Coordinating as Fast as China Commands: U.S.–Japan Cooperation on Physical AI

Executive Summary

The United States and Japan can compete with China in physical AI only if they operate as one market. That was the central conclusion of 27 leaders from government, industry, finance, research institutes, and academia at a WISC Global Dialogue in Tokyo on October 14, 2026, just days after both governments issued a Joint Statement on the Technology Prosperity Deal committing to joint cooperation on physical AI.

Six main themes emerged:

1. Scale, not subsidies. Because no single democracy has China’s market scale, domestic subsidies and tax credits cannot sustain deployment. Allies must pool demand and production.
2. Trust is a singular advantage. A seven-decade alliance provides the United States and Japan with a foundation of trust that China cannot replicate. However, that trust must be systematically engineered into joint verification, shared standards, and secure software updates.
3. Avoid fragmenting the market. Unilateral, slow, or unpredictable regulations regarding robot imports, tariffs, and onshoring risk dividing the allied market they aim to protect.
4. The standards contest has already begun. China is actively shaping international standards for AI in global forums where the United States and Japan are underrepresented—or entirely absent.
5. Cost is not the only barrier to adoption. Adoption faces barriers from cautious automakers, inertia among smaller suppliers, public skepticism regarding AI, and a shortage of test sites.
6. Physical AI is a national security priority. Physical AI represents a decisive strategic capability, and defense procurement could serve as an anchor for early commercial markets.

Physical AI now tests whether democracies can coordinate as fast as an autocracy can command. The recommendations below outline practical, actionable steps for both governments.

Major Themes

1. Scale, not subsidies, determines deployment

The discussion opened with a critical lesson from industrial policy: Japan’s traditional sequence—R&D and demonstration, followed by subsidies, tax credits, and low-interest loans—can initiate deployment, but cannot sustain it. Sustainable deployment relies on falling unit costs driven by economies of scale.

Japan learned this when it commercialized solar power but lost market share to Chinese scale. Participants warned that robotics could be next: China installed more than half of the world’s industrial robots in 2024, leveraging its domestic automotive and electronics sectors as a massive built-in market.

The conclusion was stark: no single democracy can match that scale alone. The solution is aggregating demand and production among trusted partners, making the guiding principle friendshoring, not isolated onshoring.

2. Trust is the allies’ advantage and it must be engineered

A seven-decade alliance gives the U.S. and Japan a distinct advantage where physical systems interact with people, critical infrastructure, and sensitive data.

However, this trust must be translated into technical reality. Physical robots must be secure, have clear data provenance, and run verified software updates. Participants identified three essential layers of partnership beyond basic R&D:

• • Building together: Combining Japanese precision hardware, industrial machinery, and operational factory data with American frontier models and computing infrastructure.
  • Verifying together: Allowing Japanese safety assessments to validate American models, and U.S. testing frameworks to certify Japanese physical hardware.
  • Deploying together: Harmonizing regulatory frameworks to integrate the two national markets into one.

Security considerations heighten the urgency. Given the spike in AI-enabled cyberattacks, verifying the identity of the manufacturer, operator, and software vendor directly on hardware is fundamental. Security by design must precede broad commercial deployment.

3. Rules among allies can protect the market—or fragment it

Participants agreed on restricting untrustworthy robotic systems from adversary nations. However, restrictions that include allies can also create friction: unilateral, sluggish, or unpredictable actions risk fragmenting the integrated market that scale requires.

Key pressure points include foreign robot approval mechanisms, the Section 232 investigation into robotics and industrial machinery, and strict domestic-content mandates. Participants urged policymakers to stabilize approval pathways and exempt like-minded nations from prospective tariffs or quotas. On October 28, the President found that these imports threaten national security but chose to negotiate with trading partners first, holding tariffs and quotas in reserve. They also noted that domestic-content timelines must remain realistic, as establishing supply chains abroad requires years. Poorly designed regulations risk stifling innovation before it matures.

4. The global standards contest is underway, but allies are largely absent

International standards are already being established. A focus group within the International Telecommunication Union (ITU) on embodied AI—currently chaired by China—is drafting global frameworks across mobility, logistics, healthcare, and manufacturing. Allied representation remains minimal, raising concerns that China could lock in a first-mover advantage.

Participants cited an upcoming ITU focus group meeting in Tokyo in July 2027 that provides a concrete opportunity for coordinated U.S.–Japan engagement. The consensus leaned toward forming agile coalitions of willing partners with technical capabilities, rather than waiting for global consensus. Japan is well-positioned to convene these plurilateral groups. Participants also discussed how and when to bring European partners with key capabilities into this cooperation.

5. Adoption is constrained by non-cost barriers

Participants outlined several operational barriers beyond capital costs:

• • Cautious market demand: Japanese robot installations declined in 2025 due to electric vehicle sector uncertainty.
  • SME inertia: Small- and medium-sized precision suppliers remain slow to automate operations.
  • Fragmented ecosystems: Individual firms continue building isolated data-generation and testing frameworks from scratch; the multimodal data physical AI needs, combining vision, touch, and thermal inputs, barely exists.
  • Public trust: Skepticism regarding data privacy and AI security limits workplace acceptance.
  • Infrastructure bottlenecks: U.S. electrical grid capacity constraints hamper AI scaling.
  • Shortage of test sites and metrics: Real-world testing environments alongside human workers remain scarce, as do clear national deployment targets.

