited by The Japan Timesdevelopers describe the project as physical AI: Models designed to perceive and act in the physical world, controlling robot arms and factory systems.

The four founding companies hold a stake of more than 10% in the new firm, according to Jiji Press. SoftBank and NEC will lead AI development, Honda will be the first to deploy the model in its autonomous vehicles, and Sony’s involvement covers robotics and gaming hardware. Tokyo-based AI developer Preferred Networks, a deep learning specialist with existing ties to Toyota, is also set to participate in the research and development phase.

The government money behind the physical AI push

Japan’s New Energy and Industrial Technology Development Organisation has earmarked roughly ¥1 trillion, about US$6.3 billion, in AI support over five years beginning in fiscal 2026, and the new firm is expected to apply for that program. It is considered a near-certain recipient.

The investor list also includes Nippon Steel, Kobe Steel, MUFG Bank, Sumitomo Mitsui Banking Corporation, and Mizuho Bank, which have taken stakes in the venture, placing Japan’s industrial and financial establishment behind the same bet.

Project backers say training and related data will remain in Japan and will not be processed on foreign cloud platforms, a response to what the Japanese industry has described as a digital deficit, where years of sensitive operational data have been fed into American AI pipelines with no domestic alternative.

Why physical AI, and why now

With nearly a third of the population projected to be over 65 by 2042 and a working-age population declining since 1995, the country faces a potential shortfall of 11 million workers by 2040. Physical AI is the industrial response to that arithmetic.

Japan’s Ministry of Economy, Trade and Industry announced in March 2026 that it aims to build a domestic physical AI sector and capture a 30% share of the global market by 2040, backed by an existing industrial position where Japanese manufacturers account for about 70% of the global industrial robotics market, according to the ministry’s own data. Five of the world’s ten largest industrial robot manufacturers are Japanese.

The Japanese physical AI market, valued at approximately US$307 million in 2025, is projected to reach US$6.76 billion in value by 2035, according to industry analysis. The consortium’s 2030 deployment target sits in the middle of that growth window.

What makes this venture structurally different from previous Japanese technology consortia is the division of responsibility. The resulting project is a closed-loop physical AI stack where the entity training the model and those deploying it are the same national industrial base, not a foreign platform licensing access to Japanese manufacturers.

The SoftBank contradiction is worth noting

SoftBank led OpenAI’s US$40 billion funding round in 2025, yet it is getting behind a domestic venture that aims to be independent of the American AI ecosystem it bankrolls.

The contradiction is perhaps less paradoxical than it may seem. Japan’s sovereign AI strategy, and this venture specifically, is positioned as a parallel ability to US-based alternatives, built for workloads where data confidentiality and physical control requirements make foreign dependency a liability.

Nihon Keizai Shimbun reported that as AI handling sensitive facility operations increases, concerns about overseas leakage of that information are growing, and developing large-scale domestic models is the direct response to that risk.

SoftBank’s actions are, arguably, a rational hedge bet.

What it means for the region

Japan has historically been cautious in enterprise AI adoption, a finding reflected in recent research showing Japan at intermediate agentic AI maturity compared to India’s more advanced production deployment rates.

But the Japan AI Foundation Model Development consortium is operating at the level of sovereign infrastructure. The question for other Asian economies investing in AI infrastructure is whether they are building capacity they control, or capacity they depend on. Japan has chosen its answer.