Fayette Data Center Impact
Scroll down ↓
Fayette Technology Campus · Piperton, Tennessee

Fayette Data Center Impact

A proposed data center with its own gas power plant, on about 1,085 acres at I-269 and Highway 193. Scroll to see what it could mean for noise, the economy, our water, our air and our roads, with the developer's claims, the unknowns, and the questions worth asking.

About the sound: the humming and whining you hear in the background is a recording of typical data center noise, taken from an NBC News report. It is not a recording of this site, which has not been built, and may be louder or quieter than this site would be, depending on distance, equipment and mitigation. Use the button at bottom right to turn it off.
Scroll down to explore ↓  ·  Updated November 20, 2026

The proposal

Big Watt Digital has proposed the Fayette Technology Campus, a phased data center campus with its own natural gas power plant, on about 1,085 acres in unincorporated Fayette County near I-269 and Hwy 193, close to Piperton.

DeveloperBig Watt Digital (Austin, TX)
Site~1,085 acres, I-269 & Hwy 193
PowerOn-site natural gas generation
CoolingClosed-loop (developer's description)
Permanent jobs100–200 (developer's estimate)
Sources: developer's site, Daily Memphian, FOX13. The developer launched its website on November 18.

Where it stands

Fayette County's mayor says a moratorium is in place in unincorporated areas and that Piperton would need to annex the land for the project to proceed (Daily Memphian). The site lies inside Piperton's Urban Growth Area but outside city limits (Tennessee Update). Piperton says it will hold public informational meetings but has not posted dates (town site).

To see how the land around the site is planned, see Piperton's 2023 Official Land Use Map (PDF). It shows the town limits, the urban growth boundary and the planned uses (low density residential, mixed use, commercial, industrial and others). Check its color key for how the site and its neighbors are designated.

Not yet disclosed

  • Power capacity (MW), and the number and size of turbines
  • Water volumes. The developer says wells are for initial system fill, domestic use and irrigation
  • Sound levels at the property line (it says equipment will be enclosed, bermed and buffered)
  • Expected emissions
  • Any tax incentives or payments in lieu of taxes it will request
Until these are published, anything on this site that models them uses clearly labeled placeholder assumptions.

Explore by topic

The site

Parcel 123 001.00Boyle Trust & Inv Co · ~949 ac
Parcel 123 001.01Boyle Investment Co · 131.9 ac
Combined~1,081 ac
From the TN Comptroller Property Viewer: 001.00, 001.01. Together they come close to the developer's 1,085 acres, but the developer has not published parcel numbers. The 001.01 outline is exact; the 001.00 outline is approximate along its southeast side.
Made by a Fayette County neighbor to show possible impacts fairly. It is not affiliated with Big Watt Digital, Fayette County or the Town of Piperton. Where it estimates something, it says so.
Section 1

Noise

Noise: what is known

The developer says equipment will sit deep in the interior, inside sound-attenuating structures with berms and tree buffers, and that operations are designed to meet or exceed local noise standards at property lines (developer's site). It gives no decibel figures. Tennessee's environmental agency does not regulate noise. That is usually handled by local government (TDEC), so the useful question is which local standard applies and where it is measured.

Everything below is a model with placeholder inputs, not a prediction. The turbine layout defaults to a plant the size of xAI's Southaven plant, as a reference, because no capacity has been published for this site.

Look up a point

Or click anywhere on the map.
No point selected.

Noise sources

Baseline: Mississippi permitted 41 turbines generating 1.2 GW for xAI at Southaven, about 29 MW each. Its Titan 350 units are roughly 35–38 MW. At 29 MW, the four options total about 1.2, 1.5, 1.8 and 2.4 GW.
A recording of typical data center noise plays in the background of this whole page (use the floating button at bottom right to turn it off). Here it gets quieter or louder with the estimated level at the point you select. Browsers only start sound after your first click, tap or key press. The sound is the whining section (about 2.5 seconds, played forward then reversed to loop without a seam) of one recording in an NBC News report, "This is what it sounds like to live near a data center". It is illustrative, not a recording of this site or of any specific turbine, and the source may not be a gas turbine. If turbine-loop.mp3 is missing, a synthetic sound plays instead.
Drag the DC or GT marker to move it. The red dots are individual turbines in a grid around the GT marker. Levels from all units are combined.

