Pizza Hut joins Starbucks in rolling back AI systems after poor results

Companies love a good origin story for a technology bet. They almost never love the cleanup story that follows.

For the last three years, every quarterly call from a consumer-facing chain has carried some version of the same promise. Artificial intelligence would cut labor costs, speed up service, and turn legacy operations into something investors could value like a software company. Restaurants and retailers leaned in the most of anyone.

The global restaurant automation market alone ballooned to roughly $28 billion this year, according to the National Restaurant Association. The pitch sold, hard. Every chain wanted its own version. Every analyst wanted to see one.

What none of those earnings calls covered in much detail was the messy gap between an AI demo and an AI deployment. That gap is starting to show up on the income statement, in the courthouse, and in the customer experience that AI was supposed to improve.

Two of the biggest names in American food just learned that lesson in the same week. Pizza Hut is now staring down a lawsuit from one of its largest franchisees, and Starbucks (SBUX), as covered in my TheStreet report, quietly retired the AI inventory tool that was supposed to fix its store-level shortages.

Pizza Hut franchisee accuses parent of $100 million AI failure

The plaintiff is Chaac Pizza Northeast, a Yum! Brands (YUM) franchisee that operates 111 Pizza Hut locations across New York, New Jersey, Maryland, Pennsylvania, and Washington, D.C.

Chaac filed in the Texas Business Court earlier in May, alleging that Pizza Hut's mandatory Dragontail Artificial Intelligence system caused widespread operational failures across its restaurants, according to Fortune.

More Restaurants

• 30 year old restaurant has closed all restaurants
• After bankruptcy, Hooters closes restaurants, fights for survival
• Iconic Las Vegas Strip restaurant closes without warning

The franchisee is seeking $100 million in damages for lost business and lost enterprise value.

The mechanics are almost funny if you do not own the restaurants. Yum! Brands bought Dragontail in 2021 and finished its New York rollout in 2024. The system tied kitchen operations into the DoorDash (DASH) dispatch flow, which on paper meant tighter timing and faster handoffs.

In practice, the lawsuit alleges, DoorDash drivers could see tip amounts, order timing, and other incoming orders, and many of them simply waited 15 minutes inside the restaurant to batch a second order rather than leaving with the first.

Pizzas sat out of the oven. Customers got cold food. The 30-minute delivery window stretched out.

Photo by Edward Chaidez on Getty Images

How a kitchen AI rollout handed DoorDash drivers the upper hand

Chaac used to deliver more than 90% of pizza orders within 30 minutes, the complaint alleges. Now, around half of orders take 45 minutes or longer.

Year-over-year sales growth in Chaac's New York City stores flipped from positive 10.19% to negative 9.78%, the company alleges in the same filing. The system "did the exact opposite" of what it promised, according to the complaint reported by Restaurant Dive.

Related: Starbucks delivers tough update on regional offices, cuts 100s of jobs

When I ran the deployment timeline against the litigation calendar, the chain of events looks almost engineered to backfire. The system was sold as efficiency. What it actually delivered was a free intelligence feed for the gig workers it was supposed to coordinate.

The franchisee, one of Pizza Hut's best operators before 2024, watched its numbers collapse on the exact metrics it used to lead.

The broader picture for Yum! Brands shareholders is rougher. Pizza Hut posted a 4% same-store sales decline last quarter, partially offset by stronger international results, according to the Yum! Brands Q1 2026 earnings release.

Starbucks quietly retired its own AI tool the same week

Then came Starbucks. The week of May 19, the coffee chain told store teams in an internal memo that its Automated Counting inventory tool would be retired, according to Reuters.

The tool, built with Redmond-based NomadGo, had been deployed across more than 11,000 North American stores in September 2025 under CEO Brian Niccol's turnaround push.

NomadGo marketed the system as delivering up to eight times faster results than manual counting, with 99% accuracy, according to its launch announcement. The reality was a tool that struggled to tell oat milk apart from whole milk. A promotional launch video even famously caught it skipping a bottle of peppermint syrup sitting plainly on the shelf.

A few AI deployment dates worth keeping in mind:

• Yum! Brands closed its Dragontail acquisition in September 2021, according to a Yum! Brands statement.
• McDonald's ended its IBM-built drive-thru AI partnership in mid-2024, a joint IBM statement shared.
• Pizza Hut's Dragontail rollout finished in the New York market in 2024, according to the Chaac complaint, as reviewed by Restaurant Business Online.
• Starbucks deployed Automated Counting across more than 11,000 North American stores in September 2025, Reuters reported.
• Yum! Brands announced an Nvidia (NVDA) drive-thru AI partnership covering 500 locations in March 2025, Fortune noted.

The Starbucks unwind is happening quietly. The chain framed the move as a push to "standardize how inventory is counted across coffeehouses," in a statement to Reuters. The stock is up roughly 24% year to date, but most of that has been on Niccol's broader turnaround story, not on AI returns.

What fast-food AI deployment fails mean for the broader AI trade

What strikes me when I look at these two stories together is the similarity of the failure modes.

Both companies bought the idea that an AI layer on top of store operations would deliver the gains. Both deployed across hundreds or thousands of locations. Both watched the actual people the system was meant to coordinate, whether DoorDash drivers in Queens or baristas in Cleveland, run circles around the tool's basic assumptions.

Both are now unwinding. And in each case, the customer felt it before the shareholder did.

That matters for the AI trade in two ways.

First is the simple did-we-get-our-money-back question on enterprise spend, which Yum! Brands and Starbucks investors are going to ask more loudly.

Second is the more uncomfortable one for the chipmakers riding the build-out, including Nvidia, which counts Yum! Brands as a customer through that 500-location drive-thru deal.

The data on the broader trend back the skepticism. MIT's NANDA initiative found that 95% of enterprise generative AI pilots delivered no measurable impact on the P&L, despite roughly $30 billion to $40 billion in industry spend, according to The Next Web's coverage of the study.

Two food giant rollbacks in the same week are not anomalies. They are the visible tip of a much larger AI spend that has not earned its keep.

Ajay Agrawal, a professor of entrepreneurship at the University of Toronto's Rotman School of Management, captured the underlying problem in comments to Fortune. To make AI work in restaurants, he said, "you need to redesign the whole system."

Neither chain redesigned. Both bolted the AI on top. Customers got cold pizza in Queens and empty oat milk fridges in Seattle.

Shareholders who bought the AI-at-the-counter story at face value should be looking very carefully at which of their other holdings took the same shortcut.

Related: Taco Bell keeps winning while Pizza Hut drags Yum Brands down

The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc.

This story was originally published May 26, 2026 at 3:33 AM.