Japan's worker shortage is not a forecast. It is already visible. Convenience stores are cutting hours because there is no one to staff them. Nursing homes cannot find caregivers. Farms are being abandoned. The government estimates a shortage of roughly 11 million workers by 2040, at a time when nearly 30 percent of the population will be over 65.
The response Japan announced this week is the most concrete one yet. The industry ministry has formally commissioned Noetra, a consortium of SoftBank, NEC, Sony, and Honda, to build what officials are calling a physical AI model: an AI system that lets a robot look at its environment, understand what it sees, and decide what to do, rather than just repeat a sequence it was programmed to follow. Japan's national research lab, AIST, is co-developing it.
The first version is due before the end of the current fiscal year, meaning before March 2027. Annual upgrades are planned after that. The funding for fiscal 2026 alone is around $2.3 billion, drawn from government bonds. Total funding over five years could reach $6.1 billion, but only the first two years are confirmed. After that, the government reviews progress annually and can stop funding if Noetra misses its targets.
That structure matters. The $6.1 billion figure getting shared everywhere is a ceiling, not a guaranteed spend. If the first usable model ships on time, expect the investor list to expand well beyond the current four founding companies. Nikkei reports it could reach 44 companies, spanning automotive, electronics, logistics, and finance. If it does not ship, the stage-gate process gives Tokyo a clean way to walk away.
The 2040 target of 10 million AI-equipped robots across 18 sectors is the long headline, but the nearer milestones are the ones worth watching. Japan is not trying to build a chatbot. The goal is a system that can run in a kitchen, a hospital ward, or a factory floor, processing camera feeds, sensor readings, and spoken instructions together. That is technically harder than most AI products built to date, and Japan has chosen to build it rather than depend on American or Chinese platforms.
That choice has a logic. Japanese policymakers concluded that the country's strengths in robotics, sensors, and precision manufacturing only translate into durable advantage if Japan controls the AI running those machines. Relying on a foreign AI provider for the brain of your national robot fleet creates a dependency that no government in this position wants.
South Korea reached the same conclusion one day later. President Lee Jae Myung announced over $576 billion in planned investment across chips, data centers, and physical AI, with a specific target to commercialize humanoid robots in 10 major industries by 2028. Two economies, both aging, both moving fast and in the same direction.
For businesses outside Japan and Korea, the practical signal here is about timing. The sectors named in Japan's plan: restaurants, food manufacturing, elder care, logistics, are exactly the sectors facing labour pressure across most of Europe, Southeast Asia, and parts of Latin America. Japan has historically served as an early test bed for technologies that spread globally as other nations age. What gets proven at scale in Japan over the next four years is likely to be commercially available elsewhere by the early 2030s.
For any business that relies heavily on manual, repeatable physical work and is already struggling to hire, the question worth asking now is not whether automation will arrive but whether you want to be ahead of it or behind it when it does.