Sources and methods
Where the information comes from, how we read it, and what we still don’t know.
What the result means
We check for noise complaints within two miles of your chosen location, or at a matching facility address. Each complaint is matched to a specific campus. A complaint about one site does not apply to every site that company owns.
- Red: recent, unresolved noise complaints at a nearby operating or construction site.
- Amber: older or unconfirmed reports, known equipment that could create noise, future plans, or changing operations.
- Green: no reported data-center noise concerns found in the sources loaded for this check. It also applies when equipment has not had an independent review. It does not guarantee quiet or certify a home as safe.
- Gray: the source check could not finish, or a relevant facility location is still unresolved. A failed source check cannot turn green.
Older complaints, resolved complaints and reports about stopped operations stay visible as past reports. Sources that mention noise without a direct review are labeled separately. A “Potential noise nearby” result needs site-specific information about nearby outdoor cooling equipment or generators running continuously. Facility type alone does not establish noise.
Air-permit fines are listed separately. They do not, on their own, trigger a noise warning. Reports of quieter conditions and planned noise reductions stay visible with their dates and any recorded resolution.
What we can’t tell you
Sound has not been measured at the home. We may also be missing complaints, facilities or construction schedules. Equipment, barriers, terrain, weather and background sound affect what you hear, including a hum indoors. Loudoun County’s noise guidance explains these factors.
The results do not predict decibels or determine whether a home is safe for your health. Our distance bands and review windows are screening choices, not measured sound boundaries or health thresholds.
Distance bands and review windows
The main answer uses a two-mile area and matching facility addresses. The map, full list and report details show facilities within ten miles, so a farther facility’s complaint label may differ from the answer for your address.
A report can count as recent for up to two years. Older complaints remain available as history. The underlying data also keeps an earlier “High concern” category for recent unresolved complaints within one mile; documented complaints within two miles remain history or context in that category system. The main page shows the plain-language answer instead of competing scores. These rules have not been validated as physical limits on sound.
The closest-facility tag says “Very close” within 0.25 mile, “Close” within 0.5 mile, and “Nearby” within two miles. Matching addresses have their own label. A greater distance does not prove lower noise or complete local coverage. Whether a site is operating, being built or planned also affects the result.
Where the information comes from
Facility records
Compute Atlas supplies the main facility inventory under CC BY 4.0. We filter the records and calculate distances. The list includes data centers and crypto mines with usable map points. Cancelled projects, unrelated power plants and points representing several sites are excluded.
We add reviewed campus addresses, location corrections and dedicated data-center power sites. These additions apply to both live data and the saved fallback. The power records include xAI/MZX Tech’s Stanton Road turbine site, with sources for its physical address and data-center connection. Power-site coverage is incomplete, and most imported crypto locations have not been independently audited.
Examples of reviewed corrections
Corrections include the former Bitcoin site in Dowagiac that is converting to AI, Markley’s Lowell campus, Vineland and Granbury. For Lowell, we compared the operator’s address, PeeringDB’s facility point and municipal records. PeeringDB was used for this correction; it is not a nationwide feed for the checker. See the current site-specific reviews.
Noise reports
The reviewed collection has 51 source records for 32 facilities: data-center campuses, crypto mines and a dedicated turbine power site. The collection was updated October 16, 2026; each campus and source retains its own review dates. Unconfirmed campus matches and unresolved source locations are labeled, and cannot independently trigger High concern.
Each inventory load also checks article titles and community notes supplied with facility records for noise concerns. These links are research leads until reviewed. Their publication dates may be unknown, and some describe proposals or construction. The live feed may add new links; the saved fallback keeps its dated sources.
We show reports within ten miles, with their locations and dates. This is not a complete web search or complaint history. The Google links beside individual facilities let you look beyond our collection. Finding no reports in our sources does not mean no one has complained.
Equipment and facility types
Every facility has the same set of profile fields: use, setting, cooling, power, capacity and campus-specific complaint history. Descriptions from the provider are labeled unverified. Independent site reviews have their own dates and links. A type mentioned in an article is a research lead until confirmed; missing information stays unknown.
Large hyperscale sites may have outdoor cooling and power equipment worth checking. Colocation sites may have rooftop cooling or backup generators; an urban address does not prove silence. Enterprise and edge sites also need a layout check. “Closed-loop cooling” describes a water circuit, not a soundproof enclosure.
The result uses reviewed equipment information, campus-specific complaints and separately labeled provider leads. Provider leads cannot independently trigger High concern. Many equipment fields still need independent review.
Earlier facility and complaint reviews date to October 7, 2026. Targeted October 9–12 follow-ups added Granbury, Sterling and Reston reporting and updated Aurora mitigation status; they were not new reviews of every facility. Searches may load newer facility records; the figures on this page describe the saved inventory.
Locations and distances
“Same address” means your street address matches a facility record. It does not give a distance to its equipment. Other distances run in a straight line from the matched address, a street-range estimate or your chosen map point to the facility’s recorded point.
A campus edge or its equipment may be closer than that point. We do not map campus boundaries. The provider’s “exact” and “confirmed” labels do not mean we independently verified the location. Missing or misplaced facilities can change a result, and proposed projects may never be built.
Hollow markers show locations that have not been independently verified. Known address, source, state or geography conflicts are withheld from the map and numeric proximity checks. Other approximate locations remain labeled and can still be wrong. A geography check does not certify a campus; it checks the state and county, not the building, campus edge or sound at a home.
Location audits and downloadable records
The saved inventory has 1,892 records. Of these, 1,176 campus locations still lack resolved verification. The review ledger lists the sources and gaps for each record.
Download the all-facility review ledger · Read the full sources and unresolved address candidates
The October 6 nationwide audit checked the captured data-center points for geographic consistency, compared street addresses and linked map features, and recorded sourced corrections. A newly added or changed point in the live feed needs another review and is marked unaudited.
Download the location checks · Read the full audit evidence
Registry address candidates are leads for investigation. They never replace a location automatically. A source must identify the physical campus; complaint dates and noise-reduction work are checked separately. A missing record cannot establish quiet or lower concern.
Download the ten-facility address and source review · Read all twenty Census lookup results · Read the detailed evidence and unresolved claims
Keeping reports current
Weekly research combines recent news searches with a rotating campus review, including older reporting. Findings need a readable original source, a dated event, a reviewed physical-address match and two automated source checks before publication. Uncertain matches stay in the research queue. Searches can miss reports, and automated review can make mistakes.
Publication dates and event dates are kept separate. A new article about an old complaint does not automatically mean a current problem. Reports of improvement remain attributed; they do not certify quiet.