Consumer Goods Plants Say Downtime and Rework Cost a Fifth of Revenue. They Expect a Third by 2030

American food, drink and household goods plants are, on paper, the most advanced in the world when it comes to industrial AI. They are also the most pessimistic about where their operating losses are heading. A survey released this month finds that U.S. consumer goods manufacturers already lose a fifth of their potential revenue to stoppages, scrap and rework, and that they expect to lose a third of it by the end of the decade. The technology lead and the widening losses sit side by side, and the gap between them says more about plant fundamentals than about algorithms.

A Fifth of Potential Revenue, and Rising

Schneider Electric published the U.S. figures on October 19 at PACK EXPO International in Chicago, drawing on a Censuswide survey of 252 senior manufacturing decision-makers alongside its earlier sample of more than 1,400 executives worldwide. U.S. consumer goods manufacturers estimate that downtime, waste, delays, rework and suboptimal asset utilisation consume 20% of potential manufacturing revenue, compared with a 15% global average. They project that share rising to 27% by 2027 and 34% by 2030. The global projections are lower, at 21% and 29%, but they climb just as steadily.

These are respondents' own estimates rather than audited losses, and the survey was commissioned by a company that sells automation and AI. Even so, the direction is hard to dismiss. When the people running plants expect a third of their potential output value to disappear into inefficiency within four years, they are describing pressure they can already feel on the line. "Manufacturers are being asked to increase productivity, adapt faster, and do more with the resources they have," said Greg Magdanz, vice president of OEM sales at Schneider Electric.

The wider picture points the same way. Siemens' True Cost of Downtime 2024 analysis, summarised by the Association of Electrical and Mechanical Trades, estimated that the world's 500 largest companies lose almost $1.4 trillion a year to unplanned downtime, equivalent to 11% of their revenues. The same analysis found the average large plant was losing 27 hours a month to unplanned stoppages, down from 39 hours in 2019. Plants are stopping less often than they were, yet the stoppages that remain still add up to a striking share of revenue.

The AI Lead That Has Not Closed the Gap

What makes the U.S. numbers striking is how far ahead the same plants say they are on technology. According to the Schneider release, 29% of U.S. consumer goods manufacturers have AI embedded end to end in core operations and decision-making, against 13% globally. Some 52% have deployed AI across multiple sites or processes, compared with 27% worldwide, and 67% expect AI to be core to operations by 2030, against 37% of their global peers.

If AI on its own were closing the loss gap, the U.S. figures should be moving the other way. The global study behind those comparisons helps explain why it is not. As Controls, Drives & Automation reported in April, 70% of consumer goods manufacturers worldwide said their current return on AI was below 20%, and 28.4% put it at 5% or less. Respondents also estimated that delays, downtime and equipment failure already account for 20.3% of the final cost of the products they make.

The barriers they named are mostly not about algorithms. A shortage of AI and data science skills was cited by 43.0%, outdated automation systems and infrastructure by 37.5%, and operational data that lacks context by 36.3%. Employee pushback came in at 25.7%, and security and regulatory concerns at 21.7%. "This expectation gap is the strongest signal of urgency we've seen in years," said Neil Smith, president of consumer packaged goods at Schneider Electric. Put plainly, plants are layering AI onto control systems, data and working practices that were never designed to feed it, and the losses that matter most are still happening at the filler, the changeover and the reject bin.

Where the Losses Actually Sit

The categories in the survey are familiar to anyone who has run a packaging hall: unplanned stops, slow changeovers, material waste, product held or reworked for quality, and lines running below their rated speed. None of them is new, and none of them needs a frontier model to measure. What they need is accurate, timestamped data from the equipment, put in front of the operators and supervisors who can act on it during the shift rather than in next month's report.

That is also where the skills and pushback findings bite. An AI tool that flags a quality drift is only useful if the operator trusts it and knows what to do next. Without that, plants end up with impressive pilots and unchanged scrap rates, which is roughly what a 70% share of sub-20% returns describes.

What Operations Leaders Should Do Now

The plants that bend the loss curve before 2030 will treat AI as the last layer of the stack rather than the first.

• Put a number on your own loss: break lost potential revenue down by downtime, changeover, scrap, rework and speed loss for each line, so you know which of the survey's categories is actually costing you most.
• Fix the data before the model: make sure machine states, stop reasons and quality results are captured automatically and with context, since 36.3% of manufacturers say their operational data lacks it.
• Audit the control layer: identify the ageing HMI, SCADA and PLC estates that cannot feed or act on analytics, and fund their replacement as part of any AI business case.
• Start AI where waste is visible: target quality inspection, defect prediction and material yield first, where the result shows up in scrap and rework within weeks.
• Bring the floor in early: give operators and supervisors the same live view of losses that managers see, and train them on what to do when a model raises a flag, so pushback turns into ownership.
• Measure return honestly: track AI programmes against the loss categories they were meant to reduce, and stop the ones that cannot show movement within two quarters.

The U.S. consumer goods sector has proved it can buy and deploy AI faster than anyone else. The harder test is whether it can turn that lead into fewer stoppages and less scrap before a third of its potential revenue is written off as the cost of doing business.