Summary in Plain Language
Recently, JD.com announced a major initiative called the “Physical AI Acceleration Plan,” which includes purchasing 3 million robots, 1 million driverless vehicles, and 100,000 drones within five years, collecting 10 million hours of real-world video data in two years, and establishing 80 robot manufacturing bases. The essence of this plan is that JD Cloud, which struggles to compete with Alibaba, Tencent, and Huawei in the general cloud computing market, aims to leverage its unique national logistics network to gain an advantage in the nascent field of physical AI—where robots are trained to understand the real world. This approach avoids the price wars for servers common among leading manufacturers and allows JD to reposition its business from providing computing power to developing the “brains” for robots, potentially increasing its valuation significantly. However, the plan faces significant challenges in terms of data compliance, practical application capabilities, funding, and model technology, making it difficult to implement in the short term, and its long-term success remains uncertain.
Detailed Analysis
Why Switch to Robots Instead of Focusing on Cloud?
Many may wonder why JD would abandon its successful cloud computing business to invest heavily in robots. The reason is that the cloud computing market is highly competitive: Alibaba Cloud has its own chips and operating systems, Huawei Cloud has exclusive government and enterprise clients, and Tencent has a vast consumer base. JD Cloud cannot even make it into the top ten in this market, and it has no clear advantage in terms of price, technology, or connections. Physical AI, on the other hand, is still in its infancy, and the competition is not about who has the most servers or the cheapest computing power. Instead, it’s about who can provide real-world scenarios for robot training—something JD has a distinct advantage in.
The valuation potential of physical AI is also much higher. While JD Cloud was previously valued based on the number of servers it rented out, the new focus on developing robot “brains” means its valuation could increase by several times, especially if it were to go public separately.
JD’s Strong Logistics Foundation
JD’s logistics network is a significant asset that competitors cannot easily replicate. It operates over 3,600 warehouses and 19,000 delivery points, serving more than 10 million corporate clients. Robots used in its smart warehouses have tripled the efficiency of clothing shipments and reduced logistics costs by half, with annual logistics revenue growing by over 20%. The company has also invested heavily in domestic high-performance computing clusters. This practical experience and infrastructure give JD a substantial advantage in the physical AI arena.
Controversial Data Collection Methods
A major issue is the way data is being collected. JD plans to install cameras on domestic service workers to collect video data during cleaning services. While this might seem like a convenient way to gather data, it violates privacy laws. In contrast, companies like Figure AI in the US clearly charge for collecting data from individuals.
Challenges in Home Environments
JD’s experience in logistics does not directly translate to home environments, which are highly irregular. Warehouses are standardized, but homes are not, with constant changes in layout and lighting. JD lacks the necessary experience and infrastructure to develop robots suitable for home use.
Financial Challenges
The plan requires a huge investment of over 100 billion yuan. Even if each robot costs only 10,000 yuan, the hardware alone would cost 30 billion yuan, not to mention ongoing maintenance, data training, and research and development. JD’s quarterly capital expenditure is much lower compared to its competitors, and it faces additional pressures such as high labor costs and declining profits.
Long-Term Possibilities
Although JD’s approach has potential, it faces many challenges. The physical AI market is still developing, and there is no clear leader. If JD avoids rushing into home scenarios and instead focuses on semi-public environments like offices, hotels, and hospitals, it could gradually build a competitive edge in specific verticals.
In summary, JD’s strategy to switch to physical AI has potential, but it faces significant challenges in terms of data compliance, practical application, and funding. While the long-term prospects are uncertain, a more cautious approach, focusing on gradually expanding its expertise in specific areas, could give it a chance to succeed.