Data dictionary

What each column in our data center, gas plant and power project downloads means, how it's measured and the values it can take.

Download PDF

Data centers

A row is one data center building, not a campus. A campus with six buildings is six rows, linked by the campus name.

A blank cell means we don't know the value yet. A blank is never a zero. The utility and grid fields are blank unless we know them with high confidence.

Field	Definition
Project nameproject_name	Name of the data center building. Buildings on a shared campus are numbered: Building 1, Building 2, Building 3.
Developerdeveloper	Company developing or owning the building. Not necessarily the company using it: see Tenant and End user.
Developer experiencedeveloper_experience	Whether the developer has a track record building data centers. Experienced Has at least one operating or under-construction data center, or is an established developer with a track record (for example, Tract). New entrant Has no operating or under-construction data centers.
Statusstatus	Stage of development of the building. Planned Announced or permitted, but not yet under construction. Under Construction Vertical construction is underway. Operating Fitted out, with power and all equipment installed. Canceled Abandoned by the developer.
Operating yearoperating_year	Year the building came online, or is expected to come online.
Facility power (MW)capacity_mw	Total power the building draws at full build-out, including cooling and electrical overhead.
IT capacity (MW)it_capacity_mw	Power delivered to the servers themselves, excluding cooling and overhead. Also called critical IT load.
Building size (sq ft)building_square_footage	Gross floor space of the entire building at full build-out, in square feet.
Estimated investmentestimated_investment	Modeled all-in cost of the data center, in US dollars: the powered shell (building, power and mechanical and electrical systems) plus the servers, GPUs and networking inside it. Excludes on-site power generation. Estimated at $38 million per MW of IT capacity. When only facility power is known, IT capacity is taken as facility power divided by 1.25.
Countycounty	County of the building's location.
Statestate	Two-letter state code of the building's location.
Campuscampus_name	Name of the campus the building belongs to. Buildings on the same campus share it.
Utilityutility	Retail utility whose service territory the building is in, even when the site plans to run on its own generation.
Utility parentutility_parent	The utility's parent company. Blank for co-ops and municipal and public utilities.
ISO/RTOiso_rto	Grid operator the building is in: PJM, ERCOT, MISO, SPP, CAISO, NYISO, ISO-NE, or Non-ISO.
Balancing authoritybalancing_authority	Balancing authority the building is in.
Utility EIA IDutility_eia_id	The utility's EIA utility number (EIA-861), for matching to EIA data.
Balancing authority codebalancing_authority_code	The balancing authority's EIA-930 code, for example ERCO, PJM or SOCO.
Tenanttenant_name	Company leasing capacity in the building, separated by commas when there are several. Filled only where a tenant has been publicly identified.
End userend_user_name	Company actually using the compute, when it differs from the tenant. A tenant may lease a building and resell its capacity to another company.
Latitudelatitude	Latitude of the building, in decimal degrees.
Longitudelongitude	Longitude of the building, in decimal degrees.
Source 1source_1	Primary source supporting the record. Where Cleanview has archived the document, the link is to that copy.
Source 2source_2	Additional supporting source.
Source 3source_3	Additional supporting source.

Behind-the-meter gas plants

A row is one gas plant built to serve a data center: behind the meter, on site and connected to the grid, or off site and dedicated to a named campus. Gas plants are a separate behind-the-meter subscription.

