On July 28, 2026, the Shunde District Embodied Intelligence Development Bureau was officially inaugurated in Foshan, marking the first government department in China dedicated to coordinating the entire industrial chain of embodied intelligence. At the unveiling ceremony, Chen Xinwen, a member of the Standing Committee of the Foshan Municipal Party Committee and Secretary of the Shunde District Committee, summarized the move in eight words: "Leveraging the trend for a leap, unlocking new potential with intelligence."
Just a month earlier, at Foshan's conference on accelerating the construction of a modern industrial system, Shunde had already signaled its intention to establish such a bureau. With the official launch of this department, along with the simultaneous unveiling of a key supporting initiative—the Embodied Intelligence Training and Application Industrial Park in Daliang—a clear message has been sent. In the industrial battle driven by artificial intelligence and robotics, Guangdong's sense of urgency has manifested in "extraordinary" measures within its governance toolkit.
Since 2025, a series of discreet yet significant actions have unfolded. From Shenzhen's Longgang District, which established the nation's first "AI and Robotics Administration," to the intensive creation of new bureaus in Guangzhou's Haizhu District, Zhuhai, and Huizhou, and now Shunde's pioneering bureau for embodied intelligence, an experimental project aimed at aligning "new productive forces" with "new production relations" is gradually taking shape across Guangdong.
Why is Guangdong So Actively Creating New Institutions?
To understand this governance experiment, one must first grasp Guangdong's current position in the AI and robotics industry. In 2025, the core AI industry in Guangdong surpassed a scale of 300 billion yuan, achieving a high growth rate of over 40% and accounting for a quarter of the national total. The province saw 122 large models pass national registration, with more than 430 industry-specific models deployed. The number of AI-focused specialized and "little giant" enterprises reached 147, the highest in the country, and the digital industry's revenue exceeded 8 trillion yuan, also ranking first nationwide.
On the robotics front, Guangdong's performance has been equally impressive. In 2025, revenue from robotics-related products exceeded 140.4 billion yuan, with an average annual growth rate of 8.4% over the past three years. The number of robotics companies surpassed 12,400. Production of industrial robots reached 336,300 units, accounting for 43.5% of the national total and ranking first nationwide for six consecutive years. Service robot production hit 15.18 million units, representing a staggering 81.7% of the national share. From the fact that more than one in every three industrial robots nationwide is "Made in Guangdong" to humanoid robots now moving and working on factory production lines, the province has constructed a complete industrial chain spanning core components, complete machine manufacturing, R&D innovation, and scenario application.
The Pearl River Delta is the heart of Guangdong's robotics industry, hosting 96% of the province's robotics companies and contributing nearly 96.9% of total sales revenue. In Foshan's Shunde district, which recently established its new bureau, over 100 upstream and downstream companies, including Midea Kuka and Jaten, are clustered. The district's robotics industry output approached 40 billion yuan in 2025, and 80% of its large-scale manufacturing enterprises have completed digital transformation.
Why Create New Bureaus When the Industry Foundation Is Already So Strong?
To explain this phenomenon, Du Lan, a member of the Guangdong Provincial Committee of the Chinese People's Political Consultative Conference (CPPCC) and Chief International AI Officer of the World Digital Science Academy, pointed out a long-standing structural contradiction to a Southern Metropolis Daily reporter. This is the mismatch between the cross-border, integrated nature of emerging industries and the traditional "sectoral fragmentation" of management models.
"In the past, the development of AI and robotics was dispersed across multiple departments, including development and reform, science and technology, industry and information technology, and government data management," Du Lan said. She noted that this "nine-dragon water management" pattern could be maintained during periods of steady industrial growth, but the current speed of technological iteration and industrial expansion has surpassed the responsiveness of administrative processes. "This is not a problem of any single department, but a phase-specific mismatch between the cross-border nature of emerging industries and the existing division of labor—a phenomenon common across the country."
Du Lan explained that this is particularly true for the specific field of embodied intelligence, which spans AI algorithms, hardware manufacturing, and scenario application, with technological iterations measured in months. "If an enterprise has to run through several departments just to implement a single application scenario, the opportunity could be lost along the way." In her view, Shunde's decision to directly include "embodied intelligence" in the bureau's name is more insightful than creating a generic AI bureau—the narrower the focus, the more concrete the responsibility. However, the ultimate effectiveness of a new bureau depends on whether it holds real power over budget allocation, personnel, and scenario approval.
This drive to resolve the structural contradiction is the underlying logic behind the wave of new institutions across the region. In May 2025, Shenzhen's Longgang District first unveiled the country's first Artificial Intelligence (Robotics) Administration, positioning it as a direct agency under the district government. This body consolidated functions related to planning, ecosystem development, investment attraction, enterprise services, and scenario promotion. Subsequently, the Zhuhai Artificial Intelligence Development Bureau (the nation's first city-level AI bureau), the Guangzhou Haizhu District Artificial Intelligence Development Bureau (the nation's first independently established and operationally independent district-level AI bureau), and the Huizhou Artificial Intelligence and Robotics Development Bureau were established one after another. Their common mission is to break down departmental silos and lay down dedicated "future tracks" for the high-speed train of AI and robotics.
