S09.1 · Retail & Commerce
A $1T-plus, thin-margin US category where AI-driven comparison shopping now threatens the loyalty-data moat funding retail media.
US supermarket sales hit roughly $1T in 2025 (FMI) across roughly 45,575 stores, concentrated in North America and Europe. Growth is low single digit, and Coresight expects cautious consumer spending to hold volume down in 2026 after the inflation-driven nominal gains of 2021-23. The number to watch is not the top line but the margin underneath it: at 1-3% net, grocery has no slack, and the loyalty-data asset that was supposed to fund the next leg of profit is exactly the piece AI-native shopping behavior now puts at risk.
Global grocery resists a clean single market-size figure because formats blur into mass and discount retail. In the US, Walmart, Kroger, Costco and Albertsons hold the majority of share over a deep regional and independent long tail. Grocers sit between CPG manufacturers and distributors upstream and consumers downstream, and take pressure from both directions — manufacturer pricing power on one side, mass and discount competition on the other. Net margins run 1-3%, offset by high inventory turns. The business is labor- and real-estate-intensive, but demand recurs and is largely non-discretionary, which is what makes low single-digit margins investable at scale.
A 1-3% net margin prices every operating mistake at par: shrink, labor productivity and waste each move the whole P&L. That is why grocery became one of the most operationally disciplined formats in retail without ever becoming one of the most profitable. Turns are the offsetting lever — a thin margin still compounds into an acceptable return on capital when the same shelf space sells through many times a year. The same fragility explains why the segment automated early and aggressively anywhere the customer-facing experience would not notice.
The adjacent growth vectors — convenience formats, e-commerce and delivery expansion, retail media — split by where the dollar comes from. Private label and retail media defend the thin core margin; delivery investment is offense, chasing share ceded to Instacart and Amazon. Algorithmic pricing and automated replenishment are compressing labor and shrink/waste costs now — a clean opex story, with no ambiguity about who keeps the savings.
The harder question is the loyalty-data moat itself. A decade went into converting first-party shopper data into a defensible asset, monetized directly through retail media. AI shopping agents that compare baskets across retailers on the consumer's behalf erode exactly that: the "destination" habit — a shopper committed to one store's aisles, promotions and loyalty program — is what agentic comparison shopping is built to route around. Once a basket can be re-priced and re-sourced automatically across banners, the loyalty program stops capturing the switching cost it was designed to capture, and the retail-media revenue funded by that loyalty data is exposed with it. The opex side is already moving; the comparison-shopping threat to the demand side is expected to bite within 2-5 years.