Beijing and Shanghai Now Have More Than 200 Shared Kitchens. Are They the Future of Food Delivery?

To find a Dicos near People's Square or Nanjing West Road in Shanghai, you walk ten minutes from the subway exit to No. 555 Plaza, under the North-South Elevated Road. Go down to the basement, past a Saizeriya and a small food court, and there is still no sign of it. Only when a delivery rider who knew the way led our reporter to a row of production kitchens behind a Jike Alliance sign did the Dicos logo finally appear.

This Dicos is about ten square meters. No lit menu board, no counter, no staff out front. The only way to order is to open Meituan or Ele.me and wait for someone to carry the food out to the dining area — a cramped room with fewer than 20 seats, whose design and colors look nothing like a Dicos.

Oh, and by that same method you can order from eight or nine other brands here: Master Kong's Private Kitchen Beef Noodles, Manling Porridge, Running Sushi Jiang, Haorenzhen Claypot Rice, and more.

On Ele.me the store is called "Dicos Jike Nanjing West Road." On Meituan it's "Dicos (Jike Food City)."

More and more of your late deliveries come from kitchens like this one. They are called shared kitchens: one storefront of several hundred square meters, leased out to a handful of merchants at once. Some include a small dine-in area, usually much smaller than the kitchen — Jike Alliance, for instance, sits somewhere between a pure production kitchen and a food court with generous seating. Ele.me's own Future Restaurant doesn't have a single seat.

The brands that move in include delivery-only labels like Manling Porridge — names that sound vaguely familiar but that you have certainly never seen on a street — and big ones like Wangxiangyuan and Dicos. The one thing they share on the delivery apps is that their names are usually followed by a tag: "Jike," "Panda," "Huangxiaodi." Those are the shared kitchens that give them their space.

The business has drawn attention lately. Travis Kalanick — Uber's founder, the CEO his board pushed out after he had built the company to a $68 billion valuation — has bought into it. Uber China was sold to Didi, but Kalanick's interest in the country never went away. A year after being forced out of Uber, he joined the real estate company City Storage Systems. In early February 2019, CSS's American shared-kitchen subsidiary, CloudKitchens, acquired Jike Alliance. Zhang Yanqi, one of Kalanick's former lieutenants in China, became its legal representative.

That was a jolt. Every company in this business has announced fresh funding recently. Jike Alliance's rival Panda Selected announced a $50 million round on February 22. On February 18, Huangxiaodi closed a Pre-A round; Wangxiangyuan is one of its shareholders.

"Online business has changed streets and retail environments, and created a new way of using restaurants and space," says a report on dining in "megacities" like Shanghai, written by Yeast Studio and IKEA's innovation lab SPACE10.

As more orders arrive online, a small delivery-first format in an expensive first-tier city — a stall rented inside a shared kitchen, say — is one way to cut rent and other costs and protect a margin.

There were businesses like this before shared kitchens. Lihua Fast Food and Domino's are both delivery-driven brands with no dine-in area, or only a tiny one. Delivery platforms tried shared kitchens too, and ran into the regulators: in 2016, CCTV exposed an Ele.me shared kitchen that was cooking finished meals on site, where the photos taken there didn't match what the app showed. The Beijing News reported on a shared kitchen that Baidu Waimai ran itself.

Now shared kitchens that have solved the compliance problem are back, moving out of the dark corners and into daily life. Zheng Yi of Qianshi Ventures, an investor in Huangxiaodi, says 20 percent of Beijing's delivery orders now come out of them. Within a kilometer of the Xuantejiayuan residential complex, near Shilipu station on Beijing's East Fourth Ring Road, there are three: Panda Selected, Grape Canteen and Seagull Canteen.

Beijing and Shanghai already have at least 200 shared kitchens, which add up to thousands of merchants on Ele.me and Meituan Waimai

The number is hard to pin down, because several of these companies are still taking space and expanding.