Compute scale (SpaceXAI Colossus 2 as the reference)

The developer has not said how much computing this campus would hold, and nothing published links this project to SpaceXAI; Colossus 2 is used only as a size reference. To stay with that reference, this sizes the computing from the power plant: the same share of plant output reaches the servers as at Colossus 2, and the same number of chips fit per megawatt. Equipment noise then scales with that load.

Reference points, all reported by others and not confirmed by the company: Colossus 2 is reported at about 550,000 Blackwell-class GPUs on roughly 1 GW (Introl), or about 946 MW of IT power by Epoch AI's estimate from satellite imagery (Inside Deep Tech), fed by the 41 permitted turbines rated at 1.2 GW, which is where the 79% default comes from (946 ÷ 1,200). By August 2025 it had 119 air-cooled chillers giving about 200 MW of cooling (AIComplianceCore). Those figures vary by source and are rough.
The preset below is the level for a 100 MW unit. Other sizes are adjusted by an assumed 2 dB per doubling of size (a 29 MW unit comes out about 3.6 dB lower).
These are planning assumptions, not manufacturer data. Replace with the turbine model's guaranteed sound level from the permit or spec sheet.

Combined noise contours

Gas turbine noise: what the sources say

There is no single "average" turbine. Levels depend on the model, enclosure, inlet and exhaust silencers and load. Well-treated units have been reported at roughly 55 dBA about 100 m away, with noise near the enclosure around 80 dBA (Fry, Inst. of Acoustics). Bare gas generators can reach about 122 dBA at the casing and 132 dBA in intake ducting (Iredale, Inst. of Acoustics). Open-cycle exhaust stacks are also known for low-frequency noise that A-weighted dBA understates (TRB/Hessler). This map shows dBA only.
Model: point sources, 6 dB per doubling of distance, minus air absorption and a flat terrain allowance. Ignores wind, inversions, topography and tones. Not a substitute for an acoustic study.

What neighbors of SpaceXAI's plants report

  • Reported A Southaven resident near the Colossus 2 turbines described levels from the 40s to the mid-50s dB, with a constant high-pitched hum and a low churning sound (Mississippi Free Press). Another resident documented levels above 60 dB from his backyard late at night, per CBS News (American Bazaar).
  • Reported SpaceXAI says it is investing millions in sound walls and silencers, while residents say the sound wall installed so far has given little relief (American Bazaar).
  • Concern Residents also report a constant low-frequency hum that is felt as pressure or vibration, which a single dBA number can understate (WPLN; Tom's Hardware).
  • Context These are residents' own reports, mostly of the turbine plant plus cooling equipment together, measured at unstated distances. They are not a measurement of this site.

Videos: what it sounds like

Videos from YouTube about data center noise and SpaceXAI's Colossus. Tap one to watch it here. The background sound pauses while a video plays.

These are other people's videos, embedded from YouTube and not made or checked by this site. They show other data centers and may not reflect what this site would sound like. Phone and video recordings exaggerate or flatten sound depending on the microphone and your speakers, so they are not measurements. If a title shows as "Video 2" and so on, YouTube did not return its title; it is still the linked video.

Bill Dance Lake (Herb Parsons Refuge)

  • Context The lake is a 177-acre reservoir inside a 474-acre refuge in Fayette County, renamed Bill Dance Lake State Fishing Area, and reopened with enhancements in January 2026 (TWRA). It is part of the state's Bill Dance Signature Lakes program, which funds better ramps, habitat and fish stocking (News Channel 9).
  • Evidence Studies of park visitors find that louder human-made noise makes people more likely to rate a natural area experience unacceptable, though tolerance varies by type of visitor. In one study of more than 400 visitors to Pennsylvania state forests, the noise was from natural gas compressors (Utah State University).
  • Evidence A study in Swiss nature parks found sound levels had the strongest effect on how visitors rated their recreation quality at sites with similar scenery (Austriaca). Researchers also note noise can limit the wildlife visitors can see and hear (USDA Forest Service paper).
  • Limits These are studies of parks and forests, not of anglers at this lake. We found no study of a gas power plant near a fishing lake, so the results show direction, not a forecast for Bill Dance Lake. Wind, trees and terrain change how far sound carries, and the model here ignores them.
Section 2

Economic impacts

Possible benefits

  • Developer100–200 permanent technical jobs, and up to about 20,000 total jobs (direct, indirect and induced) over the buildout (developer's site). The larger figure is the developer's own estimate.
  • PossibleConstruction work for skilled trades. The developer's site points to ironworkers and industrial electricians.
  • PossibleLocal tax revenue. A U.S. Chamber report cites a 50 MW facility in Dallas County, Texas generating about $174 million in local revenue over five years (report). That is a different state and tax regime. What Fayette would receive depends on incentives it grants.
  • DeveloperNo new load on the public power grid, because the campus generates its own power. Local electric rates are therefore less exposed to its demand.
  • EvidenceSeveral studies found no meaningful effect on nearby home values: Northern Virginia, Indiana and a Realtor.com analysis (linked below).

