Most claimed labour shortages do not survive examination, and a companion article in this cluster makes that case at length. This one does.
The difference is in how it was measured.
A shortage measured properly
The Semiconductor Industry Association and Oxford Economics published Chipping Away in July 2023, assessing the labour market gap facing the US semiconductor industry.
The projection: the industry's US workforce grows by nearly 115,000 jobs by 2030, from approximately 345,000 today to approximately 460,000. Roughly 67,000 of those new jobs, 58 percent, risk going unfilled at current rates of degree completion.
The reason that number carries more weight than a typical shortage claim is that it is not built from asking employers whether they feel short-staffed. It comes from comparing projected role creation against the rate at which the education system produces people qualified to fill them.
Employer sentiment surveys tell you how employers feel. Pipeline arithmetic tells you how many people exist. Only one of those is evidence of a shortage.
Independent structural evidence points the same way. The US Bureau of Labor Statistics projects employment of semiconductor processing technicians growing about 11 percent between 2024 and 2034, well above the average for all occupations.
And the underlying cause is unusually legible. Fabs are being built because they were funded. A fab takes years to construct and the workforce required to run it does not assemble itself on the same schedule.
The composition, which changes everything
The headline gets quoted. The breakdown almost never does, and it is where the actionable information sits.
Of the roles at risk of going unfilled:
| Share of the gap | Education required | Roughly |
|---|---|---|
| 39% | Technicians, most with a two-year degree or less | ~26,000 roles |
| 35% | Computer scientists and engineers with four-year degrees | ~23,000 roles |
| 26% | Engineers at master's or PhD level | ~17,000 roles |
The largest single block of the shortage requires the least formal education.
That inverts the usual assumption. The public conversation about chip workforce shortages tends to centre on doctoral-level engineers, because that is the most dramatic version of the problem and the hardest to solve. It is also the smallest of the three.
Two consequences follow directly.
The largest block is the fastest to address. A doctorate is a multi-year commitment with a narrow intake. A technician capable of working in a fab environment can be trained on a timescale of months from an adjacent starting point.
The candidate pool for that block is much wider than the industry treats it. People currently working in manufacturing, maintenance, process operations, the military, or any environment involving procedure discipline and clean working practice are considerably closer to fab technician work than their job titles suggest.
Why this gap is different from the usual kind
Four features distinguish it from an ordinary hiring difficulty.
The capital is already committed. The plants exist or are being built. The cost of the vacancy is not an abstraction, it is capital equipment sitting idle or running below capacity.
The skills are genuinely specific. Fab work involves cleanroom discipline, process control, metrology and equipment behaviour that people do not acquire incidentally. This is not a case of employers refusing to train for common skills.
Everyone is hiring at once. Simultaneous capacity expansion across a region means every employer is drawing from the same pool, so poaching redistributes rather than solves. The industry-level number does not move.
Geography constrains it. Fabs are where fabs are. The labour market is local, and a national surplus does not help a specific site.
The fourth point is the one that makes training-the-existing-local-workforce the structurally correct answer rather than merely a cheaper one. People already living near the plant are the supply.
The chips and quantum domain
Astra Trainer's semiconductor domain runs nine directions covering the full hardware stack: electronics engineering, semiconductor physics and devices, chip design and VLSI, semiconductor manufacturing and advanced packaging, embedded systems and computer hardware, photonics and optoelectronics, RF and wireless, quantum computing, and quantum communication and sensing.
Each is a program running from fundamentals to job-ready work rather than a course list, sequenced with the partner's own engineers and mapped onto the roles they are hiring for. Lessons are five minutes, so people in an existing production role train without leaving it. You can see the nine directions here.
Where technicians actually come from
The single most useful thing a semiconductor workforce plan can do is widen the definition of a candidate.
The traits that predict success in fab technician work are procedure discipline, comfort with measurement and data, tolerance for repetitive precision, methodical fault-finding, and willingness to work shifts in a controlled environment.
People with those traits are currently working in food and pharmaceutical manufacturing, where cleanroom and process discipline are already routine. In aerospace and automotive production. In maintenance and calibration roles. In the military, particularly in technical trades. In laboratory support roles. And in medical device manufacturing, which shares much of the regulatory and cleanliness culture.
What each of those groups lacks is the domain layer: how a wafer moves through a process, what lithography and etch and deposition actually do, what yield means and what moves it, why a particular metrology reading matters.
