TIAN Industrial Automation
EN
All articles

The shortage isn't engineers. It's competent ones.

The numbers say automation is exploding and engineering roles are going unfilled. They also say the bar for "qualified" is moving. Read together, those facts describe a competence shortage, not a headcount one — and AI is widening the gap.

Published 7 min read
  • workforce
  • skills
  • automation
  • ai

In 2024, factories installed 542,000 new industrial robots. The global operational stock crossed 4,664,000 units — up 9% in a single year and more than double what it was a decade ago (IFR World Robotics 2025).

Over the same horizon, U.S. manufacturers project a need to fill 3.8 million jobs by 2033 and warn that 1.9 million of them — more than half — could go unfilled (Deloitte & The Manufacturing Institute, 2024).

More machines than ever, and not enough people to run them. The reflexive read is “labor shortage.” That read is incomplete, and the gap it papers over is the entire problem.

A six-axis industrial robot arm tending a production line, with an engineer working at a welding station in the background

The figures, laid out plainly

Automation is still accelerating. 2024’s 542,000 installations were the second-highest annual total on record — within 2% of the all-time peak — and the installed base hit a new high. The IFR projects installations to keep climbing toward ~700,000 units a year by 2028 (IFR). Every one of those cells needs someone to specify, integrate, program, commission, and keep it running.

Demand for engineers is outpacing the average job. BLS projects U.S. employment of industrial engineers to grow 12% and mechanical engineers 11% from 2023 to 2033 — against 4% for all occupations (BLS OOH: Industrial Engineers, Mechanical Engineers). Three times the baseline.

The hole is mostly demographic. Of the 3.8 million manufacturing roles to fill by 2033, roughly 2.8 million come directly from retirements (Deloitte & MI). That’s not a wave of new positions — it’s a generation walking out the door with the tacit knowledge that never made it into a drawing.

Employers can’t fill what they have. 65% of manufacturers name attracting and retaining talent as their primary business challenge (Deloitte & MI). Globally, 63% of employers call the skills gap the single biggest barrier to transformation — the top-ranked obstacle in the WEF survey of over 1,000 companies (WEF Future of Jobs 2025).

The supply side is producing the wrong thing. In a survey of 800 HR leaders, 75% said most college educations aren’t preparing people for their jobs; 91% said recent graduates cost more to onboard, and only 24% of those graduates felt they had all the skills their role required (Hult / Workplace Intelligence). The credential is being issued. The competence isn’t reliably behind it.

And the definition of “qualified” is moving. The WEF estimates 39% of a worker’s core skills will change by 2030, with analytical thinking, technological literacy, and AI/big-data competence among the fastest-growing demands (WEF). Of any 100 workers, 59 need reskilling by 2030 — and 11 are unlikely to get it (WEF).

What the numbers actually say together

Line them up and a single story falls out. Demand for engineers is real and rising — 12% growth, millions of openings. But the openings aren’t going unfilled because the labor pool is empty. They go unfilled because employers report the available candidates can’t do the work: 63% blocked by skills gaps, 75% of educations judged inadequate, more than half of projected roles at risk of staying open. A pure headcount shortage looks like empty waiting rooms. This looks like full waiting rooms and empty seats on the floor.

I’ll be honest about which seat I started in. I came into this trade green — coursework done, a stack of certificates behind me: TIA Portal, ABB’s trainee program with its theory exam, EPLAN. None of it taught me how to stand in front of a cabinet that didn’t match its drawing and work out what was actually wired. The credential was real. The competence wasn’t behind it yet — that I had to build on site, out of my depth, with someone more experienced close enough to catch me when I fell. For my first months I was, more or less, exhibit A for this whole paragraph: a name attached to a passed exam, and not much else the floor could use.

Now layer in automation, and the gap doesn’t just persist — it widens. Here’s the mechanism the headlines miss: AI and robotics are eating the easy 80% of the job and raising the bar on the hard 20%. Copy a working block of logic, generate boilerplate structured text, auto-document a panel, first-pass a P&ID — the routine, teachable, high-volume work that used to be how a junior engineer built judgment by repetition. That scaffolding is exactly what these tools now do in seconds.

What’s left is the 20% that doesn’t automate: deciding whether the generated logic is safe, reading a machine that’s misbehaving for reasons not in any manual, knowing which corner the auto-generated solution quietly cut, owning the call when the line is down at 2 a.m. and the AI’s suggestion is plausible but wrong. That’s the work the WEF data points straight at — analytical thinking and technological literacy as the growth skills. It’s also precisely the judgment that used to be built by grinding through the 80% that no longer exists as a training ground.

So the pipeline is being squeezed at both ends. Traditional education keeps optimizing graduates for the old 80% — the part now done by tooling — while the demographic exit removes the seniors who carried the 20% in their heads. The bar for “competent” rises, the on-ramp that used to get people there erodes, and the retirement wave drains the bench at the same time. Three forces, one direction.

I’m 21, and I’ve been doing this for about a year and a half — which makes me the person that last paragraph is about, except I’m not looking back on the apprenticeship. I’m standing in the middle of it while the floor moves. My first months were the 80%: HMI screens, drawing revisions, the filler work you hand the new person. Eighteen months later, a lot of that is work I’d now pass to a tool without thinking twice. I’ve watched the on-ramp narrow in real time.

The judgment came from the other moments — the ones nothing automated. The first machine I commissioned largely on my own was a shuttle running a program I’d never seen; the I/O checks and the safety fault-finding were mine to reason through, not look up. On another job a light curtain locked itself into a muting deadlock that wasn’t in any manual — getting it back meant understanding why the safety logic was behaving that way, not just which button to press. That’s the 20%. And the only reason I could reach it is that I’d already ground through the 80% that’s now quietly disappearing for the people coming in behind me. I might be one of the last to get that apprenticeship by accident.

This is why “just hire more engineers” misreads the problem. You cannot solve a competence shortage with a headcount strategy. Add bodies and you add people trained for the part of the job that no longer needs them.

The earned conclusion

The data does not support automation hype, and it does not support doom. It supports something narrower and more useful: demand for genuinely competent automation engineers is rising, and the supply of them is not keeping pace — for structural reasons that compound rather than resolve.

The robots are real: 4.66 million and climbing. The openings are real: 12% growth, millions unfilled. The skills gap is real and employer-ranked as the number-one barrier. And the thing AI changes isn’t that it replaces the engineer — it’s that it deletes the easy work that used to manufacture one, while raising the standard for what the remaining work demands.

That’s not a labor shortage. It’s a competence shortage wearing a labor shortage’s clothes. The firms that understand the difference will stop counting résumés and start building the 20% — deliberately, because the floor that used to build it by accident is gone.

Sources