Japan commits $6.1 billion to AI model for 10 million robots

Japan's Ministry of Economy, Trade and Industry (METI) and its innovation agency NEDO have formally commissioned Noetra, a consortium majority-owned by SoftBank, Sony Group, NEC and Honda Motor, to build a national "physical AI" foundation model, AI News reported on 1 July 2026. The project runs from fiscal 2026 to fiscal 2030 and is backed by public funding of up to one trillion yen, or roughly $6.1 billion. Noetra will work alongside AIST, a national research lab, on a model that reads language, images, video and sensor data together so that a robot can interpret a room and act in it, rather than simply execute pre-programmed motions.

A funding ceiling, not a guarantee

Only the first two years of the programme are locked in. According to AI News, the current fiscal year's commission is worth about $2.3 billion on its own, drawn from a 387.3 billion yen allocation funded through Japan's GX Economy Transition Bonds. After that, funding is reviewed annually through a stage-gate process, meaning Tokyo can withdraw support if Noetra misses its milestones. An initial version of the model is due as early as this fiscal year, with annual upgrades built on data volunteered by participating manufacturers.

Why robots, and why now

Industry minister Ryosei Akazawa said the plan will "vigorously promote social implementation" across sectors including restaurants, food manufacturing and medical care, according to Let's Data Science, which cited reporting by Ary News, The Japan Times and Nikkei Asia from 30 June 2026. The government has set a target of roughly 10 million AI-equipped robots deployed across 18 industries by 2040. Behind the figure is a demographic problem with no easy fix: Japan's population has been shrinking since 2008, its working-age population is projected to fall sharply over the next two decades, and tight migration policy leaves employers with few alternatives to automation. Japan's Ministry of Health, Labour and Welfare separately projects a shortfall of 570,000 care workers by 2040, with the nursing sector currently seeing just one applicant for every 4.25 open jobs.

• Up to ¥1 trillion (about $6.1 billion) in public funding over five years, fiscal 2026-2030
• ¥387.3 billion (about $2.3 billion) already committed for fiscal 2026, from GX Economy Transition Bonds
• 10 million AI-equipped robots targeted across 18 industries by 2040
• Noetra majority owners: SoftBank, Sony Group, NEC, Honda Motor; up to 44 companies expected to join
• 570,000 care-worker shortfall projected by 2040, per Japan's health ministry

An industrial base already in place

Japan is not starting from zero. Its manufacturers already hold roughly 70% of the global industrial robotics market, according to METI data cited by Let's Data Science, giving the Noetra project an existing production base rather than a purely experimental one. The country has also spent years building robotics expertise in elder care, disaster response and manufacturing, including the long cleanup effort at the Fukushima Daiichi plant. The Noetra push is an attempt to turn that accumulated experience into a shared, exportable model rather than a patchwork of one-off products. SoftBank engineers are reportedly working alongside researchers from Preferred Networks and AIST itself, while Fujitsu and Rakuten are said to be weighing whether to join the consortium.

A regional race, not a solo effort

The timing is not incidental. South Korea announced its own robotics push within a day of Japan's confirmation, and both governments are now framing physical AI, robots that sense and act in the physical world, as the next front in a competition that has mostly played out over chatbots and cloud contracts until now. METI's earlier policy announcement, made in March 2026, set a goal of capturing 30% of the global physical AI market by 2040.

What it means for Japanese businesses

For small and mid-size Japanese manufacturers and care-facility operators, long priced out of custom robotics, a shared foundation model changes the calculus: instead of commissioning a bespoke system, they could eventually license or adapt a state-backed model already trained on sector data from larger participants. That is also the risk built into the stage-gate structure. If Noetra's first model, due this fiscal year, fails to hit its benchmarks, the funding pipeline that smaller operators are counting on could be reviewed or scaled back before it ever reaches them.