“Owning the kitchen is the most expensive way to sell a cheap lunch,” one operator put it last week, and the quarter's disclosures explain why. Alibaba's Taobao Flash Purchase has piloted Jiwa Canteen takeaway stores in Shenzhen that it does not run: local merchants cook, the platform supplies the brand, the traffic and the marketing, and rent, ingredients and labour stay with the partner. As of November 10, 2026, the trademark filing of October 14, 2026 points to a planned rollout rather than a one-off test, and the arrangement says something specific about where quick commerce supply is heading: the shelf is owned, the kitchen is rented.
Two routes to the same shelf
The same week, reporting on the coffee price war made the arithmetic explicit. Running super kitchens, front warehouses and food malls earns operating income, but the platform carries rent, inventory, spoilage and labour, so its margin converges on the retail and catering businesses it displaces. A platform that opens its own stores is referee and player at the same traffic entry, which is why Alibaba states that Taobao Convenience Store will not build warehouses, open stores or compete with merchants for profit.
Meituan took the other route. Xiaoshi Supermarket, its large-format stores and its super kitchens are funded on its own balance sheet, which gives it the most controllable supply and the heaviest asset base; management has set a long-term target of a sustainable low single-digit margin for Xiaoshi. JD built a self-operated 7FRESH ecosystem, but its share of food delivery is far smaller, so its own supply lacks the demand to fill capacity. Three platforms, three answers to one question: who pays for the shelf.
What the asset-light route buys
| Platform | How supply is held | Who carries the cost |
|---|---|---|
| Taobao Flash Purchase | Brand licence over partner kitchens | Partner holds rent, food cost and labour |
| Meituan | Own stores, kitchens and warehouses | Platform holds rent, inventory and spoilage |
| JD 7FRESH | Self-operated formats | Platform holds capacity it cannot fully fill |
The asset-light route buys speed. Jiwa Canteen's Dalang store reports more than 9,999 monthly orders, a newly opened Minzhi branch reports thousands, and dishes are priced as order fillers at 0.01 yuan with group buys just above 10 yuan and delivery waived. None of that requires the platform to lease a kitchen, and none of it builds a supply moat: a merchant can accept orders from Taobao Flash Purchase today and from Meituan tomorrow.
It also buys a quality risk the platform cannot inspect daily. Platform-branded canteens exist to close the trust gap left by unbranded small kitchens, since the brand endorses food safety and standardizes pricing. When the kitchen belongs to a licensee, that endorsement is a promise about a process the platform does not schedule.
The licence is the control point
What the platform keeps is the standard: menu structure, price bands, packaging, delivery promise. What it gives up is the ability to time prep. A licensee that runs out of a dish at seven in the evening has not broken a contract; it has optimized its own kitchen, and the stockout reaches the customer as a platform experience problem.
Spoilage is the number the model hides
Fresh food is where the two routes diverge most. A prepared dish has a shelf life measured in hours, so a wrong forecast is paid the same day. When the platform owns the kitchen it books that spoilage; when a licensee owns it, the loss is netted against the merchant's margin, and the platform sees only a missing item on a menu it thought was complete.
That asymmetry explains why cheap prepared food keeps entering quick commerce assortments while unit economics stay thin. Quick commerce revenue in China rose 45% to RMB53.3 billion in the June quarter, with profitability targeted for fiscal 2029. A business planning four years out is not optimizing one dish's margin; it is buying frequency, and a short shelf life is a very efficient way to buy it. A licensee cannot fund that frequency at 0.01 yuan indefinitely.
Where a stockout ceiling comes from
Every store-run supply model has a ceiling, and the ceiling is set before the first order arrives: how many prep stations a kitchen holds, how many batches it can cook in a shift, and how long the holding window lasts before quality drops. Those three numbers cap how many items a store can keep available at peak, and no amount of delivery capacity raises that cap.
Prep capacity, not rider count
Adding riders fixes the last mile; it does nothing for a dish that was never cooked. The operating lever is batch scheduling, where smaller and more frequent batches raise availability and raise labour cost per unit at the same time. That trade-off is why operators converge on a narrow menu with fast turnover instead of a wide one with slow turnover.
The promotion that is really a supply decision
Discounts that push a shopper to add a filler item to reach a threshold lift the basket without adding much kitchen load, because the filler is usually a drink or a side that batches well. Read that way, the 0.01 yuan item is not a pricing tactic; it is a way of smoothing prep load across the evening.
What to ask before joining a branded programme
For a brand or a chain deciding whether to join a platform-branded programme, the useful questions are narrower than the pitch. Who schedules prep, who absorbs spoilage, and what happens to the listing when a store misses a batch? If all three land on the partner while the platform keeps the customer relationship, the arrangement is a marketing channel with a supply obligation attached.
The second set of questions is about evidence. Ask for availability by store and by daypart rather than a monthly average, ask how a delisted item is replaced and how quickly it returns, and ask whether the platform's own stores compete with yours on the same search page. The model is still being written, and the operators asking these questions now will be the ones who can tell a traffic programme from a margin transfer.
FAQ
What does the Jiwa Canteen pilot actually change for a merchant?
It gives a merchant a platform brand and traffic without requiring the platform to lease a kitchen. The merchant keeps rent, ingredients and labour, and gains a standard for menu, pricing and packaging. In exchange, the merchant absorbs spoilage and bears the operational risk of missing a batch.
Why does asset-light supply not build a moat?
Because the supply is not exclusive. A partner kitchen can accept orders from one platform today and from another tomorrow, so the platform owns the brand and the customer relationship but not the capacity. Coverage grows quickly; defensibility does not.
Where does a store's availability ceiling come from?
From prep capacity: the number of stations, the batches a shift can cook, and the holding window before quality drops. Delivery capacity cannot raise that ceiling, which is why availability at peak is a kitchen question before it is a logistics question.
How should spoilage be allocated in a licensing deal?
Explicitly. If the partner absorbs all spoilage while the platform sets the menu breadth and the discount depth, the partner is funding the platform's frequency strategy. The contract should tie menu changes and discount levels to a spoilage tolerance the partner can survive.
