SpaceX filed a public notice on May 6 to build a "vertically integrated semiconductor manufacturing and advanced computing fabrication facility" in Grimes County, Texas. Initial phase: $55 billion. Full build-out: up to $119 billion. It would be one of the largest manufacturing investments in U.S. history. The press is covering the dollar figure. The harder problem is that the workforce to build and run it does not exist in the surrounding three counties.
Grimes County's population is about 30,000. Bryan-College Station, twenty miles west, adds another 280,000. That's the labor shed. A fab of this scale needs roughly 4,000 to 6,000 permanent operations staff once it's running, plus a construction workforce that will peak somewhere north of 10,000. TSMC Arizona — a much smaller fab at $40B — has been publicly fighting to staff out for three years, importing Taiwanese engineers under L-1B visas because Phoenix doesn't have the people. Phoenix is a metro of five million. Grimes County is a metro of thirty thousand.
The headline number — "thousands of high-wage jobs" — sounds like an unambiguous win for the region. It is, eventually. But the eighteen-month problem is this: every skilled trade in the Brazos Valley is already at capacity. Industrial electricians, instrumentation techs, pipefitters, millwrights, controls engineers, HVAC commissioning crews — these are the people who actually build a fab. They are not sitting on benches in Navasota waiting for SpaceX to call. They are already booked on petrochemical turnarounds in Beaumont, on the Samsung expansion in Taylor, and on the Stargate datacenter buildouts in Abilene.
When a project of this magnitude lands in a small labor shed, three things happen in order. First, wages spike — typically 30 to 50 percent above market for the skilled trades within commuting distance. Second, existing employers in the region lose people. Comal Services has seen this play out in central Texas every time a giga-project lands within a 90-minute drive. The local industrial plant that was paying $38 an hour for a senior controls tech is suddenly competing with a fab paying $58 and offering a per-diem on top. Third — and this is the part nobody plans for — the construction quality suffers, because the pipeline gets backfilled with crews flown in from out of state who don't know the local code, the local inspector, or each other.
"The single largest constraint on U.S. semiconductor expansion is not capital, it's craft labor."
— SIA workforce report, 2024
Texas has been preparing for this — sort of. Texas A&M is twenty miles from the site and has been quietly stacking up its semiconductor curriculum since the CHIPS Act passed. Blinn College in Brenham added a semiconductor manufacturing technician track in 2024. Sam Houston State runs an industrial technology program. These programs together produce maybe 400 people a year qualified to walk into a fab. Terafab will need that many in its first quarter of operations.
What this actually means, if you're running a Texas-based industrial or technical hiring function right now:
- The labor cost curve for skilled trades within a 90-minute drive of Grimes County is going to bend hard upward starting in late Q3. Bake that into 2027 retention budgets now, not when it shows up in exit interviews.
- Anybody you have today with semiconductor cleanroom, instrumentation, or controls experience is going to get a call from a Terafab-affiliated recruiter within twelve months. Have your retention conversation before they do.
- The interesting recruiting opportunity is not Terafab itself. SpaceX will use their own contingent vendors. The opportunity is the second-order pull — the plants Terafab drains, who will need to backfill at speed, and who don't have a recruiting partner that knows the Brazos Valley.
- Tax abatements get voted on June 3 in Grimes County. If you're a regional employer, that hearing is your earliest signal on how aggressively the local government will subsidize the labor competition you're about to face.
The mistake I see employers make when a project of this scale lands nearby is to assume it doesn't affect them because they're not bidding. It always affects them. It just takes about two quarters to show up in turnover numbers, and by then it's already priced in.
Capital builds the fab. People run it. Whoever figures out the people problem first in central Texas over the next eighteen months is going to be in a very different competitive position by the time the first wafer ships.