Dark kitchen in France 2026: model, profitability and accounting

The dark kitchen — a kitchen dedicated solely to delivery, with no dining room — exploded with home delivery. The model promises lighter costs: no dining room, no premium location, no service staff. But it shifts the risk onto two formidable variables: platform dependence and real margin per brand. Here is how to understand and steer a dark kitchen in France in 2026.

A dark kitchen (or cloud kitchen, ghost kitchen) produces exclusively for takeaway and delivery. It can run one brand or several virtual brands from the same production site. The model saves the rent of a visible commercial location, service staff and dining-room fit-out. But it loses or absorbs an almost total dependence on platforms (25–35% commission per sale), no direct customers (no customer database, high in-app acquisition cost) and a visibility equal to its app ranking. A dark kitchen is therefore not a "cheaper" restaurant but a different model, where margin is won on volume and food cost, not on the dining experience.

The profitability equation is tense. For a €100 sale: €25–35 go to platform commission, €25–32 to food cost, plus packaging (significant in delivery), production labour, rent (lower) and overheads. Margin only holds if food cost is tight, volume is sufficient to absorb fixed costs, and customer acquisition cost is controlled. That is why steering food cost and rigorous platform accounting are vital — far more than in a dine-in restaurant where direct margin cushions the gaps.

The trap for multi-brand dark kitchens is to reason on the overall result. Yet within the same kitchen, one virtual brand can be highly profitable and another loss-making — the average hides everything. Best practice is per-brand cost accounting: allocate to each virtual brand its revenue, food cost, packaging, commissions and platform advertising. You then get a margin per brand, allowing you to decide which brands to grow, which to renegotiate and which to drop. Without it, the operator funds losing brands with winning ones, unknowingly.

Beyond the single multi-brand operator, a second model is growing: the hub, where one operator provides production boxes to several independent restaurateurs in one equipped building. This shifts the accounting question to re-invoicing between the site operator and the kitchen tenants. Several flows must be framed from the outset: the space provision (rent or fee, subject to VAT or not depending on the contract — bare lease, fully-equipped service, occupancy agreement); utilities and charges (electricity, gas, water, refrigeration, waste) re-invoiced at cost or as a flat fee, with VAT to handle correctly; shared services (cleaning, hood maintenance, equipment upkeep, waste management) as VAT-able services; and equipment made available. The contract's qualification (bare lease vs global service) determines the applicable VAT regime and the posting account. A mis-qualified re-invoicing creates VAT gaps and distorts each operator's real margin — a point we frame at signing, not after the first audit.

In a dark kitchen, packaging is no detail: it is a cost item in its own right, often underestimated, which can represent 3 to 6% of revenue. Unlike a dine-in restaurant with reusable crockery, every delivered order consumes single-use containers, cutlery, bags, sometimes cooling elements. Two constraints add up: environmental regulation (the AGEC law, single-use plastic reduction) requires compliant, sometimes pricier containers; and packaging quality drives the delivery experience — hence the rating, the app ranking, and future volume. Steering a dark kitchen's margin must therefore include the packaging cost per order in the profitability calculation, alongside food cost and commission. Forgetting it overstates the margin of every delivered sale.

The dark kitchen follows delivery-restaurant rules: VAT 10% on food, 20% on alcohol; platform commissions as an expense with VAT often reverse-charged; collective agreement generally HCR IDCC 1979 or Fast-Food IDCC 1501 depending on the dominant activity, with the related payroll specifics (see our HCR payroll article); and a mandatory NF525 certified till where a cash system records payments. Sharing one site between several operators (the "hub" model) adds re-invoicing questions (rent, utilities, equipment) and inter-operator VAT, to frame from the outset.

Steering a delivery kitchen means tracking a handful of indicators every week, brand by brand:

• Revenue per brand and its trend (app ranking moves fast).
• Actual food cost per brand, from spec sheets and inventory.
• Average commission rate actually borne (it varies with offers and in-app advertising).
• Packaging cost per order, built into the margin.
• Acquisition cost (platform advertising relative to the orders generated).
• Average rating and prep time: they drive the ranking, hence future volume.
• Average ticket per brand, a direct profitability lever.

This consolidates into a per-brand dashboard, consistent with food cost and restaurant profitability. Without this close monitoring, a food-cost drift or a commission rise goes unnoticed until year-end.

One operator ran three virtual brands from a single kitchen, with an overall result barely at break-even. Setting up per-brand cost accounting revealed that two brands were clearly profitable and a third — launched to "fill the kitchen" — was structurally loss-making (high food cost, low volume, heavy platform advertising). Closing the loss-making brand swung the whole operation into profit, without touching the two healthy brands. The analytical data, and it alone, made the decision possible.

A dark kitchen's structural fragility is its dependence on one or two aggregators. Several levers reduce it, without illusion (platforms remain the volume channel):

• A direct ordering channel: own site or app, click-and-collect, QR code, which avoids commission and finally builds a customer database.
• An own loyalty programme, impossible on the platforms, to reactivate customers without paying for a second acquisition.
• Diversifying aggregators so as not to depend on a single algorithm.
• A strong house brand: a memorable name customers will look up directly, rather than an interchangeable brand lost in the app.

Every order captured directly saves 25 to 35% of commission, which drops to margin. It is the most profitable investment of a mature dark kitchen, and it is quantified in the project's business plan.

Unlike a food truck, a dark kitchen fits the micro-business poorly: the upfront investment (equipped premises, production equipment), the VAT recoverable on purchases and equipment, and the targeted volume almost always point to a company under corporate tax (SASU/SAS or EURL/SARL). The real regime allows deduction of actual costs — rent, energy, packaging, platform commissions — and VAT recovery, which the micro exemption forbids. For a multi-brand or multi-operator project, the company is also the right framework for inter-brand invoicing and hub re-invoicing. The precise choice (SASU vs EURL, remuneration, dividends) depends on the owner's situation and is set at launch. As soon as the kitchen runs several brands or hosts other operators, the company structure also clarifies who owns the equipment, who signs the platform contracts and how results are shared — questions a micro-business simply cannot answer.

A dark kitchen is not a discount restaurant: it is a platform-dependent model where margin turns on food cost, volume and acquisition cost. The indispensable steering tool is per-brand cost accounting: without it, losing brands eat the winners. VAT, HCR payroll and the NF525 till apply as for any delivery catering. To structure your kitchen's steering, see our restaurant accounting support, our outsourced CFO service and our complete 2026 restaurant accounting guide.

Updated 3 June 2026. This article sets out the general principles for dark kitchens; the collective agreement and VAT treatment depend on your activity and contracts. Sources: BOFiP, INSEE, Légifrance.