Why the Next 30 Days Will Reset the Restaurant-AI Pecking Order

It is 6:42 a.m. on Monday, March 3, and the only thing on my desk besides cold coffee is a paper calendar with three dates circled in red. March 5. March 6. March 18. I have been a news editor long enough to know that most months in restaurant tech are a slow drip of press releases pretending to be news. This one is not going to be. Inside the next fifteen business days, three separate events are going to land on top of each other, and when the dust settles, the pecking order in restaurant AI is going to look different than it does this morning. That is the contrarian thesis I want to plant before the rest of the industry wakes up: the pilot era is ending, and March 2025 is the hinge.

For two years, every operator deck has had a “we’re piloting AI” slide. Drive-thru voice in a handful of stores. A back-office copilot in a test market. A computer-vision camera over the fryer in one franchisee’s basement lab. That phase is closing. What is opening is the platform phase — the part where a small number of vendors get picked to be the substrate that everyone else’s AI runs on, and the rest spend 2026 explaining to their boards why they did not. My read: the events on my calendar this month are going to do most of that picking in public.

The three anchor events

The first domino is the McDonald’s edge-AI scoop the Journal has been sitting on. I expect it Wednesday or Thursday, March 5 or 6. The shape of it has been telegraphed in supplier chatter for weeks — edge hardware in stores, a named cloud partner, a multi-year rollout. It will be framed as a kitchen-and-drive-thru modernization story. The real story will be that the world’s largest restaurant chain has committed to a specific AI stack at the store level, and every other QSR’s procurement team is going to read it on Thursday morning and rewrite their 2025 RFPs by Friday.

The second is Wendy’s Investor Day on Thursday, March 6 — likely a few hours after the McDonald’s story breaks, which is either bad luck or excellent counter-programming depending on which IR team you ask. Wendy’s has been the loudest public voice on FreshAI and Google Cloud drive-thru, and an investor day is the venue where “we are testing” becomes “here is the rollout schedule and the unit economics.” Watch for a store-count target attached to a date. That is the tell that pilot has become platform.

The third is GTC on Tuesday, March 18, where Yum Brands and Nvidia are scheduled to share a stage. Yum has been quietly assembling an in-house AI platform — Byte by Yum — across KFC, Taco Bell, and Pizza Hut. A joint GTC appearance is not a science fair; it is a coming-out party. If the keynote includes a named SKU, a developer surface, or a partner ecosystem, the read is unambiguous: Yum is positioning Byte as a horizontal platform other brands can buy, not just an internal tool. That would be the loudest “pilot to platform” signal of the quarter.

Why Olo’s Feb 25 print matters more than it looks

Eight days before this morning, Olo reported FY24 Q4. Revenue up 21% year over year to $76.1 million, $29 billion in GMV across the platform, and $2.8 billion in GPV — the payments line that nobody was paying attention to two years ago and everyone should be paying attention to now. The number that matters is not the top line. It is the GPV-to-GMV ratio and the fact that it is climbing. Olo is quietly turning into a payments company with an ordering business attached, which is exactly the platform posture you want going into a year when the hardware-led POS incumbents are trying to add AI on top of fifteen-year-old stacks.

Compare that to the public-market gap that has opened up between hardware-led and platform-led POS over the last twelve months. Toast keeps adding logos — Applebee’s, Potbelly, Perkins on the enterprise side — and the platform thesis is getting priced in. Restaurant Business Online’s running list of the biggest restaurant-tech stories of 2025 has the same pattern in every entry: the winners are the ones who already look like platforms in February. My read: Olo’s print was the first proof point of the year that the platform-led side of the ledger is compounding faster than the hardware-led side, and the GTC week is going to be the second.

What to watch this month

Three things I am watching, in order of leverage. First, whether the McDonald’s announcement names a single edge partner or a portfolio. Single-partner is a platform pick; portfolio is still pilot. Second, whether Wendy’s puts a store-count number with a date next to FreshAI on Thursday. A number with a date is a commitment; an adjective is not. Third, whether the Yum/Nvidia GTC slot produces a developer-facing surface — an SDK, a partner program, anything a third party could build against. That is the difference between an internal tool and an actual platform.

I will be back in this column on Tuesday with the McDonald’s read, and again on Friday with Wendy’s. For the longer arc, a piece we later publish on Toast IQ gets into how the platform-versus-hardware framing is going to play out at the unit level, and our subsequent Bottom Line coverage of the DoorDash deal flow tracks where the marketplace money is moving while all of this is happening.

The calendar on my desk has three red circles. By April 1, two of them are going to look obvious in hindsight and one is going to surprise everyone. That is the bet.

— Hana edits The Pass. Tips: [email protected].