Hospitality Technology

Shrinking Margins and AI Fatigue Put Restaurant Tech Vendors on Notice

Full-service margins have fallen from 4% to 2.8%, and 40% of operators say tech hasn't cut costs — forcing vendors at FSTEC to sell outcomes, not AI.

Restaurant tech confronts a tough market
Restaurant tech confronts a tough market — AI-generated

Median pretax income for full-service restaurants has fallen from 4% in 2019 to 2.8% this year, with limited-service operators dropping from roughly 6% to 4%, according to National Restaurant Association data presented at last week's FSTEC conference. For the technology vendors who make up the other half of the event's audience, that margin compression cuts twice: their customers are cutting costs, and 40% of them say tech isn't delivering the savings it promised.

"The restaurant industry has a profitability crisis," said Rich Shank, VP of innovation for Technomic.

Tech isn't moving the needle

Association survey data shows restaurants are seeing some bottom-line benefit from technology, but less than expected. About 29% of restaurants said technology reduced corporate overhead, 27% said it reduced labor costs, and 20.4% said it reduced upfront costs. But more than 28% reported no significant impact on operating costs, and another 12% said technology had actually increased their costs. Even Chad Moutray, the association's chief economist, acknowledged that tech's impact on costs is "not as high as I would have expected."

The implication for vendors is blunt: they need to sharpen their ROI stories and demonstrate direct bottom-line impact, because operators in a 2.8% margin business cannot absorb tools that don't pay for themselves.

Consumers are rejecting AI ordering

Just 35% of Americans believe AI will improve their lives, according to a new Gallup poll cited by author and journalist Stephen Witt in a conference presentation. He tied that sentiment to the K-shaped economy, in which wealthy households benefit from an AI-fueled stock market while everyone else gets squeezed. "This creates a lot of political resentment of AI," he said.

The resistance shows up clearly in restaurant data. According to Technomic, just 23% of consumers find ordering from an AI drive-thru bot appealing, making it the least popular restaurant ordering method. During a FSTEC panel, Chakri Somisetti, VP of IT at Burger King, said the chain is changing its AI drive-thru strategy after finding that some customers will drive away rather than order from a robot.

That aligns with a broader swing back toward human-centric hospitality as the pandemic recedes. "There's been this pretty strong re-emergence of the importance of experience, social occasions, and service," Shank said. "How can you create that hospitality with a smile when you are building these technologies?"

Vendors retreat from the AI label

AI dominated the FSTEC vendor floor for the third straight year, yet restaurants are tired of hearing about it. AI "is a dirty word right now," said Joe Yetter, president of PAR Technology's restaurant division. PAR, one of many companies reshaping its business around AI, now sells outcomes rather than the technology itself. "There's a number of avenues where we've invested in AI, but AI is not the goal in and of itself," Yetter said.

Operators are taking the same approach to procurement. Brinker International CTO Chris Caldwell said the company starts with the pain points it wants to address, then evaluates which technologies can solve them. "Once you really narrow down the use cases that you're going after, it makes it a little bit easier," he said.

Notably, one of the standout technologies at the show contained no AI at all: the Wingstop Smart Kitchen, supplied by the Perfect Company, a gamified kitchen display system that tells employees exactly what to do and when, and which Wingstop says has been working well.

Vibe coding shifts the build-versus-buy equation

AI is also giving operators an exit from vendors entirely. Using AI coding tools, restaurants can now build their own technology at a fraction of the cost of buying it. A vendor-hosted FSTEC workshop demonstrated the approach: presenters from Revmo AI used Claude to build an AI "skill" that takes a bar's inventory from a photo of bottles on a shelf.

"The build versus buy equation has started to shift very much in the favor of building," Witt said. "That's putting pressure on vendors, because customers don't want these bells and whistles."

He cautioned that homegrown apps carry scaling risks across large organizations — "it could be easy to get trapped in a bad IT project, which is presumably the thing you're trying to avoid" — and argued the best vendors will let restaurants build on top of their platforms by de-siloing operator data.

Despite the gloom, nearly 700 operators attended FSTEC this year, and success stories were plentiful on and off the stage. Chris Padilla, CIO of Dine Brands, set the tone in the opening session: "Everybody in this marketplace has the coolest technology, right? There's no doubt about it. It's more focused on being crystal clear what outcomes that you're trying to achieve. Know those with clarity, and that will guide your decisions." As margins stay tight and skepticism runs high, that outcome-first discipline is likely to define which vendors survive the shakeout.

restaurant-technologyaifstecrestaurant-marginsroi

More from Marcus Bennett

Marcus Bennett

Show full bio

Market editor covering media and advertising at The Pass Brief.

74 articles

Pairings

« Previous article