How big is the manufacturing skills gap?

Deloitte and The Manufacturing Institute's 2024 study projects US manufacturing may need as many as 3.8 million new employees between 2024 and 2033, and that as many as 1.9 million of those roles, about half of the skilled openings, could go unfilled if skills and applicant gaps are not addressed.

How many manufacturing jobs are unfilled right now?

The BLS Job Openings and Labor Turnover Survey counted roughly 529,000 open manufacturing jobs in May 2026. The number moves month to month, but manufacturing openings have run in the hundreds of thousands for years, the gap is a present operating condition, not just a projection.

What causes the manufacturing skills gap?

Four compounding drivers: retirements removing experienced workers and their undocumented knowledge, fewer new entrants than openings, a shift in the skills roles require toward technical and digital capability, and early attrition of new hires before they reach competence. Headcount and skills are related but distinct problems.

Is the skills gap just a hiring problem?

No. Even when a body can be hired, the skills half of the gap remains: new hires arrive with less experience into roles demanding more. That is why training speed, knowledge capture, and cross-training are as much a part of closing the gap as recruiting, and they are fully within a plant's control.

What can manufacturers do about the skills gap?

Locally: build a skills matrix to expose single points of failure, capture expert knowledge before retirements, run structured operator training with defined qualification levels, cross-train against the matrix's gaps, and fix onboarding and engagement leaks that cause early attrition. None of these require solving national demographics.

Will automation and AI close the manufacturing skills gap?

Partially, and indirectly. Automation removes some tasks, but the larger near-term effect is lowering the experience required per task, guided digital workflows, searchable plant knowledge, and clear point-of-use instructions let less-experienced operators perform at a higher level sooner. Technology shortens the ramp; it does not eliminate the need for people.

The Manufacturing Skills Gap: What the Data Actually Says

3.8 million jobs, half of them at risk of going unfilled, and the part of the problem a single plant can actually do something about.

The manufacturing skills gap is the shortfall between the skills manufacturing jobs require and the workforce available to fill them. The most cited projection, from Deloitte and The Manufacturing Institute, estimates US manufacturing may need as many as 3.8 million new employees from 2024 to 2033, and as many as 1.9 million of those roles could go unfilled.

Those numbers get quoted a lot, and often loosely. This post lays out what the primary sources actually say, what is driving the gap, and, the part most coverage skips, which pieces of the problem a single plant can do something about this quarter. Because "the industry is short 1.9 million people" is not an action plan. "Our changeover knowledge lives in two heads" is.

What Do the Numbers Actually Say?

The headline figures come from two primary sources worth reading directly rather than through headlines:

3.8 million needed, up to 1.9 million unfilled. Deloitte and The Manufacturing Institute's 2024 talent study, their sixth in over two decades, projects as many as 3.8 million additional manufacturing employees needed between 2024 and 2033, combining growth and replacement of departing workers. If skills and applicant gaps are not addressed, around half of those skilled openings, roughly 1.9 million, could go unfilled (The Manufacturing Institute; Deloitte Insights).
~529,000 open jobs right now. The Bureau of Labor Statistics' Job Openings and Labor Turnover Survey counted about 529,000 open manufacturing positions in May 2026 (BLS JOLTS). The gap is not waiting for 2033.
It is the top business concern. In the National Association of Manufacturers' Q1 2024 outlook survey cited in the same study, 65% of manufacturers named attracting and retaining talent as their primary business challenge, ahead of everything else on the list.

Read carefully, the projection is two problems braided together: a headcount problem (more roles than applicants, driven heavily by retirements) and a skills problem (the open roles increasingly demand technical and digital skills the applicant pool has not built yet). They need different responses, which is why lumping them as one "gap" produces vague strategies.

The gap is four problems compounding: exits, entrants, skill requirements, and churn. A plant controls the bottom two more than it thinks.

Why Is the Gap Getting Worse?

