Kroger, one of the largest grocery chains in the United States, has written off roughly $2.585 billion tied to its automated fulfillment network after closing robot-run warehouses that were supposed to speed up delivery of fresh food. The closures, which span facilities in Wisconsin, Maryland, Florida, and Texas, have exposed a painful contradiction at the heart of food logistics: the same automation designed to move perishables faster can also strand them when systems break down or prove uneconomical. As the grocery industry races to automate, truckloads of temperature-sensitive goods risk spoiling in the gap between a failed digital system and the human capacity to pick up the slack.
Kroger’s $2.585 Billion Write-Down
The scale of Kroger’s retreat from its automated fulfillment experiment is hard to overstate. In its quarterly filing with the SEC for the period ended November 8, 2025, the company disclosed approximately $2.585 billion in impairment and related charges connected to its fulfillment network. That figure included an accrued cash termination payment of roughly $350 million to Ocado, the British robotics firm that designed and operated the automated systems inside those warehouses.
The termination payment was triggered by the closure or cancellation of certain fulfillment centers. Kroger had planned to close automated facilities in Pleasant Prairie, Wisconsin; Frederick, Maryland; and Groveland, Florida, with those shutdowns announced for January. Separately, the company also moved to close three “spoke” facilities, which are cross-docking sites that extend the reach of the automated network, in Austin, San Antonio, and Miami. These spokes serve as critical transfer points for perishable goods. When they go offline, temperature-sensitive shipments face delays that can push fresh produce, dairy, and meat past their safe handling windows.
When Robots Stop, Food Rots
The problem with concentrating perishable food flows through a handful of high-tech hubs is that any disruption, whether financial, mechanical, or accidental, can halt deliveries across an entire region. A fire at Ocado’s robot-packed warehouse in Andover, England, documented by the Hampshire and Isle of Wight Fire and Rescue Service, demonstrated how a single incident at an automation-heavy site can shut down fulfillment for days. While that fire occurred in the UK, the underlying risk applies to any facility that relies on tightly packed robotic grids to store and retrieve groceries. A conventional warehouse staffed by human workers can often reroute orders or switch to manual picking. A robotic grid that catches fire or loses its software connection cannot.
Food security researchers have flagged this brittleness directly. Analysis published in The Conversation argues that even when digital systems recover, the human ability to restart flows may be limited because automation has displaced many of the workers who once handled exceptions. Drivers, dispatchers, and warehouse staff who previously managed breakdowns with phones and clipboards have been replaced by software. When that software fails, there may not be enough experienced people left to improvise routes, reassign loads, or reconfigure storage to keep food from spoiling.
Efficiency Gains Mask Structural Fragility
Automation and artificial intelligence have genuinely increased efficiency in food supply chains, but they have also introduced critical vulnerabilities. The dominant narrative around supply chain technology tends to focus on speed, cost savings, and throughput. What gets less attention is the tradeoff: by funneling deliveries through fewer, larger, more complex facilities, companies create single points of failure that did not exist when distribution was spread across many smaller, human-operated warehouses.
Kroger’s experience illustrates this tradeoff in dollar terms. The company invested heavily in Ocado’s technology with the goal of making grocery delivery faster and cheaper. Instead, it ended up absorbing billions in losses and dismantling the very network it built. The closures do not just represent a financial setback for one retailer. They signal that the current model of centralized robotic fulfillment may not be well suited to the specific demands of perishable food, which requires constant temperature control, rapid transit, and the flexibility to reroute shipments when plans change.
That structural fragility is particularly concerning because the automated hubs are often embedded in just-in-time logistics systems that keep minimal inventory on hand. When a robot-run facility goes down, there are few buffers in place. Stores depending on daily or even hourly replenishment can see empty shelves within days, while products trapped in disabled warehouses may have to be discarded if their cold-chain integrity cannot be verified. In these scenarios, the same algorithms that once optimized routes and stock levels become a liability, locking food into digital queues that no longer correspond to physical reality.
Food Loss Baselines Already Run High
The United States already loses a significant share of its food supply before it reaches consumers. The USDA Economic Research Service publishes the loss-adjusted availability data series, which provides baseline estimates and definitions for food loss at each stage of the supply chain. That methodology tracks losses from the farm level through retail and consumer stages, offering a framework for understanding where food disappears and how much is wasted long before it is eaten.
These federal data efforts sit within a broader set of nutrition and food systems programs overseen by the U.S. Department of Agriculture. By combining production statistics, consumption surveys, and loss estimates, agencies can assess how much edible food is effectively removed from the system through spoilage, damage, or quality downgrades. Against that backdrop, any additional waste created by brittle automation is layered on top of an already leaky pipeline, rather than being offset elsewhere in the system.
Detailed consumption and diet data further illuminate how supply disruptions translate into nutritional gaps. The Agricultural Research Service maintains food pattern equivalents databases that convert what people report eating into standardized servings of fruits, vegetables, grains, proteins, and dairy. When fresh items are delayed or lost in transit, these categories are not affected evenly: perishable produce and dairy are far more vulnerable than shelf-stable grains or canned goods, skewing what ultimately reaches consumers’ plates.
On the composition side, the FoodData Central system at the National Agricultural Library provides nutrient profiles for thousands of items, from raw ingredients to branded products, through its searchable food database. Linking these nutrient data with loss estimates makes clear that it is not just calories that are being stranded when digital systems fail, but also vitamins, minerals, and other components essential for health. A pallet of spoiled leafy greens represents a different kind of loss than a pallet of snack foods, even if their dollar values are similar.
Nutrition policy also depends on stable supplies of key food groups. The Food and Nutrition Service publishes dietary patterns that underpin federal guidance and feeding programs, translating nutritional science into recommended amounts of fruits, vegetables, grains, and other categories. When robot-dependent warehouses falter, the shortages that appear in stores can make it harder for households to follow these patterns, especially in communities that already have limited access to fresh food. In that sense, technical failures in logistics can ripple into public health outcomes.
Designing Automation for Resilience, Not Just Speed
The lesson from Kroger’s write-down is not that automation has no place in food logistics, but that its design must prioritize resilience alongside efficiency. Distributed networks of smaller, semi-automated facilities may offer a better balance than a handful of giant hubs. Maintaining a core of skilled human workers who can revert to manual operations during outages is another form of insurance. So is investing in interoperable systems that allow inventory and orders to be shifted quickly between sites when one node goes down.
Regulators and policymakers, armed with detailed loss and nutrition data, have a role to play in setting expectations for how critical food infrastructure should perform under stress. That could include requiring contingency plans for large automated warehouses, encouraging redundancy in regional distribution, or supporting research into hybrid human-machine systems that fail gracefully rather than catastrophically. For retailers, the financial shock of multi-billion-dollar impairments may be a powerful motivator to rethink how much risk they are willing to concentrate in any single technology or location.
As the grocery sector continues to experiment with robotics and AI, the central question is shifting from “How fast can we move food?” to “How reliably can we keep it moving when things go wrong?” Kroger’s experience, combined with mounting evidence from food security research, suggests that the answer will depend less on the sophistication of individual machines and more on the humility of the systems that surround them: the willingness to preserve human expertise, accept some redundancy, and measure success not only in throughput, but in the meals that actually make it to the table.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
