Dark, Cloudy Ghost Kitchens
Before we dive into this week’s post: It is said that “when the guns roar the muses are silent.” It is hard to write about business models and food when you know how much suffering people are experiencing in another part of the world, currently under war. I have no comparative advantage in writing about the war (even though I probably know more about combat than 99.9% of the people waxing poetic about it on Twitter).
The last thing I wanted to do was to write about the supply chain implications of this war, knowing that this is not what we should care about at this point, and may even be viewed as a distraction from the real suffering.
So I chose this as my small corner of escapism.
If you don’t follow this industry, you may be wondering: What are cloud kitchens? Are they different from ghost kitchens or dark kitchens? And why are supermarket chains interested in this new line of business?
These are all good questions, and I am more than happy to answer them!
Let’s start with the first one: What is a cloud kitchen?
“A cloud kitchen—also referred to as a “ghost kitchen” or “virtual kitchen”—is a commercial kitchen space that provides food businesses the facilities and services needed to prepare menu items for delivery and takeout.”
I am a huge fan of cloud kitchens, or at least their concept.
If you think about it, restaurants are a bundle of services and attributes:
Food production: the kitchen
Sourcing of raw material: the acquisition of ingredients that are often not available to the public.
Food discovery: the waitstaff or the online menu, available via the QR code or the iPad on your table.
Delivery: food shuttling from the kitchen to the table
Real estate rental: “leasing” a seat for anything between 30 minutes to 3 hours.
Social network: the framework which allows for people to meet and create strong ties (those at your table), or weaker ones (those at surrounding tables).
Entertainment: the background music or ambiance that is supposed to enrich your experience, while you do all the rest.
The timeless statement of Jim Barksdale is true here as well: “there’s only two ways to make money in business: One is to bundle; the other is unbundle.”
So like any other bundle, there are clearly positive externalities among these different services. But there are clear opportunities to unbundle them.
For example, online platforms unbundle the need to have food production in the same place as the other services. Restaurants that allow you to order on an iPad, as in many airports these days, unbundle the food discovery service.
But until recently, most of the unbundling activities didn’t change the most important aspect, the fact that food production remained where “real estate” and “social network” services were being offered. While the changes enabled the two services to be done in a different location, the food production remained at its original place.
Here came the idea of cloud kitchens, which started to emerge a few years ago.
What are the benefits?
Reduced real estate cost: Unlike traditional dine-in restaurants, cloud kitchens allow food businesses to create and deliver food products with minimal overhead.
Flexibility by pooling resources: In the current structure, a brick and mortar restaurant must commit to all aspects of the “experience”. You are either a French restaurant or a burger place. You cannot be both (unless you are a really bad, suburban Cheesecake Factory). Cloud kitchens don’t have to commit to a specific type of cuisine, and that’s where the name “cloud” came from. It’s similar to cloud computing. Not having your own servers allows AWS, or Google Cloud to flexibly align supply and demand.
Planning layout for modern food consumption: you can make sure that everything from logistics to delivery is ready and focused on delivery rather than “entertainment,” for example. This is not true in standard dine-in places which are usually based where customers are, and not where it’s most convenient to deliver the ingredients or pick up the food.
And indeed, over the last 3 years, we have seen significant growth in this domain.
Has Covid Changed Anything?
What surprised me was that there was no rapid expansion of the cloud kitchen model during COVID, even though many restaurants suffered from lockdowns or people's reluctance to sit in closed spaces.
And there may be multiple reasons for this:
While the idea of cloud kitchens may sound economical, most of the costs are variable, limiting savings significantly.
In particular, different calculations show that raw material and kitchen labor account for approximately 60% of the costs, and since they are highly linear, no significant savings occur from moving them to a different location. It may lead to reduced real estate costs, but that’s not enough.
Another reason for the limited growth could be system rigidity: dine-in restaurants have already invested in a fully-equipped kitchen, so moving to another location is not all that simple.
On the other hand, restaurants serve as part of the “food discovery” and advertising process. They are part of the brand. With no brick-and-mortar location on a “main street,” everyone is left with very slim margins, and competition on DoorDash, for example, becomes cut-throat.
But let me add one more reason: delivery is very costly. And the farther away, the more expensive it becomes. In order to reduce the delivery cost, food must be transported quickly and in close proximity. For example, according to GridWise, DoorDash drivers make $15.57 per hour, $1.04 per mile, and $7.97 per trip. Given that the average order size and the delivery fee are independent of the delivery cost, transporting food over long distances is actually quite expensive, thus turning building massive kitchens on the outskirts of the city or away from customers into a disadvantage.
There are multiple new business models trying to disrupt the current one by addressing the issue of proximity, but all of them come with their own set of pros and cons.
Proximity as a Service
Reef, a “proximity as a service” business model, started as a parking lot technology company named ParkJockey. They tried converting parts of parking lots into small enclosures surrounded by trailers, offering various services such as cloud kitchen, child care, and package delivery. SoftBank backed it in 2018 and Reef used much of the $1.2 billion it raised to become the largest parking-lot network in the US.
While competitors in the cloud kitchen market tend to rely on large shared kitchens for numerous restaurant brands or types of food, Reef’s strategy focused on installing trailer-sized kitchens in parking lots. In these trailers, Reef prepares and sells food with its own workers and pays the restaurant brands a percentage of each order.
For example, according to the WSJ, Wendy’s signed a deal with Reef to open and operate up to 700 locations in North America and the U.K.
But running a kitchen means playing with fire…literally:
“Since the summer, local officials in New York City, Houston, Detroit and Chicago have suspended operations at some or all of Reef’s fleets of trailers for violating regulations, totaling more than 25 closures. Many of the suspensions were for kitchens that were operating without permits, while others were for failing to tow the trailers to a central commissary every day, a requirement for food trucks in many cities.”
