360° View on U.S.–Japan Cooperation on Physical AI: Scale, Rules and Trust

Six Priorities to Build Physical AI Together

By Mark Kennedy, NYU DRI Director, WISC Founding Director

China installed about 59 percent of the world’s industrial robots last year, more than the rest of the world combined. Neither the United States nor Japan can match that scale alone. Together, and with trusted partners, they can.

That conviction ran through the WISC Global Dialogue we convened in Tokyo on October 14 with Japanese and American officials, executives, and researchers. It also frames a pressing question for Washington. On October 28, a presidential proclamation under Section 232 identified robotics imports as a national security threat and opened a 180-day window to negotiate with trading partners before potentially imposing tariffs. How the allies use that window will decide whether physical AI becomes the foundation of a shared industrial base or another market dominated by China.

We asked six contributors one question: what single step, taken by which government body in the next year, would most help the two countries deploy AI-enabled robots and machines at scale and strengthen the industries that build them? Their answers outline a coordinated strategy.

The first priority is scale. Tatsuya Terazawa argues that no single country, not even the United States, can reach the scale needed to compete alone, and calls for exempting applied products from Section 232 measures. Jeffrey Kucik warns that U.S. domestic-content requirements risk excluding Japanese components, and proposes recognizing content from trusted allies. Jerry Haar extends this approach to Mexico and Brazil, where joint development finance can make trusted automation the commercially rational choice.

Achieving scale also depends on common standards and stronger security. Akira Igata urges the allies to coordinate in international standards bodies, where Chinese organizations already hold key leadership roles. Jun Osawa calls for joint safety and cybersecurity guidelines built on security by design. Danny Steed proposes that the U.S. and Japanese coast guards make physical AI resilience exercises routine.

Together, these six priorities point to a unified strategy: aggregate allied demand, align the rules, and build security in from the outset. China can command its industries to scale. Democracies must coordinate theirs, and the next 180 days are an opportunity to demonstrate that coordination can move just as fast.

Global Scale is Crucial to Remaining Competitive in Physical AI

By Tatsuya Terazawa, Chairman and CEO, The Institute of Energy Economics, Japan; former Vice-Minister for International Affairs, Ministry of Economy, Trade and Industry, Government of Japan

Deployment is an essential element in harnessing the potential of physical AI. Governments can provide subsidies to support deployment, but such support requiring substantial funding will not be sustainable without cost reduction. To reduce costs, economies of scale and going down the learning curve are necessary. Previously, countries could have achieved these within their domestic markets, but this is no longer possible. China’s success in realizing global scale in areas such as solar panels and batteries has made its products extremely cost competitive, leading to market domination.

The competition for physical AI has just begun. Yet if we do not take the right actions now, we will repeat the mistakes made in the realms of solar panels and batteries. China’s strong position in manufacturing will also provide a firm basis for it to achieve global scale in physical AI. No single country, not even the United States, will be able to realize the scale to compete with China. To remain competitive, we need to aggregate demand and production to ensure global scale. Onshoring is bound to face limits by failing to ensure competitiveness. Friend-shoring with allies is the realistic strategy to ensure global scale and remain competitive.

To do so, the Department of Defense should grant exceptions to physical AI from allies in a fast and predictable manner. Section 232 tariffs and quotas should exempt products from allies.

The FCC’s rules on foreign-produced robots include onshoring conditions, a domestic-content requirement of more than 65 percent, rising to 75 percent in 2029. Such conditions should be lifted. In the meantime, the pragmatic approach is to allow a long-term plan for onshoring. Investment in the United States by players from other countries can only happen after they have brought down costs and built enough demand in the U.S. market to justify it.

We should not lose sight of the strategic goal to enhance security by developing a supply chain of physical AI that can compete with China. That requires an alliance not only militarily, but also for the physical AI supply chain.

