Methodology
Where every piece of this site's data comes from, how it is put together, and what it can't tell you. The data itself is on the downloads page.
Sources
- Virginia DEQ, Air Sites (daily). Every site DEQ flags as a data center on its Air Sites layer (198 on the last refresh), with its name, address, location and operating status. Data centers need a DEQ air permit for their backup generators, so a project appears here once it applies for one.
- Virginia DEQ, issued air permits for data centers. Permit numbers, dates and programs come from DEQ's list of issued air permits for data centers. That page blocks automated requests, so it is saved by hand from a normal browser as a dated snapshot (currently as of Sep 21, 2026) and joined to the Air Sites records by registration number.
- Virginia DEQ, county and city boundaries. Which county or independent city a site is in comes from DEQ's boundary layer: a point-in-polygon calculation by this site.
- Loudoun County land applications (LOLA). Rezonings, special exceptions, other land-use applications and site plans that mention a data center, from Loudoun's public land application records. Sub-applications are folded into their umbrella application so each proposal appears once. The fields that name reviewing staff are never downloaded, and names of people are removed from descriptions.
- Public schools. Names and locations from the National Center for Education Statistics' school locations (2024-25). Common names (such as “Rachel Carson Middle School” for NCES's “Carson Middle”) are added by hand.
- ZIP codes. Center points of ZIP Code Tabulation Areas from the U.S. Census Bureau's 2025 Gazetteer file.
- County meeting agendas. The meetings page lists agenda items that mention a data center, from the agenda feeds that Fairfax, Loudoun and Prince William publish (Fairfax County · Loudoun County · Prince William County). Those servers ask automated programs not to fetch them in their robots.txt files (Fairfax County · Loudoun County · Prince William County), so this site doesn't fetch them automatically: the agendas are saved by hand from a normal browser and refreshed periodically. The page says when they were last checked.
Stages
The stage shown for each data center is DEQ's own operating status on the Air Sites layer, renamed only for readability. This site does not judge a site's stage itself.
| DEQ status | Shown as |
|---|---|
| Planned | Planned |
| Under Construction | Under construction |
| Operating | Operating |
| Temporarily Shutdown | Temporarily shut down |
County filings show the county's own status (for example “In Review” or “Approved”), with a plain-language label. Grouping filings into land-use applications (decided by the county board, usually after a public hearing) and site plans (reviewed by county staff) is this site's own classification of the county's application types.
Distances
Distances are straight-line (“as the crow flies”) between two points, using the haversine formula on a sphere with the Earth's mean radius (3,958.8 miles), not driving distance. A ZIP code search measures from the ZIP code's Census center point, which can be a few miles from a given home in it; a school or a dropped pin gives a precise point. The calculation is the same in the build (geo.py) and in your browser (geo.js), and tests check the two agree. A data center's point is DEQ's location for the site; a county filing's point is the center of its parcel outline, calculated by this site.
The change log
Every morning the build compares the day's records with the previous day's and logs what changed: new sites, stage changes, new air permits, renamed or removed sites, new county filings and filing status changes. The log started on Sep 28, 2026; it shows when this site noticed a change, which can be later than the date the agency made it. It feeds New this week, its RSS feed and the changes download. If a source looks broken (for example, far fewer data centers than the day before), the build refuses to publish and the previous day's data stays up.
Our own classifications and estimates
- The locality of each data center (point-in-polygon on DEQ's boundaries, above).
- Land-use application versus site plan for county filings, the plain-English application types and status labels, and the filing's point (the parcel center).
- Which schools and ZIP codes get a page: those with a data center within 2 miles of the school or 5 miles of the ZIP code's center, or a county filing within 2 miles.
- Common school names, added by hand to NCES's official short names.
- Which agenda items are about data centers (the item's text says “data center” or names a tracked case number).
Known issues
- Manassas city line. 3 sites near the Manassas airport (74240, 74118, 74246) fall in Prince William County on DEQ's boundary layer, but DEQ's permit list says Manassas City. The two official sources disagree, so the site shows both rather than pick one.
- Missing addresses. 2 DEQ records have no street address and 1 record no ZIP code on the Air Sites layer; pages say “not listed by DEQ”.
- Permits not yet on the list. 15 sites on the Air Sites layer have no row on DEQ's issued-permit list yet, probably because the permit is still in progress; and the list itself is a hand-saved snapshot, so the newest permits can lag.
- Old applications still “In Review”. A few Loudoun applications filed years ago still show “In Review” on the county's records (STPL-2018-0019, SPAM-2018-0039). The site shows the county's status as it is.
- County filings are Loudoun only so far. Fairfax and Prince William filings are not included yet, so a project there appears only once it is on DEQ's records.
- Agenda items can be missed. An agenda item about a data center project that doesn't say “data center” or name a tracked case number isn't found; Fairfax's agendas are PDF board packages and aren't searched.
Spotted a mistake? Each item on the site links to its official record; if the site differs from the record, the record is right. Tell us and it will be fixed. The full source notes, including dates checked, are in the project's source notes. Reuse terms are on the terms page.