On August 4, 2026, Westlake's Planning & Zoning Commission approved the site plan for the Circle T Data Center Campus — four buildings, more than a million square feet, on roughly 90 acres along U.S. 377, backing directly onto Keller's Marshall Ridge neighborhood. Keller residents get the noise, the traffic, and the view. Westlake gets the tax revenue. This page lays out what is actually known, what is genuinely contested, and where residents still have leverage.
Every figure below comes from the developer's own presentation, town documents, or the Town of Westlake's public statements. Sources are linked in Section 3.
| Project name | Circle T Data Center Campus (called "Circle T Data Center" in Westlake town documents) |
| Location | Along U.S. Highway 377 in Westlake, on Circle T Ranch land — part of Hillwood's 27,000-acre AllianceTexas development |
| Developer | American Real Estate Partners (AREP) and PowerHouse Data Centers. Jeff Parsons, SVP of development and construction, presented to the commission. |
| End tenant | Not publicly disclosed. No operator has been named in any public filing or news coverage to date. |
| Size | Roughly 87–90 acres (figures vary by document); four data center buildings totaling more than 1 million square feet |
| Also on site | A private electrical substation / Oncor switchyard, backup generators, a security building at the main entrance, and a pump & water treatment building serving two storage tanks holding 5.5 million gallons |
| Power | Oncor switchyard sized for 350 MW. The data center is to use 300 MW; the remaining 50 MW is described as available for use within Westlake. |
| Water | 54 million gallons per year. Average daily demand 165,000 gallons; peak daily demand 1.5 million gallons. Westlake says the cooling system runs waterless roughly 85% of the year. |
| Employment | About 75 employees on site at any given time, per Westlake's project FAQ |
| Distance to Keller homes | 506 feet from the building footprint to the property line next to homes on Llano Court. Westlake's minimum requirement is 150 feet. The nearest Westlake residence is about 1,200 yards — two-thirds of a mile. |
| Buffers | 220-foot landscape buffer on the south (Keller) side; 75-foot buffer along U.S. 377; a landscaped berm along the highway |
| Claimed benefit to Westlake | Westlake officials have said the project could increase the town's tax revenue by roughly 14% |
This is the single most important thing to understand: the zoning fight already happened, years ago. The land was zoned for commercial-industrial use with a data center among the permitted uses before Westlake updated its development code to address data centers specifically. The August 4 hearing was narrower than most people in the room realized — the commission was only deciding whether the submitted site plan complied with the development standards in Town Ordinance 307.
Whether or not you accept that characterization, it defines the battlefield. Arguments that the project is a bad idea in principle have very little purchase in a site-plan review. Arguments about enforceable conditions, state permits, and public money have a great deal.
Note what is missing. None of these conditions include a numeric noise limit measured at the Keller property line, a monitoring requirement, a reporting requirement, or a remedy if the mitigation does not work. "Complies with the noise ordinance" is a promise about a measurement method, not a promise about what residents will hear. Getting measurable, enforceable, monitored limits into the development agreement is the most winnable fight left. See Section 4.
This section is deliberately honest. Some concerns about data centers are strongly supported by evidence; others are not, and repeating weak claims at a public hearing hands the other side an easy win. Each item below is labeled so you know which arguments will hold up under cross-examination.
Data center noise comes from cooling equipment that runs continuously, 8,760 hours a year. On hyperscale facilities, cooling tower fans produce roughly 70 to 85 decibels at 50 feet depending on fan size and speed. Air-cooled chillers, cooling towers, air handling units and server fans can reach up to 100 dBA at the source. Diesel backup generators — which must be periodically test-run — can hit around 110 dBA.
The critical point is that "complies with the noise ordinance" and "you won't hear it" are different claims. When noise was measured at Northern Virginia data centers that were the subject of resident complaints, levels on residential property came in between 40 and 59 decibels — quiet enough to pass local ordinances, but within the range the EPA associates with reduced ability to sleep, work and concentrate. Residents there reported sleep disruption, headaches, and an inability to escape the sound inside their own homes.
The reason is physics. Data center noise is heavily weighted toward low frequencies, which travel farther than higher-pitched sound and pass through walls, windows and insulation far more easily. Standard A-weighted (dBA) measurement, which most municipal noise ordinances rely on, deliberately discounts low frequencies. A facility can be fully compliant on paper and still produce a constant hum that residents cannot tune out. Ask Westlake directly which metric its ordinance uses, at what location it is measured, and who does the measuring — and get the answer in writing.
