TMTPOST — At the 2026 World Artificial Intelligence Conference, China’s internet nobility adhered to a familiar playbook. Baidu, Alibaba, and Tencent filled their exhibit halls with consumer-facing software, presenting autonomous AI agents, smart consumer hardware prototypes, and cloud service platforms. They competed fiercely for screen time, eager to secure the ultimate user gateway of the artificial intelligence era.A short walk away, JD.com presented a markedly different vision. The country’s premier e-commerce operator displayed head-mounted data collection gear, mock residential spaces, and heavy-duty stock-sorting robots. Rather than promoting conversational consumer assistants, JD showcased tools designed specifically to interact with physical reality.According to Duan Nan, Vice President of JD Group and Deputy Head of the JD Explore Academy, the company plans to build the world's largest physical world operating center. Mr. Duan explained that JD is aligning its foundation model matrix, dubbed JoyAI, around embodied intelligence—the capacity for machines to perceive, reason, navigate, and execute complex physical tasks within three-dimensional space.This strategic divergence highlights JD’s distinct position in China’s technology ecosystem. As its peers compete for lightweight consumer software interfaces, JD is anchoring its AI strategy to its vast, capital-heavy supply chain. It represents an ambitious attempt to turn an expensive operational footprint into a unique structural advantage, even as it exposes the firm to deep financial and organizational vulnerabilities.To understand JD’s current approach to artificial intelligence, one must look past its consumer storefront to its underlying infrastructure. While tagged primarily as a retailer of computers and consumer electronics, the company’s core identity remains its physical delivery network. E-commerce serves merely as the customer touchpoint, while the supply chain functions as the structural foundation of the business model.Shortly before the June 18 shopping festival in 2025, JD founder and Chairman Richard Liu made this operational focus explicit. He noted that JD is not a diversified conglomerate, asserting that every business within the group centers entirely on the supply chain and that he avoids any venture unrelated to it.That focus required immense capital expenditure over two decades. Company disclosures indicate that JD Logistics operates more than 3,600 warehouses spanning over 34 million square meters. This integrated infrastructure allows JD to fulfill 95 percent of its self-operated e-commerce orders within 24 hours, achieving minute-level delivery in nearly 100 cities.This network provided an operational moat during periods of aggressive competition. When rivals like Pinduoduo expanded rapidly through ultra-low-cost, unbranded merchandise, JD retained its high-value consumer base due to strict quality control, guaranteed authentic goods, and reliable fulfillment.JD applied the same asset-heavy model to new verticals, including food delivery. In July 2025, the company launched Qixian Xiaochu, a joint catering platform opening JD’s upstream supply chain to restaurant merchants. The goal was not to profit from consumer meal orders directly, but to generate margin through underlying food distribution. Within a year, JD’s food delivery initiatives reached 240 million cumulative users, capturing over 15 percent of the market while individual kitchen sites averaged over 500 orders daily after three months.Mr. Liu explained the strategic logic behind the move, noting that while the public views the venture as a direct clash with food-delivery incumbent Meituan, the underlying focus remains the fresh food supply chain. He emphasized that front-end meal sales carry slim margins, whereas controlling the upstream supply chain provides sustainable earnings.Yet this physical infrastructure carries heavy financial liabilities. The capital required to build and maintain facilities, alongside a massive operational workforce, severely dampens overall profitability. In 2025, JD recorded a net profit margin of just 1.5 percent. By comparison, Pinduoduo’s asset-light marketplace model generated a net profit margin of 23.01 percent during the same period.Investors have responded coolly to these financial dynamics. As veteran investor Duan Yongping famously observed, high sales volume yields little value if the underlying operations fail to generate meaningful profits. Shares of JD traded in New York have dropped more than 70 percent from their 2021 peak, reflecting Wall Street's preference for scalable software platforms over thin-margin logistics operations.As artificial intelligence shifts from cloud-hosted text generators to autonomous software agents, tech conglomerates are vying to build the primary consumer interface. In January 2026, Alibaba integrated its Qwen AI assistant across its service network, including