Opening a ghost kitchen: what it actually costs in 2026

What does it cost to open a ghost kitchen as of August 2026?

Between USD 8,000 and 22,000 in a shared space with equipment included, and between USD 35,000 and 90,000 if you fit out your own unit from scratch, priced across Latin American markets as of August 2026.

The low band buys a station inside a hub that already solved extraction, grease traps and health licensing, so you bring smallwares, a point of sale and first-quarter working capital. The high band covers electrical work, a hood with its own duct, a walk-in cooler, municipal permits and the two months of advance rent no equipment vendor mentions, because he doesn't invoice them. That famous USD 3,000-and-a-fryer story describes the first-time operator's outlay, not that of someone planning to last. Imported cooking equipment tracks the exchange rate, so work only with quotes under sixty days old. Every band buys a different level of risk, not simply more square meters.

What each investment band includes, line by line?

From USD 8,000 to 12,000 you take a shared station of 6 to 9 meters with the hub's griddle and fryer, pay a two-month deposit, spend roughly USD 1,500 on smallwares and build brand, food photography and initial packaging.

From USD 12,000 to 22,000 you can sustain two virtual brands, add a blast chiller or small cold room, a point of sale integrated with the apps and around USD 4,000 in working capital. Above USD 35,000 equipment stops being the heavy line: electrical upgrades, extraction ducting, grease trap, civil work and licences absorb between 55% and 70% of the budget. And past USD 60,000 we're talking about your own unit with two cooking lines, capacity for three brands and a dispatch area separated from the dirty zone. In a 25-meter ghost kitchen, equipment accounts for 30% to 45% of total investment, and the rest disappears into invisible line items.

Equipment is never the big line item

An electrical panel that can't carry two fryers and a griddle at once costs you USD 2,800 in a new service feed, a figure that never shows up in the vendor's quote because he sells stainless steel, not wiring. Roof-vented extraction on a second floor runs USD 3,500 to 7,000 depending on building height; a code-compliant grease trap, USD 900 to 2,200 installed. Add permits, drawings and fire-department visits: rarely under USD 1,200, and in some cities eleven weeks of waiting. The classic mistake is signing a lease with the equipment quote in hand, convinced the budget is closed, when barely a THIRD of it is on the table. A 29% commission on gross ticket weighs more than any rent you might negotiate. Run the numbers: a USD 12 average ticket leaves USD 8.52 after the platform takes its cut; subtract food cost at 32% —which climbs in delivery because packaging enters the plate's variable cost— and roughly USD 4.68 remain to cover labor, rent, utilities and your profit.

Commission sinks your margin, not rent

List rates don't bend for small volume: reduced plans appear above 250 weekly orders per brand and some markets demand exclusivity. Anyone who opens thinking «I'll negotiate later» is financing the learning curve at 30%. Channel size explains why platforms don't budge: Uber Eats moved roughly USD 74.6 billion in gross bookings during 2024, per its Form 8-K filed with the SEC, and an operator doing 40 orders a day doesn't bend that leverage. Five variables explain almost all the spread between USD 8,000 and 90,000. First comes the state of the unit: a space that already operated as a kitchen saves you 30% to 45% of the construction work, whereas a former warehouse with no gas line detonates the budget. Second, building height, since every floor of ducting adds USD 600 to 900. Third, the number of virtual brands: going from one to three demands another cooking line and 20% to 25% more working capital.

Five factors that move the price and how much each weighs

Fourth, the municipal health framework, which in strict-inspection cities tacks on USD 1,200 to 3,000 in drawings, extinguishers and modifications. And fifth is the exchange rate, because imported cooking equipment is quoted in dollars and a 10% devaluation shifts your equipment line by nearly the same proportion. Your real market is the three or four kilometers the app assigns you, and that geography matters more than any recipe. A cheap unit in an industrial zone with thin residential density gives you USD 400 rent and demand that won't sustain 40 daily orders; one at USD 900 surrounded by office towers and housing multiplies achievable volume. Before signing, count how many restaurants in your category already appear in the app within that radius and at what rating: if fourteen competitors carry more than 400 reviews each, you land at the bottom of the list and will pay geotargeted ads simply to exist.