From 'Managing Industries' to 'Building Ecosystems': How Much Drive Can New Institutions Generate?
Unveiling the plaque is just the start. The true measure of success for this reform lies in how these new institutions actually operate. Examining the operational logic of these "nation's first" agencies reveals a reshaping of the relationship between government and industry. On the day of its unveiling, Shunde's Embodied Intelligence Development Bureau simultaneously launched three major initiatives. The most notable is the "100 Intelligent Robots Plan," which aims to implement 100 real-world application scenarios in Shunde's factories, parks, shopping malls, and public spaces within two years. This action targets the core bottleneck of industrialization: a shortage of scenarios. The most critical need for AI companies transitioning from technology to large-scale deployment is not funding, but "testing grounds." Shunde's approach involves the government opening up scenarios, state-owned capital taking the lead in application, and using large-scale application to drive down costs—a strategy more compelling than simple subsidies.
Shenzhen's Longgang AI and Robotics Administration operates on a similar principle. Its "Robot Voucher" subsidy policy uses a model of "government issues vouchers, enterprises hold them, institutions provide services, and settlement happens post-use," with each voucher worth 100,000 yuan. This directly tackles the dilemma faced by small and medium-sized enterprises in digital transformation, which are often hesitant and lack the know-how or funds. More notably, Longgang pioneered the creation of the world's first Robot 6S Store and AI 6S Store, attracting approximately 100,000 visitors per month, 30% of whom are international guests. This model, integrating display, sales, customization, and data feedback, effectively builds a "last mile" bridge for technology to reach the market.
In Zhuhai, the AI Development Bureau has prioritized "maximally opening up scenarios for AI enterprises" and has used the Special Economic Zone's legislative power to enact data resource development and utilization regulations, providing legally compliant data support for enterprises. The Guangdong-Hong Kong-Macao Modern Agriculture Demonstration Park serves as a prime example. This intelligent glass greenhouse, invested in by the Zhuhai Science and Technology Industry Group, uses its self-developed G-Lab system platform to precisely control the internal environment, achieving tomato yields 10 to 15 times higher than traditional facility agriculture.
While these explorations across Guangdong follow different paths, they converge on the same goal: the government is transforming from an industry "manager" into a "scenario co-creator." Compared to past methods of using land and tax incentives to solve "settlement" issues, these new institutions are focused on solving the "application" problem—ensuring AI is not only manufactured in Guangdong but also utilized first in Guangdong.
How to Align Institutional Reform with Industrial Development? 'Setting Up a Bureau' Is Only the First Step
Behind this series of institutional adjustments in Guangdong lies an undeniable reality: the province's core AI industry has surpassed 300 billion yuan in scale, its industrial robot production accounts for over 40% of the national total, and its service robot production commands more than an 80% share. When an industry reaches this magnitude, the absence of a dedicated coordinating department makes institutional shortcomings frequently apparent at government-enterprise interfaces.
Furthermore, from a deeper perspective, the logic of competition among regions and industries is shifting. As the marginal effectiveness of traditional policy tools like land and tax incentives diminishes, the ability to offer lower institutional transaction costs—faster approval responses, smoother scenario connections, and a more secure environment for data flow—will determine who gains the upper hand in the next wave of industrial layout.
From this viewpoint, establishing specialized institutions is essentially a signal to the outside world that the local government is willing to break conventions for this industry. The growth of QSI (Guangzhou) Co., Ltd. provides a fitting example. The company went from breaking ground to production in just 18 months, with a monthly production capacity exceeding 60,000 wafers, attracting nearly 150 upstream and downstream enterprises to cluster in Guangzhou. In the future, sustaining such speed will rely not on single policies but on a service process reorganized around the industry. When approval, service, and regulatory functions scattered across various departments are consolidated into a single, dedicated body, the friction of industrial operations decreases.
Of course, "setting up a bureau" is only the first step. Hai Yubao, Director of the Shunde District Embodied Intelligence Development Bureau, frankly admitted that from hanging the "nation's first" plaque to seeing 100 real-world scenarios deliver tangible results in two years, there are countless specific problems that need to be solved. The answers to these problems lie not in the moment of the unveiling, but within the workshops, industrial parks, and shopping malls waiting to be transformed by real-world applications. This, perhaps, holds more weight than the title of "nation's first." As Director Hai noted, the true measure of this reform's success will not be the novelty of the plaque or the number of projects signed, but the number of robots that have truly stepped off the exhibition floor and onto the production line two or three years from now, and the number of traditional enterprises that have gained new competitiveness through AI.
Comments