Huang Xianxing, founder of Huangxiaodi, told QDaily that Shanghai has roughly 100 shared kitchens. By the companies' own figures, the count of outlets has already passed 100: Jike Alliance and Huangxiaodi each run about 50 in Shanghai; Panda Selected has 25 open there, and Shiyunji about 10. And the roughly 500 food courts in older office buildings and shopping centers inside Shanghai's Middle Ring Road keep turning into shared kitchens.

That doesn't count Ele.me's Future Restaurant, which has more than 300 outlets nationwide, more in Shanghai than in any other city.

Each shared kitchen holds at least three of the storefronts you see on Ele.me and Meituan Waimai. Some hold more than 10.

They pick similar sites. IKEA's innovation lab SPACE10 and Yeast Studio wrote a report on eating in Shanghai and cities like it, Food and the Megacity: How Technology and Urbanization Are Changing the Way China Eats. It calls shared kitchens "dark kitchens," and describes where they cluster: more than 3 kilometers from the main thoroughfares and core commercial centers; close to dense housing, where rent is relatively cheap; a "middle zone" that suits delivery.

Compared with traditional restaurants, shared kitchens are more spread out and more alike in where they land, and they don't line up neatly with a city's commercial centers. That is what Gu Qin and Luo Zixin, graduate students at Tongji University's College of Architecture and Urban Planning, found after scraping location data from Dianping and Ele.me. Their conclusion: the third-rate patches of a first-rate shopping district — ordinary side streets, or failed commercial developments with cheap storefronts going spare — are the ideal sites for a shared kitchen.

Take 698 Plaza on Nanjing West Road, or the old Shimao Plaza at People's Square: Shiyunji and Jike Alliance have opened two outlets in each. Both are older malls, off the main roads and away from the big shopping centers, with thin foot traffic; their advantage is the office workers nearby. Then there are 158 Fang and No. 555 Plaza, which sit inside the Huaihai Road and Nanjing West Road shopping districts but aren't close to any subway station. In that limited stock of space, Future Restaurant and Jike Alliance moved in.

Huang Xianxing said Huangxiaodi's rule for picking a site is more than a million delivery orders a month within a 3-kilometer radius. "Add up the online traffic and the offline traffic, and go wherever the number is bigger," Wu Hao, founder of Shiyunji, said of his own logic in an earlier interview. What his company wants is business parks and office towers.

They all promise to handle every step so a merchant can just move in. For now, shared kitchens are mostly sublandlords

To attract small and midsize merchants, shared kitchens like to riff on Jack Ma's line: there is no restaurant business too hard to run.

Jike Alliance, Panda Selected, Shiyunji, Huangxiaodi and the rest mostly call themselves "the WeWork of dining" or a "restaurant aggregation platform."

"The WeWork of dining" gives away a lot about the model. They are all sublandlords: find a space, cut it into small stalls, sublet them to companies or individuals. (Some charge no rent and take a share of sales instead.) Their selling points are much the same — lower upfront cost, because the sublandlord does one standardized fit-out, and help getting a restaurant license. Seen that way, a shared kitchen isn't very different from a food court like Food Republic.

As delivery takes a bigger share of restaurant revenue and the business grows more data-driven, shared kitchens use delivery as the thing that separates them from a food court like Food Republic. "Warehouse and store in one" is a phrase they repeat often. Because their sites are chosen partly as warehouses, and because they pull in customers through the delivery apps, they don't depend on passing foot traffic the way a food court does.

They say they will sign merchants up on the delivery platforms, market them online, and run their delivery operations for them — data analysis, menu design, promotions. On store layout, a few, Shiyunji among them, have proposed ways to make delivery faster: a pickup area separate from the dining room, a conveyor belt for handing off orders. (Not every outlet actually has one.)

As the restaurant industry industrializes and the division of labor gets finer, an opening appears, and nearly every shared kitchen claims it can solve a merchant's supply chain and hiring. Shiyunji and Huangxiaodi go further and say a merchant can move in "with nothing but a suitcase."

