Public EV Charging Is a Counted Number. Home Charging Capacity Never Has Been

Why nobody has published this join before

Public charging infrastructure is exhaustively mapped. AFDC updates its station locator continuously, from utility filings, state programme records and network operator data, and publishes exact port counts down to the charging level. It is the single most complete infrastructure dataset the EV transition has produced, because a public charger is a fixed asset with a permit, a meter and an operator who wants it found.

A home charging circuit has none of those. It sits behind a wall, on a residential meter indistinguishable from every other residential load, installed by any of thousands of licensed electricians under a permit that a county records as "electrical work," not as "EV charger." No agency aggregates it, because there is nothing to aggregate from — the data was never generated in a countable form in the first place.

HYRE analysis. That asymmetry has a predictable effect on which number gets reported. A public port count is a press release: a governor announces a NEVI corridor, a network operator announces its network size, a trade group publishes a milestone. A home capacity figure has no announcement to attach to, so it does not get one — not because it matters less, but because nobody who could publish it has a reason to. This page assembles the closest federal substitute available and states plainly, throughout, where it is a count and where it is a proxy.

The public side: 255,551 ports, and where they actually are

DOE’s Alternative Fuels Data Center counted 81,541 public charging station locations and 255,551 individual charging ports nationwide as of 6 September 2026. Level 2 (240-volt) ports account for 179,368 of the total — 70.2% — and DC fast charging 75,472, or 29.5%. Level 1 (120-volt) public ports are a rounding error at 699, because Level 1 is rarely worth deploying commercially when the equipment cost difference to Level 2 is small relative to the site work.

California alone holds 66,704 ports — 26.1% of the national total — followed by New York (21,541), Florida (15,022), Texas (12,538) and Massachusetts (11,699). Those five states hold 49.9% of every public port in the country. AFDC’s scope note states these figures exclude residential charging equipment entirely — this is exclusively the counted, public side of the ledger.

The public build-out runs on a different logic than home capacity

HYRE calculation. Across all 51 jurisdictions, a state’s EV-adoption intensity (registrations per 1,000 occupied homes) and its public-port intensity relative to that adoption (ports per 1,000 registered EVs) move in opposite directions: Pearson r = -0.50. California leads the country on EV adoption at 93.5 per 1,000 homes, yet sits near the bottom on ports-per-EV at just 53.1. North Dakota sits at the opposite end on adoption — fewer than 3 EVs per 1,000 homes — yet tops the ports-per-EV ranking at 291.

HYRE analysis. This is not public charging failing to keep pace with demand — raw port counts and raw EV counts by state correlate strongly (r = 0.97), because both simply scale with population. The ports-per-EV ratio is the more informative number, and it says something different: federal and state corridor-charging money — the National Electric Vehicle Infrastructure (NEVI) formula programme chief among it — is allocated to cover interstate highway miles and geographic gaps, not to track where EV owners already live. A rural state with few EVs and a long interstate corridor gets stations built for through-traffic and future adoption; a dense state with many EVs has already absorbed a large fleet against a public network sized for a smaller one. Both are defensible policy choices. Neither is a measurement of home-charging capacity, which is the point: the two sides of this ledger are being built by different logics entirely.

The home side: the only federal proxy for charging capacity

EIA’s Residential Energy Consumption Survey (RECS) asks a national sample of homes a single question relevant here: does this home have an attached garage. It asks it only of single-family detached and single-family attached homes — 84.52 million of the 123.53 million occupied homes RECS counted in 2020 — and explicitly does not ask it of apartments or mobile homes, a combined 39.01 million homes RECS labels "not asked" rather than "no."

Of the 84.52 million single-family homes asked, 51.79 million report an attached garage — 34.18 million of them a 2-car garage, 11.27 million 1-car, and 6.34 million 3-or-more. 32.72 million single-family homes report none. A home without an attached garage is not necessarily unable to host a charger — a driveway or a detached garage can still work, and RECS does not ask about either — but it is a home for which the wiring path is materially harder to establish from survey data alone, which is exactly why this page reports the number as "uncertain," not "excluded."