And the pressure is not only coming from regulators, but from the brands themselves. For example, increasing reports regarding food-safety issues, caused Reef to lose David Chang's Fuku, to Kitchen United.
I don’t think we’ve seen the last of this model, but it’s definitely messier than it seems. The model also doesn't really utilize the idea of “the cloud” in the sense of flexibility and pooling. It has the potential to do so, but it doesn’t.
Don’t Come to Us, We will Come to you
Another solution to the problem I outlined above comes from Wonder:
“Ingredients are prepped and packaged at a large central kitchen before they’re distributed to smaller kitchen hubs, which are accessed throughout the week by Wonder’s vehicles.”
“Wonder also wants customers to receive their food when it’s piping hot. So it finishes off the final meal prep inside of vans that have been outfitted with kitchen appliances, once the driver arrives at a destination. Each Wonder van has one trained chef onboard and is dedicated to only one restaurant.”
Wonder came out of stealth in December, and was working with 17 restaurants and servicing about 17,000 households in New Jersey at the time.
Looking at the unit economics, it’s hard to spot the savings. There are some advantages in prepping the food in a central kitchen, and some savings on delivery (since the route can be pre-planned and delivery can start before the food is ready). This means pooling the cooking and the delivery time, which will reduce the costs of the actual delivery, but again, you need a lot of density for this, and since every kitchen (i.e., the van) is dedicated to a single restaurant, you are gambling with the demand for that specific restaurant.
The founders argue that, like Netflix, customers would enjoy exclusivity on the content they choose, but I don’t really see it.
And judging from the latest issues Netflix is facing, this is not the case, and neither will it be the case for food, where it’s easy to multi-home. I don’t mind getting something from Bobby Flay, but I don't want everything from his kitchen.
To make the unit economics work, you need to have either a high average order value, or a lot of density, and indeed:
“For a Wonder van to break even for the evening, it needs to make about $100 in sales per hour, according to Lore and Hilton.”
However, there are benefits that go beyond costs. Potentially, the food quality can be better than that of a traditional food delivery since it’s being prepared on the way to the customer, but it’s hard to say how much “willingness to pay” exists for this. Zume Pizza tried, but shut down 2 years ago. So…
I am the last person to bet against Marc Lore, but contrary to what the article claims, this is not a winner-takes-all market. Preferences are clearly different among people (cost, variety, speed), multihoming is easy, and network effects become exhausted after a while, as we can see from DoorDash’s flat take-rate model.
Supermarkets: Already There
Until now, all cloud kitchen solutions brought the food to you.
And this is where supermarkets come in:
“Located in the Kroger store at 1035 N. Shepherd Dr. in Houston, the new kitchen center enables customers to place orders online or through a mobile device or an in-person ordering kiosk using Kitchen United’s MIX platform and then pick up their meals on-site or have them delivered. Restaurant staff prepare the orders, with delivery service fees set by third-party providers.”
One of the main advantages of supermarkets is that shoppers can mix orders from multiple restaurants and have their purchases grouped in one receipt, something that is currently not possible in the traditional restaurant business model (or even with Reef and Wonder). Currently, customers can choose from 15 eateries, including local and national restaurant brands such as Bad-Ass Breakfast Burritos and Bad Mutha Clucka (these are their actual names).
But it’s not only Kroger, Walmart, which is partnering with Ghost Kitchen Brands, is also joining in.
At Walmart, groceries account for 56% of revenue, and the chain also captures about a fifth of all consumer food and beverage spending in the United States. Any small shift in spending away from groceries will have a significant impact on the retailer’s bottom line.
It is clear what Walmart gains from this and what competitive advantages this brings.
“... most customers purchasing from these ghost kitchens on-site are ordering at the end of their shopping trip on their way out the door. At the start of their trip, they may be goal-oriented, eager to get their shopping over and done with, but as they leave, they may have more time on their hands, and they may have worked up an appetite as they shopped.”
Since you are already shopping, why not order food on the way home? Or even better, why not take it further? Order when you start your shopping and pick it up freshly made when you are done. As stores have more sensors and cameras, they can figure out where you are and start packing things when you join the cashier (or just pick up products in an Amazon-Go-like store). I know it may sound futuristic (and maybe even a little creepy), but the future is not all that far.
So there are numerous benefits: saving time, reducing real estate costs (compared to traditional brick and mortar restaurants), increased flexibility and variety (once again compared to traditional models), and reduced delivery costs (compared to DoorDash, and the like).
Essentially, the supermarkets are trying to cling to their customers, who are moving to order more online and prepare less food at home, leveraging their assets (real estate and distribution).
There are many other models that have emerged. From WoodSpoon, which allows chefs to use their own kitchen and deliver the meals using DoorDash and UberEats, to many others (albeit this is more a P2P model than a cloud model).
Which model will prevail?
I don’t think only one model will because, as mentioned before, it has largely to do with preferences.
Sometimes I like to buy prepared meals, sometimes I like to prepare them myself.
Sometimes I know what I want and choose it from the menu, sometimes I want the waitstaff to help me decide.
Sometimes I want to pick it up, sometimes I want to have it delivered.
Sometimes I like to eat at home, sometimes in a restaurant.
There are people with whom I enjoy eating, and others with whom I don’t (even if I enjoy speaking with them about other things).
In other words, since preferences are so diverse, I do not see a convergence to one model.
But this is just the beginning.
Thanks for reading Gad’s Newsletter! Subscribe for free to receive new posts and support my work.