Skip to main content

Methodology

How we collect, estimate, and verify data on data center proposals in Central Texas.

On this page

Our approach

The Central Texas Data Center Tracker is a volunteer-run civic organization. Our project tracker is built on publicly available information — no inside sources, no leaked documents. Every data point we publish must be traceable to a specific source, and we identify clearly when a figure is a developer claim vs. an independently verified fact.

When sources conflict, we show the discrepancy rather than picking a number. When a figure is unavailable, we say so rather than estimating. We will always err on the side of transparency about uncertainty.

Where our data comes from

Primary sources (highest confidence)

  • Permit applications and government filings — Flood hazard development permits (Hays County), zoning applications and Preferred Scenario Map amendments (City of San Marcos), and TCEQ air quality permit applications.
  • Developer disclosures — Official press releases, investor announcements, regulatory filings, and statements made at public hearings on the record.
  • Utility and regulatory agency records — Crystal Clear Special Utility District, Bluebonnet Electric Cooperative, Pedernales Electric Cooperative, Edwards Aquifer Authority, ERCOT, PUC of Texas, and TCEQ.
  • City and county meeting records — City of San Marcos council agendas and meeting recaps, Hays County Commissioners Court minutes, Guadalupe County and Caldwell County public records.

Secondary sources (verified journalism)

We rely on investigative reporting from local and regional outlets that have directly covered these proposals. Approved outlets by county:

AI-generated content aggregators (e.g., CitizenPortal.ai, Patch AI) are not approved secondary sources. Social media posts (LinkedIn, X/Twitter) are acceptable only when they constitute an official developer or government disclosure, and must be labeled as such; a general company feed is not a citable source. Industry registries (datacentermap.com, baxtel.com) may fill NULL fields only when no approved source exists, and must be labeled as industry registry data.

Reference standards and research

  • Edwards Aquifer Authority — drought stage triggers, permitted withdrawal data
  • Texas Water Development Board — 2022 State Water Plan (Hays County supply projections)
  • NRDC, “Thirsty Data Centers” (2022) — data center water use methodology benchmarks
  • Siddik et al., “The environmental footprint of data centers in the United States,” Environmental Research Letters (2021)
  • Global Energy Monitor (GEM Wiki) — power plant tracking

How we estimate water use

Water figures are the hardest data to obtain. Texas law does not require data center operators to publicly disclose projected or actual water consumption. We use a tiered approach:

  1. Developer-disclosed figures— Some developers have submitted water use estimates in permit applications or restrictive covenants (e.g., Highlander SM One's covenant cap of 75,000 gal/day). These are labeled as developer-stated.
  2. Industry benchmarks— For facilities that haven't disclosed figures, we apply the U.S. industry-average Water Usage Effectiveness (WUE) of 1.8 L/kWh (Lawrence Berkeley National Laboratory, DOE report LBNL-1005775, 2016), scaling to 2.5 L/kWh for Texas summer heat (Uptime Institute, 2016). At 80% utilization, this translates to:

1 MW × 1,000 kW/MW × 24 hrs × 0.80 utilization × WUE ÷ 3.785 L/gal

At 1.8 L/kWh → 9,000 gal/day/MW

At 2.5 L/kWh → 13,000 gal/day/MW

A 100 MW facility: ~900,000–1,300,000 gal/day

Some modern hyperscale operators (Meta, Microsoft) achieve far lower WUE (0.2–0.3 L/kWh) through advanced cooling. Where a developer publicly discloses a lower figure, we use that instead. Where they claim “waterless” or “closed-loop” cooling without independent verification, we apply the industry benchmark — Texas law imposes no disclosure requirement, and self-reported developer figures have not been independently audited by TWDB or TCEQ.

  1. Water-for-power estimates— We separately estimate the water consumed by the power source using Scanlon et al. (2013): 0.26–0.55 gal/kWh for the ERCOT grid blend (range covers natural gas combined-cycle at 0.19 gal/kWh to steam turbines at 0.52 gal/kWh), and 0.3–0.5 gal/kWh for on-site natural gas generation. This water is consumed at power plants, not at the data center itself. Note: ERCOT's growing share of wind and solar (~36% in 2025) reduces the actual grid water intensity below the historical range; we use the higher figures conservatively.

When a figure is a developer claim, we say so. When it is our estimate, we show the methodology. When we have no basis for a figure, we show “Not disclosed” rather than guess.

