Defining Physical AI: The Fusion of Intelligence and Physicality

Physical Artificial Intelligence, often interchangeably called Embodied AI or Embodied Intelligence, marks a pivotal evolution in artificial intelligence. This field focuses on AI systems embedded within physical bodies—such as robots, drones, or autonomous machines—that interact directly with the real world. Unlike purely software-based AI, which processes data in virtual environments, Physical AI learns through sensory perception, movement, and environmental feedback loops. For instance, a humanoid robot might grasp objects by combining visual recognition, force sensing, and adaptive motor control, mimicking human-like dexterity.

In China, this domain has gained traction due to its potential to revolutionize industries like manufacturing, healthcare, and logistics. The term 'embodied intelligence' (具身智能) underscores the idea that true intelligence emerges from bodily interaction, a concept rooted in cognitive science and robotics research dating back to the 1970s with Japan's WABOT-1, the world's first humanoid robot. Today, advancements in large language models, vision transformers, and reinforcement learning have accelerated progress, enabling machines to perform complex tasks like dancing or caregiving.

Major Chinese universities are responding to this shift by restructuring curricula to blend AI algorithms with mechanical engineering, control systems, and materials science, ensuring graduates can bridge theoretical models with practical deployment.

Government Backing Fuels University Expansion

China's central government has positioned embodied intelligence as one of six 'new quality productive forces'—key future industries alongside quantum tech, biomanufacturing, and 6G. The 2025 Government Work Report explicitly urged accelerated development, prompting a nationwide push in higher education. This aligns with broader strategies under the 14th Five-Year Plan, where over one-fifth of academic programs were adjusted in recent years to prioritize strategic fields like AI.

Authorities are establishing talent cultivation mechanisms that sync education with industrial demands, including vocational-general education integration and cluster training for high-skilled workers. During the plan period, higher education produced 55 million graduates, with vocational systems supplying over 70% of new skilled labor for modern sectors. At least 535 universities now offer AI-related majors, and top institutions are expanding enrollment—Peking University added 150 spots in 2025 for strategic needs, while Renmin University increased AI places by over 100.

This policy ecosystem not only funds labs and platforms but also fosters industry-academia ties, addressing fragmented curricula and outdated facilities criticized in current programs.

Shanghai Jiao Tong University Pioneers China's First Embodied AI Major

Leading the charge, Shanghai Jiao Tong University (SJTU) announced China's inaugural undergraduate major in embodied intelligence in late 2025, housed in its School of Artificial Intelligence. This four-year Bachelor of Engineering program, directed by Lu Cewu—a prominent researcher and co-founder of Noematrix (backed by Alibaba)—aims to produce interdisciplinary experts capable of real-world AI deployment.

The curriculum systematically integrates AI, mechanical engineering, computer science, and robotics, emphasizing hands-on training. Students engage in projects involving perception-decision-action loops, from sensor fusion to dexterous manipulation. SJTU plans collaborative platforms with industry leaders for practicum centers, tackling gaps in practical skills and real-scenario detachment. Facilities include advanced robot labs simulating factories and homes, preparing graduates for roles in humanoid development.

"Current curricula fail to meet industry demands," SJTU noted, highlighting fragmented disciplines and weak engineering focus. Graduates are poised for high-demand positions, reflecting the program's strategic alignment with national goals.

Zhejiang University and Other Powerhouses Join the Fray

Zhejiang University, alma mater of DeepSeek AI founders, is rolling out embodied intelligence courses and majors, leveraging its robotics strengths. Beijing Institute of Technology (BIT) plans to enroll 120 undergraduates annually, targeting employment at Huawei, Tencent, state space firms, and automakers—70% for further study, 50% direct workforce entry.

Beihang University, renowned for aeronautics, and Xi'an Jiaotong University are also launching programs, focusing on aerospace and intelligent manufacturing applications. These initiatives build on existing AI labs at Tsinghua and Peking Universities, where research in AI-Newton (autonomous physics discovery) and cognitive reasoning complements embodied efforts. Collectively, seven top universities are adapting, creating a robust talent pipeline.

