By our reporter |
On July 11, JD.com broke ground on its very first RoboBase industrial park project in Guangzhou—a major step into the embodied AI race.
This is the first physical project under the strategic cooperation agreement between the Guangdong Provincial Government and JD Group, with an investment of around 1 billion yuan. Expected to be fully completed by the end of 2028 and hit full production by 2030, the base is projected to generate an annual output value of 1.75 billion yuan. The park will focus on bringing in companies working on embodied AI, industrial robots, service robots, and their upstream and downstream supply chains.
JD’s bet on embodied AI has now become a core strategic priority. The group has identified robots and AI as a key strategic direction for the next five years. Founder Richard Liu revealed earlier this year that the group is planning to set up over 80 RoboBase robot bases across the country. Meanwhile, JD Logistics plans to purchase 3 million robots, 1 million unmanned vehicles, and 100,000 drones over the next five years, aiming to fully upgrade its supply chain from warehouses to last-mile delivery.
You can also see this commitment in their capital moves. Since 2025, JD has led or strategically invested in a number of companies in the space, including Humanoid, Zhiyuan Robot, Zhongqing Robot, Pacinian, and RoboScience, essentially building a bench of core players from the ground up.
The newly launched JD Property RoboBase project is JD Group’s first robot base globally. Liang Mian, Deputy General Manager of JD Property South China, told reporters that RoboBase will serve as a concrete platform for the group’s robot strategy, connecting local resources and companies. The park will cover everything from R&D and design, to manufacturing, pilot production, corporate headquarters, product showcases, and production support. It’s also a flagship project for the group’s strategic shift from traditional logistics warehousing to high-end manufacturing infrastructure.
But here’s the thing: JD isn’t building robots itself. Instead, it’s stepping in as a platform player. Liang explained that the group will push forward from four angles: supply chains, technology, real-world validation, and capital support. Right now, the embodied AI industry is struggling with three big bottlenecks: a lack of real-world application scenarios, stable sales orders, and sustainable funding. On one hand, JD Property runs a ton of modern infrastructure assets across the country, with solid know-how in park planning, project management, and smart operations. On the other, backed by JD Group, it has the ecosystem advantage of cross-industry collaboration. This could help break through the current bottlenecks facing embodied AI companies.
“On the supply chain side, we’ll integrate the group’s big retail and industrial system resources to help partner companies cut procurement costs and open up sales channels both at home and abroad. On the tech side, we’ll open up the JoyInside large model and our national big data collection center to share algorithms and data capabilities with industry partners. For real-world scenarios, we’ll open up JD MALL, logistics parks, and health pharmacies as long-term testing and validation grounds for robot companies. And for capital, we’ll team up with our own strategic investment funds and JD Finance to cover funding needs,” Liang said.
Right now, JD’s mature e-commerce ecosystem is already a powerful engine for robot companies to break through sales and accelerate commercialization. Liang gave an example: Songyan Power launched its product on JD and sold over 500 units in just two days. This year, the goal is to hit 10 billion yuan in robot sales on the platform. Over the next three years, they aim to help 100 brands each reach over 1 billion yuan in sales. By optimizing the shopping experience in consumer scenarios, they hope to quickly build brand recognition and buying habits among customers.
But breaking through on sales is just the start. Beyond potential orders, robot companies desperately need diverse, open scenarios to collect high-quality data for iterating their models.
Professor Yang Yaodong from Peking University’s AI Institute and Chief Scientist at Lingchu AI recently pointed out at an event that embodied AI hasn’t had its ‘ChatGPT moment’ yet, mainly due to data constraints. Academics estimate that a model needs about 100 million hours of human operation data to make a leap in capability. Take hand operation data, for example: in China, collecting it costs about 300 yuan per hour. So even though companies are pouring in huge sums, the existing data reserves are still far from enough.
To tackle this “data gap,” Liang said they’ll set up a data collection center, a co-creation space, an OPC conversion base, and a pilot platform within the RoboBase project to lower the barrier for companies to get data.
On July 11, JD Property held a signing ceremony with ecosystem partners, including Zhiyuan Robot, Leju Robot, Qingtong Vision, GAC’s Huilun Technology, and other robot companies, as well as component suppliers like Qingtong Vision and Wuba Intelligent.
Cao Ye, COO of Zhiyuan Robot’s product line, told reporters that the two sides have already developed a set of reusable application models within the RoboBase park scenarios, from autonomous patrols and smart tours to material handling. Once these scenarios prove successful, they can be scaled up, and the park itself can become a terminal for product display and sales.
Recently, a joint venture between Zhiyuan and Guangzhou Metro officially set up shop in Guangzhou’s Huangpu District. You can also see Zhiyuan’s G2 robots in use at some Guangzhou Metro stations. Cao Ye believes that the city’s diverse scenarios—rail transit, industrial parks, exhibition venues, and manufacturing plants—provide fertile ground for Zhiyuan to validate its tech and close the business loop.
Local companies are also eager to join this “friends circle.” Zhang Aimin, Chairman of GAC Robot, told reporters that this year, Huilun Technology plans to launch its security patrol robots and dexterous hands on JD for market validation, and explore applications of embodied AI robots in high-frequency scenarios like logistics, commercial spaces, and offline retail.
Zhang noted that the security patrol robot started pilot applications in 2025 and plans to achieve small-batch trial production this year, with 300 to 500 units. Next year, they aim for mass production, with sales exceeding 1,000 units and an estimated output value of 50 million yuan. Overall, they’re targeting 5,000 units sold within three years. In the short term, they’ll focus on the security sector, and after three years, they plan to expand into areas like auto aftermarket and healthcare, such as car 4S shops, car washes, and car sales.
Beyond scenarios, JD is also targeting a more hidden pain point: after-sales service. Liang mentioned that the embodied AI industry is showing similarities to the early days of the new energy vehicle industry—R&D and manufacturing barriers have dropped significantly, but the backend network for customer maintenance and repair is still a weak spot. The group plans to leverage the thousands of service personnel from “JD Home Services” who have door-to-door repair and maintenance capabilities, to help robot companies build a nationwide after-sales service network.
Liang also revealed that beyond Guangzhou, JD Property is also eyeing Shenzhen for a RoboBase location. Site selection is based on four factors: industrial foundation, talent, policies, and JD’s operational capabilities.
Back in 2015, JD and Guangzhou signed their first strategic cooperation deal. At that time, with the internet boom, they focused on cross-border e-commerce, accelerating e-commerce logistics infrastructure, and supply chain development.
This time around, the cooperation has been elevated from the city level to the entire province, and the scope has expanded to include robot manufacturing and smart manufacturing. On July 11, the Guangdong Provincial Government and JD Group signed a comprehensive strategic cooperation agreement. The two sides have outlined nine key areas for collaboration: consumer development, modern logistics, digital economy, smart healthcare, home services, supply chain construction, modern agriculture, government-enterprise cooperation, and regional headquarters development.
