Marc Lore Says AI Will Let Anyone Open a Restaurant. Builders Should Pay Attention.

Marc Lore thinks you will soon open a restaurant the same way you spin up a GitHub repo: with a short description and a lot of automation. His company, Wonder, wants to turn its robotic kitchens into what he calls AI-powered restaurant factories. The pitch is simple. You describe a virtual food brand, maybe Tuscan vegan bowls or Nashville hot chicken, and the system handles the rest. Robots prep the food. Algorithms price the menu. Delivery pipes get the order to the customer. Lore told TechCrunch that this shift is coming fast, and he tends to know when logistics are about to bend under software pressure.

He has done this before. Lore built Diapers.com into a logistics monster that scared Amazon enough to buy it. Then he founded Jet.com and sold that to Walmart for over three billion dollars. His bets usually involve taking a physical business that everyone assumes needs human hands, then replacing the coordination layer with code. Restaurants are the next target. The industry runs on thin margins, high turnover, and chaotic supply chains. AI agents and robotic kitchens want to eat that chaos.

Ghost kitchens already tried the virtual-brand playbook during the pandemic, and many of them collapsed. The problem was not demand. It was operational insanity. Multiple unrelated brands shared one kitchen, one tablet stack, and one exhausted crew. Orders got lost. Food quality cratered. Customers had no idea who was actually cooking. Lore is betting that better software and actual robots can fix that coordination mess. But even if the kitchen hardware works, someone still has to build the digital layer that makes a hundred virtual restaurants feel like real brands instead of spreadsheet rows.

Build the Software, Skip the Grill

This is where builders come in. The romantic vision of the restaurant owner is shifting from a chef in an apron to an operator with a laptop and a fleet of prompts. That operator will need tooling. They will need a dashboard to manage fifty virtual brands at once. They will need dynamic pricing that reacts to inventory and local demand. They will need customer-facing apps that do not look like cookie-cutter templates, because in a world where anyone can launch a brand, trust and design become the actual moat. The grill might be robotic, but the user experience is still handmade.

If you are an indie hacker or a small team, the hardware arms race is not your fight. Wonder and its competitors will spend billions on kitchen robots. Your opening is the software that connects those robots to the rest of the world. Think order-routing APIs that send pasta orders to one kitchen and burger orders to another based on real-time capacity. Think inventory systems that predict when a robotic fryer needs more potatoes before it runs out. Think loyalty programs that follow the customer across twenty different virtual brands without them noticing. These are full-stack problems, and they need to ship fast.

The Backend Is Where It Breaks

A virtual restaurant brand lives or dies on backend speed. At 6:45 PM, a TikTok influencer can drive ten thousand hits to a ghost brand's menu in under a minute. If your database chokes on that spike, dinner service is over before it starts. You need reactive queries that push order updates to kitchen screens instantly. You need durable workflows that handle payment failures, driver cancellations, and inventory refunds without dropping state. And you need vector search if you want an AI agent to suggest the right upsell based on what is actually in the walk-in freezer right now. Most no-code tools crumple under dinner rushes.

The ghost-kitchen crash exposed a dirty secret. Most operators built their stack from duct tape. Generic website builders, brittle third-party ordering APIs, and spreadsheets pretending to be ERPs. The next wave of restaurant factories will not tolerate that fragility. Builders who show up with real backends, real-time orchestration, and actual deployed products will run these operations. The prompt might create the brand, but the database keeps the lights on.

Lore is predicting a future where opening a restaurant costs little more than a good idea and a prompt. That future will create an avalanche of virtual brands, and most will fail. The ones that survive will run on software that can handle real complexity, real traffic, and real money. If you are a builder, the kitchen floor is not your job. The code that runs the floor is. And that is where you should start cooking.