Black kitchen: the definition nobody gets right (and the myth that will cost you cash)

What exactly is a black kitchen?

Production only, no dining room, no counter, nobody facing the public: that is what the words black kitchen designate, a format assembled to push out whatever lands through the delivery apps or your own ordering channel.

Asia stuck the label on it a decade back, when fast-food chains leased burners in units with no commercial window. English and Spanish both inherited the name loosely, and behind that looseness slipped the jumble of tags traders now pass around as equivalents: dark, ghost, cloud, virtual restaurant. Equivalent they seldom are, and there sits the trap, because every variant builds you a different fixed-cost structure. Strictly read, the phrase names a property and a function, cooking without receiving guests, and never the brand or the lease behind it. Whose name sits on the deed, not which noun you use, governs your costing. Diego F.

Black kitchen vs dark kitchen: the difference that actually changes your costing

Parra checks that clause before anyone signs inside a shared facility, because ownership of the space dictates whether rent stays flat or rides your volume, whether a cleaning crew gets split with the neighbours down the corridor, and whether the building manager can move your commission from one month to the next without notice. The label in common use nearly always points at a module leased from a third party; dark kitchen works as a wide umbrella and swallows even the seated restaurant you converted; English settles the rest under ghost kitchen. Masterestaurant enters these openings through costing, never through format enthusiasm, and that clause — your own lease against a third-party manager — predicts whether the business reaches month twelve. Physical construction, none: a virtual restaurant, or virtual brand, amounts to a BRAND without premises, housed in the kitchen of a restaurant already trading and fed by the idle capacity of its off-peak hours.

What about the virtual restaurant? That's a different animal entirely?

Both formats sell through delivery alone, which is where the confusion never ends, yet the capital each one demands runs in opposite directions. Putting up the module forces you to lease or build, buy dedicated equipment and carry a monthly rent;

launching the virtual brand over burners you already fired up costs, in most cases, a line cook's time at three in the afternoon plus one more account on the aggregator. Owners who mix the two compare the return on a six-figure investment against something that added no rent at all, and that crooked arithmetic inflates every expectation about pure-delivery margin. Strip out commission, packaging and minutes of prep: whatever survives that line, rather than the layout or the logo, tells you if the business holds. Uber Eats closed 2024 with 26.1% of the US delivery market and Grubhub with barely 6.3% (Earnest Analytics 2024); at that concentration, the dominant aggregator's cut in each territory becomes the heaviest fixed cost the model carries, heavier even than the module's rent.

What does it actually cost to run an order through a black kitchen?

Spain splits the board between Glovo at roughly 31% and Just Eat at around 26% (Ken Research 2025), so an operator trading on both pays two separate commission structures against one average ticket and almost never works out the weighted figure.

There sits the question nobody raises before signing for space: what does EACH order cost once commission is gone, not what does a month of rent cost. Pulling the tables out of a restaurant does not produce a delivery kitchen; it produces a mutilated restaurant. The menu has to be redrawn in full so it withstands half an hour in transit without surrendering texture or temperature, and the dish that shines in the dining room, sauce on the side, crisp garnish, plating done at the last second, collapses inside a box; fixing that recipe once the doors are open runs dearer than thinking it through on day one.

The most common mistake: treating a black kitchen as a restaurant with the dining room removed

Food cost does not vanish either just because no guest walks in: the 32% ceiling per dish still governs, with packaging as the aggravating factor, an input plenty of operators never cost, capable of adding 3% to 6% onto each order's direct cost. Read "no dining room" as "no costing discipline" and you have explained most of the closures before the first anniversary. Two separate businesses sit under one roof, and it pays to know which one you got. The upstairs one, real estate, puts up a building with twenty modules and leases them to twenty brands: that is the cloud kitchen. The downstairs one, operations, is your module, your three cooks and your menu. Independents took 61.7% of cloud kitchen market revenue in 2025 (Grand View Research, Cloud Kitchen Market 2025), a sign that small operators leasing space still outweigh the chains building infrastructure of their own.

Cloud kitchen isn't the same as black kitchen, even though English lumps them together

Mexico City went past 1,200 active dark kitchens in 2025, a jump of more than 40% against 2023 (CANIRAC 2025), and most of those new modules live inside somebody else's building. Placing yourself on the right floor completely changes which risks you can control and which ones belong to your landlord. Open one when volume in your area already asks for dedicated space and pulling those tickets off the dining-room line hands back capacity at peak; never to feel out delivery with no demand data. Regional numbers look friendly — Latin America's meal delivery market clears USD 39 billion by 2027, per Statista 2024 — though that figure promises nothing about the ten blocks where your rent falls due. My read is blunt: without three months of your own order history, putting up the module from zero is betting blind against a new fixed cost. Suppose demand fails to show in month four; rent keeps running, the aggregator returns nothing, the equipment is already bought.

When does opening a black kitchen make sense, and when doesn't it?

Prove it first with a virtual brand on the kitchen you have, then decide, ticket and frequency in hand, whether volume pays for a space no customer will ever enter.

A third party holds the building and leases out modules: that arrangement is what people label a black kitchen. Dark kitchen stretches far wider and covers any kitchen with no dining room, your own converted storefront included, while English speakers settle almost everything under ghost kitchen. Markets blur all three, though the contract (whose name is on the space) genuinely reorders your fixed costs. With a virtual restaurant the nature of the thing changes: what you have is a BRAND with no premises, layered onto a working restaurant's kitchen and fed by idle off-peak capacity. Nothing gets built and no module gets rented; a second menu simply runs on burners you already pay for. Cloud kitchen: that's what infrastructure operators call it, since it sounds like technology instead of a warehouse with a hood. Run the costing either way and the label makes no difference to the arithmetic.