How it works

Sources, rings, and the estimates

Short version: we keep a list of data center projects with coordinates and a status, measure the straight-line distance from the address you type, and then do arithmetic you can check. Here is the long version.

Where the projects come from

• County and city planning records: rezoning and special use applications, planning commission and board of commissioners agendas, permit approvals.
• Development of Regional Impact filings with the state, which large projects must make before local approval.
• Public Service Commission dockets and utility filings, including large load agreements and integrated resource plans.
• Local reporting from outlets that cover the county, which is often the first place a code-named project gets a real name.
• The operating baseline: the IM3 Open Source Data Center Atlas from Pacific Northwest National Laboratory, built from OpenStreetMap building footprints, which maps operating facilities nationwide. This site includes data from the atlas and from OpenStreetMap contributors under the Open Database License.

Each project carries a status: rumored (named by a credible outlet, not yet on a public record), filed (an application is on the record), approved (permission granted, no visible construction), under construction, operating, on hold, withdrawn, or denied. The location is the parcel when we have it and the town center or county seat when we do not, and the page says which. Every project page links its sources.

The two tiers matter when reading a result. Operating facilities are tracked in every state. Planning filings are researched county by county and cover a growing list of states; in a state without filings coverage a quiet result means less, and the page says so. The coverage page lists both tiers exactly.

Finding your address

Street addresses are placed by the US Census Bureau geocoder, which also tells us the county. ZIP codes and town names are placed by OpenStreetMap's Nominatim at the center of the ZIP or town, so distances for those are approximate and the page says so. We keep the result of each lookup so repeated searches and shared links do not hit those services again.

The three rings

Under 2 miles is the range where you might hear cooling fans and generator tests, see the buildings and the lights, and share a road with the construction traffic. Two to 10 miles is the same water system, the same substation, the same school district. Ten to 25 miles is mostly about the utility: the whole territory pays for new generation and transmission through rates, whether or not you can see the building. Distances are straight-line from the point above, not driving distance.

The bill estimate

We map your county to its primary electric provider and show that utility's rate matters on record. When a matter has a requested or approved percentage, we multiply it by a typical bill (yours if you enter it) and show a range of plus or minus 15 percent around that, because riders and tiers move the real number. When the utility has published a dollar figure for the typical bill instead, we use that. Base rate freezes, fuel cost riders and storm cost riders are different things and we try to say which one is moving. This is arithmetic on public filings, not a forecast and not advice.

The noise estimate

We start from a reference sound level at 500 feet from the fence line, scaled by the size of the project: 62 dBA for under 50 MW, 67 for 50 to 200, 71 for 200 to 500, and 74 above that. Those are in line with measurements reported around operating campuses, where continuous cooling noise at the property line commonly falls in the 60s and 70s. From there the level drops 6 dB for every doubling of distance plus a small allowance for air absorption. We then compare the result to the World Health Organization's 40 dBA guideline for outdoor noise at night. Terrain, tree cover, which way the buildings face, the cooling design and wind all change the real number, and backup generator testing is louder and periodic. Treat the figure as a rough scale, not a measurement.

The water estimate

For projects reported as evaporative cooled, and for projects whose cooling design has not been reported, we show what evaporative cooling would draw: 0.5 to 1.8 liters of water per kilowatt-hour of IT load (1.8 is Lawrence Berkeley National Laboratory's 2016 figure for evaporative cooling; newer designs run lower), applied to the project's reported megawatts running around the clock. We convert that to gallons per day and to an equivalent number of households at 200 gallons per household per day. Closed loop and air cooled designs use a small fraction of that water and more electricity instead.

Open for comment

We list hearing dates and comment deadlines that are on the public record for projects within 25 miles of you and for the utility's rate matters. Dates move; the linked record is the authority, not this page.

Alerts

When you subscribe we note the address's coordinates and the projects already within 25 miles. From then on, one email when a new project appears inside that circle or an existing one changes status. Nothing else. Unsubscribe from any email.

Corrections

Send the filing or the article to [email protected]. We would rather fix it than argue about it.