Taobao, Alipay, and Fliggy, allowing users to order meals, purchase goods, and book travel using natural language commands. By June 2026, Tencent began public beta testing for WeChat AI, enabling third-party mini-programs to run through conversational prompts directly within China’s primary messaging app.Lacking a dominant social network or universal consumer portal, JD opted not to build a standalone consumer AI assistant. Instead, it surrendered the front-end user interface to third parties and opened its network to external platforms.When Tencent opened WeChat AI testing, JD was among the first to integrate its e-commerce, food delivery, and logistics infrastructure into the platform. By mid-July 2026, JD connected its AI capabilities to Tencent’s Yuanbao assistant, allowing users to query JD’s product catalog and complete purchases directly within Yuanbao’s chat interface.This strategy carries clear operational risks. As consumer queries migrate toward unified AI assistants, standalone applications risk becoming invisible utilities. JD gains transaction volume but surrenders direct traffic, user attention, and customer touchpoints, reducing its role to that of a back-end fulfillment engine.In response, JD is pursuing a three-stage AI framework consisting of digital intelligence, attached intelligence, and embodied intelligence. The initial phase uses digital intelligence to optimize existing retail operations and logistics routes. The second phase uses attached intelligence to secure touchpoints across smart home hardware. The final stage deploys embodied intelligence, enabling robots to perceive, navigate, decide, and execute complex physical tasks inside warehouses and real-world spaces.By steering clear of direct model parameter battles and software portal wars, JD aims to become the physical execution layer that carries out tasks ordered by third-party AI agents. As Mr. Duan highlighted, JD’s vast operations across retail, logistics, health, and industrial sectors offer a real-world testing ground for embodied models that pure software companies cannot replicate.The shift toward embodied intelligence introduces a delicate social and management challenge regarding the fate of JD’s frontline workforce.JD has systematically increased automation across its operational chain. Automated sorting equipment handles the vast majority of warehouse throughput, autonomous long-haul trucks have completed over 700,000 test kilometers, and last-mile delivery buggies have logged millions of operational miles. In November 2025, Mr. Liu announced that JD would launch its first fully automated, uncrewed delivery station by April of the following year.As autonomous systems gradually replace traditional couriers and sorters, managing staff reductions presents operational, social, and reputational challenges.To address this, JD introduced internal retraining initiatives. At the APEC CEO Summit China Forum in June 2026, Mr. Liu outlined the Nirvana Plan, a program aimed at transitioning blue-collar logistics staff into technical roles. Under the initiative, JD plans to establish more than 80 regional robotics bases across China to train frontline workers in maintenance, repair, and operational care for commercial robotics.Mr. Liu addressed this transition during an internal talk, stating that while future deliveries will inevitably be handled by machines, the company remains committed to providing new career pathways for its 700,000 frontline employees.This retrain-and-retain policy aligns with JD’s traditional organizational approach of maintaining high operational control over its workforce rather than relying on gig-economy contractors. However, whether commercial demand for robot maintenance can absorb hundreds of thousands of low-skilled workers remains an open question.JD’s AI strategy presents a stark contrast to traditional tech approaches. While software-focused competitors scale operations with minimal overhead, JD is linking its technical execution directly to real-world infrastructure—warehouses, truck routes, automated sorters, and human technicians.This approach creates significant operational leverage if AI development transitions from processing online data to executing physical tasks. If autonomous logistics, home robotics, and automated commercial fulfillment become central to e-commerce, JD’s physical network could serve as a valuable operating backbone.Conversely, if the main financial benefits of AI remain concentrated in software platforms, foundation models, and digital traffic distribution, JD’s heavy asset base will continue to weigh on its operating margins, leaving the company vulnerable to digital disintermediation.JD’s AI transition depends on a fundamental balance: determining whether its expensive physical footprint and large workforce remain an operational bottleneck, or evolve into a necessary foundation for embodied intelligence.更多精彩内容，关注钛媒体微信号（ID：taimeiti），或者下载钛媒体App