Huawei's moves in the robotics sector have recently accelerated significantly. At the 2025 Huawei Developer Conference, Huawei Cloud launched the CloudRobo Embodied Intelligence Platform, introducing three major models: embodied multimodal generation, embodied planning, and embodied execution, and clearly stating its commitment to "not manufacturing robot bodies, but leaving the brain and platform to partners." Prior to this, the Shenzhen Global Embodied Intelligence Industry Innovation Center commenced operations, signing the first batch of 16 companies, including Leju, ZHAOWEI Machinery, Topstar, Efort, and Hechuan, covering key links such as robot bodies, dexterous hands, servos, and control systems.
From the Pangu Large Model and Ascend computing power to HarmonyOS collaboration, and further to the cloud-side training platform, Huawei is making the "underlying foundation for robotics" increasingly robust.
The foundation lies in the cloud, in Shenzhen, and in Songshan Lake. However, to mass-produce core components and high-end equipment, a sufficiently large hardware factory is required. For this task, Huawei has turned to Tangxia, Dongguan.
7.2 Billion CNY: Huawei's "Asset-Heavy Mother Factory" in the Robotics Era
In June 2023, Dongguan Jimu Machine Co., Ltd. was registered and established, with Huawei Technologies holding a 100% stake. The legal representative and chairman, Li Jianguo, simultaneously serves as a Managing Director of Huawei and President of the Manufacturing Department.
A month later, Jimu secured an industrial land parcel in Keyuancheng, Tianxin Community, Tangxia: with a total area of 599,982.08 square meters, equivalent to approximately 900 mu; zoned for M1 industrial use, with a plot ratio of 3.4 and a total floor area of approximately 2.04 million square meters; the transfer requirements stipulate a total investment of no less than 7.2 billion CNY, with the industry locked into next-generation information technology, new materials, and high-end equipment manufacturing; construction must commence before August 27, 2024, and be completed before August 27, 2027, with phased delivery and phased move-in.
This is no ordinary land acquisition.
In December 2024, Jimu's registered capital increased from 870 million CNY to 3.89 billion CNY; in December 2025, it further increased to 4.6891 billion CNY, jumping two levels in just over a year, all subscribed by Huawei. The market generally interprets Jimu as Huawei's resource integration platform for humanoid robots, with directions covering machine vision, natural language processing, electronic components, and intelligent manufacturing. In January 2024, the No. 1 Energy Building was topped out; since then, the main factory building framework has been advancing, while dormitories and supporting facilities continue to be finalized, with the construction schedule reverse-engineered to meet the 2027 completion milestone. For a technology company renowned for its R&D and software capabilities, the willingness to invest 7.2 billion CNY and build a 2.04 million square meter factory in a key town adjacent to Shenzhen is itself a signal that the robotics strategy is shifting from "building models" to "building products."
Huawei Draws the Dual Axes of R&D—Pilot Production—Mass Production in Dongguan
Songshan Lake is Huawei's explicit thread in Dongguan: focusing on terminal R&D, cloud and system ecosystems, talent communities, and headquarters-level support, leaning towards "from 0 to 1."
Tangxia Jimu is the implicit but heavier thread: a 900-mu large factory, electronic component manufacturing, engineering research, machine vision, and high-end equipment, leaning towards "from 1 to 1 million."
The difficulty of embodied intelligence does not lie in demo videos, but in three things:
First, sensing components must be stable; cameras, optical lenses, and sensors determine whether the robot can see and see accurately;
Second, control and execution must be reliable; joints, servos, and controllers determine whether the robot moves steadily and understands commands;
Third, mass production must reduce costs; yield rates, supply chains, and automated production lines determine whether laboratory prototypes can be transformed into deliverable products for industrial sites.
The electronic components and engineering R&D track where Jimu operates fits perfectly into these three links. Coupled with Tangxia's proximity to Guanlan, Longhua, Shenzhen, its location at Dongguan South Station, its connection to Songshan Lake in the north, and the Shenzhen Embodied Intelligence Center in the south, R&D feedback, prototype verification, and batch supply can all be completed within a single commuting circle.
This is also what distinguishes Huawei's approach from simply "buying bodies and assembling demos": the cloud-side CloudRobo handles training and planning, the Shenzhen Center handles partner integration, Songshan Lake handles system and terminal collaboration, and Tangxia handles the delivery of core components and high-end equipment. Combining these four locations creates a full-link chain for embodied intelligence.
Not Just Adding Another Factory, but Rewriting the Robotics Supply Chain Adjacent to Shenzhen
With Jimu's arrival, the first thing to change is the supporting radius. The robotics industry chain is long and iterates quickly: reducers, servo systems, torque sensors, vision modules, optical lenses, controllers, dexterous hand parts, structural components, PCBs (Printed Circuit Boards), packaging and testing, industrial software, HarmonyOS adaptation... Any company manufacturing robot bodies or core components fears that modifying a prototype and sending samples once could take a week. With the chain leader located in Tangxia, suppliers setting up warehouses, small-batch production lines, and joint laboratories nearby will shorten the closed loop of "Shenzhen provides solutions—Tangxia produces samples—Keyuancheng surroundings deliver batches" to within a few dozen kilometers.
