Marc Lore Says AI Will Let Anyone Open a Restaurant. Wonder Has the Cash to Try

I was in the back of the room at the WSJ Future of Everything on Tuesday afternoon, watching Marc Lore lean into a stage mic in a black T-shirt and tell a half-skeptical, half-curious audience that the future of the restaurant industry is a prompt box. Not a kitchen. Not a brand. A prompt. The line the room woke up for — and that TechCrunch wrote up the same afternoon — was that AI will soon let anyone open a restaurant. Lore named the tool: “Wonder Create.” He said it would do the menu engineering, the pricing, the supplier sourcing, the marketing rollout, the unit-economics model. From a prompt.

The contrarian read before we get any further: this is not a software launch yet. It is a roadmap pitch. Lore did not demo Wonder Create on stage. He did not give a ship date. He gave the category a name and put Wonder’s $7B balance sheet behind the claim that the category will exist. Treat the announcement as a directional signal about where Wonder is pointing its capital, not a product you can license next quarter.

What Lore actually said, in order

The framing on stage was sequential. First, the macro pitch — opening a restaurant is hard, the failure rate is brutal, and most of the failure happens upstream of the kitchen, in the menu and pricing and operating-model decisions nobody who can cook is well-positioned to make. Second, the Wonder pitch — the company has now assembled, through a $600 million raise that values it at $7 billion, a stack that includes Grubhub on the demand side, Tastemade on the content side, and Sweetgreen’s former Infinite Kitchen on the automation side. Third, the AI pitch — given that stack, the missing piece is an authoring tool. Wonder Create is, in Lore’s framing, the authoring tool. Prompt in, restaurant concept out. He used the word anyone. He meant it as a thesis, not a literal market.

Mark the framing. The category Lore is naming is agentic restaurant operations — AI agents doing the work that today requires culinary, finance, supply-chain, and marketing teams together. The category is real. Restaurant Business Online’s 2025 retrospective catalogues the build-up — Toast, Square, Olo, and a handful of well-funded private players pushing into different slices of the same agentic stack through 2025. Wonder is not first. Wonder is the first to give the category a consumer-facing name on a WSJ stage with $7B of valuation behind it.

The stack Lore is sitting on is the actual news

The thing that makes this pitch worth taking seriously is not the tool. It is the stack. Wonder owns Grubhub, which gives the company demand-side visibility into roughly a third of US third-party delivery volume. Wonder owns Tastemade, which gives the company a content and culinary-IP engine that no other operator-tech company has. And Wonder owns the Sweetgreen Infinite Kitchen technology — the original captive automation play the Sweetgreen long-form story traces in detail — which gives the company a make-line automation platform it can deploy across multiple brands.

That is a vertically integrated stack of a kind that nobody else in operator-tech has assembled. Toast has the point of sale and the operator software. Square has payments and a thinner operator stack. Olo has digital ordering and a B2B SaaS posture. None of them own demand, content, and automation the way Wonder now does. Interpretation flag — Wonder Create, if it ships close to the way Lore described it, is the user interface on top of that stack. The stack is the moat. The prompt box is the demo.

Compare it to the path the Toast Sous Chef → ToastIQ piece traces from the other direction: Toast spent two years quietly piloting a chat box that nudged operators about pricing and menu mix, then rebranded it as ToastIQ once they had figured out what the surface area needed to be. Wonder is doing the inverse — announcing the surface area first, on a WSJ stage, before the product exists. The Toast approach is the operator-tech playbook. The Wonder approach is the consumer-tech playbook. Lore comes from consumer-tech. The choice tracks.

What operators should actually do

Nothing, on Tuesday’s news. The product is not shipping. There is no SKU, no API, no pilot. PYMNTS is expected to publish a deeper write-up later this week with more granular sourcing on the Wonder Create internals; treat that as forthcoming detail, not ship date.

What operators should do is mark the category. Agentic restaurant operations is the line the next eighteen months of operator-tech roadmaps will be drawn against. Every operator-facing AI tool you evaluate between now and the end of 2026 will sit somewhere on a spectrum between “narrow assistant for a single workflow” and “end-to-end concept author.” Wonder Create stakes the far end. Toast, Square, and Olo will stake intermediate points. The Sweetgreen operator stack we walked through in the Sweet Growth Transformation piece and the Q4 Pass companion read sits closer to the narrow-assistant end — coherent, defensible, slower. The strategic question for any operator deciding where to put 2026 software dollars is which end actually compounds for their category and store count.

I left the conference hall a little after four. Two operators I know were standing by the coffee station laughing at the Wonder Create idea — not derisively, just the laugh you do when something is plausible enough to be unsettling. They are right to laugh and right to be unsettled. Wonder Create is not a product yet. The category it names is. Mark the category. The product will follow somebody’s roadmap, and the operators who have been paying attention to which roadmap will be ready when it does.

— Samuel hosts the Service podcast for TableTransfers. Tips: tips@tabletransfers.com.