Is any of this worth what they say it is?

We interviewed three chain merchants and two independent owners. Upfront costs really are much lower. Beyond rent, the efficiency gains aren't guaranteed, and many of the advertised services haven't arrived.

U Ding Maocai, a sister brand of Haidilao, seems fairly happy. It has two outlets inside Panda Selected locations in Beijing, and four or five inside Jike Alliance in Shanghai. Its brand manager, Liu Huai'an, said outlets open faster because there is no fit-out and no licensing to wait through. She told QDaily that U Ding pays about half the rent it would pay outside a shared kitchen, and half the labor cost.

Simon, founder of the Shanghai milk tea and light-meal brand Seewant, said the same thing about revenue per square meter rising as rent and labor fall. Seewant has eight outlets in Shanghai; one is inside a Panda Selected location in United Plaza, and the rest are mostly street-facing shops.

A staff member at a merchant inside Grape Canteen, a shared kitchen near Shilipu in Beijing, told QDaily that rent there is 10,000 to 20,000 yuan a month, against 70,000 to 80,000 for a spot in a food court. From Grape Canteen, deliveries also reach several nearby residential compounds and the Yuanyang International offices by the Fourth Ring Road.

There are exceptions. GuaKa, a Shanghai light-meal brand, says the upfront cost of its one street-facing shop came to roughly 50 to 60 percent of what a food court or shared-kitchen location would have cost. "A street shop can be a lot cheaper — if you can find one," said co-founder Sun Bo.

GuaKa has still put more than nine outlets inside shared kitchens or food courts. Sun Bo gave two reasons for going into a shared kitchen: the merchant doesn't have to get licensed on its own, and he liked the shopping district it sat in. Getting in cost him an entry fee of more than 40,000 yuan on top of rent — mostly the price of the location and the license.

Licensing is a serious undertaking for a restaurant in a first-tier city. In Beijing, you need a hygiene permit, health certificates for the staff, a food business license and a fire-safety permit, and the shop must be at least 60 square meters. The process usually takes three months, during which the fire, health and food-and-drug authorities each come to inspect once. Malls have already done the preliminary fire and hygiene work, which makes approval easier; a shop outside a mall is harder, and paying an agent found on Taobao to handle the whole thing costs 20,000 yuan.

In Shanghai, a restaurant needs a business license, a food business license and health certificates for the staff, and the address must be at least 20 square meters. In principle Shanghai issues one food business license per address, but registering under a food court's or shared kitchen's address is still allowed. In 2016 Shanghai also became the first place in China to create a dedicated license for online food delivery, with rules on packaging, logistics and delivery times.

Shared kitchens usually take over food courts or restaurants that already hold licenses, and they can get the online-delivery license too. Sun Bo said it took GuaKa two months to license its street shop, and he knows someone who waited half a year, while "getting licensed through a shared kitchen is quick and easy — they have more experience and more contacts."

Beyond location and licensing, Sun Bo said shared kitchens have run customer follow-up calls and prize draws for GuaKa, and done a first round of hiring screens, though the hires were mediocre. He wants them to solve staff housing and to fix delivery. Neither exists.

On February 15, with many delivery riders still not back in Shanghai before the Lantern Festival, deliveries crawled. GuaKa's People's Square outlet lost more than 50 orders to cancellations during that day's lunch rush. The loss ran to more than a thousand yuan, and it still bothers him.

"Efficiency mostly comes down to the merchant. They're still just sublandlords," Sun Bo said.

Here is how shared kitchens explain their efficiency: put several restaurant brands in one place, and a rider can collect several orders in one stop. That only works if the shared kitchen is full and the orders are there.