The Longfeng New Area has already written this chain into its planning: covering six towns and 653 square kilometers, positioned as the core zone of the global intelligent manufacturing center; by 2035, the area's GDP will exceed 480 billion CNY, with the proportion of high-tech manufacturing exceeding 45%; the high-end equipment direction clearly focuses on robotics and electronic information manufacturing equipment, cultivating joints and controllers, introducing precision reducers and high-performance sensors, and jointly building robotics innovation centers and testing verification platforms.
Tangxia is designated as the main comprehensive service center of the area, with its main body incorporated into the core area of southern integration with Shenzhen development. The Keyuan-Daping group focuses on semiconductors, high-end equipment, and smart terminals, while the Fenghuanggang group focuses on high-density smart terminal manufacturing. Jimu's 7.2 billion CNY project landing in Tianxin is equivalent to injecting a strong chain leader into this plan: relying not on policies to shout directions, but using orders, factories, and certification standards to pull in the upstream and downstream sectors.
Collaboration in Shenzhen-Adjacent Areas Upgrades from "Supporting Role" to "Community"
In the past, the industrial division of labor between Dongguan and Shenzhen was often simplified as "Shenzhen writes code, Dongguan turns screws." This logic is no longer sufficient in the era of embodied intelligence. The new formula given by the Longfeng plan is "R&D in Shenzhen, manufacturing in Dongguan, and joint expansion of applications." Smart terminals, HarmonyOS intelligent manufacturing, robotics, and semiconductors are all coordinated in this manner; the area's GDP in 2025 has reached 210.676 billion CNY, with 3,217 industrial enterprises above designated size and 1,914 high-tech enterprises. Over 60% of industrial enterprises above designated size are deeply embedded in the Shenzhen chain, and there are over 120 supplier enterprises for Huawei and Apple.
In other words, Tangxia taking on Jimu is not about attracting investment from scratch, but upgrading on the foundation of mature electronic manufacturing: factories that originally produced mobile phone modules and terminal structural components can switch to robot vision modules; factories that originally produced precision hardware and automation can switch to joint structural components and assembly fixtures; teams that originally did software outsourcing can switch to HarmonyOS robot applications.
More crucial is the talent structure. Jimu will drive not only production line workers but also mid-to-high-income "intelligent manufacturing white-collar workers" in fields such as machine vision algorithm engineering, optical design, servo debugging, robot system integration, quality systems, MES, and industrial software. With such people moving in, Tangxia's rental housing, commerce, childcare, commuting, and office buildings will no longer be configured merely as "industrial worker camps," but will be rebuilt as an "engineers' new city." Years ago, Songshan Lake changed the city's energy level by relying on R&D headquarters; Tangxia Jimu may change the industrial texture through hard-tech mass production—one leaning towards a CBD and the other towards an intelligent manufacturing mother factory. Only by combining these two threads can we see Huawei's complete chassis in Dongguan.
Ignore Rumors About Unit Volumes, Focus on Four Hard Indicators
Robotics projects are most easily written about as "mass-producing tens of thousands of units immediately," but industrial articles must remain restrained. To judge whether Tangxia Jimu has entered the realization period, it is recommended to focus only on four items:
First, before the August 2027 completion milestone, which buildings will be delivered first and equipped with electronic component/vision/high-end equipment production lines;
Second, whether Jimu's ecosystem partners set up supporting factories around Keyuancheng, Daping, and Fenghuanggang, especially for servos, sensors, lenses, and structural components;
Third, whether Huawei's CloudRobo, Pangu, and HarmonyOS-related robotics solutions are piloted in industrial scenarios in Tangxia or Dongguan, rather than just being demonstrated at exhibitions;
Fourth, whether the commerce and leasing around Tianxin, Tangxia, and Dongguan South Station shift from "factory support" to a dual customer base of "engineers + technical workers."
These four signals are more accurate than any leaks. Because they correspond respectively to: actual factory delivery, actual supply chain aggregation, actual technology implementation, and actual population upgrading.
The first half of Huawei's story in Dongguan belongs to Songshan Lake—that is the ceiling for R&D, cloud, terminals, and talent; the second half is now belonging to Tangxia—7.2 billion CNY, 900 mu, 2.04 million square meters, sinking the perception, control, components, and high-end equipment of embodied intelligence to the front line of manufacturing adjacent to Shenzhen.
When CloudRobo trains models in the cloud, when the Shenzhen Center collaborates with partners for body integration, and when Songshan Lake polishes the system ecosystem, Tangxia Jimu incorporates all of this into factories, production lines, and supply chains. In the future, when mentioning "Huawei Robotics," one should not only think of press conferences and demos, but also the construction site filled with tower cranes in Tianxin, Tangxia: a place that does not hype concepts, but only turns models into parts, prototypes into production lines, and pushes the Shenzhen-adjacent area of Dongguan into a hard-tech closed loop of "R&D in Shenzhen, mass production in Tangxia."