Field	Definition
Plantplant_name	The plant's name in its permit, siting case or announcement. Otherwise, the data center's name followed by "power plant."
Developerdeveloper	Company that builds and owns the plant. Often not the data center developer: Williams builds Meta's plants in Ohio.
Data centercampus_name	Name of the data center campus the plant powers. Matches the campus name in the data center download.
Statusstatus	Status of the plant's most advanced unit that isn't canceled. A plant with an operating first stage and a second stage under construction is Operating. A plant is Canceled only when every unit is. Announced Announced or applied for. No air permit or siting approval has been issued. Permitted An air permit or state siting approval has been issued, but construction hasn't started. Under Construction Work on the generating equipment is underway: foundations, equipment on pads, stacks. Clearing and grading don't count. Operating Generating power for its data center, including temporary bridge power. Canceled The project won't be built.
Statestate	Two-letter state code of the plant's location.
Countycounty	County where the plant's generating equipment is.
Nameplate (MW)nameplate_mw	Number of machines times the manufacturer's rating, summed over the units that aren't canceled. This is the size of the power plant, not the data center's load, and it's usually larger than the load because of backup and redundancy.
Firm output (MW)firm_output_mw	Power the plant is contracted, permitted or stated to deliver to its data center. Filled only when a document states it.
Firm output basisfirm_output_basis	The kind of document behind the firm output. contracted A power purchase or supply contract. permit limit The output limit in the plant's permit. net capacity The plant's stated net capacity. developer stated A figure the developer has given publicly.
Configurationconfiguration	Kinds of generation at the plant, separated by semicolons. Combined cycle Gas turbines whose exhaust heat drives a steam turbine. Simple cycle Gas turbines without heat recovery. Reciprocating engines Gas-fired piston engines. Fuel cells Fuel cell servers running on natural gas.
OEMsoems	Equipment manufacturers at the plant, separated by semicolons. Blank until the equipment is public.
Modelsmodels	Equipment models at the plant, separated by semicolons.
Unitsmachines	Number of turbines, engines and fuel cell servers that aren't canceled. A minimum when a unit's count isn't public.
Yearfirst_operating_year	Year the plant's first unit began generating, or Cleanview's estimate of when it will.
Announcedannouncement_date	Date of the first public record of the plant (YYYY-MM-DD).
Air permitair_permit_url	Link to Cleanview's archived copy of the plant's issued air permit, or its latest amendment. Blank until the plant is permitted.
Latitudelatitude	Latitude of the plant's generating equipment, not the data center campus.
Longitudelongitude	Longitude of the plant's generating equipment, not the data center campus.

Grid-connected power projects

A row is one project in a grid operator's interconnection queue, or one generator in EIA's monthly survey of power plants (EIA-860M). The same plant can appear in both.

Each download covers one status. Planned projects come from the interconnection queues and EIA-860M, and operating projects from EIA-860M.

Field	Definition
Projectproject_name	Project or power plant name, as the source gives it.
Developerdeveloper_name	Company developing or owning the project, as Cleanview has identified it from queue records, interconnection agreements and regulatory filings. Often the parent company of the project LLC named in the queue. Blank where we haven't identified it yet.
Capacity (MW)capacity_mw	Capacity of the project or generator in megawatts.
Dateoperating_date	For operating projects, the date the generator began operating. For planned projects, the date it's proposed to come online.
Technologytechnology	Standardized technology type, for example Solar, Wind, Battery, Natural Gas, Nuclear or Hybrid - Solar + Storage.
Countycounty	County of the project's location.
Statestate	Two-letter state code of the project's location.
Gridbalancing_authority	EIA-930 code of the balancing authority the project connects to, for example ERCO (ERCOT), PJM, MISO, CISO (CAISO), SWPP (SPP), NYIS (NYISO), ISNE (ISO-NE) or a utility's own, such as SOCO or DUK.
Statusstatus	Stage of development. Each download covers one status. Planned Active in an interconnection queue, or reported to EIA as planned. Operating Generating power, as reported to EIA.
IA signedhas_ia_signed	Whether the project has signed an interconnection agreement with its grid operator (true or false). Planned download only. Blank for EIA-860M rows.
Under constructionunder_construction	Whether the project has started construction (true or false). Planned download only. Blank for the CAISO, NYISO and SPP queues, which don't report it.
IR dateir_request_date	Date the project's interconnection request was filed (YYYY-MM-DD). Planned download only. Blank for EIA-860M rows.
Latitudelatitude	Latitude of the project, in decimal degrees.
Longitudelongitude	Longitude of the project, in decimal degrees.
Sourcesource_table	Where the record comes from. eia_860m EIA's monthly survey of power plants (EIA-860M). *_queue A grid operator's interconnection queue: caiso_queue, ercot_queue, miso_queue, ne_iso_queue, nyiso_queue, pjm_queue or spp_queue.
Source IDsource_id	The record's ID in its source: the queue position for queue projects (for example, AB2-089 in PJM), or the EIA plant ID and generator ID for EIA-860M rows (for example, 61179_JCCS3). Source and Source ID together identify a row.