That layer is teachable. It is exactly the kind of structured, cumulative, vocabulary-and-model knowledge that a sequenced program delivers well, and it is the difference between someone who can follow a work instruction and someone who can tell when the process is drifting.
The roles behind the numbers
Worth naming, because workforce plans written in the abstract tend to miss whole categories.
Process technicians, running and monitoring the equipment that performs each fab step. The largest technician population.
Equipment maintenance technicians, keeping extremely complex tools running. Chronically short and expensive to replace, since much of the capability is tool-specific.
Metrology and inspection, measuring what the process produced and interpreting it.
Process engineers, owning a step and the variables that move its yield.
Packaging and assembly, an area growing in importance as advanced packaging becomes a performance lever in its own right rather than a back-end afterthought.
Facilities, the least discussed and genuinely critical. Ultrapure water, gases, chemical delivery, air handling. A fab is a chemical plant wrapped around a cleanroom and it needs people who understand that.
What a pipeline looks like on a realistic timescale
Now to six months. Widen the candidate definition and start converting adjacent workers. Build the domain layer through structured training while people remain in their current roles, then move them. This is the only lever that produces people inside a year.
Six to eighteen months. Deepen. Technicians who understand process behaviour rather than only procedure become the people who catch excursions early, and that is where the value sits. Start the internal progression into process engineering for the ones who show the aptitude.
Eighteen months and beyond. The four-year and postgraduate layers, which need education-system partnerships and cannot be accelerated much. Worth starting now precisely because they are slow.
Running all three concurrently is the point. Organisations that begin with the postgraduate layer, because it feels like the serious problem, spend three years before anyone arrives while the technician gap, which is larger, stays open the whole time.
On credentials and safety. Fab work involves hazardous chemicals, high-voltage equipment and processes with real safety consequences. Structured training builds the knowledge that makes someone effective and safe on the floor, and it sits alongside mandated safety certification and supervised qualification on specific equipment rather than replacing either. Any workforce plan treating a training program as equivalent to tool qualification is wrong in a way that will surface badly.
Three mistakes that keep getting made
Recruiting from each other. At regional scale this moves people between employers at rising cost and does not create any. It feels like progress on a site level and does nothing at industry level.
Starting with the smallest block. The doctoral gap is the most discussed and the least tractable. Twenty-six percent of the problem, multi-year lead time. The technician layer is 39 percent and moves in months.
Ignoring retention. A pipeline that fills roles while experienced people leave is a treadmill. The AGC construction survey found 48 percent of firms reporting new hires failing to show up or quickly quitting, which is the same pattern in a different industry: the intake is not the only leak.
What to take from this
This shortage is real and the evidence is structural rather than sentimental, which is rare.
Thirty-nine percent of the gap is technicians needing a two-year degree or less. That is the biggest block, the fastest to close, and usually the last to be addressed.
The candidate pool is much wider than the industry's job descriptions imply. Food manufacturing, pharmaceutical production, maintenance, military technical trades and medical devices all produce people with the right underlying traits.
The missing piece is the domain layer, and that is teachable on a timescale of months while people stay in their current roles.
And because fabs are geographically fixed, the local existing workforce is not merely the cheaper source of people. It is the only source that scales.
How big is the semiconductor workforce shortage?
The SIA and Oxford Economics projected about 115,000 new US semiconductor jobs by 2030, with roughly 67,000, or 58 percent, at risk of going unfilled at current degree completion rates.
Is the gap mostly PhD engineers?
No. Twenty-six percent of the at-risk roles need a master's or doctorate. Thirty-nine percent are technicians with a two-year degree or less, which is the largest single block and the fastest to address.
Who makes a good fab technician?
People with procedure discipline, comfort with measurement, methodical fault-finding and tolerance for precise repetitive work. That describes many people currently in food and pharmaceutical manufacturing, maintenance, military technical trades, laboratory support and medical device production.
Can training replace a degree for these roles?
For technician positions, structured domain training on top of relevant experience is a recognised route and the industry's own analysis places most of those roles at two-year education or less. It does not replace mandated safety certification or supervised qualification on specific equipment.
Where do programs for this start?
Usually with one direction and a small cohort drawn from adjacent roles. Astra Trainer's semiconductor domain covers nine directions from device physics to fab and packaging, and you can see them here.