FCC Data Hub Licensing and Tower Data

The regulator's file cabinet turns out to be the deepest single well of American tower ground truth. data.fcc.gov hosts the Universal Licensing System bulk archive under /download/pub/uls/, and inside it the Antenna Structure Registration - the federal ledger of every registered tower, mast and rooftop-mounted structure in the United States and its territories.

Scale first. a_tower.zip (208 MB zipped) unpacks into pipe-delimited .dat files totaling roughly 8.7 million rows: CO.dat with 1.49 million coordinate rows in NAD-27 degrees-minutes-seconds plus earth-centered X/Y/Z, EN.dat with 2.77 million entity rows naming registrants down to contact and address, HS.dat with 2.49 million history rows, RA.dat with 1.47 million registration actions, and AT.dat's 59,727 application attachments. Beside it, d_tower.zip (56 MB) packages the obstruction-evaluation flow - 1.09 million determination-of-no-hazard cases spread over 3.4 million DE/DP/DR rows, each with position, heights and registrant.

Within Datadory's catalog of 1,744 datasets averaging 7.81, this record scores 9 out of 10 - the highest in the telecom tower REITs slice - on verified definitions, shipped samples and public-domain rights. Get a sample of this dataset cut to your states before anything else.

One line, one positional record, no header row:

DE|89-ASO-2235-OE||36|4|5.20||129845.20|89|20|31.20||321631.20|100.6|57.9|158.5|01/05/1990|||01/22/1990||DYERSBURG|TN|MAXCELL TELECOM PLUS, INC.
CO|APP|A0099939|1204700|3482621|T|47|48|9.0|N|172089.0|103|16|52.7|W|371812.7||
AT|APP|A0463140|1250054|2918248|O|Completed FAA 7460-1|09/08/2005

Read the anatomy rather than the raw strings. Each line opens with a file code - DE for a determination, CO for coordinates, AT for an application treatment, EN for an entity, HS for history - and everything after the pipes lands in fixed order. The determination row carries the case number (89-ASO-2235-OE), the DMS position, three height measures, two dates, city, state and registrant: a complete vertical fingerprint of one structure in one line. The coordinate row adds the earth-centered triplet, so a GIS load needs no projection math.

Datadory ships these flattened into typed tables - one row per determination, per coordinate set, per entity - with the dictionary unchanged and decimal-degree conversions alongside the survey originals. The Dyersburg line above is also the oldest record in the working set, filed January 5, 1990 by MAXCELL TELECOM PLUS, INC.: thirty-six years of American vertical infrastructure in one continuous format.

Ten fields carry every registry record, definitions verified against actual .dat rows during the August 2026 pass:

• file_code - the record type switch: AT, CO, EN, HS, RA, DE. Parse order follows it.
• file_number - the ASR identifier in the AA-XX-YYYY-OE pattern. One string joins a structure's coordinates, entities, history and determinations.
• latitude / longitude in DMS - NAD-27 degrees, minutes, decimal seconds and hemisphere, the survey-form position every conversion starts from.
• geocentric X/Y/Z - the earth-centered triplet appended in CO.dat, pre-computed so distance and visibility math skips the projection step.
• registration_date - MM/DD/YYYY, the axis any build-out series runs on.
• city / state - uppercase locality from the determination files, the coarse filter before coordinates take over.
• overall_height_ft - height above ground level alongside ground elevation and height with appurtenances.
• registrant_name - the landlord of record from EN.dat and DE.dat.
• treatment_code - FAA 7460-1 obstruction paperwork per application.

Folded under additional fields on request: HS history rows, RA actions, environmental notices, condition codes, the l_cell.zip cellular-license tables and the a_lease.zip spectrum-manager leases that tie tenants to structures. Derived decimal-degree columns ride along on every delivery.

Geography - the United States and its territories, every FCC-registered antenna structure. Set against the trade's roughly 158,500 purpose-built US towers, the registry nets far more: buildings, rooftops, water tanks and poles that never surface in commercial inventories still register here.