Possible concerns

  • ConcernHome values. A Berkeley–Harvard paper estimates a 7.2% average decline within one mile of a newly announced large data center, with larger declines at hyperscale and first-of-their-kind sites. See the home-value tool below.
  • ConcernTax breaks. Tennessee's data center sales tax exemption has no sunset (tracker). A League of Women Voters of Tennessee slide deck, citing the Department of Revenue, reports about $77 million (state share) and $27.5 million (local share) forgone in FY2025 (slides). We could not check this against the Department's own report.
  • ConcernLocal payments depend on negotiated incentives such as abatements and PILOT deals (Oklahoma State Extension). Gallatin's 20-year PILOT with Meta is one Tennessee example.
  • ConcernBenefits skew urban. A 2026 study found data center growth and job gains are most pronounced when the local economy is more urban and developed (Fortune).
  • Context100–200 permanent jobs on 1,085 acres. For comparison, Meta's Gallatin campus was expected to create 40 full-time jobs plus around 100 daily subcontractors (Data Center Knowledge).

Home values near the site

Load nearby homes

Home locations come from OpenStreetMap and are incomplete: only buildings tagged as houses/residential, or with a street number. Larger radii take longer.

Value assumptions

Default is about the Zillow typical Fayette County value ($377k). Per-parcel assessor values need the state's authenticated service, so every home here uses this one number.

Summary

Load homes to see totals.
The evidence conflicts. A Berkeley–Harvard paper estimates a 7.2% average decline within one mile of a newly announced large data center, and about 3% on average within three miles over five years (Newsweek summary). The 1–3 mile figure here (2.5%) is our back-calculation assuming homes are evenly spread, not a number from the paper. Studies in Northern Virginia and Indiana and a Realtor.com analysis found no meaningful effect. Rural, first-of-their-kind sites, like this one, are where the Berkeley–Harvard paper found larger declines. These are scenarios, not appraisals.
Section 3

Water

What the developer says

The campus uses closed-loop cooling and its own on-site wells and private storage tanks, and draws nothing from Piperton or Fayette County systems. The developer says the wells are for initial system charging, domestic use such as restrooms, and irrigation (WREG). Baseline monitoring and third-party audits are described as planned.

What is not known

  • Well depths, locations, and whether they tap the Memphis Sand aquifer or shallower water
  • Pumping rates, and how much makeup water a closed loop will need over time
  • What the gas plant needs. Combined-cycle plants can use significant water. Simple-cycle turbines use little.

In Tennessee, anyone withdrawing 10,000 gallons a day or more from groundwater must register with the state, but registration does not authorize or limit the withdrawal (TDEC).

Water demand calculator

Placeholder: follows the computing load estimated in the Noise section (turbine plant × share to computers) until you edit it. The developer has not published a capacity.

Estimated demand

Fairness note: the developer says it will use closed-loop cooling, so the evaporative option is a "what if" for comparison, not its stated design. At this scale evaporative figures are very large; please do not quote them as the project's expected water use.

Why the aquifer matters here

NASA-supported work with Protect Our Aquifer identified high-recharge locations in Fayette County (NASA). Fayette County is the recharge area for the Memphis Sand aquifer, which supplies drinking water across the Memphis region (Memphis Flyer / CAESER). The aquifer is reported to be losing water faster than it is replenished near the xAI site in Memphis (Planetizen), and the unconfined recharge zone means surface contamination can reach it more directly. This calculator shows demand only. It does not model drawdown, which needs well locations, pumping rates and aquifer properties from a hydrogeologic study.
  • Per-MW evaporative range: Data Center Knowledge
  • Low-water case: EPRI (150 MW at about 340,000 gal/day)
  • Household use ~300 gal/day: Orenco
  • Reference: TVA once proposed up to 3.5 million gal/day from the Memphis Sand (Memphis Flyer)
Section 4

Air pollution

What gas turbines emit

Gas turbines release nitrogen oxides (smog-forming), carbon monoxide, fine particulates (PM2.5) and formaldehyde, a carcinogen (SELC, Earthjustice). The developer says generation will be phased within state and federal Clean Air Act standards through TDEC and will use dry low-NOx controls (developer's site).

