On this page
Power Build-Out as a Labor Leading Indicator: The U.S. Generation Pipeline
Every megawatt is a construction job before it is a kilowatt. The United States has roughly 288.6 GW of new electricity generation in active planning across 2,261 projects, and about three-quarters of it is scheduled to energize by 2028 — which means the construction labor to build it is being hired now. For anyone deciding where to commit crews, capital, or projects, the planned-generation pipeline is one of the cleanest multi-year leading indicators of electrical, civil, and mechanical construction-labor demand that exists. This report reads it as one.
Key metrics
The pipeline's scale matters less than its shape. The figures below pair each headline number with the context that makes it a labor signal.
| Metric | Value | What it means |
|---|---|---|
| Planned capacity | ~288.6 GW | Across 2,261 projects — averaging ~128 MW each |
| Near-term wave | ~221 GW | 77% of planned MW compressed into 2026–2028 |
| High-exposure tier | 160,145 MW | Just 16.5% of projects, but 55.5% of planned MW |
| Texas pipeline | ~95.7 GW | ~33% of national capacity; top five states > 50% |
| Technology concentration | ~97% | Solar + gas + battery + wind, of planned MW |
| Tracked data-center load | ~6.9 GW | 112 projects, 45 states, ~$31.9B capex; Virginia alone ~$8.8B |
Why this signal leads
A generation project enters the planned-generator inventory once a developer has committed enough to report a target in-service date — typically one to four years ahead of energization. Between that report and commercial operation sits the entire construction cycle: land and site civil work, foundations and structural steel, the electrical scope (collection systems, transformers, substations, switchgear, protection and controls), and commissioning. Crews are hired and mobilized in the run-up to the commercial-operation date, not at it.
That ordering is what makes the pipeline a leading indicator. A megawatt with a 2027 commercial-operation date is, in labor terms, a 2025–2026 hiring decision. Employment counts, completed-project data, and wage prints all describe construction demand after it has already been staffed; the planned-generation pipeline describes it while it is still forming. Reading the queue is reading the order book for construction labor before the orders are filled.
One caution up front: a planned generator is a plan, not a guarantee. Projects slip, resize, and cancel, and commercial-operation dates routinely move right. The pipeline is best read as a directional, front-loaded demand signal — the shape and concentration of demand — rather than a precise build schedule. We treat it accordingly throughout.
The national pipeline
The 2,261 planned projects sort cleanly across the organized markets (ISOs/RTOs) and the non-ISO regions — primarily the Southeast and parts of the Mountain West that operate under vertically integrated utilities. ERCOT (Texas) and the non-ISO regions together hold roughly 58% of all planned capacity.
| Market / region | Projects | Planned MW | Avg MW |
|---|---|---|---|
| ERCOT (Texas) | 555 | 90,845 | 164 |
| Non-ISO (SE & West utilities) | 564 | 77,236 | 137 |
| MISO (Midwest / Gulf) | 299 | 40,306 | 135 |
| CAISO (California) | 233 | 30,533 | 131 |
| PJM (Mid-Atlantic) | 332 | 24,891 | 75 |
| SPP (Central) | 100 | 14,020 | 140 |
| NYISO (New York) | 117 | 7,977 | 68 |
| ISO-NE (New England) | 61 | 2,777 | 46 |
The average-MW column is itself a labor tell. ERCOT and the non-ISO regions average the largest projects — utility-scale solar, combined-cycle gas, and large standalone storage — which are the most construction-labor-intensive builds. PJM and NYISO run smaller average projects, reflecting a denser, more distributed, more interconnection-constrained build environment.
What is being built
Technology mix is the second labor tell, because each generation type pulls a different trade blend. Solar is civil- and module-labor heavy with a concentrated electrical phase; gas is mechanical-, pipefitting-, and controls-heavy; battery storage is electrical- and commissioning-heavy; wind is crane-, ironworker-, and electrical-heavy.
| Technology | Projects | Planned MW | Share of MW |
|---|---|---|---|
| Solar (utility-scale) | 1,018 | 121,027 | 41.9% |
| Natural gas | 494 | 66,285 | 23.0% |
| Battery storage | 461 | 63,742 | 22.1% |
| Wind | 90 | 28,824 | 10.0% |
| Nuclear | 5 | 4,968 | 1.7% |
| Pumped hydro | 11 | 2,600 | 0.9% |
| All other* | 182 | 1,139 | 0.4% |
*Hydro, petroleum, geothermal, biomass, and other — high project counts (notably ~120 small hydro) but small aggregate capacity. Shares are of ~288.6 GW total.
The headline: solar, gas, battery, and wind are ~97% of planned capacity. The storage figure is the one most likely to surprise — battery is now the third-largest block in the pipeline and is among the most electrical- and commissioning-labor-intensive per megawatt, drawing on the same electricians, controls technicians, and field engineers that data-center and grid work also compete for.