Temporal - back to 1990, with the earliest working-set registration dated January 5, 1990. Freshness is structural rather than aspirational: the complete snapshot rebuilds weekly - files stamped August 15-16, 2026 during cataloging - and per-weekday change packages land between rebuilds, giving deterministic delta handling no scraped page can match.

Granularity - one row per registration action, coordinate set, entity or hazard determination. Nothing arrives pre-aggregated, so collocation counts, state-by-state height histograms and registrant league tables are group-bys rather than estimates.

Where it ranks among its neighbors: best telecom tower REITs datasets, with the pooled view on the telecom tower REITs data hub.

API, files, or your warehouse. Daily, weekly, or hourly.

The upstream archive rebuilds weekly with daily change files; your delivery cadence gets set separately, to match the decision being fed. Rows arrive typed and keyed on the ASR file number, so this quarter's pull joins last quarter's without hand-mending - and when a column shifts or a registrant renames, the change ships flagged rather than discovered mid-model.

Every delivery flattens the headerless positional extracts into tables with the field dictionary above unchanged, decimal-degree conversions beside the DMS originals, and sample rows for validation before the first join. Name the states, record families and cadence when you request a sample of this dataset; the ongoing arrangement follows once the rows validate.

Six jobs the registry earns its keep on:

• National inventory construction - every registered structure with coordinates, heights and owners in one keyed table. It anchors the best telecom tower REITs datasets ranking for exactly this reason: it is the only record attributing structures to entities at national scale.
• Registered-versus-active analysis - registry steel joined against observed radios, the disagreement itself the finding.
• Ownership rollups - 2.77 million entity rows grouped by registrant into who-holds-what tables, or worked as outreach lists.
• Collocation screening - height bands, localities and determination status shortlist structures before an engineer ever opens a CAD tool.
• Build-signal research - registration activity through time as physical-capex evidence beside REIT disclosure.
• Competitor mapping - rivals' filed structures visible ahead of earnings commentary.

Each job maps to a persona below, and the sample validates whichever one you came for.

Competitive intelligence and product teams see filings before commentary: rival structures, heights and registrations land in the registry quarters before they reach an investor deck.

Investors and quants read registration flows as infrastructure-build evidence - a public-domain series continuous since 1990.

Data scientists get stable coordinate-height-entity features on a single identifier spine.

Market researchers count structures by state and height band with federal denominators. Sales teams turn entity rows into call lists, and journalists, academics and students cite government data with no licensing friction.

The common thread: everyone who needs to know what steel exists and who filed it starts here, whatever they join onto it next.

This record anchors the slice; the honest differences are about shape and scope:

• FCC Open Data Portal - Wireless Datasets - the query-first sibling: a SODA API over about 56 hosted tables, JSON and CSV out, but no ASR registry. The head-to-head runs in the Data Hub vs Open Data Portal comparison.
• TowerMaps US Cell Tower Database - 600,000-plus consolidated commercial sites with owner-side commerce: availability, zoning, sales agents. The market's own book beside the regulator's ledger.
• OpenCellID Bulk Downloads - millions of observed cells worldwide; proves which filed structures actually carry live radios.

If the job starts 'enumerate American steel,' start here; if it starts 'query one table cheaply,' the portal; if it starts 'who owns and sells the structure,' TowerMaps.

Four things, stated plainly.

First, the files are headerless and positional. Parsing them yourself means writing a positional reader against undocumented layout pages - www.fcc.gov returned 403 to scripted clients throughout research, so field meanings here were verified by inspecting real rows. Datadory ships the parse done, with the dictionary attached.

Second, two access caveats survive at the edges: the legacy License View JSON API now redirects to www.fcc.gov and returns Akamai Access Denied, and some l_*.zip license files respond 302 to plain curl clients. The bulk zips remain the reliable path.

Third, cells of paperwork are not towers - a structure accrues many rows across CO, EN, HS and DE files, and deduplication onto physical structures is part of the delivery, not your homework.

Fourth, rights are clean: general public catalog under 17 USC 105, redistribution permitted without attribution. Get a sample scoped to your states and record families before anything is scheduled.