Food Away From Home Cools to a 3.7% YoY Climb. The 'AI Premium' Theory Dies a Little.

I read the August CPI release the way most operators read a P&L: bottom to top, looking for the line that ruins the rest. The BLS print landed this morning, and the line I keep circling is food away from home — up 3.7% year over year, a deceleration from the spring, and a number that quietly undoes a thesis a lot of investor decks have been running on.

The thesis goes like this: AI takes labor and prep cost out of the back of house, operators bank the spread, and consumers keep tolerating menu hikes because the experience is improving. I have never liked it, and I said so in a forthcoming May piece that picked the assumption apart line by line. Today’s release is the first month where the macro print, not the vibes, is doing the arguing for me.

The decoupling nobody wants to name

Inside the 3.7% headline is the part operators should be staring at. Full-service meals are up 4.2% over twelve months. Limited-service is up 3.2%. That gap — a full percentage point — is not noise. It is the structural divergence that operators have been pretending wasn’t there since the spring.

For most of 2024, the two segments moved in something like a lockstep. Drift, yes; decoupling, no. What August shows is full-service still pulling pricing while QSR has visibly stopped. The chains that own the value end of the market read the room a quarter ago and pulled back on menu hikes. Independents and full-service groups — the ones least able to absorb a wage step — kept pushing. The FRED series for food away from home makes the shape obvious if you pull a five-year chart and stop staring at the most recent dot.

The contrarian read: the “AI premium” was never a real thing. It was a story that gave operators permission to keep pricing through a softening consumer. Limited-service walked away from that story first because their elasticity is brutal and immediate. Full-service is walking away from it next, just more slowly, and the August print is what the early innings of that walk look like.

Grocery is not the relief valve you think it is

The other half of the picture matters. Food at home rose 2.7% over the year. That’s the slowest gap-to-FAFH spread we’ve seen in months, but the composition is ugly. Meats, poultry, fish and eggs are up 5.2% over twelve months. Cereals and bakery are up 1.6%. The middle of the plate is hot; the carbs are cold.

What that means for operators is mark — the protein spec on your menu is being repriced by a different macro than your bread program. Burger groups and steakhouses are eating the hottest line on the CPI table. Bakery cafés and pasta-forward concepts get a softer ride. If you are a 12-unit café group running fast-casual price points, as I sketched out in an upcoming May piece, you have more flex than a midscale steakhouse right now, and you should be using it to hold the line on a Q4 hike rather than chasing one.

What I’d act on this week

Three things. Mark — these are the moves I’d make Monday morning if I ran a multi-unit P&L.

First, freeze the planned Q4 menu hike at full-service if you’ve got one queued. The pricing power your model assumes is decaying in the actual data, and a 4.2% segment YoY does not survive a soft holiday season without a traffic hit. Second, audit your protein-heavy SKUs against the 5.2% line and decide whether you re-spec, reprice, or quietly 86. Third — and this is the one operators hate — stop telling your board the AI deployment is going to fund the hike. The math has to stand on its own. If it doesn’t, you don’t have an AI story; you have a margin problem with a chatbot taped to it.

The August print is not a recession signal. It is something more useful: a reminder that the consumer is doing what consumers do at the end of a long pricing cycle. They notice. And the segment-level CPI lines are starting to notice them noticing.

— Hana edits the newsroom for TableTransfers. Tips: [email protected].