Why America's grocery prices are higher than they need to be

Of every dollar Americans spend on food, the farmer who grew it keeps about twelve cents, down from nearly fifteen two decades ago. At the grocery checkout the share is a little higher, closer to eighteen cents, but the picture is the same: a thin and shrinking cut for the people at the start of the chain. More than eighty cents on the dollar is absorbed somewhere in the long passage from field to shelf, and how much of that is earned, against how much is simply wasted, has quietly become one of the more important questions behind the price of food.The usual explanations are inflation and corporate greed, and neither is wrong. Grocery prices climbed by roughly a quarter between late 2019 and early 2023, the sharpest stretch in four decades, and while the rate has since eased to under 3 percent a year, the cumulative rise still stings at the register. But a meaningful part of that eighty-cent middle is neither inflation nor profit. It is waste of the ordinary, fixable kind, the sort that builds up when the machinery linking the companies that make food to the stores that sell it has been left a generation behind.Over the past decade the food industry has poured money into everything a shopper notices: slick online ordering, targeted marketing, robots on the warehouse floor. The systems that connect a brand to the retailers selling its products, and that track what was shipped, stored, and actually paid for, have been left largely untouched. The friction that produces is a persistent and largely invisible drag on the cost of food.Closing that gap has become the work of Leon Micheel-Sprenger, a researcher and entrepreneur at Stanford's Graduate School of Business, who has spent his time there at an unusual intersection: the frontier of artificial-intelligence research coming out of Silicon Valley on one side, and the decidedly unglamorous question of how food actually reaches American shelves on the other. His thesis, developed over more than a year of research and published with the venture investor Scott Brady, is blunt. The food business, he argues, has modernized everything its customers can see and allowed almost everything they cannot to fall behind, and shoppers end up paying for the difference. It is an argument that has turned him into something of a translator between two worlds that rarely meet, and a name that senior leaders at large food companies now seek out for a single question: where artificial intelligence genuinely belongs in their operations, and where it is merely the latest fashion.His starting observation is a contradiction. "Almost every company I work with runs an automated warehouse and markets with AI," he says. "Then you look at how the same business tracks a shipment, or squares a payment against what actually arrived, and it's a spreadsheet, an inbox, and someone downloading files from a portal by hand."The clearest way to watch the gap in action is the deduction. A retailer takes in a shipment, decides something was wrong with it, and subtracts the disputed sum from the supplier's next payment. Picture an order of 500 cases: the retailer records 475, deducts for the 25 it says are missing, and pays the smaller amount; weeks later the cases turn up in its own warehouse, but the window to dispute the charge has closed. Industry analysts estimate that between a tenth and a fifth of such charges are invalid, and that across the grocery business they run to 5 to 15 percent of a supplier's gross sales. Most go uncontested anyway, because no finance team can afford to litigate a $25 charge, so the small ones accumulate. The same disconnection surfaces everywhere else, too: a late shipment nobody flags until the penalty arrives, inventory stranded in one region while another runs short, a forecast built on numbers that were stale before anyone read them. "A lot of this money is genuinely owed back to the brand," Micheel-Sprenger says. "Companies don't chase it because proving the claim costs more than the claim is worth."The imbalance is also structural. The largest retailers run their side of these transactions through automated systems and keep staff dedicated to defending them. Walmart charges suppliers 3 percent of the cost of goods on orders that miss its on-time, in-full targets; Target's penalty reaches 5 percent. A regional producer rarely has the people to contest such charges one at a time, and so absorbs them. Washington has begun to notice. Last year the Federal Trade Commission revived enforcement of the Robinson-Patman Act, a 1936 law against discriminatory pricing that had sat largely dormant for decades, and in January, in a rare moment of bipartisan agreement, senators from both parties pressed regulators to examine how the buying power of dominant grocers affects prices. An association representing independent grocers warned that such power was already surfacing as higher prices, fewer competitors, and shrinking food access in smaller communities.That squeeze falls hardest on the producers with the least room to absorb it, the regional brands and the farms another link or two back. Micheel-Sprenger grew up inside that math, on a farm where every dollar of avoidable loss registered, and he sees the companies he studies as facing a version of the same pressure.What has changed, he argues, is not the problem but the cost of solving it. The data needed to run a tighter operation has always existed, scattered across shipping records, invoices, and portals; stitching it together used to be prohibitively expensive, and artificial intelligence has finally made that cheap enough to be worth doing across an entire business rather than one corner of it. His message to the executives who consult him is not that AI is magic, but that most of them are pointing it at the wrong targets, automating the work customers can see while the expensive friction sits untouched out of sight. He points to one large food company that, after years of handling problems only once they had happened, connected its commercial operations this way. Disputes resolved faster. Out-of-stocks dropped. The finance team stopped writing off money it was owed. Nothing about the underlying operation had changed, he says. What changed was that everyone could finally see the same picture at the same time.None of this is theoretical for Micheel-Sprenger. He is a co-founder of a company bringing this kind of AI innovation he describes to enterprise food brands in America, on the conviction that the gap is real, large, and worth closing. The economics are not complicated: waste taken out of the system is waste that no longer has to be priced into the product.Whether those savings would ever reach the shopper is harder to say, and the effect on any single receipt would be small. The larger prize is supply chains that are sturdier and fairer: ones that lose less to friction and leave more of the dollar with the people who actually grow the food. The same logic holds far beyond the United States, in India and in every market where modern retail meets a crowd of smaller suppliers. The farmer, for now, still keeps about twelve cents. What is finally beginning to change is how much of the rest survives the trip to the shelf.