Aligning U.S. Domestic Content Rules with Allies

By Dr. Jeffrey Kucik, WISC Global Fellow, Adjunct Professor of Political and Law at the University of Arizona

The United States and Japan cannot compete effectively with China in physical AI deployment if trade restrictions divide their supply chains. Yet recent regulatory changes run precisely that risk. The Federal Communications Commission (FCC) has expanded its “Covered List” of market restrictions to include foreign-produced advanced robotics. Now, robots must satisfy the Buy American Act domestic-content standard requiring that more than 65% of component costs—rising to 75% in 2029—originate in the U.S. And on October 28, a presidential proclamation under Section 232 found that robotics imports threaten national security.

The FCC rule contains no exemption for trusted allies, and the Section 232 outcome keeps the door open for additional tariffs. These policies risk pushing valuable contributions from countries like Japan out of the U.S. market.

A different approach is required—one that balances domestic security concerns against the benefits of allied cooperation. To that end, the FCC, in coordination with the Department of Defense, could establish a “Trusted Allied Content” credit mechanism within the domestic-content test so that robots with substantial allied content are not added to the Covered List.

The incentives for these carveouts are clear. First, Japan produces some of the world’s best precision robotic hardware. Treating Japanese components as non-qualifying foreign content creates inefficiencies for American robotics integrators, forcing them to replace high-quality allied components with domestic alternatives that may not be ready for prime time.

Second, U.S. industrial policy is right to prioritize removing untrusted adversary technology from critical supply chains. But it should not penalize key security partners—and blanket restrictions deny the technologies and expertise that partners can offer. Carve-outs for verified components produced in treaty-allied nations like Japan will safeguard U.S. manufacturing resilience without fragmenting transpacific technological integration.

With those two considerations in mind, creating a trusted allied rule of origin will lower production costs, accelerate deployment across factories, and ensure that democracies build the physical AI ecosystem together.

Make Trusted Automation the Rational Choice in Latin America

By Dr. Jerry Haar, WISC Global Fellow

As robot prices fall and the threshold for adoption drops, how can the United States and Japan partner with manufacturing hubs in Latin America, such as Mexico and Brazil, so that their adoption of automation strengthens trusted supply chains rather than defaulting to cheaper Chinese systems?

The best approach is not to try to exclude Chinese robots by price, but to make trusted automation the commercially rational choice in Mexico and Brazil: lower lifecycle costs, financeable systems, credible security standards, local integration capacity, and guaranteed access to U.S. and Japanese markets.

China’s scale is formidable—Chinese factories accounted for nearly 60% of global industrial-robot installations in 2025—so a purely “buy allied” rule would be costly and politically difficult.

As for Latin America, the United States and Japan should treat automation in Latin America as supply-chain infrastructure, not simply as equipment procurement. Japan brings industrial robotics hardware, integration know-how, and industrial-data practices; the United States brings AI/physical-AI software, cloud and cybersecurity capabilities, finance, and market access.

Japan’s automotive manufacturing base in Mexico is substantial: roughly 1.3 million light vehicles annually—about one-third of Mexico’s production—alongside approximately $18 billion in accumulated automotive investment and several hundred automotive-related manufacturers and suppliers. That makes it a natural U.S.–Japan opportunity because Japanese companies already operate inside an integrated North American production system, rather than simply exporting finished vehicles from Japan. The October 2026  The U.S.-Japan Technology prosperity Dealexplicitly identifies this division of labor—Japanese hardware and industrial robotics plus U.S. physical-AI models—and includes joint R&D, grand challenges, and supply-chain mapping.

Mexico and Brazil are the natural first anchors: Mexico is deeply tied to North American manufacturing and accounts for an estimated 40-45% of Latin American robot installations as of 2026, while Brazil’s demand is more diversified across automotive, agricultural equipment, food processing, and consumer goods. The binding constraint is less the purchase of robots than the ability to integrate, operate, maintain, and continuously improve them.

Undoubtedly, the single most useful step that partners the U.S. and Japan could take within the next year should focus on development finance. The U.S. Development Finance Corporation and the Export-Import Bank, together with Japan’s JBIC and NEXI, should launch a joint trusted-automation financing facility for Mexico and Brazil. Such an undertaking would finance complete systems, including robots, integration, training and maintenance. These actions would be carried out according to terms that beat Chinese offers on lifecycle cost, provided they meet shared U.S.–Japan security standards. Providing this initiative is paired with clear access to U.S. and Japanese markets, it would turn trusted automation from a political preference into the better business decision.

