How Physical AI Is Reshaping the Competitiveness of Chinese Manufacturing

AFFILIATE_PRODUCTS:

From Exporting Products to Exporting Capabilities

Physical AI is reconstructing the underlying logic of China’s manufacturing industry, driving a leap from the “world’s factory” to the “factory of factories,” and further to an industrial infrastructure model. A post by Zhong Dingjing and Sun Li in Tsinghua Management Review points out that global trade structures and manufacturing competitive orders are undergoing a period of restructuring.

In recent years, China’s exports of consumer goods have slowed, while exports of intermediate and capital goods have increased significantly. Emerging manufacturing centers in ASEAN, India, Mexico, and elsewhere are taking on final assembly processes, while remaining dependent on China for the supply of core production elements such as high-precision components, automation equipment, and industrial control systems. In early 2026, exports of electronic and mechanical products recorded a year-on-year increase of 21.8%. What supports this growth is not low-value-added manufacturing, but an enhancement of equipment manufacturing capacity and industrial system support capabilities.

According to estimates by the McKinsey Global Institute, the average geopolitical distance of world trade fell from 3.5 in the early 2010s to 3.1 in 2023. This figure indicates that trade relations are concentrating in economically and politically closer blocs. Meanwhile, the average transport distance of trade activities has remained at approximately 5,200 kilometers, meaning nearshoring has not progressed on the large scale that was expected. Despite a trend of political contraction, the spatial structure has not shrunk.

Within this structural divergence, China’s position in the global manufacturing system is at a turning point. A transition is underway from being a supplier of finished products to a provider of production factors and system capabilities. Even if U.S. tariff measures reduce the scale of direct U.S.-China trade, Chinese manufacturing continues to participate in the production processes of other economic blocs through exports of intermediate goods and equipment.

The Definition and Significance of Physical AI

Physical AI refers to technology that directly embeds artificial intelligence into physical systems like robots, drones, and smart production lines, enabling perception, decision-making, and execution to be completed in dynamic real-world environments. Compared to traditional automation systems reliant on predefined rules, it leverages multimodal large models (covering vision, language, and action) to reduce dependence on human instruction and equip systems with the abilities for continuous learning and situational adaptation.

Under these technological conditions, the methods of factory operation are also transforming. The production model that relied heavily on a large workforce is shifting toward an organizational form centered on “robots + data.” Through technical approaches like physical information embedding, smart systems can now make judgments and adjustments that account for physical constraints such as gravity, friction, and material changes, moving beyond mere image recognition. In scenarios demanding flexibility, such as complex assembly, operational stability and efficiency are improved, with performance on some tasks approaching that of human labor.

China’s advantage in this field does not stem from a single technological breakthrough. Instead, it manifests as a systemic capability supported by its industrial structure and production organization forms—backed by the world’s most complete industrial classification, the richest application scenarios, and the fastest industrial coordination capacity.

Different Development Paths for the U.S. and China

The United States and China are adopting different technological strategies in physical AI. The U.S. follows a path of “model-driven innovation,” emphasizing large-scale models, world models, closed-source systems, and high-end computing resources. The integration of software and hardware with a closed-system approach tends to ensure system reliability and intellectual property protection, but requires high R&D investment and lengthy validation cycles.

China practices “scene-driven innovation,” prioritizing manufacturing, deployment, cost reduction, and rapid dissemination. It focuses on more concrete production scenarios, relying on open platforms and industrial coordination to pursue continuous adjustment and optimization in practical applications. A comparison of leading companies shows that Tesla aims to build advantages through “vertical integration + computing power supremacy + unified infrastructure,” while Xpeng Motors is developing along the direction of “algorithm optimization + multi-terminal collaboration + scene breakthroughs.”

In 2024, China recorded approximately 295,000 new installations of industrial robots (54% of the global total), far exceeding the approximately 34,000 units in the United States. China currently operates over 2 million industrial robots, providing a rich practical foundation for the real-world deployment and application verification of physical AI systems.

Transformation at Three Levels

The automation capabilities that physical AI forms within industrial scenarios are driving advancements in manufacturing automation and organizational management reform across three levels.

First, the change in automation logic. Traditional automation relied on fixed rules, setting programs in advance for “how to do things” and executing according to fixed flows. Physical AI enables systems to acquire cross-scene transfer and generalization capabilities, advancing automation from “rule control” to “embodied intelligence.”

Second, the change in organizational management logic. With the deployment of physical AI systems on the production floor, decision-making authority disperses to the site. Frontline personnel can make quicker adjustments based on real-time data, and organizational structures move toward flattening.

Third, the improvement of ESG and sustainability. Physical AI systems can more precisely control energy and material consumption, reducing defect rates and production accidents. This supports the improvement of manufacturing companies’ ESG performance while contributing to reduced compliance costs and brand risks.

These three levels of change interact synergistically, pushing manufacturing enterprises from “equipment automation” toward “overall system intelligence.”

Three-Stage Evolution of the

Manufacturing Paradigm

Over the next 30 years, the global manufacturing paradigm is predicted to evolve through three stages.

The first stage, from 2025 to 2035, will see the popularization of a hybrid production model of “robots + humans.” Manufacturing powerhouses like China will take the lead, significantly improving production line efficiency and quality stability, and replacing human labor with machines in some dangerous or high-intensity occupations.

The second stage, from 2036 to 2045, will involve the maturation of swarm intelligence technology, enabling automation systems to acquire the ability to organize, schedule, and maintain other robots. An autonomous production model where “robots organize robots” is expected to take shape.

The third stage, from 2046 onward, will occur against the backdrop of the essentially established global physical AI infrastructure. The “adaptive manufacturing” paradigm will gradually become mainstream. A manufacturing system will be established that relies on unified equipment operation standards, real-time industrial operating systems, and smart grids, capable of autonomous adjustments according to circumstances.

After 2045, the site selection for global manufacturing will no longer hinge on labor costs as the core rationale. An era will arrive where a region’s competitiveness depends on its ability to provide the lowest combined “smart computing cost + energy cost + logistics cost.” It is seen that if China can maintain its advantages in the robot industrial chain, smart production lines, and industrial data, and promote the world’s first practical application of physical AI technology, it could transform from a cost center to a profit center within the global manufacturing value chain.

Editorial Opinion

In the short term, China’s physical AI strategy will have the effect of deepening relationships with emerging manufacturing bases like ASEAN and Mexico. Participation through exports of intermediate goods and equipment, in an indirect form, helps China’s manufacturing influence bypass tariff barriers and serves as a buffer against geopolitical risks. In the short term, this supply-dependent structure is likely to strengthen.

In the long term, as the competitive axis of manufacturing shifts from labor costs to systemic capability, the decisive factor will be whether one can control the trinity of software, hardware, and data. Whether China can maintain a scene-driven, open technological deployment will be a critical variable in its competition with the U.S.’s closed model.

An untested point is the impact of physical AI’s proliferation on employment structures. Will labor replacement by robots be concentrated in specific occupations, or will it spread to the broader labor market? Furthermore, how will organizational flattening redefine the role of middle management? How human capital is redesigned alongside technological adoption will determine the sustainability of the manufacturing industry.

References

• ” 物理AI重塑中国制造的全球竞争力：从世界工厂、工厂的工厂到工业基础模型的跃迁 ”, by 清华管理评论© — 虎嗅网, 2026-08-13T22:09:22.000Z (ARR)
• Source URL: https://www.huxiu.com/article/4883014.html?f=rss