Shiyunji's outlet at 698 Plaza on Nanjing West Road sits right beside HKRI Taikoo Hui and SMG, and business is bad. A security guard on the ground floor told me that before Shiyunji took over the old food court and renovated it in April 2018, more than a hundred riders came through in a day to collect orders, and Ele.me kept three dispatch stations nearby. "It's far fewer now." Four months of renovation later, this "upgraded" food court has lost more than half its customers. Of its 12 stalls, three have long-term tenants and two are pop-ups — occupancy under half, nowhere near enough for the promised gain of a rider collecting several orders at once.

The big chains are moving in too, using shared kitchens to open small outposts and cover more ground

Beyond small and midsize restaurants, more and more big chains are trying shared kitchens: Wangxiangyuan, Dicos, Zhen Kung Fu, Xiao Nan Guo. For a big chain, the 100,000 to 200,000 yuan it costs to test a format this small is nothing, unlike for an independent or a small chain. And because shared kitchens want the endorsement a big name brings, they offer better terms to big tenants.

A traditional restaurant with several hundred square meters of dining room has very few places it can go. Nobody orders from a Wangxiangyuan 10 kilometers away. Through shared kitchens, these chains can pick sites more freely, open quickly, and reach places where a full-size restaurant would never fit.

Industrialization helps: many dishes at a chain now need only a small space for reheating or simple assembly. That is what makes the small format possible at all.

Xiang Xiaodi, Wangxiangyuan's delivery brand, plans to open more than 1,000 outlets in the next three years. Unlike a Wangxiangyuan restaurant, where people share dishes around a table, Xiang Xiaodi is built for one person eating alone. It is going after the largest piece of the delivery market: weekday lunch, nearly 70 percent of orders. Shared kitchens are one part of that expansion. Yuanwei Group, Wangxiangyuan's parent, has invested in Huangxiaodi. Wangxiangyuan president Zhao Jianguang wrote to QDaily: "Looking at restaurants today, everything from picking a site to licensing to fit-out and opening is a complicated job. It takes a long time and real expertise, and not every entrepreneur can do it alone. So there is bound to be demand for this kind of service."

Ting Hsin and Zhen Kung Fu both declined to answer questions about shared kitchens. Zhen Kung Fu said its shared-kitchen project is still being tested.

Small delivery-first shops are clearly where things are going. But a shared kitchen that only sublets has no moat

For a restaurant industry that cares more about delivery every year, the value of a shared kitchen is obvious. It gives brands large and small a cheap way to reach more people, fast.

But Federico and Roy Lin, the authors of Food and the Megacity: How Technology and Urbanization Are Changing the Way China Eats, argue that shared kitchens have not changed how restaurants work, online or off, or how people eat. They told QDaily that in delivery the real power still sits with the platforms, Ele.me and Meituan — more than with the restaurant brands themselves.

Delivery has already rebuilt the whole chain. The storefront a shopping mall used to lease out was not just somewhere to run a restaurant; to a large degree it determined how many people walked past. That is what commercial real estate is worth.

In delivery, the platforms determine the traffic. Ele.me and Meituan decide which restaurants a user sees on opening the app. Their algorithm weighs distance and quality, and also how much the restaurant pays for advertising and whether it uses the platform's own couriers, which changes the commission it owes.

Renting land, fitting it out, chopping it up and subletting it to delivery restaurants gives you nothing like that hold over what a customer picks. Nobody orders from a restaurant because it happens to sit inside a Panda Selected or a Jike Alliance. What decides the order is the platform ranking, the brand, the taste, the price — or the subsidy.

In the chain that runs from fresh ingredients to delivery platforms to restaurants to shared kitchens, the shared kitchen has the least leverage of all. It cuts a restaurant's operating costs and helps it reach more people, and there is no great barrier to doing that.

Scale won't keep rivals out. No new company is likely to start a delivery platform today, because it could not build a comparable courier network or take users from Ele.me and Meituan. But opening a few more shared kitchens does nothing to stop a competitor from finding another space in the same neighborhood and opening one of its own.

