What Environmental Regulators Hold About San Antonio’s Data Centres
The building permit tool on this site refuses, in words, to answer the questions people actually ask about data centres. No figure for electricity. No figure for water. Nothing about jobs, cooling, generators or the grid. That refusal is correct, and it stands: the City permit file has no answer to any of them.
This page asks whether any other government record does. It publishes what federal environmental regulators already hold about the same facilities, and it publishes the gaps with equal weight. Nothing here is modelled and nothing is estimated.
This is not a count of data centres, and nothing on this page is a power figure. It counts 8 federal registrations carrying a data-centre classification in the San Antonio metro. Some of those are places the building permit tool already publishes and some are places the corroboration tool already publishes. Adding any two of the three totals counts the same buildings twice.
There is no megawatt figure, no water figure and no jobs figure anywhere on this page, because no machine readable government record carries one for these facilities. Nothing here is estimated from floor area or from anything else.
The City has no category for a data centre. The federal register does.
The whole argument of the building permit tool is that the City record carries no occupancy type, no use code and no flag identifying a data centre, so the only handle is free text somebody typed into a project name. The federal environmental register is different: it classifies facilities by NAICS code, and two of those codes describe this industry directly.
Searching by classification is not the same as searching by name, and that is the point. A keyword list is our judgement about which companies to look for. A NAICS code is the regulator’s judgement about what a facility is. Three of the addresses below were tested against the City permit record in August and returned no permits at any building number; the classification reaches them anyway.
Every classified facility, and what each sibling record says about it
Two independent verdicts travel with every row, because a gap in one record is not a gap in the other. The address test runs first where the register carries a street address, and distance is the fallback only where it does not.
| Facility as the register names it | Address | Air permit | What the City permit record says |
|---|---|---|---|
| EX WEST GREEN DATA CENTER | THE SE CORNER OF SENTRY GATEWAY AND WEST MILITARY | none in the register | no street address in the register; a permit site is within 400 m |
| MICROSOFT | 5120 ROGERS RD | Title V, major source | a different address on the same street, 196 m away |
| MICROSOFT SAT CAMPUS SAT09 13 40 46 | 15434 LAMBDA DR | Title V, major source | no City permit site within 400 m |
| MICROSOFT SAT15 | 3555 WESTOVER LINK | Title V, major source | no City permit site within 400 m |
| SAT2 SAT3 SAT4 | 9554 WESTOVER HILLS BLVD | Title V, major source | the same address was applied for and never issued |
| STA1 | 9999 WESTOVER HILLS BLVD | Title V, major source | no City permit site within 400 m |
| VANTAGE DATA CENTER TX2 | 5207 ROGERS RD | Title V, major source | the same address carries issued City permits |
| VANTAGE TX11 | 14720 OMICRON DR | Title V, major source | no City permit site within 400 m |
Highlighted rows are the ones with a street address, no matching address in the City permit record, and no permit site within 400 metres. One row is deliberately not highlighted despite being a near miss: a facility registered at 5120 Rogers Rd sits 196 metres from a permit site at 5150 Rogers Rd. Those are two different house numbers on one street, and the permit tool never normalises a house number, so the evidence does not settle whether they are one place. It is reported as unsettled rather than counted either way.
Major sources that report no emissions
Those two numbers are not a contradiction, and this page will not present them as one. A Title V major source classification describes the potential to emit criteria pollutants if the equipment ran to its permitted limit. The greenhouse gas programme records actual reported carbon dioxide equivalent against a separate threshold. Different pollutants, different tests. A facility can cross the first and sit under the second with nothing inconsistent about it.
What the pair does support is narrower, and it is the reason this page exists: the standby equipment at these sites is large enough to be regulated as a major source, and the federal record contains no evidence that it runs much. Neither statement is a measure of how much electricity a data centre uses. Standby generators are what run when the grid fails.
What no government record will tell you
Each of these was checked on 2026-08-30, and the negative result is published rather than filled in.
| The question people ask | What the record holds |
|---|---|
| How many megawatts? | Nothing machine readable. Engine capacity appears in TCEQ permit documents as PDF only, and a search of the state open data catalogue returns other jurisdictions’ portals rather than Texas air permits. The air permit number is published in the table above so a reader can retrieve the document instead of taking an estimate from us. |
| How much water? | Nothing. The San Antonio Water System publishes no machine readable consumption data; its open data portal returns no datasets, and the City GIS layers carrying its name are outages, hydrants and distribution sites. |
| How many jobs? | No government record found. Not estimated. |
| How much cooling? | No government record found. Not estimated. |
| What did they actually emit? | One of the 8 facilities has a National Emissions Inventory record at all, and the federal Envirofacts service exposes no facility level table for it. |
| How big is the substation? | The City GIS substation layer returns no queryable fields and no feature count. The register does record that utility substation construction was filed in the data centre cluster, but a substation is provisioned headroom and serves a network, not one customer, so this page reports the filing and never what it powers. |
Why there is no estimate here
The obvious way to answer the power question is to take a recorded floor area and multiply it by a published watts per square foot figure. It would produce a number. It is refused, for a reason that is worth stating plainly.
The only floor area in the permit record is the largest single permit’s stamped area at a site, and that figure exists in the first place because the same area is repeated across sibling permits for one project. At a site carrying more than a hundred permits it is a floor of unknown tightness. Multiplying it by a density we choose produces a result whose precision is entirely our own assumption, and publishing that beside figures taken from government records would make the two indistinguishable to a reader. A range would not fix it. It would only show how wide our uncertainty is while still putting a megawatt number on the page.
The federal classification is not tidy either, and that is published rather than smoothed: one facility here carries a manufacturing NAICS code alongside its data centre code, from two different federal systems. A facility with no Title V permit is likewise not a facility with no generators. It is one the register does not classify as a major source, which can mean smaller equipment, a different permit type, or a filing this search did not reach.
Source: United States Environmental Protection Agency facility register, retrieved 2026-08-30. Study area is the San Antonio and New Braunfels metropolitan statistical area, a published federal boundary. County comes from each facility’s coordinates, never from the county string on the filing. Machine readable copies: JSON API and CSV.
The rest of this investigation
Four pages read four different public records to ask four different questions about the same buildings. A place can appear in two of these records and still be one building, so no figure on any of them can be added to a figure on another.
- Data Centers: The anchor of this set, data centre sites found in the City of San Antonio’s own building-permit file, published with the candidates that were rejected and why each one failed.
- Data Center Zoning: How much land would still be eligible, under the rules Council votes on in October, by council district. It measures acres, not buildings, and scores a proposal that has not been adopted.
- Data Center Corroboration: Places the federal environmental register records where the City permit file has nothing, corroboration of where the City record stops, not an inventory.
Other San Antonio tools
Same method, a different public record. Each one states what its record cannot answer.
Crime Map Building Permits Bar & Restaurant Revenue Aggressive Driving Trail Accessibility Licence Plate Readers Data Centers Data Center Zoning 311 Response Times Data Center Corroboration Data Center Environmental Record Bond Projects
All 12 are free, with no signup and no email wall, and the public API behind them needs no key.