6. Physical AI is a core national security issue

Physical AI must be evaluated through a national security lens as a strategic capability altering military and industrial balances. While advanced technologies typically transition from defense to commercial markets in the U.S., Japan has historically integrated advanced technology into defense last. Defense and government procurement can serve as vital early-stage demand anchors.

Key Open Questions

• How can allied demand be aggregated in practice? What specific legal and budgetary mechanisms—such as joint procurement or mutual recognition of safety certifications—are required?
• How can allies balance onshoring with friendshoring? How can domestic-content requirements be structured to strengthen domestic manufacturing and allied cooperation?
• What baseline criteria qualify a “trusted producer?” Which technical frameworks for verifying hardware provenance, software supply chains, and update mechanisms can both governments accept?
• Who should lead international standards coordination? Which specific agencies in Washington and Tokyo should lead efforts within the ITU and related bodies to maximize impact ahead of the July 2027 Tokyo meeting?
• How should European partners be integrated? Is a dedicated U.S.–Japan core working with select European partners more practical than attempting a broader transatlantic framework?
• Should security and safety dialogues include non-allied nations? Should safety standards involve broader international participation, or should efforts remain strictly confined to allied nations?
• How can defense procurement best anchor early adoption? Which public-sector applications are ready for immediate physical AI integration, and how can Japan’s historic policy barriers between civilian and military technology be addressed?
• How can the physical AI data bottleneck be resolved? What mechanisms can best collect, manage, and share complex multimodal datasets (combining vision, touch, and thermal data) across borders securely?

Recommendations

The following recommendations draw on the discussion, but are WISC’s own. Each proposal identifies lead entities and a 12-month decision window.

1. Establish fast, predictable regulatory pathways for allies

Create a streamlined qualification framework allowing manufacturers from trusted allied nations to achieve certification through clear criteria and predictable timelines. Use the Section 232 negotiations to formally exempt trusted partners tariffs or quotas on robotics, and align domestic-content rules with realistic manufacturing capacity timelines.

• • Lead Entities: U.S. (DoD, FCC, Commerce, USTR); Japan (METI).
  • Timeline: Within the 180-day Section 232 negotiating window, with an update due by late April 2027, and during early FCC rule implementation.
  • Trade-off: Requires confidence in foreign assessments and reduces tariff leverage.

2. Implement “verify once, deploy in both markets” frameworks

Leverage the Joint Statement’s standards dialogue to establish a unified test and verification process for physical AI within 12 months, evaluating safety, data governance, and software update security for mutual recognition.

• • Lead Entities: NIST (U.S.); AIST and AI Safety Institute (Japan).
  • Timeline: Initiate at the inaugural standards dialogue meeting; target full implementation within 12 months.
  • Trade-off: Requires both nations to delegate a degree of direct regulatory control over domestic safety assessments.

3. Coordinate bilateral engagement in international standards bodies

Formulate a unified strategy for the ITU focus group on embodied AI and related standards organizations. Target the July 2027 ITU meeting in Tokyo for joint policy proposals while recruiting like-minded partners.

• • Lead Entities: U.S. (State, Commerce); Japan (MIC, METI).
  • Timeline: Immediate action to shape meeting agendas well in advance of July 2027.
  • Trade-off: Requires ongoing administrative resources; failing to coordinate risks further market fragmentation.

4. Shift funding toward real-world deployment and test beds

Direct joint R&D funding toward industrial deployment—focusing on system integration, workforce training, and shared datasets. Link Japan’s planned Center of Excellence for AI robotics with counterpart U.S. facilities.

• • Lead Entities: U.S. (OSTP, science agencies); Japan (METI, NEDO).
  • Timeline: FY2027 budget cycles in both countries; align with the launch of Japan’s Center of Excellence.
  • Trade-off: Risks subsidizing specific commercial firms; programs must focus strictly on open assets and performance.

5. Leverage government and defense as early buyers

Identify immediate public-sector opportunities—naval shipyard operations, military maintenance, and civil infrastructure inspection—and launch joint U.S.–Japan procurement pilots to aggregate demand.

• • Lead Entities: U.S. (DoD, infrastructure agencies); Japan (Ministry of Defense, infrastructure ministries).
  • Timeline: Incorporate into upcoming defense budget and procurement cycles.
  • Trade-off: Integrating civilian tech with defense remains politically sensitive in Japan, and procurement rules naturally favor domestic suppliers.

6. Publish security-by-design guidelines and sectoral milestones

Issue joint security-by-design standards covering software supply chain transparency and update verification, alongside target sectoral deployment milestones to guide private investment.

• • Lead Entities: U.S. (NIST, CISA); Japan (Cabinet Secretariat, Digital Agency).
  • Timeline: Issue within 12 months, synchronized with Recommendation 2.
  • Trade-off: Compliance increases overhead for smaller businesses, while public targets risk driving subsidy competition over organic adoption.