This multi-institutional approach ensures regional coverage, from Shanghai's tech hub to Beijing's policy center, fostering innovation clusters.

Bridging the One-Million Talent Gap in Physical AI

China's Physical AI sector faces a stark talent shortage—estimated at one million professionals by BIT—amid explosive growth. Cumulative global humanoid robot financing hit 32.8 billion yuan by September 2025, with over 80% (120 of 140 events) in China. The market is projected to reach 400 billion yuan by 2030 and exceed 1 trillion yuan by 2035, per the State Council's Development Research Center.

Job postings from ByteDance, JD.com, Huawei, and BYD underscore demand, with embodied intelligence algorithm engineers commanding 25,000-90,000 yuan monthly—average annual salary 333,400 yuan, surpassing general AI's 290,900 yuan (Liepin data). Broader AI talent gap exceeds five million. Universities' response integrates theory with practice, producing graduates ready for robotics R&D, autonomous systems, and smart factories.

• High salaries reflect specialized skills in multimodal AI and hardware-software fusion.
• Industry needs versatile talent versed in physics-AI hybrids.
• Vocational tie-ins ensure scalable skilled labor.

Curriculum Overhaul: From Theory to Real-World Robotics

New programs emphasize step-by-step skill-building: foundational AI (neural networks, reinforcement learning), physical modeling (dynamics, kinematics), sensor integration (vision, LiDAR, tactile), and end-to-end systems design. Students progress from simulations to physical prototypes, using platforms like ROS (Robot Operating System) and advanced manipulators.

Interdisciplinary electives cover biomechanics for humanoid gait, ethical AI for deployment safety, and scalable manufacturing. Labs feature outdated upgrades to state-of-the-art humanoid testbeds, enabling projects like collaborative robots in warehouses. This contrasts prior fragmented offerings, now unified for holistic training.

Stakeholders praise the shift: academics gain industry relevance, enterprises access plug-and-play talent, and students benefit from high employability. For aspiring professionals, exploring academic career advice can prepare you for these competitive fields.

Industry-Academia Synergies and Career Pathways

Partnerships are central: SJTU's Noematrix collaboration exemplifies funding and internships, while BIT links to Huawei and automakers. Companies like Unitree and UBTech, dominating humanoid sales, recruit heavily, applying EV-scale strategies to robotics.

Graduates target roles in R&D, system integration, and deployment, with pathways to research assistant jobs or faculty positions. China's 6,000+ AI firms (1.2 trillion yuan core industry 2025) amplify opportunities. International students eye China for cutting-edge exposure, boosting global talent flow.

Explore university jobs in China or China academic opportunities to join this boom.

Challenges in Scaling Physical AI Education

Despite momentum, hurdles persist: high setup costs for robot labs, faculty shortages in hybrid expertise, and ethical concerns over job displacement. Curricula must evolve rapidly with tech like vision-language-action models. Regional disparities—concentrated in coastal hubs—risk uneven development.

Solutions include government subsidies, online AI platforms training 500,000+ from 2,000 unis, and international exchanges. Balanced views highlight workforce augmentation over replacement, with Physical AI addressing labor shortages in aging societies.

Photo by Spencer Gu on Unsplash

Global Implications and Future Outlook

China's push positions it as a Physical AI leader, challenging U.S. dominance amid robotics surges. By 2030, embodied AI robots could hit $77 billion market (IDC), with China central. Universities' innovations—fusing info-physical-bio intelligence (Tsinghua vision)—promise breakthroughs in elderly care, disaster response, and space exploration.

Future trends: standardized benchmarks, ethical frameworks, and AGI pathways via embodied learning. For educators and job seekers, this era demands adaptability. Visit higher ed jobs, rate my professor, and career advice for resources. Post a job at our recruitment page to attract top Physical AI talent.