The Real Cost of Manufacturing Downtime in 2025
Every operations leader knows downtime is expensive. But when I show them the actual numbers from recent Siemens research, they're usually shocked. Here's what the data says—and what you can do about it.
Founder & CEO at Omira Technologies
Last spring, I sat down with a plant manager in Tennessee to look at his maintenance records. He'd been in the job for twelve years, knew his operation cold, and had always thought of downtime as just one of those things you manage.
We went through the numbers together. Every unplanned stop, every breakdown, every time the line went quiet when it shouldn't have. When we tallied it up, his face changed. $4.2 million. Gone. In a year.
His facility does about $80 million in revenue. That's more than 5% of his top line evaporating—not into wages or materials or anything you can point to, just into silence when machines should be running.
He wasn't an outlier. I've had this conversation enough times now to know that most operations people dramatically underestimate what downtime is actually costing them.
The Scale of the Problem
Siemens published their True Cost of Downtime 2024 report, and the numbers stopped me cold. Across the world's 500 largest manufacturers, unplanned downtime now consumes about 11% of annual revenue. That's $1.4 trillion globally.
Here's what's unsettling: it was $864 billion in 2019-2020. The problem is getting worse, not better. I have some theories about why, but first let me share what an hour of stopped production actually costs:
Hourly Downtime Costs by Industry (Siemens 2024)
I know. That automotive number looks like a typo. It's not. When a modern car plant stops, everything stops—every station, every robot, every worker standing around. And with just-in-time inventory, there's nothing in the pipeline to absorb the shock.
That number was $1.3 million in 2019. It's climbed 77% in five years—far outpacing the 19% US inflation during the same period. The global energy crisis and increasingly complex supply chains have made downtime exponentially more expensive.
The Costs Nobody Puts on a Spreadsheet
Those hourly numbers capture the obvious stuff—the direct loss of what you would have produced. But I've learned that the secondary effects often hurt more.
The emergency premium. When something breaks at 2 AM, you're not shopping around for the best repair rate. You're paying whatever it takes to get running. I've seen emergency repair costs run 3-4x what planned maintenance would have cost. Overnight shipping for a part that's usually $200? Now it's $800 and you're grateful to get it.
The catch-up scramble. Production lost on Tuesday still needs to ship by Friday. So you run overtime—time-and-a-half, double-time on weekends—with tired people making more mistakes than usual. I talked to a food processor who figured his real downtime cost at 2.5x the hourly production value once he factored in recovery.
The quality wobble. Equipment that's been hastily repaired doesn't run the same. Neither do operators who've been working 12-hour shifts to catch up. Defect rates climb. Sometimes the problems don't show up until you get the call from your customer.
The relationship damage. Miss one delivery and you get a concerned email. Miss two and someone from procurement wants a call. Miss three and you're in remediation—or you're being replaced. I know a tier-two supplier who lost a $12 million contract over reliability issues that traced back to equipment they kept patching instead of fixing.
You Probably Have More Downtime Than You Think
Most plants I visit undercount their downtime by at least 30%. Sometimes 50%. The reasons are always the same.
Micro-stoppages don't get recorded. A five-minute jam gets cleared, someone hits the reset button, and nobody writes anything down. But if that happens fifteen times a shift, that's over an hour of lost production that never shows up in any report.
Changeovers are considered "planned," so they don't count as downtime even when they take twice as long as they should. Speed losses are invisible—if a machine runs at 85% of rated speed, that's 15% hidden downtime that compounds every hour of every shift.
Industry surveys show an average of about 800 hours of unplanned downtime per year for a typical facility. That's 15 hours a week. More than a full shift, every week, of paid non-productive time. According to the Siemens report, the average facility now experiences 25 downtime incidents per month—actually down from 42 in 2019, but the cost per incident has risen dramatically.
Why It's Getting Worse
You'd think with all our sensors and software that we'd be getting better at this. We're not. A few things are working against us:
Equipment has gotten complicated. Modern machines have more controllers, more sensors, more software than ever—and more ways to fail. When they do fail, diagnosis takes longer because you might need a specialist who isn't available until Thursday.
A lot of infrastructure is aging. Capital investment got deferred during COVID, and again during the supply chain chaos, and now there's a backlog of maintenance coming due all at once.
The people who know how to fix things are retiring. There's no good way to say this: the maintenance workforce is shrinking and getting less experienced. The tribal knowledge—the stuff that lets a veteran tech hear a bearing and know it has two weeks left—that knowledge is walking out the door.
What I've Seen Actually Work
The plants that take downtime seriously—that really attack it—are seeing meaningful results. Here's what the successful ones have in common:
They measure everything. Not just the big stops. Every micro-stoppage, every speed loss, every extended changeover. You can't fix what you can't see.
They predict instead of react. The U.S. Department of Energy has documented 70-75% reduction in breakdowns for facilities using predictive maintenance, along with 25-30% cost reduction. Vibration monitoring, thermal imaging, oil analysis—catching problems before they become emergencies.
They do root cause right. Not "the bearing failed, replace it" but "why did the bearing fail, and what changes so it doesn't happen again?" This requires discipline and time, but it's the only way to actually improve.
A steel manufacturer I worked with invested in predictive maintenance—sensors, software, training—and logged $850,000 in savings the first year. Equipment reliability improved 85%. The investment paid back in eleven months.
A 90-Day Starting Point
Days 1-30: Actually Measure
Track every stop. Every one. Duration, cause, equipment. Yes, even the five-minute jams. Get real data on what's actually happening.
Days 31-60: Find the Patterns
Pareto your data. What equipment causes the most downtime? What time of day? What failure modes? The patterns will surprise you.
Days 61-90: Fix One Thing
Pick your number one problem and attack it properly. Root cause analysis. Preventive measures. Prove the approach works before trying to fix everything at once.
My Honest Take
Every plant has downtime. The question is whether you're managing it or just enduring it.
The facilities I see making real progress share a common trait: they stopped treating downtime as inevitable. They started measuring it rigorously, understanding it deeply, and investing in prevention systematically.
Is it easy? No. Does it require investment? Yes—in sensors, in training, in changing how people think about maintenance. But the math is hard to argue with. At $250,000 an hour average cost, even a 20% reduction adds up fast.
That plant manager in Tennessee? Last time we talked, he'd cut his unplanned downtime by 35% in eight months. He said the hardest part was admitting how bad it had been—and how much he'd been leaving on the table by treating it as normal.
If you want to run the numbers for your operation, we built a free ROI calculator that can help you estimate what downtime reduction could mean for your bottom line.
Ready to tackle your downtime problem?
I help manufacturers implement predictive maintenance and condition monitoring systems. Book a free 30-minute call to discuss your specific challenges.
Book a discovery callFounder & CEO at Omira Technologies. I help manufacturers implement AI automation that actually works—predictive maintenance, computer vision, and operational efficiency.More articles →