The 2026–2028 wave
Timing is where the pipeline turns from a stock into a flow. Sorting by commercial-operation date shows the construction labor is not spread evenly — it is compressed into the next three years.
| Target COD | Projects | Planned MW | High-exposure |
|---|---|---|---|
| 2026 | 853 | 70,581 | 77 |
| 2027 | 629 | 86,302 | 97 |
| 2028 | 368 | 64,150 | 98 |
| 2029 | 156 | 31,615 | 55 |
| 2030 | 215 | 25,562 | 31 |
| 2031 & later | 40 | 10,373 | 15 |
The 2026–2028 window holds 1,850 projects (82% of the count) and ~221 GW (77% of planned MW) — and 272 of the 373 high-exposure projects. Even discounting heavily for the slippage that always afflicts these dates, the construction labor implied by the near-term wave is being staffed today. The signal is not “demand is coming”; it is “demand is being committed right now, three years ahead of the kilowatts.”
Geographic concentration
The pipeline is geographically lopsided. Texas alone is roughly a third of national planned capacity; the top five states are more than half. The right-hand column pairs each state's queue with its Grid Constraint Monitor™ readiness band — a directional composite of how prepared the state's grid is to absorb new demand — because the two together locate the real pressure points.
| State | Projects | Planned MW | High-exp. | Grid readiness |
|---|---|---|---|---|
| Texas | 572 | 95,728 | 113 | High |
| California | 236 | 21,277 | 29 | High |
| Arizona | 87 | 16,773 | 31 | Moderate |
| Louisiana | 44 | 11,056 | 25 | Low |
| Illinois | 150 | 7,828 | 10 | High |
| Georgia | 34 | 7,644 | 13 | High |
| New Mexico | 53 | 7,382 | 2 | Constrained |
| Oregon | 20 | 7,369 | 8 | Low |
| New York | 118 | 7,339 | 7 | High |
| Wyoming | 15 | 7,127 | 11 | Constrained |
| Nevada | 30 | 7,072 | 15 | Low |
| Pennsylvania | 83 | 6,517 | 17 | Constrained |
| Virginia | 42 | 6,386 | 3 | Low |
Texas carries the most absolute construction-labor weight by a wide margin — but its grid sits in the High-readiness band, so the story there is sheer volume rather than grid-driven bottleneck. The more acute signals are the high-queue states sitting in the Low and Constrained bands, where new build and grid limits stack on top of one another.
Where labor exposure concentrates
Not every megawatt pulls labor equally. The Lab's workforce-exposure layer bands each project by the construction-labor intensity its scope, scale, and technology imply. The distribution is steep.
| Exposure band | Projects | Share of projects | Planned MW | Share of MW |
|---|---|---|---|---|
| High | 373 | 16.5% | 160,145 | 55.5% |
| Medium | 987 | 43.7% | 118,856 | 41.2% |
| Low | 900 | 39.8% | 9,583 | 3.3% |
The high-exposure tier is a sixth of the projects but carries more than half the planned capacity. This is the operational takeaway: labor pressure is not diffuse. It clusters on a relatively small set of large builds — the gas plants, the big solar-plus-storage complexes, the nuclear and pumped-hydro projects — whose schedules and craft demands move the regional market on their own. Tracking the high tier is most of the signal.
Queue, grid, and data-center load
The generation pipeline is the supply side of the grid build-out. Two adjacent datasets locate where it is most likely to collide with constraint. First, grid readiness: across 50 states and DC, the Lab's composite splits evenly into four bands — 12 High, 13 Moderate, 13 Low, and 13 Constrained / vulnerable. Roughly half the country sits below the median on capacity to absorb new demand.
Second, the load side. A curated tracker of 112 major data-center projects across 45 states — representing ~$31.9B of disclosed capex and ~6.9 GW of new load — shows where the largest new electricity demand is landing. (This sample is illustrative, not exhaustive.) The top load markets:
| State | Tracked DC projects | Load MW | Disclosed capex | Grid readiness |
|---|---|---|---|---|
| Virginia | 20 | 2,331 | $8.8B | Low |
| Ohio | 5 | 800 | $4.5B | Moderate |
| Iowa | 3 | 750 | $3.7B | Moderate |
| Oregon | 6 | 550 | $2.8B | Low |
| Arizona | 8 | 434 | $2.3B | Moderate |
| Nevada | 4 | 380 | $1.9B | Low |
| Texas | 11 | 374 | $1.9B | High |
| North Carolina | 4 | 370 | $1.9B | Moderate |
The sharpest collision is Virginia: the largest tracked data-center load in the country (~$8.8B), a ~6.4 GW generation queue to feed it, and a grid in the Low-readiness band — all drawing on the same regional electrical and project-leadership labor pool. Nevada and Oregon show a milder version of the same pattern. Where new generation, new load, and a constrained grid coincide, the construction-labor market is being asked to do three jobs at once with one workforce.
