China’s delivery boom has long been powered by people—riders on scooters, drivers in vans, and gig workers who absorb the daily churn of demand. But in a warning that lands like a marker on a timeline, the boss of JD.com has suggested that robots will eventually take over a large share of that work. Speaking in the context of China’s accelerating automation push, the company’s leadership said delivery automation could replace roughly 700,000 delivery workers “sooner or later,” a figure that immediately reframes what many observers have treated as a gradual technological transition into something closer to a structural labor shift.
The statement matters not only because of the number. It also matters because it comes from a major logistics operator whose business model depends on speed, reliability, and cost control—exactly the levers that robotics and AI target. When a firm that already runs sophisticated fulfillment networks signals that automation will keep expanding, it implies that the next phase of change may be less about pilots and more about scaling. And when scaled, automation doesn’t just change how goods move; it changes who earns a living from moving them.
What makes the warning particularly sensitive is the nature of the jobs at stake. Delivery work in China—like much of the gig economy elsewhere—is often characterized by flexible arrangements, but also by limited bargaining power, thin margins, and a workforce that can be large enough to feel invisible to policymakers until displacement becomes undeniable. If automation reduces the need for human couriers at scale, the impact won’t be confined to one employer or one city. It will ripple through subcontracting networks, platform-based dispatch systems, and the informal economies that cluster around delivery routes.
At the center of the debate is a question policymakers are increasingly asking: how do you manage productivity gains without triggering social instability? In China, where logistics is both an economic backbone and a political priority, the stakes are amplified. The country’s rapid adoption of technology—across manufacturing, retail, and services—has created a pattern: new systems arrive quickly, and labor markets adjust more slowly than machines. That mismatch is now colliding with the gig economy’s most visible segment: last-mile delivery.
To understand why the JD.com warning resonates, it helps to look at what “delivery automation” actually means in practice. It is not a single invention. It is a stack of technologies that can reduce labor needs at multiple points in the chain. Warehouses increasingly rely on automation for picking and sorting. Route planning and dispatch software optimize how deliveries are assigned. Some operations use autonomous or semi-autonomous vehicles for longer segments. Others deploy robotics for loading, unloading, and internal movement. Even when the final handoff still involves a person, automation can reduce the number of workers needed per package by improving throughput and minimizing idle time.
In other words, the replacement figure is not necessarily a claim that every courier will be replaced by a robot tomorrow. It is more likely a projection of cumulative effects: fewer humans required for each unit of delivery volume as systems become more efficient and as tasks are redistributed. Over time, that can translate into a large reduction in demand for delivery labor, especially in dense urban areas where routes are predictable and delivery density is high.
This is where the “sooner or later” phrasing becomes revealing. It suggests that the transition is not purely technical. It is also operational and regulatory. Companies can build prototypes, but scaling requires integration with existing infrastructure, safety standards, insurance frameworks, and public acceptance. It also requires business decisions about whether automation is cheaper than labor under current wage levels and contract structures. When a CEO frames replacement as inevitable, it implies that those constraints are expected to loosen—either because technology improves, costs fall, or policy becomes more accommodating.
Yet the labor question is not simply about whether robots can do the job. It is about what happens to the people who do it today. Delivery work is often a stepping stone for migrants and others seeking income quickly. It can also be a fallback option for workers displaced from other sectors. If automation reduces opportunities, the immediate effect may be lower earnings for those who remain employed, more intense competition for available shifts, and a shrinking pool of entry-level roles. Over time, it can also change the geography of work—routes that once supported thousands of riders may consolidate into fewer, more automated corridors.
Policymakers, therefore, face a dilemma. On one hand, automation can strengthen economic competitiveness, reduce delivery times, and improve service quality. On the other hand, if the transition is too fast or too concentrated, it can create a social cost that is difficult to measure in quarterly results but significant in human terms. The concern is not abstract. In many countries, automation has historically arrived unevenly, hitting certain occupations harder than others. Delivery work is particularly exposed because it is repetitive, route-based, and increasingly measurable—qualities that make it attractive for automation.
The JD.com warning also highlights a broader trend: logistics is becoming a proving ground for AI-driven labor substitution. Once a system can reliably handle packages in controlled environments—warehouses, depots, and certain segments of last-mile delivery—it becomes easier to justify expansion. That expansion can then influence adjacent sectors: customer service, scheduling, maintenance, and even parts of retail fulfillment. The result is a cascading effect where automation doesn’t just replace one job category; it reshapes the entire ecosystem around delivery.
