Why American Retailers Are Spending Billions on Cameras That Watch Themselves

Walk into any large American big-box store this year and look up. The number of ceiling-mounted cameras has roughly doubled in eighteen months, and almost all of them now feed real-time computer-vision systems trained to detect theft, mis-scans and process failures at self-checkout. The story matters because organised retail crime and self-checkout shrink have become a board-level issue for every u.s. retailer above a certain footprint.

The economics are brutal The largest U.S. retailers report shrink rates that have nearly doubled since 2019, with self-checkout estimated to account for a meaningful share of the increase. Computer-vision systems that can flag a suspected mis-scan or missed item in real time are paying for themselves in months in the highest-loss stores.

The privacy debate is real but quiet U.S. retailers have largely avoided the European-style scrutiny of in-store vision because the systems are framed as loss prevention rather than personalisation. That framing is unlikely to hold forever; the same camera infrastructure is already being repurposed for store-operations analytics and marketing.

The cameras started out as loss prevention. Within a year they will be the most valuable behavioural-data asset any large retailer owns.

What to watch next Expect the line between loss-prevention vision and behavioural-vision analytics to dissolve quickly — and a new wave of regulatory attention to follow. For operators and investors, the read-through is clear: physical retail's data infrastructure is being rebuilt at speed, mostly out of public view.