Because the exits and the entries are mismatched in both number and knowledge. Manufacturing's experienced cohort is retiring through the 2020s, and each departure removes decades of accumulated, mostly undocumented know-how, the setup tricks, the failure histories, the "it sounds different before it breaks" judgment this site calls tribal knowledge. Their replacements, when they can be hired at all, arrive with less experience into roles that meanwhile demand more: the same Deloitte/MI research emphasizes rising demand for technical and digital skills alongside the traditional trades.

The result is a knowledge-transfer race run on a shortening track. A retiring machinist's replacement might have months of overlap or none. Whatever is not captured, documented, or trained across before the exit is gone, and gets re-learned later at the cost of scrap, downtime, and near misses.

The shaded gap between the lines is where plants live for the next decade. Knowledge capture pulls the top line's know-how down to the next crew; faster ramp-up raises the bottom line.

What Can a Single Plant Actually Control?

A plant controls four levers that determine how hard the national gap bites locally: how much expert knowledge it captures, how fast new hires reach competence, how many people can cover each critical skill, and how many trained people it keeps. Here is the sequence that turns those levers into a program:

Find your single points of failure. Build a skills matrix: every critical skill, every person, honest proficiency ratings. Any skill with one qualified name next to it is a resignation letter away from being a crisis.
Capture before it walks. Prioritize the retiring and senior experts on critical skills. Record them doing the job, turn it into work instructions and searchable know-how, the full playbook is in our tribal knowledge guide.
Shorten time-to-competence. A structured operator training program with defined qualification levels beats shadowing-by-osmosis. Plants that cut ramp time from months to weeks effectively hire people they never had to recruit.
Cross-train against the matrix. Use the matrix's gaps to set a training queue: deepest coverage need first, not whoever asked most recently.
Fix the leaks. Onboarding quality and day-to-day experience drive early attrition, see the first 90 days and engagement in manufacturing. Losing a hire at month four returns you to the back of the hiring line.
Lower the experience floor of every task. The less a task depends on memory and folklore, the wider the pool of people who can do it safely. Clear instructions at the point of use, guided digital workflows, and searchable plant knowledge make a two-year operator effective in ways that used to take ten.

That last lever is where connected-worker tooling earns its keep: when a newer operator can search the plant's accumulated knowledge in plain English and get cited answers at the station, the way Harmony indexes SOPs, tribal knowledge, and plant data the gap between a veteran and a second-year stops being a wall. The demographics are national; the ramp curve is yours.

What Does the Data Not Say?

Three cautions before the numbers get repeated in a budget meeting. First, the 3.8 million and 1.9 million figures are projections with explicit conditions, "as many as," over a nine-year window, contingent on growth materializing and gaps going unaddressed. They are credible planning numbers, not measurements; quoting them as settled fact invites a correction from anyone who reads the study. Second, the projection is national and your labor market is not. A plant near a community college with a strong industrial program, or in a region with several plants competing for the same maintenance techs, faces a very different local curve than the national average, which is why the plant-level diagnostics above matter more than the headline. Third, an unfilled opening is not always a missing person. Some of the gap is wage and shift-structure mismatch, some is qualification inflation in job postings, and some is genuine skills scarcity. The study is careful about this; secondhand summaries usually are not.

None of that softens the operational reality, half a million open roles and a retiring expert cohort are visible from any plant floor. It just means the useful response is local and specific, not a recitation of national numbers.

What Should You Take From the Data?

Three things. First, the shortage is real, current, and measurable, half a million open roles today, with credible projections it compounds through 2033. Second, it is two gaps, not one: bodies and skills, and the skills half is the one plants can attack internally. Third, the winning local strategy is unglamorous: capture the knowledge you have, train faster than you lose people, and make every task less dependent on scarce experience. The plants that do that will hire from the same thin pool as everyone else, and need less from it. The gap is real; whether it becomes your plant's crisis is still, mostly, a set of decisions you control.