But there is another layer that deserves attention: the gig economy’s structure can amplify disruption. Gig work often relies on subcontractors and platform intermediaries. When automation reduces demand, the first casualties may not be direct employees of a major brand. Instead, it may be the smaller operators who depend on volume contracts. Those operators may have less capacity to retrain workers or absorb revenue shocks. That means the displacement risk can be distributed downward, affecting workers who are least able to adapt.
This is why the conversation about automation cannot be limited to corporate responsibility alone. Governments and regulators play a crucial role in determining how transitions occur. If policy focuses only on encouraging innovation, the labor consequences may be treated as collateral damage. If policy instead builds transition mechanisms—such as training subsidies, wage insurance, or pathways into other roles—the same technological shift can be managed more smoothly.
However, designing effective transition support is difficult. Retraining is often discussed, but it can fail when it is generic, short-term, or disconnected from actual hiring demand. Workers displaced from delivery may need skills that are not easily transferable, such as operating and maintaining automated systems, working in logistics control centers, or taking roles in quality assurance and safety monitoring. These are not necessarily “high-tech” jobs in the way people imagine, but they do require structured training and credible employment pipelines.
A unique challenge in China’s context is the scale and speed of change. Training programs that might work in slower transitions can become overwhelmed when displacement occurs rapidly. That raises the question of timing: should governments intervene early, before displacement becomes visible? Or should they wait until the labor market shows clear signs of strain? The JD.com warning suggests that waiting may be risky. If automation is already on a trajectory to scale, then policy needs to anticipate rather than react.
There is also the question of how companies should measure success. If the only metric is cost reduction, automation will naturally expand wherever it is profitable. But if companies are evaluated on broader outcomes—such as workforce stability, worker earnings, and the creation of new roles—then incentives may shift. That could mean companies investing in internal mobility, offering training to existing workers, or partnering with vocational institutions to create job pathways tied to real vacancies.
Another angle that often gets overlooked is that automation can create new types of work even as it eliminates old ones. For example, as delivery systems become more automated, there is increased demand for technicians, system supervisors, fleet managers, and safety compliance staff. There may also be roles in data operations—monitoring performance, handling exceptions, and improving routing algorithms. The problem is that these jobs may not be accessible to displaced workers without targeted training and support. So the issue is not whether new jobs exist, but whether displaced workers can realistically access them.
The “700,000” figure also invites scrutiny about how replacement estimates are made. Replacement numbers can be interpreted in different ways: direct substitution of couriers, reduction in demand for delivery labor due to higher throughput, or a combination of both. Without full methodological detail, any number should be treated as a directional signal rather than a precise forecast. Still, even if the figure is approximate, it communicates the magnitude of concern. It suggests that the transition is not marginal; it is potentially large enough to affect labor markets beyond individual firms.
That magnitude is precisely why the warning is likely to trigger policy discussions. In China, where economic planning often emphasizes stability, large-scale labor displacement can become a political priority. Policymakers may respond with measures that range from labor protections to incentives for companies to retain workers during transition periods. They may also consider regulations that require companies deploying automation to contribute to training funds or to demonstrate workforce impact assessments.
Workplace protections are another area where the gig economy’s vulnerabilities come into focus. Delivery work often involves variable hours, performance-based pay, and limited benefits. If automation reduces the number of available routes, workers may face more volatility. Stronger protections could include clearer contracting terms, minimum earnings guarantees, or mechanisms to ensure that workers are not abruptly cut off without support. But protections can also be controversial if they raise costs for employers and slow innovation. The challenge is to design policies that protect workers without freezing technological progress.
There is also a cultural and social dimension. Delivery work is not just an economic activity; it is part of the daily rhythm of cities. When automation changes who performs that work, it can alter perceptions of fairness and opportunity. If the benefits of automation accrue mainly to companies and consumers while workers bear the costs, public sentiment can turn. Conversely, if workers see credible pathways into new roles and if communities observe that automation improves services without eroding livelihoods, acceptance may grow.
The JD.com warning therefore sits at the intersection of technology, economics, and social contract. It is a reminder