Americans purchase a nearly uncountable amount of circuit boards, but almost none of them are made here anymore. The fabs that survived offshoring stayed bunched in a few California and Oregon metros, hundreds of miles from where the demand sits.
I wanted to see where that gap is widest. So I took the 365k U.S. manufacturer directory I cleaned up last week, pulled out the firms that buy or make boards by name, and scored every county four ways. The default view shows the gap itself: darker counties have more unmet local demand.
Two maps that don't overlap#
Board demand follows what people are building. Medical devices in Minnesota. Auto electronics in Michigan. Aerospace in Washington and the Mountain West. Instruments and machinery across the Midwest.
Board fabrication doesn't. It stayed where the 1980s left it - California and Oregon. The gap between those two maps is where a new domestic operation would have customers waiting.
County demand÷supply#
This is the gap itself. For each county I measure two things within ~150 km - board demand and fab supply - and take the ratio. Dark counties have demand the local fabs can't meet. Light counties are either thin on demand (the Mountain West, the Plains) or already balanced by nearby supply.
The darkest band runs through the industrial Midwest: Indiana, Ohio, Illinois, Michigan. The accent markers pick out the metros with the biggest gaps - Chicago, Detroit, Atlanta, New York, Minneapolis - with smaller spikes around Denver, Salt Lake, Phoenix, and the Southeast.
Build-quality factors#
This layer ignores demand and asks where a fab could run well. Three inputs, weighted equally: industrial electricity price (EIA 2023 - fabs are hungry), nearby engineering talent (Carnegie 2025 R1 universities), and the existing electronics-supplier base.
The Ohio belt stands out on power. Columbus, Cleveland, and Cincinnati all sit around 7¢/kWh - roughly half coastal California's rate. Chicago, New York, and Indianapolis round out the top. Not all of these are top demand-gaps; the interesting question is what's dark on both layers. That overlap is the industrial Midwest.
Combined - bare board vs assembly#
The combined layer multiplies demand÷supply by build-quality. Dark only where you have both. The two buttons rank metros for two different businesses.
Bare-board fab is wet copper-etching: capital-heavy, water-intensive, chemical permitting. It scores best in New York, Chicago, Minneapolis, Baltimore, and Atlanta.
Assembly / EMS is surface-mount work: light regulation, mostly talent and floorspace. Chicago, Minneapolis, Atlanta, Detroit, and Miami lead.
The combined view weights demand and build-quality equally. That's a judgment call, so the three source layers stay separately clickable above.
Methodology#
County demand÷supply. I sum board-consuming demand and PCB/EMS fab supply over a Gaussian kernel (~150 km) around each county centroid. Close firms count fully, distant ones fade. The score is half the local demand magnitude (log) and half the demand-to-supply ratio (log), z-scored and squashed through a logistic so the gradient stays continuous. No per-fab capacity constant.
A light highway tie-breaker (weight 0.12) nudges scores toward counties near the Interstate network, using the U.S. Census TIGER/Line 2023 Primary Roads shapefile. It's a tie-breaker, not a driver - the correlation with the un-tweaked score is 0.99 - and a demand-access proxy, not literal shipping time.
Build-quality. Three inputs, equally weighted: EIA 2023 industrial electricity price (state-level, because utility territories ignore county lines), Carnegie 2025 R1-university talent (county Gaussian, 150 km), and the supplier base from this dataset (county Gaussian, 200 km).
Combined is demand÷supply × build-quality, equally weighted. Metros are official Census CBSAs (2023 crosswalk).
Caveats#
A few things this map cannot do.
Every firm was classified by name, not by anything anyone measured. "Acme PCB Assembly" is easy; "Acme Electronics" is a guess. I dropped ~23% of raw matches that weren't actual manufacturers. Each firm counts once, and the five consuming industries (medical devices, auto electronics, aerospace, instruments, machinery) are weighted equally. "None identified nearby" means no fab name matched within range, not a verified zero. The totals: ~6,700 board-consuming firms, 106 PCB/EMS firms (37 bare-board fab, 69 assembly).
Chicago tops the demand-gap ranking with 295 firms, but the directory over-represents Illinois by roughly 2x. The city's base is real; its #1 placement is partly an artifact. Read Detroit, Atlanta, and Minneapolis as closer peers.
None of this captures water access, permitting, labor markets, or local incentives - the things a real site decision turns on. The map is a starting point for the question, not the answer.
What I'd take from this#
The industrial Midwest is the only place where heavy unmet demand and decent build conditions overlap at scale. Inside that band, the metro you pick depends on what you're building.
For a bare-board fab, New York leads on the combined map; Chicago and Minneapolis follow once you discount the Illinois bias. For an assembly shop, Chicago and Minneapolis lead, then Atlanta and Detroit.
This isn't a recommendation - it's what the map shows after the correction. If anyone opens a new domestic operation in the next five years, it would be in one of these metros.
Co-authored with Claude for post scaffolding and prose extraction.