Where charging actually happens, by DOE’s own projection

Source fact, reused from this site’s renter-lockout study rather than re-derived. DOE’s Vehicle Technologies Office, citing NREL’s June 2023 report The 2030 National Charging Network, states that "the majority (64%) of EV charging is estimated to take place at single family homes using L1 and L2 charging," with Level 1 and Level 2 combined — home, workplace and public L2 together — reaching 80 per cent of all charging, and DC fast charging the remaining 20 per cent. This is a modelled projection for the year 2030, built for a scenario of 33 million EVs on the road, not a measurement of today’s mix. That page also traced and could not verify a shorter, unsourced version of this claim — "about 80% of charging happens at home" — that appears on some of DOE’s own consumer pages with no study behind it, and flagged it rather than repeating it. This page carries the same distinction forward rather than inventing a new one.

HYRE analysis. Even the sourced, dated 64-per-cent figure is a projection of behaviour, not a count of infrastructure — and infrastructure is what this page counts. The two questions are related but not the same: a majority-at-home charging pattern is compatible with a public network that looks, by port count, like the dominant story, because ports are what gets photographed, funded and announced. The next section is why that photograph is misleading about how hard those ports are actually working.

The part nobody photographs: how hard the public network actually works

Source fact. NREL’s The 2030 National Charging Network states plainly that "analysis of historical EVSE data tends to find relatively low utilization rates (e.g., less than 10%)" across the public charging network, citing real-world analysis of tens of thousands of charging ports. For network-sizing purposes, the same report uses the PEAK-hour utilization observed in that data — not the average — reporting peak-hour rates of 60 per cent for workplace Level 2, 55 per cent for public Level 2, and just 20 per cent for public DC fast charging, each measured at the busiest window in a large sample of real-world ports rather than the busiest single port.

HYRE analysis, and it is not a criticism. A 20 per cent peak-hour figure for the most expensive, most photographed form of public charging sounds like waste until the report’s own reasoning is read: DC fast charging sessions arrive unpredictably through the day, so a network sized to its average load would queue drivers during genuine peaks, and queueing is the failure mode operators are most trying to avoid. Level 2 charging — mostly workplace and long-dwell public parking — runs at a much higher peak utilization precisely because its demand is predictable: people arrive and leave on schedules. Low DC utilization is a designed margin against an unpredictable peak, not evidence the stations are unneeded.

What the finding actually says. The public network’s headline number — port count — measures capacity that exists. It does not measure how much of that capacity is doing anything most hours of most days. A homeowner deciding whether to fund a home circuit is solving a different problem than a state agency deciding where to fund the next DC corridor station, and reading one number as evidence for the other question is the mistake this page exists to head off.

Method

Public-charging source. US DOE Alternative Fuels Data Center, state-by-state station and port counts, retrieved 6 September 2026. Station counts are physical locations; port counts are individual charging outlets, several of which can sit at one station. AFDC states these figures exclude residential charging equipment.

EV registration source. DOE AFDC/NREL, "Electric Vehicle Registrations by State," all-electric (BEV) vehicles only, data as of 31 December 2023, workbook last updated September 2024, derived from Experian Information Solutions data. Reused from this site’s renter-lockout study rather than re-downloaded, so the two pages cite an identical figure.

Home-capacity source. US EIA, Residential Energy Consumption Survey 2020, tables HC1.1 (housing unit type) and HC2.1 (structural and geographic characteristics, including the attached-garage question). Preliminary release March 2022, final release March 2023. National totals only; RECS does not publish the attached-garage breakdown by state.

Utilization source. NREL, The 2030 National Charging Network: Estimating U.S. Light-Duty Demand for Electric Vehicle Charging Infrastructure (June 2023), Section 2.3.3, citing Borlaug, Yang, Pritchard, Wood and Gonder (2023), "Public Electric Vehicle Charging Station Utilization in the United States," Transportation Research Part D 114: 103564. Retrieved as the PDF published by DOE’s Joint Office of Energy and Transportation at driveelectric.gov. The underlying peer-reviewed paper sits behind a publisher paywall; the figures quoted here are as reported in NREL’s own document, not independently re-verified against the original paper.

Calculations. Ports per 1,000 EVs = state ports ÷ (state EV registrations ÷ 1,000). EVs per 1,000 homes = state EV registrations ÷ (occupied housing units ÷ 1,000), reused from the renter-lockout study. The correlation is a standard Pearson coefficient across all 51 jurisdictions (50 states plus DC) on the two per-capita rates. Every input is a public federal figure; the joins and ratios are ours and are shown as arithmetic, not modelled.

Reproducibility check. The 51 state port figures used here sum to 255,551, matching AFDC’s published national total of 255,551. The 51 state EV figures sum to 3,555,445, matching the 3,555,445 national total this site’s renter-lockout study already verified. Puerto Rico and other territories are excluded from both AFDC totals used here.