Concept visual: Wild Bite Club.
After a year of subsidised meals and drinks, China’s largest delivery platforms are moving money into dark stores and neighbourhood warehouses. Food bought the habit; broader retail must now make it pay.
A cheap lunch disappears in twenty minutes. The delivery habit it buys can last much longer. That was the wager behind China’s food-delivery price war: subsidise a coffee, a burger or a bowl of noodles today, then use the returning customer to sell groceries, flowers, cosmetics and electronics within an hour tomorrow.
For roughly a year, Meituan, Alibaba and JD.com poured money into coupons, free delivery and merchant incentives. Consumers saw astonishingly low prices. Restaurants saw a flood of app orders. The platforms saw something else: a high-frequency route into daily life. Food was not simply the category being fought over. It was the cheapest training product for a much larger retail behaviour.
By September 2026, the giveaway phase was easing. Regulators had attacked irrational competition, company profits had been bruised and the leaders were talking less about coupons and more about dark stores, supermarkets, “lightning warehouses” and order-level economics. The price war did not end because the appetite for speed disappeared. It ended because the bill had to become an asset.
The order that was worth more than the meal
Food delivery is a powerful customer-acquisition engine because hunger repeats. A television may be bought once every several years; lunch is ordered again tomorrow. Cheap drinks are even more useful: standardised, easy to compare and naturally frequent. Heavy subsidies can therefore purchase not only an order, but a place on the customer’s home screen and a new expectation about how quickly a city should respond.
The platforms paid dearly for that repetition. Reuters reported that Meituan swung to a loss during the competition, Alibaba’s profitability declined and JD.com’s profit almost vanished. Merchant economics were distorted too. Luckin Coffee said comparable sales at self-operated stores fell 5.3% in the second quarter of 2026, against 13.8% growth a year earlier, partly because the earlier period had been inflated by platform subsidies. When the coupon is removed, yesterday’s spectacular demand becomes today’s difficult comparison.
Beijing also forced the cost into view. Draft rules issued in June would stop platforms from forcing merchants into subsidy campaigns, shifting subsidy costs onto them, exploiting capital advantages for unfair competition or selling below cost. The intervention matters because a “cheap meal” can be financed by several invisible parties: the platform, the restaurant, the courier and investors willing to tolerate losses. The consumer sees one low price; the system carries four separate liabilities.
From coupon budget to concrete shelves
The second phase changes what the money buys. A coupon disappears with the order. A dense fulfilment node, a trained courier network and a better inventory model can serve thousands of future orders. Meituan is expanding supermarkets, while Alibaba and JD.com are building online-only dark stores and compact neighbourhood warehouses designed to fulfil within an hour.
This is the non-obvious pivot. The delivery war looked like a fight over takeaway market share, but the strategic prize is instant retail: the ability to move many categories through the same urban network. Meals and drinks supplied frequency. Broader retail supplies basket size, higher-margin categories and more occasions across the day.
Official company results show the shift from volume theatre to economic repair. Alibaba reported in August that its quick-commerce business was improving unit economics while maintaining market share. Reuters calculated that Alibaba’s instant-retail revenue rose 45% year on year to 53.3 billion yuan in the June quarter. JD.com said losses in its new-business segment narrowed significantly, driven largely by lower losses at JD Food Delivery; its marketing expenses fell 24.8%, while fulfilment expenses rose 10.4% as it continued investing in delivery and operational infrastructure.
That combination is revealing. Marketing spend falls while fulfilment spend rises. Money is moving away from persuading someone to place an artificially cheap order and toward making the order physically work. The subsidy war bought attention; the warehouse phase must buy reliability.
The market became larger than food
China’s Ministry of Commerce research projected instant retail at 1.2 trillion yuan, about $178 billion, by the end of 2026, with average annual growth of 12.6% through 2030. In the second quarter, Analysys data cited by Reuters put Alibaba’s share at 45.7%, Meituan at 45.3% and JD.com at 7.7%.
Those numbers do not describe a settled market. They describe three companies trying to define what an urgent purchase is. A missing ingredient, a phone charger, a bouquet and a late-night snack become operationally similar when all must reach a customer within sixty minutes. The app no longer waits for hunger. It waits for any small gap between intention and possession.
Food remains essential because it keeps the network warm. Restaurants create orders at predictable peaks, couriers circulate through dense districts and customers learn to trust the timer. Non-food retail can then use the capacity between meal occasions. A warehouse holding household goods may earn more from the network than another discounted coffee, but the coffee helped establish the reflex.
This is why the platforms can justify losses that would look irrational inside a standalone restaurant marketplace. They are not valuing only the gross profit on one meal. They are valuing shopping frequency, cross-category conversion, logistics density and the chance to become the default interface for immediate consumption.
Who paid for the new reflex
The consumer was the obvious winner during the discount phase, but the gain was temporary and unevenly financed. Restaurants could receive more orders while losing control over pricing and customer ownership. Couriers carried the promise of speed. Platforms absorbed subsidies and watched margins collapse. Investors funded a race whose winner remained unclear.
Regulation has now changed the acceptable transfer of those costs. China’s market regulator has repeatedly pushed against the race to the bottom, and the June draft rules explicitly protect merchants from coerced participation. That makes the second phase harder but more honest. Platforms must prove that fast delivery creates enough genuine value to support real prices.
Meituan’s return to profit in the second quarter after three consecutive loss-making quarters suggests how quickly economics can recover when subsidy intensity falls. JD.com likewise reported improving food-delivery losses. But recovery does not automatically validate the warehouse strategy. Cutting promotional spend can repair a quarterly result before a new retail model has earned durable returns.
The habit may be rented, not owned
The strongest counterargument is simple: customers may love the subsidy, not the service. Market shares could move again once discounts disappear. Orders may fall below the density required to make sixty-minute delivery economical, particularly outside the largest cities. Dark stores also introduce inventory risk, rent, picking labour and waste—costs that a restaurant marketplace can leave with merchants.
There is a second constraint. Instant retail can cannibalise the same retailers and restaurants that make the platform useful. If commissions, forced promotions or opaque ranking systems weaken merchants, the network may gain transactions while degrading its supply. Regulation can prevent the harshest transfers, but it can also make the model’s true cost more visible.
The thesis would weaken if reduced subsidies produce sharply lower order frequency, if cross-category purchases remain rare, or if dark-store expansion fails to improve order-level margins. It would strengthen if customers continue using one-hour delivery at normal prices and if fulfilment nodes raise basket size without recreating the losses of the coupon war.
The receipt was a map
China’s delivery battle is often described as irrational competition. At the level of the subsidised meal, it was. But the spending also drew a map of urban demand: who orders, when, how frequently, from which neighbourhood and with what tolerance for waiting. The next contest is to turn that map into profitable infrastructure.
The platforms no longer need to prove that people will accept a cheap drink at the door. They need to prove that the same customer will pay a sustainable price for immediacy across the rest of life. The meal was the invitation. The warehouse is where the bill arrives.