By Mark Esposito
Japanese and US industry are both going to need large numbers of robots but success depends on more than quantity.
For different reasons, Japan and the United States will need to install industrial robots on a large scale if they are to compete internationally. But maximizing the benefits is as much about process design and management as it is about the technology, as this article explains.
1. Introduction
Humanoid robots may sound like a technology of the future, but Japan’s interest in them is driven by a present-day economic problem: it is running out of workers. The Ministry of Economy, Trade and Industry (METI) projected in 2025 that Japan would need 4.98 million workers with AI and robotics skills by 2040, but that current training trends would supply only 1.72 million, leaving a shortfall of about 3.3 million (METI, 2025; Xinhua News Agency, 2025). The ministry also expected a shortage of 2.81 million manufacturing workers. Against this backdrop, robotics is no longer an attractive possibility; it has become an economic necessity.
It is tempting to read Japan’s push as proof that the United States is falling behind and, on robot density or public spending, it is. But those figures measure what goes in, not what comes out. What matters is whether robots let firms and their workers produce more in an hour. Buying the machines is the easy part. Someone has to redesign the workflow around them, train the technicians who keep them running, and decide which tasks stay with people. My argument is that the gains will go to the countries and firms that get this organizational work right, not to those that install the most robots.
2. Two Kinds of Scarcity
Japan faces an absolute labor shortage. Its population peaked in the late 2000s, and official projections indicate it will fall from about 126 million in 2020 to around 87 million in 2070. The working-age population is expected to decline even more rapidly (National Institute of Population and Social Security Research, 2023). In practical terms, Japan will have fewer people available to work.
Japan and South Korea have so many robots per worker because they are aging fast and because they are good at building and using machines.
America’s problem is not a lack of people. Its population is still growing. What has shrunk is the share of adults who work or look for work: 61.6 percent in August 2026, almost six points below the 2000 level (USAFacts, 2026), and the Bureau of Labor Statistics expects an aging population to push it lower over the next decade (U.S. Bureau of Labor Statistics, 2026). Yet manufacturers, logistics operators, and builders still say they cannot hire. The workers exist, but often with the wrong training, or in the wrong place, for the jobs on offer. The United States thus has two problems stacked on each other: too few people participating, and a poor fit between those who do and the work that needs doing.
That difference shapes why each country wants robots. In Japan, automation fills jobs that nobody is available to do. In the United States, its value depends more on whether training institutions and labor markets adjust alongside the technology.
3. Japan’s Strategy
Japan’s robotics policy predates the current excitement about artificial intelligence by years. The 2015 New Robot Strategy already treated robots as a way to cope with labor shortfalls in manufacturing, agriculture, infrastructure, and elder care (Headquarters for Japan’s Economic Revitalization, 2015). More recently, the government’s attention has moved from the hardware itself to the AI systems that make machines more autonomous.
In June 2026, METI selected Noetra, a consortium backed by SoftBank, Sony, NEC, Honda, and the National Institute of Advanced Industrial Science and Technology, to build foundation models for physical AI. The project received ¥387.3 billion in its first year and could receive up to ¥1 trillion over five years, depending on how the consortium performs at each review stage (TechWire Asia, 2026). It sits within the AI Robotics Strategy launched in March 2026, which targets more than 30 percent of the global AI robotics market by 2040 and the deployment of roughly 10 million AI-powered robots across 18 sectors, including logistics, nursing, and food service (METI, 2026; NVIDIA, 2026). The scale of the bet follows from the demographics. A country whose workforce is shrinking can only hold output steady by producing more with fewer people.
4. The Automation Gap
According to the International Federation of Robotics (IFR), industrial robot installations in the United States rose 11 percent in 2025, to 38,000 units (IFR, 2026a). Even so, the country trails several other large manufacturing economies. South Korea runs 1,220 industrial robots per 10,000 manufacturing workers, Japan 446, and the United States 307 (IFR, 2026b). Worldwide, about 542,000 units were installed in 2024, more than double the figure of a decade earlier, and China alone accounted for 295,000 of them, or 54 percent of the total (IFR, 2025).
These gaps are not simply a matter of national technological ambition. Acemoglu and Restrepo (2022) show that aging societies adopt robots faster, because scarce labor raises the payoff from automation. Put simply, Japan and South Korea have so many robots per worker because they are aging fast and because they are good at building and using machines. Their industrial mix helps too: car plants and electronics lines, which loom large in both economies, are among the easiest places to automate. Robot density, then, reflects demography, specialization, investment, and policy together, and it is a poor proxy for technological leadership. What is clear is that automation is now a routine part of how economies compete.
5. What Robots Do to Jobs
Research on automation points to two opposing forces. Robots can displace workers from tasks people used to do, but the productivity gains they bring can also lower costs, expand output, and create demand for new kinds of work (Acemoglu & Restrepo, 2019).
How these forces net out varies widely by country and industry. In US local labor markets between 1990 and 2007, each additional robot per thousand workers reduced the employment-to-population ratio by about 0.2 percentage points and wages by about 0.4 percent (Acemoglu & Restrepo, 2020). Evidence from 17 countries tells a different story: robots raised labor productivity without substantially cutting total hours worked, though they did shrink the employment share of less-skilled workers (Graetz & Michaels, 2018). Germany followed yet another path. Manufacturing job losses there were largely offset by growth in services, and many affected workers were moved into new roles rather than laid off (Dauth et al., 2021).
