The Ceiling Is a Person, Not a Machine
I have sat across from owners with a backlog they were proud of and a worry they didn't say first. The work was there. The quotes were good. What they weren't sure about, when the room got quiet, was who was going to run it. Not the machines. The machines were fine. The people to stand in front of them were the thing they couldn't seem to buy.
That's the constraint most fabrication shops actually live under, and it's the one they're slowest to name because it doesn't sit on the balance sheet. You can finance another laser or another brake and have it on the floor in a quarter. The certified welder or the CNC hand who makes it earn its keep takes years to grow and, right now, is getting genuinely hard to find at any price.
The numbers behind the empty station
This isn't a feeling. The American Welding Society has been putting figures to it, and they're stark. The average welder in the US is now about 55 years old. AWS data has 21 percent of the welding workforce sitting over 55, more than 157,000 people close enough to retirement to plan for it, while under 10 percent of that workforce is younger than 25. The pipeline is emptying from one end faster than it's filling from the other. One way it's been put is that for every five skilled workers walking out to retire, only two are walking in.
Stack that against demand and the math gets worse. AWS projects welding occupations will need to fill roughly 80,000 openings a year through 2029, and manufacturing overall was carrying more than 400,000 unfilled jobs in mid-2025, with welders, CNC machinists, and maintenance techs among the hardest seats to fill. So the trade needs more skilled hands every year, and it's losing its most experienced ones on a schedule you could set a watch to.
For a shop doing $2 million or $80 million, the effect is the same in shape, different only in scale. Your throughput isn't capped by how much steel you can buy or how many machines you own. It's capped by how many people you have who can turn that steel into a part that passes inspection. Add a shift you can't staff and you've added rent on idle iron. The machine was never the ceiling.
What walks out with them
The retirement problem is quietly two problems. One is the headcount, and that's the one everybody talks about. The other is what leaves the building inside the person's head. The 40-year hand doesn't just run the weld. He knows which fixture warps on the long parts, how to read the material that came in a little off, which customer's prints always mean something other than what they say. None of that is written down anywhere, and none of it shows up in the ad you post to replace him.
When that person retires, you don't lose one worker. You lose a station's worth of output and a warehouse of judgment that was holding up your quality without appearing on any org chart. The shop feels it as a slowdown it can't quite explain. Scrap ticks up. The hard jobs take longer. The remaining hands are covering ground that used to belong to someone who made it look easy.
Widen the neck while you still can
You treat a people constraint the way you'd treat a machine constraint, except the lead time is measured in years, so you can't start when it's already biting.
Start by finding out who your building leans on. Which one or two people, if they gave notice tomorrow, would drop your capacity in a way you'd feel on the ship dates. That's your constraint wearing a name. Get what's in their heads onto paper and into the next hand: written setups, apprentices standing next to them now instead of after they're gone, cross-training so a single retirement doesn't take a whole capability with it.
Then think hard about automation, not as a headcount cut but as a multiplier on the hands you can't hire. Shops are putting in robotic welding cells that let one operator oversee the work several used to do, which is less about replacing the welder and more about stretching the scarce one across more output. If you can't add people, you make the people you have reach further.
And plant the pipeline yourself, because nobody's going to hand it to you. The shops that will have welders in 5 years are the ones running apprenticeships and reaching into local trade programs today, while the ones waiting for the labor market to fix itself are bidding against everyone else for the same shrinking pool.
Go find the person your shop can't run without. Then spend this year making sure that when he finally hangs up the hood, he doesn't take your capacity with him.