In the end, subletting is a property business. The shared kitchen pays to take the space and rebuild it, and carries the risk if occupancy is thin or rents fall. Beijing and Shanghai have only so many good locations, and near any good location there is usually more than one fading commercial property available to rent out as restaurant space. As money keeps pouring in, the 200-plus shared kitchens already open will multiply, and competition is unavoidable.

If a shared kitchen runs out of cash, the restaurants inside it may not get their deposits back, and may lose their stalls to the building's owner.

Policy risk sits with the shared kitchen too. Beijing and Shanghai have each rewritten the rules for delivery restaurants several times, always tighter. Whether that turns on shared kitchens specifically, nobody can say.

Even the size of the market is unclear. By Meituan's own numbers, in the first half of 2018 the company — which calls itself China's largest delivery platform — handled 2.77 billion delivery orders, an average of 3.79 million a day. Elevators packed with riders at noon in Beijing and Shanghai office towers are not how the country eats. And for now, shared kitchens are almost all in Beijing and Shanghai.

The startups running shared kitchens know how low the barrier is. They are adding services — group catering, managed delivery operations — and some are pushing toward retail: a forward warehouse, or a study of Freshippo's model, which is what Huangxiaodi is doing. Some are investing in restaurant brands and incubating them; Jike Alliance backed Youtou Fenmian, built on Guizhou rice noodles.

Better-armed competitors have already arrived. Freshippo is supporting small restaurant brands and pulling in big ones like Xibei to open delivery counters. Unlike a shared kitchen, which offers nothing but a cheap shop, Freshippo's app can also supply customers. Food Republic is working on delivery too: it has opened a combined storefront for its delivery brands on Ele.me, so a customer can order from several stalls at once the way they would in person, and it has signed a partnership with Ele.me.

Ele.me has not been idle either. It plans to use its own data to run its own restaurants — think of it as Taobao Xinxuan is to Taobao, or a supermarket's house brand. Shanghai Zhuanshan Catering Management Co., Ltd., an Ele.me subsidiary, holds 55 registered brands, among them Yixiayike, Xiaomage Braised Pork Rice, Super Hungry Bento and Tailang Pickled Fish. Three to five of them are combined inside each Future Restaurant.

As former Ele.me CEO Zhang Xuhao first imagined it, house brands would set Ele.me apart from Meituan. "Every Future Restaurant we open adds 150 to 200 SKUs to the platform. Those SKUs are distinctive, they're scarce," he said in an internal video. The other logic is the one retailers use for private labels: house-brand restaurants have fatter margins.

In July 2018, Ele.me announced that Future Restaurant had passed 300 outlets, growing through open franchising. Even with Ele.me tilting traffic its way, it isn't doing well. Through this February, Future Restaurant outlets in Beijing and Shanghai were generally taking only 200 to 300 orders a month, a tenth of what a good restaurant nearby takes, with gross profit too small to cover staff wages. Last November, a franchisee brought his losses to Ele.me's headquarters and made a scene.

Huang Xianxing thinks that while Ele.me's house restaurant brands carry higher margins, they are also harder to manage. "A shared kitchen is an offline business, and a heavy one. You have the product before you have the orders, and it all comes down to operations. It's nothing like doing things online."

But the closest thing to the heart of the problem may be the comparison Y Combinator founder Paul Graham likes to make between startups and restaurants: "Most of the time, the reason startups fail and the reason restaurants fail is the same — their product, the food, isn't good. If a restaurant has good food, its location can be bad, it can be expensive, the service can be terrible, and people will still come. If the food isn't good, it has to do something very, very special to get anyone through the door."

Ele.me's Future Restaurant serves mediocre food, literally, and that is proof that having data does not mean you can run a restaurant. For shared kitchens, which also hope to bring in brands and get into the restaurant business themselves, the problem is exactly the same.

Charts: Feng Xiuxia, Luo Cong, Gu Qin, Luo Zixin

Cover photo: ChrisGoldNY on Visual Hunt / CC BY-NC; The Mercury News; Dong Zhifei