Korea's chip industry is running short of the engineers it needs

Government and industry projections point to a shortfall of tens of thousands of semiconductor workers over the coming decade, and the university seats created to close it are competing with medicine for the same students.

Semiconductors are the load-bearing wall of the Korean export economy, accounting for something close to a fifth of outbound shipments in a typical year and rather more of the country’s trade surplus in good ones. The industry’s expansion plans are correspondingly large: new fabrication complexes in Gyeonggi Province, packaging and test capacity, research campuses. What the plans consistently assume, and what the country is least confident of supplying, is people.

The projections have been circulating for several years and vary with methodology, but they point in one direction. In a 2022 assessment, Korea’s industry ministry estimated that the semiconductor sector would need roughly 127,000 additional workers over the following decade and would fall short by somewhere in the region of 56,000. Industry associations have produced comparable figures. All such forecasts are exercises in extrapolation and should be read as orders of magnitude rather than counts, but no serious version of the exercise produces a surplus.

The policy response has been to build a pipeline. A 2022 government plan set a target of training 150,000 semiconductor-related workers over ten years, spanning vocational programmes, undergraduate expansion and doctoral fellowships. More consequentially, the education ministry allowed universities to enlarge enrolment quotas in semiconductor-related departments, a partial exception to the long-standing cap on expansion in the Seoul metropolitan area. Chipmakers underwrote contract departments at institutions including KAIST, Yonsei, Korea University and Sungkyunkwan University, in which students receive tuition support and, in most designs, a guaranteed offer of employment.

The programmes fill. What they do not do is settle the underlying competition for talent, which the chip industry has been losing at the top of the distribution. For two decades the highest-scoring science-track students in Korea’s college entrance examination have concentrated in medicine, dentistry and pharmacy, and the pattern is strong enough that admissions analysts treat medical placement rates as the primary ranking metric for science high schools. Contract semiconductor departments have found that a portion of admitted students withdraw to retake the entrance examination for a medical seat. The government’s 2024 decision to enlarge medical school admissions, whatever its merits for healthcare provision, added capacity to the destination that competes most directly with engineering for the same applicants.

The shortage is also uneven by seniority, and the seniority that matters most is the hardest to manufacture. Undergraduate expansion addresses entry-level hiring; the acute gap reported by firms is in doctoral-level process, device and packaging specialists, the people who diagnose yield problems on a line that costs tens of millions of dollars a day to idle. That expertise accumulates over ten to fifteen years, which is why the same engineers attract offers from abroad — and why Korea has prosecuted a steady run of industrial technology leakage cases involving departing staff, a symptom of scarcity as much as of malice.

Underneath all of this sits arithmetic no training programme can revise. Korea’s total fertility rate was 0.72 in 2023, according to Statistics Korea, the lowest recorded nationally anywhere, and the cohort now entering universities was born when the rate was already below 1.3. Every sector competing for technical graduates is drawing from a pool that shrinks each year, and semiconductors compete not only with medicine but with finance, platform firms and overseas employers.

None of this suggests that Korean chipmaking is about to stall. It does suggest that the binding constraint has shifted from capital to labour, and that the two operate on different clocks. A quota expansion produces graduates in four years. It produces a senior process engineer in fifteen.