The most useful fact in this whole document comes from Westlake's own commission chairman. Before the August 4 vote, P&Z chairman Terry Wilkinson drove 2.3 miles to the Meta data center campus in Fort Worth's Alliance area and walked its perimeter. His words: "there was an obvious noise generated from that campus." He added that berms greatly mitigated the sound where they existed — which is why he asked for more of them. That is the chair of the approving body confirming, on the record, that a comparable facility is audible from outside its fence line.
Three hundred megawatts is not an abstraction. Running continuously for a year, a 300 MW load consumes roughly 2.1 to 2.6 billion kilowatt-hours, depending on utilization. The average Texas home used about 13,152 kWh in 2024, per EIA data. That puts this single campus in the range of 160,000 to 200,000 Texas households. Keller has about 47,500 residents at an average household size of 2.89 — roughly 16,400 households, or about 216 million kWh a year. This single campus will draw on the order of ten times the entire residential electricity consumption of the City of Keller, on ninety acres, for about 75 jobs.
Westlake's FAQ says residents' electricity bills will not increase because the developer paid Oncor to bring infrastructure to the west side of town at no cost to residents. That is a real and meaningful commitment as far as it goes. But it addresses only the local drop — not the regional transmission buildout that serving loads of this size requires.
Under ERCOT's structure, equipment connecting a specific customer is typically charged to that customer, but larger transmission upgrades that strengthen the regional grid go into a systemwide pool recovered from all electricity customers. In CenterPoint's territory, residential consumers used 33% of the electricity but paid 49% of transmission costs. Texas Senate Bill 6 (2025) directed the PUCT to examine whether the four-coincident-peak methodology fairly distributes those costs.
The counter-evidence: a May 2026 independent report by the consulting firm E3 found no historical evidence that data centers have driven increases in residential electricity costs under existing rate structures. Be aware that opponents of your position will cite it. The honest framing is forward-looking: PUCT's own transmission cost review concluded that rapidly growing computational loads are driving infrastructure investment that could raise residential and small-business bills under current policy — which is precisely why Abbott directed the PUC to require data centers to fully fund the infrastructure needed to serve them, and why cost-allocation rules are being rewritten right now.
Westlake says the campus will use less water than an average golf course and less than 1,000 homes. Run the arithmetic and that is roughly defensible: 54 million gallons a year works out to somewhere under 500 typical North Texas households on an annual basis. If you argue at the podium that this project will drain the water supply, you will be corrected, publicly, with the town's own numbers.
The stronger version of the argument is about timing, not volume.
The ask: a binding commitment that the data center's water use is curtailed on the same schedule, and by a comparable proportion, as residential customers when drought stages are triggered. If Keller families can be told they may not water their lawns, a 300 MW campus can be told to lean on dry cooling. Ask for this in writing in the development agreement.
This is the argument residents reach for first and it is the weakest one in the file. The formal research does not currently support it:
If you lead with property values at a public hearing, expect the developer's consultant to read the GMU conclusion back to you and move on. The defensible version: the research base is thin, drawn almost entirely from a region where data centers arrived alongside enormous job growth, and it measures sale prices rather than time on market, buyer pool, or disclosure obligations. A seller on Llano Court in 2029 will have to answer questions about the hum. That is a real cost to a real family even if it never shows up in a regression.
Here is the part that gets lost in the noise-and-water argument, and it is the cleanest case Keller residents have.
The costs land in Keller. The revenue lands in Westlake.
The nearest Keller home is 506 feet from the building line. The nearest Westlake home is two-thirds of a mile. Westlake officials project the project could raise Westlake's tax revenue by roughly 14%. Not one dollar of that flows to Keller's budget, Keller's roads, or Keller's residents. Keller has no vote, no zoning authority over the site, and no share of the proceeds. This is the entire argument in two sentences, and it does not depend on any contested study.
Even for the host jurisdiction, the economics are far worse than the headline investment numbers suggest. Data centers are the most capital-intensive and least labor-intensive major development a town can host. Good Jobs First has documented the pattern across states:
| Deal | Public subsidy | Permanent jobs | Cost per job |
|---|---|---|---|
| Apple — North Carolina | $321 million | 50 | $6.4 million |
| Google — Columbus, Ohio | $54.3 million property tax abatement | 20 | $2.7 million |
| Illinois statewide (since mid-2019) | $468 million in sales tax exemptions and credits | 339 | ~$1.4 million |
Good Jobs First further reports that in 10 of the 20 states that disclose annual costs, data center subsidy programs exceed $100 million a year, and that Georgia localities alone are projected to lose an estimated $1.1 billion in 2026 and $1.4 billion in 2027. In Oregon, data centers owned by Amazon, Apple, Alphabet and Meta collected $616 million in property tax abatements between 2016 and 2025, with annual program costs rising 762%.
