Physical AI, which combines artificial intelligence with robots and sensors to interact with the physical world, is arriving faster than most governments anticipated, yet fewer than 10% of US economic leaders say their states are prepared for this transformation. The technology is no longer confined to research labs; it's already reshaping factories, logistics centers, and manufacturing floors across the globe, particularly in China, where humanoid robots are now performing semiconductor assembly and other precision tasks that were once exclusively human work. What Exactly Is Physical AI, and Why Should Governments Care? Physical AI refers to artificial intelligence systems that operate in and interact with the physical world, rather than existing only in software or digital environments. Unlike traditional AI that processes data in data centers, physical AI combines AI models with sensors, actuators, and control systems that allow machines to perceive their surroundings, reason through complex problems, and take action in real time. This represents a fundamental shift from "bits to atoms," as one industry analyst described it. The implications are staggering. According to Deloitte research, physical AI could have a transformative impact on 41% of companies within three years, yet only 3% have extensively integrated it into their operations today. For governments, this creates an urgent policy challenge: how do you prepare your workforce, infrastructure, and regulatory frameworks for an economic shift that's happening faster than anyone expected? "Physical AI is different in that it's domain specific, requiring separate adoption by end markets that each have unique requirements," explained analysts at Citigroup, noting that spending patterns will be defined not by major tech companies' investment plans, but by the pace of adoption in each industry sector. Citigroup Research Team How Are Governments and Businesses Preparing for Physical AI Deployment? Preparation for physical AI requires a multi-layered approach that goes beyond simply purchasing robots. Industry leaders and policymakers are focusing on several key areas: - Digital Twin Development: Creating virtual representations of physical processes and manufacturing environments to test and train AI systems before deploying them in real-world settings, reducing costly errors and accelerating adoption timelines. - Real-World Data Infrastructure: Building edge computing networks and sensor systems that gather continuous data from physical environments, enabling AI systems to learn and adapt to local conditions and variations. - Simulation and Training Environments: Investing in high-fidelity physics modeling and photorealistic rendering to dramatically reduce model training times and make simulation a primary training ground rather than just a testing tool. China has taken an especially aggressive approach. The country's 2026 Government Work Report explicitly pledged to establish mechanisms to boost investment in embodied AI, alongside quantum technology, brain-computer interfaces, and 6G. This represents a renewed effort after embodied AI was first elevated as a national priority in 2025. What Real-World Results Are Companies Already Seeing? The economic impact is no longer theoretical. At Chinese carmaker Nio's smart manufacturing plant, embodied AI technology is being used to navigate automated storage systems, retrieve parts, and assemble vehicle bodies. The results speak for themselves: production efficiency increased by more than 30%, labor costs dropped by 25%, and defect rates fell by 40%. These aren't marginal improvements; they're transformative changes to the economics of manufacturing. In Hefei, Anhui province, a 1.66-meter-tall humanoid robot called Lingshu is performing semiconductor assembly, a task requiring extreme precision where even a fraction of a millimeter error means failure. According to its developer, Youibot, one such robot can work as efficiently as 8 to 12 human workers per shift while operating 24 hours a day. The company has already deployed Lingshu robots in electronics factories and logistics centers across multiple Chinese cities. Global market data shows the momentum is accelerating. International Data Corporation reports that global shipments of embodied AI industrial robots reached 18,000 units in 2025 and are expected to exceed 50,000 in 2026, with China accounting for more than 45% of the market. UBTech's Walker S2 humanoid robots are being deployed in factories across South China, with the company securing orders worth over 100 million yuan and planning to deliver more than 1,000 units in 2026. "Over the past five years, China has made rapid progress in embodied AI, especially humanoid robots, reaching the top international tier and even taking a leading position in some areas," stated Yao Qizhi, a Turing Award winner and academician at the Chinese Academy of Sciences. Yao Qizhi, Turing Award Winner and Academician, Chinese Academy of Sciences Why Are Governments Falling Behind on Physical AI Readiness? Despite the rapid deployment of physical AI technology, government preparedness lags significantly. According to BCG's AI Maturity Matrix, while 88% of US economic and workforce-development leaders recognize that AI is crucial to their state's competitiveness, fewer than 10% say their states are actually ready for this transformation. This gap between awareness and readiness represents a critical vulnerability. The challenge isn't just about understanding the technology; it's about the systemic nature of physical AI adoption. Unlike generative AI, which primarily impacts data centers and digital infrastructure, physical AI is domain specific. Each industry sector, from logistics to healthcare to manufacturing, has unique requirements and adoption timelines. This means governments can't simply copy one state's playbook; they need customized strategies for their regional economy. "For China, it is not a single technological breakthrough, but a systematic project," explained Yao Qizhi, emphasizing that embodied AI represents a convergence of computing power, algorithms, hardware, and real-world data that requires coordinated national effort. Yao Qizhi, Turing Award Winner and Academician, Chinese Academy of Sciences What Obstacles Still Stand in the Way of Widespread Adoption? Despite impressive progress, significant technical and policy hurdles remain. Lin Yonghua, chief engineer at the Beijing Academy of Artificial Intelligence, noted that more work is needed to achieve stable, high-quality control of humanoid robots, improve their dexterous manipulation capabilities, and overcome constraints in power supply and heat management. These aren't minor engineering challenges; they're fundamental limitations that affect how long robots can operate and how precisely they can perform delicate tasks. Global competition is intensifying as well. The United States, Japan, and Germany are ramping up investment in embodied AI, creating pressure on other nations to accelerate their own efforts. For China and other countries, this competition is also about securing industrial resilience and supply chains, not just technological leadership. He Xiaopeng, CEO of electric vehicle maker Xpeng, called for greater national-level research and development funding and standardized frameworks similar to autonomous driving classifications to accelerate commercialization. Without clear standards and regulatory guidance, companies face uncertainty about how to scale their physical AI deployments. What's the Market Forecast for Physical AI Over the Next Decade? The financial opportunity is enormous. China's Development Research Center of the State Council forecasts that the domestic embodied AI market could reach 400 billion yuan, approximately $55 billion, by 2030 and surpass 1 trillion yuan by 2035. These projections suggest that physical AI will drive productivity gains across logistics, manufacturing, and services sectors at a scale comparable to previous industrial revolutions. Industry analysts at EY estimate that physical AI could be five to six times the market size of agentic AI within five to six years. This projection underscores why governments can no longer treat physical AI as a distant concern; it's reshaping the economic landscape right now. The bottom line for policymakers is clear: physical AI adoption is accelerating faster than government readiness. States and nations that fail to prepare now risk falling behind in workforce development, infrastructure investment, and regulatory frameworks. The robots are already on the factory floor. The question is whether governments will catch up in time.