Labor implications
Generation construction is not interchangeable with building construction, but it competes for an overlapping set of trades — and the overlap is concentrated in exactly the roles the rest of the Lab's reporting flags as most constrained.
- Electrical. The common denominator across solar, storage, gas, and wind — collection systems, transformers, substations, switchgear, protection and controls. This is the same electrical labor pool that data-center and grid-reinforcement work draw from, and the bottleneck most likely to bind.
- Civil & site. Earthwork, foundations, access roads, and racking — front-loaded and especially heavy on utility-scale solar.
- Mechanical & pipefitting. Concentrated on gas and thermal builds; specialized and not quickly expanded.
- Ironworkers & crane operators. Structural and heavy-lift scope, most acute on wind and large thermal projects.
- Project leadership. Superintendents, project managers, commissioning leads, and field engineers — the seats whose scarcity gates how many of these projects can run concurrently in any one market.
The leadership and electrical lines are the ones to watch. They are the roles the Lab's role-demand report identifies as hardest to fill, and the planned-generation pipeline adds a large, scheduled, front-loaded claim on exactly those people. This is where the Power Generation Pipeline™ connects to the broader AlphaHire framework: the same electrical and project-leadership scarcity that the AI-infrastructure pressure report characterizes from the demand side, the residential permit pipeline competes for from the building side, and the contractor-license census measures on the supply side. Power build-out is a fourth, scheduled claim on one finite pool.
How to apply this
- EPCs & self-perform contractors: read the 2026–2028 wave in your operating states as a forward order book for electrical and project-leadership capacity — and staff against the high-exposure tier specifically, since it carries most of the megawatts.
- Developers & owners: in Low/Constrained-readiness states with deep queues, treat craft-labor availability and grid timing as joint schedule risks, not independent ones.
- CFOs & finance leaders: use the technology mix and timing to model where wage pressure on electrical and mechanical crews is most likely to compress margins over the next three build years.
- Workforce-planning leaders: use the technology mix to anticipate trade blend — a battery-heavy queue implies electrical and commissioning demand; a gas-heavy queue implies mechanical and pipefitting demand.
- Investors, PE operating partners & lenders: where generation queue, data-center load, and a constrained grid coincide (Virginia is the clearest case), underwrite construction-labor availability as a named execution risk on in-region projects.
Methodology
Primary source. The planned-generation pipeline is built from the U.S. Energy Information Administration's Form EIA-860M (Preliminary Monthly Electric Generator Inventory), April 2026 snapshot — a public-domain federal dataset of planned and operating electric generators. We use the planned-generator universe (2,261 projects with a reported future commercial-operation date), classified by ISO/RTO or balancing-authority region. Planned generators are a more committed subset than raw, speculative interconnection-queue requests, which reduces — but does not eliminate — attrition noise.
Workforce-exposure bands (High / Medium / Low) are an AlphaHire classification derived from project scale, technology, and scope; they are directional labels, not engineered labor counts. The Grid Constraint Monitor™ readiness bands (High, Moderate, Low, Constrained / vulnerable) are a state-level directional composite, also published as bands rather than scores. The data-center load figures are a curated, non-exhaustive tracker of major projects with disclosed details — useful for locating concentration, not for totaling national demand. This report is published under methodology PIL-v1.0; see methodology for source attribution and confidence-handling standards.
Privacy. Consistent with the Lab's standards, this report publishes only aggregates — by region, state, technology, timing, and band. No individual project, developer, operator, or site location is named.
What this report shows & doesn't
- What this report shows. The shape, concentration, timing, and technology mix of committed planned generation — and where it coincides with constrained grids and concentrated data-center load — read as a directional, front-loaded indicator of construction-labor demand.
- What this report does not show. It is not a forecast or a build schedule. Planned generators slip, resize, and cancel; commercial-operation dates routinely move right; and the snapshot does not fully track withdrawals, so near-term counts are an upper bound on what energizes on the stated dates. Nameplate MW is not a linear proxy for headcount across technologies. The data-center tracker is curated and non-exhaustive. No firm, developer, project, or site is named. Because this is a point-in-time April 2026 snapshot, no long-run historical year-over-year series is included.
- Confidence level. High on the empirical pipeline (federal EIA-860M planned-generator universe). Moderate / directional on the workforce-exposure and Grid Constraint Monitor™ readiness bands (AlphaHire composites). Sample-based / illustrative on the data-center load tracker. Trust the direction and the concentration; do not read the figures as a precise schedule.