AI in restaurants: the real numbers for 2026

AI in restaurants is real but still early: 26% of US restaurant operators say they use AI-related tools, and only 6% of restaurants use AI for customer orders, according to the National Restaurant Association [1].

You'll see much bigger numbers online, and some don't say who was asked. Here's what the credible data shows, which statistics to be wary of, and where AI already helps in a pizzeria or a quick-service restaurant, starting with the phone.

How many restaurants use AI?

Two recent US sources measure two different things:

• Actual use: about one in four. In its State of the Restaurant Industry 2026 report, the National Restaurant Association found that 26% of operators use AI-related tools, and only 6% of restaurants use AI for customer orders [1].
• Openness: much higher. In Toast's 2026 survey of 676 decision-makers at restaurants with 16 or fewer US locations (a blind survey of Toast and non-Toast customers, April 2026, margin of error ±4%), 87% said they feel comfortable using AI, and nearly 9 in 10 said they're experimenting with it [2]. Toast sells restaurant software, but it publishes its method.

Feeling comfortable with AI isn't the same as using it every day, which is why the two numbers are so far apart.

Official data elsewhere tells the same story. In France, 12% of accommodation and food service businesses with 10 or more employees used AI in 2025, up from 2% in 2023, according to Insee, the national statistics office [3].

Are diners ready to order from an AI?

Many are. In TouchBistro's 2025 U.S. Diner Trends Report, based on a Harris Poll survey of 1,500 US diners in October 2024, 41% said they were comfortable with AI phone ordering, 32% uncomfortable and 28% neutral [4].

Asked more broadly, people are even more open. In a 2025 Harris Poll survey of 2,065 US adults for Hostie, which sells AI guest communication tools to restaurants, 89% said they would be open to using an AI agent for tasks related to interacting with a restaurant [5]. Both surveys were paid for by companies that sell restaurant technology, and both measure what people say, not what they do: read them as a trend.

One condition matters: telling callers they're talking to an AI. In the European Union, it's the law: since August 2, 2026, Article 50 of the AI Act requires that people be informed they're interacting with an AI system, at the latest at the first interaction [6]. Elsewhere, saying it upfront is still the simplest way to keep trust. Our guide to what diners want from an AI phone agent covers the surveys and a greeting that works.

Which AI statistics should you be wary of?

Statistics about AI in restaurants spread fast, and some don't hold up once you check who was asked.

• “98% of operators use AI.” It comes from a 2026 trends report on Australia by Lightspeed, a point-of-sale company: “Nearly every operator (98%) is now using AI in some form” [7]. The public page gives no sample size and no method, so there's no way to know what it measures.
• “68.8% of restaurateurs use generative AI.” It comes from the barometer of Food Hotel Tech, a hospitality technology trade show in Paris: 337 restaurateurs surveyed among the show's visitors on April 14 and 15, 2026 [8]. The organizer itself notes that respondents were already aware of technology issues and came to the show looking for solutions [9]. Trade-show visitors aren't a sample of restaurant owners.

Vendor surveys aren't wrong by default: Toast's survey, above, publishes its sample, dates and margin of error. Before you trust any AI statistic, ask three questions: who was asked, how many people, and who paid for the study?

What does AI actually do in a restaurant?

It doesn't cook or deliver. It takes repetitive tasks off your team's plate:

• Writing: menu descriptions, social posts, replies to reviews.
• Translating: your menu in Spanish or French in a few minutes.
• Admin: supplier emails, schedules, first drafts.
• Answering the phone: picking up when the team is busy and taking the order.

The phone is the clearest example, because that's where AI handles a whole task. It's 7 pm on a Friday, the oven is full, the phone rings. An AI voice agent picks up, says it's an AI, takes the order from your menu (items, options, pickup or delivery), reads it back and gives the pickup time you set. It can offer a drink or a dessert, once. The ticket prints in the kitchen, with a text and email alert. When you're closed, it says so and gives your hours. That's what Soleir does: try the live demo.

Few restaurants do this yet: just 6% use AI for customer orders [1]. Meanwhile, every call that rings out during the rush can send an order elsewhere; see what a missed call really costs.

Can AI help with the staffing shortage?

It won't replace a cook or a driver. But it can take one job off a short-staffed team.

Hiring is still hard. In the National Restaurant Association's April 2026 report on hiring and staffing, 47% of operators said they had job openings that were difficult to fill in 2025 [10]. Turnover keeps the pressure on: the industry's average turnover rate was 122% in 2025, according to the same report [10]. In Toast's survey, hiring is the second challenge operators name (22%), right after inflation (27%) [2].

When a position stays open, someone does two jobs: they make food and they answer the phone. That's where AI can help: it takes the calls during the rush while your team stays on the orders in progress. To put numbers on that time, see how many staff hours your phone eats during the rush.

To get started, pick one task that costs you time every week, try a tool for a month, and keep it only if it saves time or brings in orders.