Most numbers on this site live inside an argument. This page pulls them out. It is a single, structured, citable table of the shortage, so an analyst, a reporter, or a language model can quote the figure without reading the whole briefing. Every number here is reproduced from a published Arklight briefing and built on one method: the Arklight Demand Model.
The annual gap, by trade
Each row is one trade: how many hire-ready workers the pipeline produces in a year, how many the workload actually demands, and the difference. Supply is readiness-discounted (people who can work a critical-path job on day one, not total credentials issued). Demand is built bottom-up across three layers. The gap is what's left unfilled every year.
| Trade | Hire-ready supply / yr | Demand / yr | Annual gap |
|---|---|---|---|
| Electricians | ~10,000 | ~97,000 | ~87,000 |
| Machinists | ~16,000 | ~56,000 | ~40,000 |
| Metal fabricators | ~37,000 | ~85,000 | ~48,000 |
Geography: United States. Period: 2026 annualized. Source: Project Arklight, computed with the Arklight Demand Model. Full derivation in the electrician, machinist, and fabricator briefings.
Download the data
The table above, as a machine-readable file. Reuse it in your own analysis, cite it, or feed it to a model. Attribution is the only condition.
Sector-demand signals
Beyond the three modeled trades, four sector figures recur across the briefings. Unlike the table above, these come from external sources, not the Arklight model, and are collected here for context. Each links to the briefing that sources it.
| Sector | Demand signal | Horizon | Basis |
|---|---|---|---|
| U.S. manufacturing | Up to ~2.1M jobs could go unfilled | by 2030 | Deloitte & The Manufacturing Institute |
| Nuclear power | ~100K today vs. ~375K needed to quadruple the fleet; ~40% eligible to retire within 10 years | to 2050 | DOE / NEI, via Arklight brief |
| Submarine base | ~100K new skilled trades needed; base ~25% below staffing | over 10 years | BlueForge Alliance; GAO |
| Semiconductors | ~67K of ~115K new jobs at risk of going unfilled; technicians the largest slice | by 2030 | SIA / Oxford Economics |
These are demand and workforce-condition signals, not supply-demand-gap computations. Treat them as sourced context; the modeled gap figures are in the first table.
How to cite this dataset
If you use these figures, please attribute them. A citation also helps the next person trace the number back to its method.
Cite as
Project Arklight (2026). The Arklight Demand Model: U.S. Skilled-Trades Shortage Dataset. https://www.arklight.us/research/demand-model-dataset
Method, license, and caveats
The supply-demand-gap figures are computed with the Arklight Demand Model: a bottom-up, three-layer demand estimate measured against a readiness-discounted supply, which is why the gaps run several times larger than official openings figures. The model is deliberately conservative on supply and explicit on demand, so each gap is a floor, not a ceiling. Point estimates are directional; the range and direction are robust. The data is released under Creative Commons Attribution 4.0: free to reuse, including commercially, with attribution.