No emissions estimate has been published for Fayette. The Southaven figures below are a reference for what a very large plant can emit, not a forecast for this site.

Where exposure would likely be highest

The map shades a relative exposure index out to 20 miles. It is a simple illustration, not a dispersion model: exposure falls off with distance from the plant, and if you pick a wind direction it stretches the zone downwind. Real zones depend on the plant's stack heights, emissions and local weather, which the permit's air modeling should show.

Wind changes through the year and the day. Ask for the wind rose and the dispersion modeling (often AERMOD) used in the permit application, and for predicted concentrations at homes, schools and the parcel line.

Reference: xAI's (now SpaceXAI's) Southaven plant

  • The permit application for 41 turbines shows more than 1,300 tons a year of health-harming air pollutants and over 6 million tons of greenhouse gases (reported).
  • A Reuters analysis of 30 of xAI's temporary turbines estimated potential emissions of about 2,500 short tons of nitrogen oxides, 4,000 of carbon monoxide and 22 of formaldehyde a year, assuming continuous operation. These are potential emissions from equipment specs, not stack measurements. Caveat: those temporary turbines may have been at xAI's Memphis site (Colossus 1) rather than at Southaven; we have not confirmed which, so treat this as a figure for xAI's turbines in the area, not specifically the Southaven plant (summary).
  • The Southern Environmental Law Center's modeling estimated $30–44 million a year in health damages from the 41 permanent turbines (SELC). SELC is an advocacy group that opposed the permit, and the estimate has not been independently reviewed here.
  • Mississippi issued the permit and has said temporary turbines do not need one. The NAACP and SELC dispute that and say the permit underestimates pollution (NAACP).

How it would be regulated in Tennessee

  • TDEC's air division regulates turbines and other emitting equipment, but does not approve a data center as a whole.
  • Sources under 5 tons a year of any regulated pollutant are exempt from permits. Major-source rules (Title V) generally start at 100 tons a year.
  • Davidson, Hamilton, Knox and Shelby counties run their own air programs. Fayette is not among them, so as far as we can tell the state program would apply.
  • A TDEC permit does not override local zoning or a moratorium.

Questions worth asking

  • How many turbines, of what size and type, and will any run in simple cycle?
  • What are the permitted annual limits for NOx, CO, PM2.5 and formaldehyde, and what controls will be installed?
  • Will there be continuous emissions monitoring, and will the data be public?
  • Will a backup or phase-in fleet of temporary turbines be used, as at Southaven?

Bill Dance Lake (Herb Parsons Refuge)

  • Evidence Nitrogen oxides from burning fuel can settle into lakes. Research across more than 2,000 lakes found that airborne nitrogen shifts the nutrient balance, making algae less nutritious for the tiny animals that feed fish, and lowering diversity (Umeå University). Experiments added nitrogen to Rocky Mountain lakes and caused eutrophication responses, from more algae to changed algae communities (Colorado State).
  • Evidence In China's Lake Taihu, nitrogen deposition was concentrated in summer, and the authors say it may matter for how long algal blooms last (AGRIS).
  • Limits Those lakes are mostly remote alpine or already badly polluted. How much nitrogen would reach a small reservoir like this one depends on the plant's emissions, stack heights and the lake's existing nutrient load from its watershed. No deposition estimate exists for this project. Ask for one in the permit review.
  • Enjoyment We found no study that measures how gas turbine air pollution changes fishing enjoyment. The noise studies in the Noise section are the closest evidence on the experience of visitors.
Section 5

Traffic and roads

The biggest traffic effect of a data center usually comes during construction, not operation. The map highlights the major roads within 20 miles that trucks and workers could use.

Not disclosed We have not found a published traffic study, haul routes, construction schedule or workforce size for this project. The estimator below uses placeholder inputs you can change.

