Japan doubles down on 'physical AI': $6.1b bet powers robot surge and national AI factory push

Japan's Ministry of Economy, Trade and Industry (METI) is spearheading a landmark $6.1 billion investment over five years into the Noetra consortium, a public-private alliance anchored by industrial giants such as SoftBank, Sony, NEC, and Honda. This initiative aims to develop domestic multimodal foundation AI models that integrate diverse data types—including images, video, audio, and physical properties—to power 'physical AI' applications controlling robots. By targeting sectors of national strength like healthcare, disaster response, manufacturing, and nuclear decommissioning, Japan seeks to reduce its reliance on foreign AI technologies and enhance both industrial competitiveness and national security.

The Noetra consortium plans to release its initial foundation model within fiscal year 2026, with annual iterative improvements driven by manufacturer data, creating a robust feedback loop that ties R&D directly to real-world deployment. This approach is bolstered by complementary subsidies totaling approximately JPY72.5 billion to expand domestic GPU cloud infrastructure, enabling low-latency, on-premise AI compute that eases data governance and integration with robotics hardware. Such a combined strategy not only accelerates adoption but also mirrors successful global public-private partnership models emphasizing practical data-sharing agreements and service-level commitments critical for industrial uptake.

Beyond technology, Japan’s sovereign AI strategy is underpinned by a concrete deployment target of 10 million AI-powered robots across 18 sectors by 2040, including food manufacturing, restaurants, and medical services. This ambitious goal leverages Japan’s commanding 70% share of the global industrial robotics market, providing a credible production base and unique data sources to sustain a continuous engineering demand curve for domain-specific AI solutions. As Noetra President Hironobu Tanba warns, reducing dependence on foreign AI models is vital to safeguard business continuity and protect sensitive corporate information from international disruptions.

Japan’s sovereign AI initiative also serves as a powerful economic engine by fostering local talent, innovation, and entrepreneurship within a supportive ecosystem. By building AI models that incorporate local language and cultural context, the country aims to create a flywheel effect where government subsidies and industry collaboration stimulate a thriving domestic AI landscape. This strategic focus not only enhances national security but also positions Japan to lead in specialized AI applications aligned with its societal and industrial priorities.

Japan’s ambitious plan to deploy 10 million AI-powered robots by 2040 across 18 sectors—including restaurants, food manufacturing, and medical fields—reflects a strategic response to its acute labor shortages driven by demographic challenges. Industry Minister Ryosei Akazawa emphasized that this target is not merely aspirational but a concrete government mandate aimed at vigorous social implementation, signaling a shift from optional automation to essential policy-driven deployment.

This large-scale robotics deployment leverages Japan’s dominant position in the global industrial robotics market, where its manufacturers hold roughly 70% share, providing a credible production base and rich data ecosystem to support the initiative. However, scaling to millions of reliable AI-powered robots requires overcoming unique physical AI challenges such as on-device perception, real-time control latency, safety validation in regulated sectors, and effective simulation-to-reality transfer, underscoring the complexity beyond traditional cloud-based AI models.

The government’s binding timeline for deploying 10 million robots creates a sustained engineering demand curve that extends beyond research into real-world applications, driving the development of edge inference solutions, domain-specific datasets, and validated integration stacks. This approach integrates physical AI solutions not only to mitigate labor shortages but also to enhance national security, reflecting a comprehensive strategy that unites startups and industrial leaders in advancing robotics as a cornerstone of Japan’s sovereign AI ambitions.

Japan has pioneered the world’s first national AI factory, a landmark government-backed initiative launched in partnership with Nvidia and leading domestic industrial players such as Mitsubishi Heavy Industries and Hitachi. This facility, powered by an immense array of 27,500 Nvidia Rubin GPUs and 13,750 Vera CPUs, is designed to serve as a strategic national asset, elevating Japan’s AI compute capabilities to rival those of global tech giants and underscoring the country’s commitment to large-scale AI infrastructure as foundational as highways or power grids.

Set to become operational by June 2028, Japan’s AI datacenter will harness Nvidia’s next-generation Vera Rubin GPU architecture, contingent on timely GPU shipments starting in fall 2026 and robust power grid support to meet the facility’s substantial energy demands. This ambitious project exemplifies cross-border and cross-industry collaboration, with South Korea’s SK Group developing an AI-specialized datacenter in Japan, while Mitsubishi Heavy Industries and Hitachi innovate advanced cooling, power systems, and even floating datacenter concepts to optimize infrastructure for the intense workloads of physical AI applications.

Japan’s national AI factory is uniquely optimized for physical AI—technologies that interact with and manipulate the physical world through robotics, autonomous systems, and manufacturing—leveraging the country’s industrial strengths in logistics, healthcare, and automation. This centralized, government-coordinated approach to AI infrastructure stands in sharp contrast to the primarily commercial, decentralized investments seen in the US and China, positioning Japan as a potential global pioneer in integrating AI compute resources directly with industrial innovation and national strategic priorities.

Leading industrial giants such as Fanuc, Kawasaki, and Stellantis are spearheading a transformative wave of AI integration on factory floors by forging strategic partnerships with Silicon Valley tech leaders like Google, Nvidia, and Accenture. These collaborations focus on embedding embodied AI systems that combine perception, reasoning, and physical action to tackle pressing manufacturing challenges including labor shortages and the need for flexible, high-mix production. For example, Fanuc’s alliance with Google aims to create smarter, adaptive robots capable of reducing human intervention during changeovers, while Stellantis leverages digital twin technology to optimize production workflows, collectively addressing the 'great margin squeeze' in manufacturing.

The rise of imitation learning as a practical training method is revolutionizing how robots acquire skills on the factory floor, moving beyond rigid, hand-coded instructions to learning by observing human actions. This approach, however, demands high-quality data, advanced force sensing, and robust production-grade hardware to ensure reliability at scale—needs that are being met through innovations like time-of-flight 3D imaging sensors championed by companies such as IDS. These cost-effective sensing technologies enhance robotic perception in complex environments, enabling more nuanced interactions and adaptability critical for industrial applications.

Cyberdyne’s launch of a $62 million corporate venture capital fund in partnership with Silicon Valley’s Pegasus Tech Ventures exemplifies Japan’s strategic push to bridge traditional robotics engineering with cutting-edge generative AI development. This fund aims to secure early access to next-generation AI 'brains' that will empower Cyberdyne’s robotic 'limbs,' reflecting a broader trend where startups provide advanced AI intelligence to industrial leaders—a model mirrored in global partnerships like Agility Robotics with BMW. As Pegasus CEO Anis Uzzaman notes, overcoming the cultural divide between established engineering and fast-paced AI labs is key to advancing physical AI, especially in human-assistive roles driven by Japan’s acute labor shortages and aging population.

Fujitsu’s collaboration with leading Japanese robotics firms to integrate Nvidia’s technology underscores the growing synergy between domestic startups, industrial leaders, and international tech companies in advancing physical AI. This partnership highlights how leveraging global AI compute infrastructure and expertise accelerates the development of intelligent robotics capable of addressing Japan’s workforce challenges and enhancing national competitiveness in the burgeoning physical AI sector.