Your delivery radius decides the market before your menu does

Contrast that with a channel that keeps growing —Statista projects USD 1.51 trillion in online food delivery for 2026— and the paradox becomes clear: the market expands, your radius doesn't. Negotiate three specific fronts and you trim 15% to 30% off the opening budget. Rent first: ask for three grace months while construction runs, backed by the weeks the health paperwork actually takes, and commit to a twenty-four-month term in exchange. Equipment second: buy the griddle and fryer used with a technical inspection —that saves 40% to 55%— but never compromise on refrigeration, because a faulty cold room ruins inventory every single week. Third, start in a shared hub for six months before touching the USD 35,000 of your own unit, and use that half-year to measure real average ticket, repeat rate and acquisition cost per order. At Masterestaurant we cost these operations with the same hard rule as always, food cost at 32% per plate maximum, and Diego F.

How to negotiate and cut the outlay without crippling the operation?

Parra puts it plainly: a ghost kitchen is a logistics business with a kitchen inside, not a cheap restaurant. Open with USD 22,000 and nothing left to operate, and you close before month five even with a good product.

The sequence is predictable: you launch at 12 orders a day because nobody knows you, the app settles at fourteen or twenty-one days, you pay suppliers cash, and by week six you're buying ingredients on a card at 2.5% monthly. So you cut the ad spend that was bringing new orders, and the decline turns circular. Set aside USD 4,000 to 8,000 untouchable for payroll, supplies and advertising across the first ninety days; that's the line most people delete to buy better equipment. The global dark kitchen sector is projected at USD 171.3 billion by 2033, according to Global Growth Insights, but that number doesn't cover your October payroll.

What happens if you open with no working capital?

Open a separate account this week and park the working capital there before signing anything. Equipment is not the big line item. In a 25-square-metre ghost kitchen, equipment accounts for 30% to 45% of investment;

the rest goes to electrical upgrades, extraction, the grease trap and permits — lines no equipment supplier ever quotes you, because they do not sell them. An electrical panel too small for two fryers and a griddle can cost you 2,800 USD that never appeared in the spreadsheet. Platform commission does not bend for small volume. An operator with one brand and 40 daily orders pays list price; reduced plans appear above 250 weekly orders per brand, and in some markets they demand exclusivity. Anyone opening on the assumption that «we will negotiate later» is financing their learning curve at 30%. The delivery radius defines your real market and almost nobody measures it before signing.

The differences that move the final number

A ghost kitchen serves roughly 3.5 kilometres before delivery time degrades the rating; if that polygon holds no residential density or offices, the cheap location turns out to be the expensive one. The Masterestaurant rule is blunt: count households inside the polygon, not square metres inside the unit. The virtual brand model lets you test without construction. You can launch a wings brand from an existing kitchen for 900 to 2,500 USD — photography, packaging, platform onboarding, first campaign — and measure real demand over eight weeks before committing 50,000 to a unit of your own. That is by far the route that destroys the least capital, and the one fewest people take. Dark kitchen and physical restaurant do not compete; they subsidise each other. A dining room with an idle kitchen between 3 and 6 pm has installed capacity already paid for, and running a virtual delivery brand there reuses the same hood, the same walk-in and part of the same payroll.

The differences that move the final number — in practice

If you already own a restaurant, opening a separate ghost kitchen is usually the worst option on the table. The printed menu does not disappear when delivery arrives. If your operation has a dining room, keep the physical menu alongside the QR menu: the printed card controls the guest experience — service pace, menu narrative, suggestive selling — while the QR adds price updates, accessibility and analytics. A pure ghost kitchen has no printed card, which is precisely why it loses the most profitable upselling lever in the trade.