Construction and operation are very different

  • Concern Traffic planners say data centers do not fit standard assumptions: the surge comes from the temporary construction workforce and frequent deliveries of heavy equipment and materials, which typical traffic studies, built around peak commuter hours, may not capture (Kittelson & Associates).
  • Concern Steel, concrete, generators, transformers and cooling equipment can arrive day after day for years, and local governments often lack the tools to evaluate multi-year freight surges (Next City). A gas power plant adds turbines and other large equipment to that list.
  • Context Once built, traffic is mostly the daily commute of a small staff. Large data centers commonly employ about 25 to 150 permanent workers (Eno Center for Transportation). The developer's estimate is 100 to 200.

What happened elsewhere

  • Evidence Near Meta's data center construction in Holly Ridge, Louisiana, traffic has nearly tripled and crashes around the site have risen more than 600% (Louisiana Illuminator). That is a much larger project on rural roads, so it is a warning sign, not a forecast.
  • Evidence In nearby Southaven, Mississippi, congestion has grown on Stateline Road around the xAI power plant and data centers, and a local business owner says customers are avoiding the area (WSMV).
  • Context The region is also absorbing other large projects. TDOT expects about 5,800 jobs at the Ford campus in West Tennessee to raise commuter travel demand, and is adding roadway improvements in Fayette, Haywood and Tipton counties (TDOT).

Rough trip estimator (placeholder inputs)

Every input is an assumption. Look up real counts for your road in TDOT's traffic history data, and ask the developer for its workforce and delivery plan. Trips are counted one way: a worker driving in and out is two trips.

Questions to ask

  • Will a traffic impact study be required, and will it cover construction as well as operation?
  • Which roads will be haul routes, and what hours will deliveries run? Are school bus routes and church or farm traffic considered?
  • Who pays to repair roads damaged by heavy loads, and will the developer post a bond?
  • Will turn lanes, signals or interchange work be needed at I-269 and Highway 193, and who pays?
  • How will oversize loads, such as turbines and transformers, be scheduled and escorted?
  • Is there a plan for emergency vehicle access during construction?
Section 6

Get involved

What you can do

  • Ask Piperton and Fayette County to publish meeting dates and the annexation and zoning timeline. Piperton lists its Administration Center at 3300 Highway 196 South, phone 901-853-4830, and a "Notify Me" sign-up on its site. That listing is general city contact information, not a contact for this proposal.
  • Ask the developer to publish capacity, turbine count, water volumes, noise levels and emissions.
  • Watch for any TDEC air or water permit application and its public comment period.
  • Request records: noise ordinances, annexation and rezoning applications, any incentive or PILOT agreements.

About this site

This site was put together by a Fayette County resident who lives near the proposed campus. The goal is to show neighbors and local government the possible impacts, good and bad, with the sources beside each claim, so people can judge for themselves. It is not affiliated with Big Watt Digital, Fayette County, the Town of Piperton or any agency.

How it tries to be fair: the developer's own claims are labeled as the developer's; benefits are listed next to concerns; studies that found no effect on home values are shown alongside the one that found a decline; and anything modeled uses stated placeholder assumptions, because the developer has not yet published capacity, turbine, water, noise or emissions figures. Figures from other places are examples, not predictions for Fayette. Estimates are planning-level illustrations, not appraisals, engineering studies or legal advice.

If the developer or a public agency publishes real numbers, they should replace the placeholders here.

Limits of this site

  • No capacity, turbine count, noise level, water volume, emissions or traffic figures have been published for this project. Every model here uses placeholder inputs, labeled as such.
  • Southaven and Colossus figures are references for a project of similar size, not predictions. Some come from secondary sources and vary between them. Nothing published links this project to SpaceXAI.
  • Advocacy groups (SELC, NAACP, Earthjustice) and residents' own reports are cited for what they say, and labeled. Studies that found no effect on home values are shown alongside the one that found a decline.
  • The noise and air maps are simple illustrations, not an acoustic study or a dispersion model. Real permit modeling should replace them.
  • The background sound and the videos are illustrative and are not measurements.

Method notes

  • Parcels come from the TN Comptroller Property Viewer. Homes come from OpenStreetMap, which is incomplete in rural areas.
  • The noise model is a simple point-source model with placeholder levels and no terrain, wind or tonal effects.
  • Home values use a single typical value for the county, not individual assessments.
  • Roads on the Traffic map come from OpenStreetMap. The trip estimator uses placeholder inputs, not project data.
  • The background sound is a short clip from an NBC News report (credit: NBC News), used for illustration. It is not a recording of this site, may not include gas turbines, and its volume on your device says nothing about real levels.