Ghost kitchens — delivery-only restaurants operating from commercial kitchens with no dine-in service — were supposed to be the future of food. Investors were excited. Commercial real estate companies built entire facilities for them. Existing restaurants opened delivery-only brands from their kitchens.

Two years later, most of the early operators have closed or pivoted. But the concept hasn’t died. It’s evolved, and the survivors offer useful lessons about what works and what doesn’t in Australian food delivery.

What went wrong

The first wave of ghost kitchens made several assumptions that turned out to be wrong.

Assumption 1: Low overhead means easy profits. Yes, ghost kitchens have lower rent than traditional restaurants. But delivery platform commissions (25-35 percent per order) eat into margins just as aggressively. A ghost kitchen paying 30 percent commission to Uber Eats has similar cost pressures to a traditional restaurant paying higher rent but selling at full margin.

Assumption 2: You can create brands from nothing. Many ghost kitchen operators launched multiple “virtual brands” — essentially fictional restaurants, each with a different menu, all cooking from the same kitchen. The problem was that customers figured it out. When three different “restaurants” on Uber Eats all have the same address and suspiciously similar packaging, trust evaporates.

Assumption 3: Food quality doesn’t matter as much for delivery. Wrong. Australian consumers turned out to be quite discriminating about delivery food quality. The novelty wore off and expectations rose. If your burger arrives soggy or your salad arrives warm, you don’t order again.

The closure wave

Kitchen United, one of the larger multi-brand ghost kitchen operators, scaled back its Australian operations in 2025. Several standalone ghost kitchens in Sydney and Melbourne shut down. CloudKitchens, backed by former Uber CEO Travis Kalanick, has been quiet about its Australian performance, which usually means the numbers aren’t great.

Industry estimates suggest that about 60 percent of ghost kitchens launched in Australia between 2021 and 2024 have closed.

Who’s surviving and why

The operations still running tend to share some characteristics.

Established restaurant brands. Restaurants that launched delivery-only extensions of their existing brand have fared better than pure ghost kitchens. Gami Chicken in Melbourne, for example, runs delivery-focused operations from some locations that benefit from existing brand recognition and customer loyalty. People order because they already know and trust the food.

Single-concept operators. Ghost kitchens focused on one thing done well outperform multi-brand operations. A ghost kitchen that makes excellent fried chicken and nothing else builds a reputation faster than one offering six different cuisines from the same kitchen.

Direct ordering. The operations that invested in their own ordering systems, reducing reliance on third-party delivery platforms, have better margins. Commission rates on direct orders are zero or minimal, which changes the economics entirely.

Smart technology use. The surviving ghost kitchens use data aggressively. Order patterns, delivery radius analysis, menu optimisation based on what travels well versus what doesn’t. The ones treating it as a technology-enabled food business rather than a restaurant without chairs are the ones still operating.

The technology evolution

This is where things get interesting for the future. Ghost kitchens generate enormous amounts of data about what people order, when they order it, and how far food can travel before quality degrades. Some operators are using this data to make increasingly precise decisions about menus, pricing, and delivery zones.

AI development company Melbourne teams have been working with food delivery operators on demand prediction and kitchen workflow optimisation. The application of machine learning to ghost kitchen operations — predicting order volumes, timing food preparation to match driver availability, dynamically adjusting menu offerings based on ingredient availability — is genuinely useful.

What this means for Australian dining

Ghost kitchens aren’t going away. They’re just finding their actual niche, which is narrower than the hype suggested.

The realistic future looks like this: delivery-only operations as an extension of established food brands, focused on menus specifically designed for delivery, using technology to optimise operations, and gradually building direct customer relationships instead of depending entirely on platforms.

It’s a quieter, less exciting story than “ghost kitchens are replacing restaurants.” But it’s honest. And the food coming out of the surviving operations is, frankly, getting better.

For consumers, the advice is simple: if you order delivery regularly, look for operators with their own ordering platforms. You’ll often get better prices (no platform markup), better service, and you’ll be supporting a business model that might actually be sustainable.