What this means for you, concretely. No tax abatement has been publicly reported for the Circle T project. That is not the same as none existing. Watch for three separate approvals, each with its own hearing and its own vote:
If any public money is on the table, the ROI conversation changes completely — and unlike the Westlake site plan, Tarrant County and the school district are bodies Keller residents actually vote for. That is real leverage. Abbott's August 3 audit directive also requires PUCT and ERCOT to verify each project's public financial assistance, which means this information is going to have to surface.
Westlake says operational traffic impact will be minimal, with about 75 employees on site at a time. That is probably accurate for the operating phase. The real traffic burden is the multi-year construction period — heavy equipment, concrete, transformers and generator deliveries on U.S. 377 — which is rarely addressed in site plan review and should be addressed in the development agreement (hours, routes, staging, dust control, road repair bonding).
On air quality: the campus includes backup generators. Statewide, at least 38 Texas data centers have received minor-source air permits since 2024, covering more than 2,100 diesel backup generators. Generators emit nitrogen oxides and particulate matter during testing and outages. The permitting path matters enormously for your rights — see Step 3.
Everything cited above, plus the official channels you will need. Verified as of August 8, 2026.
Start with an honest premise: stopping this project outright at the Westlake Town Council is unlikely.
The land was already zoned for it, staff found the site plan compliant, and the town has said publicly there was
no legal basis to deny it. A strategy built on "vote no on August 18" is a strategy built to fail.
But "we can't stop it" is not the same as "we have no leverage." There are four separate places where Keller
residents can still change outcomes, and three of them are outside Westlake's control entirely. Work all four
at once.
The development agreement is a contract. Unlike the site plan, its terms are negotiable, and the council has broad latitude over what goes in it. Every hour spent asking the council to reject the project is an hour not spent getting binding protections written into a document that will govern this facility for thirty years.
Scheduling conflict — plan for it now. The City of Keller's own Pre-Council and Regular Council meeting is also scheduled for Tuesday, August 18 at 5:00 p.m. Residents cannot be in both rooms. Split deliberately: send your strongest technical speakers to Westlake, and send a second group to Keller's council to demand the baseline acoustic study and a formal resolution (Step 6). Do not let the two meetings cannibalize each other by accident.
On August 3, five days before this project's site plan was approved, the Governor ordered a comprehensive verification and audit of every data center moving through ERCOT's interconnection process — effectively pausing approvals. Projects that fail the transparency requirements or state law will be denied connection to the grid.
A 300 MW campus cannot operate without an interconnection. This is the one venue where the answer can be no.
The audit is required to verify each project's:
Two of those five — public financial assistance and community protections — are exactly what Keller residents have been asking about and not getting answers on. Put your questions into that record.
This is the most commonly missed opportunity, and timing is everything. Whether you get any public rights depends entirely on which permitting path the developer chooses:
At least 38 Texas data centers have received minor-source permits since 2024, covering more than 2,100 diesel backup generators statewide. Do not wait to find out which path this project takes.
Keller residents have no vote on Westlake's council. You do have a vote on Tarrant County Commissioners Court and on your school board. Any tax incentive requires a separate approval, with its own public hearing, from bodies that answer to you.
Set up agenda alerts for all of these bodies. Public Citizen's Texas guide (linked in Section 3) walks through each mechanism and how to intervene. If public money is involved, the entire ROI conversation reopens on terms far more favorable to you — and Abbott's audit now requires that public financial assistance be disclosed and verified regardless.
There is a documented factual dispute between two cities about who knew what and when, a former commissioner testifying she was removed from the P&Z the night before a key session, and a project PowerPoint that the town reportedly "could not find." Records requests are cheap, fast, and they generate exactly the kind of material that keeps reporters engaged.
Keller's political interest is aligned with residents here, so its records are likely to come back faster and more completely. Under the Texas Public Information Act, a governmental body generally must produce records promptly or seek an Attorney General ruling within ten business days.
Keller's leadership has expressed support for residents. Statements of support cost nothing. Convert that into things that carry legal and evidentiary weight — and note that Keller's council meets the same evening as the Westlake vote, which makes it the natural place to press these asks:
If you only do three things this week:
The TCEQ air permit for the backup generators, the state interconnection audit, and any tax incentive from the county or the school district all come later — and each one is a separate hearing with its own deadline. Leave an email and we'll tell you when the next one lands. That's all it's used for.
No newsletter, no fundraising, no list sharing. Notification of hearings and deadlines only.