Chip plants are going up faster than the technicians to run them

TSMC and Intel are turning wafers in Arizona. Samsung is preparing to start Taylor, Texas. Micron is building in Idaho and New York. SK Hynix is building advanced packaging next to Purdue. The CHIPS Act moved the buildings. It did not move the people.

It did not move people. A McKinsey analysis for the SEMI Foundation, circulating among fab HR desks this month, puts the 2030 shortfall at as many as 157,000 semiconductor workers. Only about 3 percent of American engineering graduates choose the chip industry in a given year. The plants will open on a calendar set by construction. They will run on a calendar set by technicians.

157k

Projected U.S. chip-worker gap by 2030

3%

Engineering graduates entering semiconductors

$35bn

Samsung’s Texas commitment, competing for the same pool

What the jobs actually are

A modern fab is vacuum, chemistry and software. The people who keep a $200 million tool in spec on a night shift are process technicians, equipment engineers, yield analysts, and facilities specialists who understand ultra-pure water and the sub-fab. Those job titles barely appear in university course catalogues. Where they appear, it is usually in community colleges that have been underfunded for a generation.

Samsung has been flying Korean commissioning teams to Texas because the local bench is thin. Tool vendors are running their own academies, because waiting for the education system would mean waiting out the decade. SK Hynix’s chief executive called staffing “really the critical issue.” He was not talking about vice-presidents.

You can accelerate a cleanroom. You cannot accelerate a technician who has never seen an EUV tool. The bottleneck has moved from permits to people, and people do not yield to a press release.

Hiring between the plants

Moving people from Intel to TSMC to Micron raises wages. It does not add technicians. The plants that will start on time are treating the workforce as a capital project: paid apprenticeships that begin before the roof goes on, community colleges that teach vacuum and metrology rather than a generic science course, and a willingness to hire people who never went to a four-year school.

A capex plan with no named owner for the workforce is a schedule risk.