When political and technology leaders gathered in Paris for the 2025 AI Action Summit, the headlines belonged to a familiar contest: Emmanuel Macron's multibillion-euro investment pledge and the widening rivalry between the United States and China. Far less noticed was a quieter signal from Tokyo. Japan became the first country in Asia to sign the Council of Europe's landmark convention on artificial intelligence, a commitment less about raw capability than about trust. It pointed to a shift now reshaping the industry: as AI moves beyond the chatbot, leadership will be defined not by who builds the most powerful model, but by who can put the technology safely and profitably to work. Few countries are better placed for that contest than Japan.
It is a claim that invites immediate objection, because Japan is also, by common consent, a digital laggard. The same country that leads the world in factory automation still runs much of its corporate life on fax machines, personal seals and software written decades ago. Both things are true at once, and the tension between them is the key to understanding where Japan fits in the next phase of AI.
Two Reputations, One Country
Seen through the lens of the past three years, the doubts are easy to state. Japan has not produced a dominant consumer AI platform. It does not command the hyperscale computing power or the venture capital ecosystem of its largest rivals. Measured by model size, funding rounds and consumer reach, the standard scorecard of the AI race, it is not among the front runners.
The second doubt is harder still, because Japan admits it. The government's own ministries have warned for years of a 2025 digital cliff, the point at which ageing, black-boxed IT systems could cost the economy as much as 12 trillion yen a year. Roughly four in five companies still depend on legacy core systems, many of them more than two decades old, and cashless payments accounted for only about a third of transactions as recently as 2022. On the enterprise-software layer where digital transformation is usually measured, Japan trails much of the developed world.
Yet both doubts describe the same thing: the layer where the previous race was scored. They say little about the layer where the next one will be. As businesses move from experimenting with AI to depending on it, the decisive questions are no longer about model size or consumer buzz. They are about reliability, safety, measurable return and the ability to embed intelligence into the machines, factories and hospitals where work actually happens. That is the layer where Japan already leads.
The Layer Where Japan Already Leads
Nowhere is that clearer than in physical AI, the fast-growing field in which software controls robots and industrial machinery. Japan enters it from a position of dominance. Its manufacturers account for roughly 50 percent of the global industrial robotics market, and five of the world's ten largest robot makers, among them FANUC, Yaskawa, Kawasaki and Nachi-Fujikoshi, are Japanese. In 2026 those companies are folding modern AI into that installed base. FANUC has partnered with NVIDIA to put robots that respond to spoken commands on the factory floor, while Yaskawa's newest systems learn tasks by physical demonstration, capturing the tacit skill of a veteran craftsman before it retires with him.
Government policy is pointed squarely at this advantage. Building on the 2025 AI Promotion Act, which favours adoption over restriction, Tokyo has set a target of capturing 30 percent of the global physical AI market by 2040 and has allocated 387.3 billion yen to domestic AI models and physical AI within a 1.23 trillion yen package for artificial intelligence and semiconductors. The plan is not to out-spend Silicon Valley on foundation models, but to lead where AI meets the physical world.
Even Japan's most advanced model-builders reflect this instinct. Sakana AI, the Tokyo lab now valued at around 400 billion yen and backed by MUFG and NVIDIA, has become the country's most valuable AI start-up by rejecting the brute-force race outright. Rather than chase ever-larger models, it builds smaller, efficient, specialised systems tuned to Japanese language, industry and values, and is expanding in 2026 into manufacturing, finance and the public sector. It is a reminder that Japan's reflex, even at the frontier, is to make AI useful before making it large.
That reflex runs through the wider ecosystem. Mitsubishi Electric applies its Maisart AI to smarter manufacturing, and Fujitsu advances AI platforms for healthcare and enterprise operations, while a newer generation of specialists, from HACARUS in explainable visual inspection to PKSHA Technology in enterprise automation, targets concrete industrial problems. The pattern repeats where Japan's demographic pressure is sharpest. In healthcare, LPIXEL deploys AI medical imaging across Japan and six ASEAN markets, and Waseda University's AIREC robot pioneers AI-assisted caregiving. In services, logistics and finance, AI now handles multilingual support, demand forecasting, route optimisation and fraud detection, allowing shrinking workforces to hold service standards that would otherwise slip.
A Model the World Is Starting to Need
Here the story turns outward, and the commercial logic becomes clear. The pressures forcing Japan to make AI practical, an ageing population, a shrinking workforce, rising healthcare costs and stubborn productivity gaps, are no longer Japan's alone. They are arriving across the developed world. Technology proven under Japan's demanding conditions is therefore built for export, not merely for home. Japanese firms are already acting on it: FANUC earns a large share of its revenue abroad, Yaskawa is expanding across China and Southeast Asia, LPIXEL is live across ASEAN, and Sakana has declared its intent to pursue partnerships and acquisitions for long-term global growth.
What Japan's applied-AI leaders have often lacked is not capability but visibility. The world still tends to score Japan on the race it is not trying to win, and to overlook the one it is. As global attention shifts from AI experimentation to proven enterprise adoption, the companies that can demonstrate trusted implementation, measurable gains and deep industry expertise will define the market's next phase. Many of them are Japanese, and the international audience deciding whom to partner with, invest in and buy from is only beginning to notice.
The first AI race rewarded invention. The next will reward implementation. That is where Japan's opportunity begins.