Winning the Rules Race: Japan-U.S. Cooperation on Embodied AI Standards

By Akira Igata, Project Lecturer, Research Center for Advanced Science and Technology, The University of Tokyo

Embodied AI, which includes AI-enabled robots, vehicles, industrial machinery, and other systems operating in the physical world, could become one of the most consequential sources of industrial growth over the next decade. In sensitive sectors such as defense, critical infrastructure, advanced manufacturing, economic security, and cybersecurity concerns will increasingly favor domestic suppliers or suppliers from trusted countries. But the much larger global market, particularly in emerging economies, will also be shaped by something considerably less visible—international technical standards.

Japan and the United States should coordinate their positions and engagement in international standard-setting bodies for embodied AI, bringing together METI, MIC, MOFA, AISI, industry, and academia with U.S. counterparts (including OSTP and NIST). The October 2025 U.S.-Japan Technology Prosperity Deal already commits both governments to cooperation on AI standards. Embodied AI should be a natural next step in translating commitment into practice.

Standards shape interoperability, interfaces, testing methodologies, safety requirements, and ultimately whether products can scale efficiently across multiple markets. The United States has historically provided much of the technical capacity underpinning the international standards system, from which Japan has benefited. But that capacity cannot be taken for granted. NIST is facing significant staffing constraints and the administration has proposed further funding reductions for FY2027. Japan should therefore view greater participation not simply as an industrial-policy objective, but as alliance burden-sharing in the provision of a global public good.

The urgency is already visible. ITU-T established its Focus Group on Embodied AI in February 2026. Its chair is from China’s CAICT, and Chinese organizations hold several important working group leadership positions. The group is already examining the technical foundations from which future standards may emerge: architectures, datasets, interfaces, interoperability, benchmarking, and evaluation. Japan is participating, but Japan and the United States should coordinate far more systematically.

China’s rise in 5G demonstrated how sustained engagement in seemingly technical standards bodies can reinforce commercial competitiveness and global technology diffusion. The lesson for embodied AI is simple. Democracies should work together to shape the rules before others shape them for us.

Ensuring the Safety and Cybersecurity of Physical AI: A Priority for U.S.–Japan Cooperation

By Jun Osawa, Senior Fellow, Nakasone Yasuhiro Peace Institute (NPI). He is also Senior Fellow at Sasakawa Peace Foundation, Executive Advisor, IT Security Center, Innovation Promotion Agency, Research Director, DENTSU SOKEN Center for Economic Security Research, a board member of Kajima Peace Institute.

Physical AI integrates artificial intelligence with physical systems, enabling humanoid robots, autonomous vehicles, and industrial robots to operate and interact alongside humans. AI, which has largely been confined to the digital realm, is now entering the physical world where people live and work. By augmenting human capabilities and complementing human activities, physical AI has the potential to contribute to a more efficient and productive society.

However, as physical AI systems increasingly interact directly with humans, ensuring their safety—particularly preventing their actions from causing physical harm—becomes critically important. Although physical AI systems will rely heavily on edge computing for local processing and decision-making, completely isolating them from external networks (including the Internet) through air-gapped architectures will be difficult. Ensuring their safety therefore requires robust cybersecurity measures based on the principle of “security by design,” including protection against cyberattacks originating from external networks.

In the digital realm, cyberattacks leveraging increasingly sophisticated AI capabilities are a growing threat, making cybersecurity more urgent than ever. The fundamental principle of cybersecurity—identifying and remediating vulnerabilities as quickly as possible—remains unchanged. At the same time, defenders are increasingly expected to leverage advanced AI capabilities to autonomously identify vulnerabilities and remediate them in real time without disrupting system operations, where feasible.

Similar cybersecurity capabilities will be essential for physical AI systems as they become more widely deployed. Japan and the United States should work together to develop international guidelines establishing common safety and cybersecurity standards for physical AI before its widespread adoption. These guidelines should incorporate the principle of “security by design,” with particular emphasis on enabling real-time vulnerability detection and remediation without interrupting operations, as well as detecting and preventing the manipulation or poisoning of AI training data.