Japan’s evidence points to more complementarity between robots and workers. From 1978 to 2017, falling robot prices raised both robot adoption and employment (Adachi et al., 2024). In nursing homes, robotic systems have been linked to higher employment and better staff retention, because taking over some physical and repetitive tasks frees caregivers for the interpersonal side of the job (Lee et al., 2024). Where vacancies are already hard to fill, there are fewer workers to displace. For labor-constrained sectors such as logistics, warehousing, and elder care, Japan may therefore be a better guide than America’s own earlier experience with industrial automation.
6. The Japanese Paradox
If robot density alone lifted productivity, Japan would sit near the top of the world’s league tables. It does not. In 2024, Japan ranked 28th of 38 OECD members on hourly labor productivity and remained the least productive economy in the G7 on the same measure (Japan Productivity Center, 2025). Average annual productivity growth slowed from 1.2 percent between 2000 and 2008 to 0.3 percent between 2019 and 2024, and services productivity continued to trail manufacturing (OECD, 2026).
Japanese manufacturing is heavily automated, but the benefits have not spread across the economy. Service companies and smaller firms, which employ a large share of the workforce, have adopted robots more slowly. Investment per worker has also fallen since 2000, holding back both the complementary technologies and the organizational changes needed to use robots well (OECD, 2026). The result is a country with a great deal of automation and modest productivity to show for it. This exposes the basic weakness of robot-density figures: they count the machines installed, not how productively those machines are used.
7. The Organizational Bottleneck
Economic history offers an explanation. Factories that simply swapped electric motors in for steam engines did not capture the full benefits of electricity; the larger gains came later, once companies redesigned factory layouts and production processes around what electric power made possible (David, 1990). Brynjolfsson et al. (2021) describe a similar productivity J-curve, in which investments in training, process redesign, and other intangible assets can depress measured productivity at first while laying the ground for larger gains later. Earlier waves of information technology followed the same pattern: firms earned the highest returns when they changed how work was done, not merely which tools they used (Bresnahan et al., 2002).
Humanoid robots may well take the same route. They can move materials, inspect equipment, and handle repetitive tasks, but none of this removes the need for human oversight. Firms still need maintenance schedules, safety procedures, ways of handling the unexpected, and a clear division of responsibility between workers and machines. Automation may absorb routine work while making judgment and problem-solving more valuable (Autor, 2015). Today’s robotics also depends on a good deal of labor that is easy to overlook, from the people who generate training data to the supervisors who monitor automated systems (Kim, 2026; O’Donnell, 2026). This echoes earlier research on the hidden labor behind digital platforms (Gray & Suri, 2019).
Japanese elder-care facilities show what happens when workplace organization is ignored. In several cases, care robots added to staff workloads instead of reducing them, and facilities dropped the systems once the pilot programs ended (Wright, 2023). A technically capable robot can still fail if it complicates existing routines, demands too much supervision, or addresses a task that was never the facility’s real constraint. The binding barrier, in other words, is usually organizational rather than technological.
8. Consequences for the United States
Japan’s experience suggests that robotics is as much a management and implementation problem as an engineering one. Designing capable machines is hard; fitting them into workplaces in ways that raise productivity may be harder. Yet public policy tends to concentrate on developing the technology and gives less attention to deployment, organizational adaptation, and the everyday conditions in which robots actually operate.
Bloom et al. (2012) found that American companies got larger productivity gains from information technology than their European counterparts, mainly because they were better managed. That finding may matter even more for robotics. Once organization is the main constraint, the ability to redesign work and manage technological change will count for more than the number of robots installed.
Despite calls from industry groups, the United States still has no national robotics strategy (Association for Advancing Automation, 2025). An effective American strategy should differ from Japan’s in three respects.
First, it should put more weight on deployment and integration. Small and medium-sized enterprises often lack the expertise and capital to evaluate robotic systems, reorganize production around them, and absorb the initial disruption. Testbeds, technical advisory services, and implementation support could help close that gap.
A technically capable robot can still fail if it complicates existing routines or demands too much supervision.
Second, the United States should strengthen the technical workforce that maintains, integrates, and supervises robotic systems. Wider adoption will require AI specialists and robotics engineers, but also technicians who can diagnose faults, adapt systems to particular workplaces, and coordinate automated equipment with human staff.
Third, policy should judge success by outcomes rather than by the number of robots installed. Purchase counts say little about whether the machines have improved performance. Evaluation should instead track changes in productivity, costs, output, and job quality, along with the human labor needed to keep automated systems running.
Government can fund research, training, testing, and standards, and it can lower the risks that keep smaller firms from experimenting. In the end, though, firms determine whether automation succeeds. Their management practices, investment decisions, and capacity to reorganize work will decide whether robots become productive assets or expensive, underused equipment.
9. Conclusion
Japan and the United States face labor shortfalls for different reasons, yet both see robotics as part of the answer. The evidence suggests that robots can support both employment and output, particularly where labor is already scarce. Japan’s experience also shows, however, that adopting robots does not by itself guarantee faster productivity growth.
As robots become more capable and cheaper, access to the technology will matter less. The larger differences will come from how organizations use it. Firms that rework their workflows, build the necessary technical skills, and coordinate human and robotic labor well will be best placed to benefit. The future of robotics will depend not only on what the machines can do, but on whether organizations learn to use them well.
