The Home EV-Charging Gap: 3.6 Million EVs, 20.7 Million Homes That Cannot Plug In

The join nobody had run

Public-charging maps are everywhere: every major charging network, several state energy offices, and AFDC itself publish maps of where stations sit. A residential installation-demand map is not — because it requires two datasets that live in different agencies and were built for different purposes. The Alternative Fuels Data Center, run by DOE with NREL, counts registered vehicles. The Census Bureau, in a table built for housing and urban-planning research, counts who rents in what size of building. Put together, they answer a question neither one was designed to answer: how big is the population that, independent of anything to do with electric vehicles at all, cannot host a home charger?

This site’s renter-electrification-gap study asked a related but different question — how many renters live in homes old enough to carry a hazard-era electrical panel. That study used construction era (table B25036) as its axis, because the hazard it measures is about panel vintage. This page uses building size (table B25032) as its axis, because the constraint it measures is about whether a dedicated parking space with rewireable electrical service exists at all. The two studies share a subject — renters — and answer genuinely different questions from genuinely different Census tables.

The reason this join is worth publishing rather than filing away is who reads it. State energy offices, utilities sizing make-ready programmes and EV-charging trade press already use AFDC’s registration counts on their own; none we found publishes them next to a housing-stock constraint. A figure a utility planner or a state programme office can cite directly — how much of a state’s rental stock is structurally excluded from home charging, regardless of what happens to EV prices — is what this join exists to put in front of them.

Where the two counts agree, and where they clearly do not

HYRE calculation. Across all 51 jurisdictions, EV intensity and the locked-out-renter share move together only weakly — Pearson r = 0.37 — which is itself the honest finding: a state’s housing stock is not a strong predictor of its current EV adoption, because adoption today is driven far more by state purchase incentives, charging-network build-out and vehicle availability than by how many of its renters live in large buildings.

California sits high on both measures — 93.5 EVs per 1,000 homes and 48.2% of its renters in 5-plus-unit buildings — because both its EV market and its dense coastal rental stock are simply the largest in the country. North Dakota is the sharpest counter-example: lowest EV intensity in the nation at 3.0 per 1,000 homes, yet its renters have the second-highest locked-out share of any state, 62.2% — a state with almost no current EV demand and a rental stock unusually concentrated in large buildings for reasons this dataset cannot determine. Vermont runs the other way: a top-15 EV-intensity state (29.0 per 1,000 homes) with one of the lowest locked-out shares in the country (37.3%) — a rural, largely single-family housing stock that puts comparatively few of its renters behind this specific barrier.

A limitation worth stating plainly: EV registrations require car ownership, which this page does not measure. New York ranks only 22th of 51 on EV intensity despite having the fourth-highest locked-out renter share (62.3%), driven substantially by New York City, where overall car ownership runs well below the national norm for reasons that have nothing to do with charging access. A low EV-per-1,000-homes figure in a dense urban state understates residential charging-access need rather than measuring it accurately, because many of its households were never going to register a car of any kind.

Method

Sources. DOE Alternative Fuels Data Center and NREL, “Electric Vehicle Registrations by State,” data as of 31 December 2023, workbook last updated September 2024, derived by NREL from Experian Information Solutions data; all-electric vehicles only. US Census Bureau, American Community Survey 2023 5-year estimates, table B25032 (Tenure by Units in Structure), for the United States and the 50 states and the District of Columbia. Both retrieved 6 September 2026.

Retrieval route. The AFDC workbook was downloaded directly from its published URL. An unauthenticated request to api.census.gov returned a missing-key redirect, so ACS figures were taken from the public table-based Summary File at www2.census.gov — the same substitution this site’s other Census-based studies record, changing no figure.

Derivation. Renter-occupied units in structures of 5-plus units sum table lines 018-021 (5-9, 10-19, 20-49, 50-or-more units); the 2-plus width adds lines 015-016 (2 units, 3-4 units); the 20-plus width uses only lines 020-021. Margins of error are the square root of the sum of the squared component margins — the Census Bureau’s published method for derived sums — at 90 per cent confidence. EVs per 1,000 occupied homes divides each state’s registration count by its total occupied housing units (table line 001) and multiplies by 1,000. The Pearson correlation coefficient is computed across all 51 rows from the state EV-intensity and locked-out-share columns in the table above.

What this page reuses, and what it does not. This is the first HyreElectrical study to use table B25032. The renter-electrification-gap study uses table B25036 (tenure by year built) for a construction-era question and is not the source of any figure here.

Reproducibility. Both inputs are public files at the URLs in the sources list below. Anyone with the same two downloads can rebuild every number on this page.