Project status and urgency

Each project carries two labels that we update as the approval process evolves:

Status

Status reflects where a project is in the regulatory and construction pipeline:

  • In Planning — Credible reports of developer interest; no official filing yet
  • Application Filed — Formal permit, rezoning, or development agreement application submitted to a government body
  • Government Review — Active regulatory review underway; public comment periods and hearings may be open
  • Under Construction — All required permits granted; construction authorized or underway
  • Partially Operational — Some buildings or phases are online; campus expansion is ongoing
  • Fully Operational — Facility is at or near full buildout and operational capacity

Urgency

Urgency reflects how much community action can still influence the outcome:

  • Urgent — A public hearing, vote, or comment period is imminent; direct civic action is most effective right now.
  • Engage — No immediate vote, but active regulatory proceedings are underway; contacting officials is productive.
  • Monitor — No active proceedings; we are watching for new filings or developments.

Urgency is set manually by our data team. If you believe a project's urgency is outdated, contact us.

How we score environmental impact

Each data center project can receive an environmental score from 0 to 100 — translated to a letter grade (A through F) — computed by our /score-projects skill from structured fields in our database. Not every project has a score yet; scoring requires that structured power and cooling data has been researched and entered. For deeper background on the dimensions behind the score, see How Data Centers Are Powered & Cooled.

Rubric weights

Higher scores indicate less environmental impact. Weights sum to 100%.

  • Water (35%) — Direct cooling intensity (gal/day/MW), indirect water for power, recharge-zone penalty, and water source type.
  • Power Source (25%) — Carbon intensity (gCO₂e/kWh) and additionality-adjusted renewable percentage. On-site nuclear scores better than on-site gas; unbundled RECs do not count as additionality.
  • Cooling Efficiency (15%) — PUE, WUE, cooling-type classification, and a bonus for waste-heat reuse (district heating, greenhouse, industrial process).
  • Backup & On-site Combustion (10%) — Backup generator count × annual runtime cap × fuel emissions factor. Diesel-heavy, uncapped generator fleets score poorly.
  • Grid Impact (10%) — Transmission upgrade requirements, peak-load coincidence, and demand-response program participation.
  • Land & Siting (5%) — Acreage on the Edwards Aquifer recharge zone and distance to the aquifer.

Letter grade bands

  • A — 90–100
  • B — 80–89
  • C — 70–79
  • D — 60–69
  • F — below 60
  • Not Rated (NR) — score not yet computed; typically because structured power and cooling data is not yet available for the project.

Rubric weights are editorial and transparent. Any future revisions will be noted in the project changelog with a correction timeline event on affected projects.

Verification

Projects carry an “Unverified — review in progress” badge until a member of our data team has independently confirmed the core facts against primary sources. Verified projects show the name of the reviewer and the date of last review.

Verification does not mean every figure is perfect — it means a human reviewer has read the underlying sources and believes the data is a fair, good-faith representation of what is publicly known.

What we flag as unverified

We explicitly label developer-stated claims that have not been independently confirmed. Current examples:

  • CloudBurst / Evolve Holdings— The $10 billion investment figure comes from the developer's Guadalupe County tax abatement filing. The 480–600 permanent jobs projection is from the same filing. The “closed-loop, zero-evaporation” cooling claim has not been independently verified; the water feasibility study was declared confidential.
  • Edged Energy— The “waterless cooling” claim is based on ThermalWorks technology. ThermalWorks is a sister company of Edged under the same parent (Endeavour) — not an independent third party. No peer-reviewed or third-party audit of ThermalWorks performance has been published.
  • Highlander SM One— The developer's tax revenue projections ($9M/yr to the city, $15.2M to the school district) and permanent job count (~50 positions) are from hearing presentations and have not been independently modeled.
  • Tract— The 800 permanent jobs figure is from the Hays-Caldwell Economic Development Partnership's announcement, citing full-buildout projections from the developer.

How to submit a correction or tip

We are a small volunteer team and we make mistakes. If you have a primary source that corrects or adds to our data:

Corrections are reviewed within 7 days. If accepted, we publish a correction timeline event on the affected project so the change is visible in the project history.

Key references

Sources for water benchmarks are marked with ✦.

  1. ✦ Shehabi, A. et al., “United States Data Center Energy Usage Report,” Lawrence Berkeley National Laboratory (DOE/LBNL-1005775), 2016 — establishes 1.8 L/kWh as the U.S. industry average WUE
  2. ✦ Lawrence Berkeley National Laboratory, “2024 United States Data Center Energy Usage Report,” December 2024 — most recent federal update; U.S. data centers consumed 17 billion gallons directly in 2023
  3. ✦ Scanlon, B.R. et al., “Controls on Water Use for Thermoelectric Generation: Case Study Texas,” Environmental Science & Technology, 2013 — Texas grid water intensity: NGCC 0.19 gal/kWh, steam 0.52 gal/kWh
  4. ✦ Orr, R. & Klesner, K., “Don't Ignore Water Consumption,” Uptime Institute Journal, June 2016 — hot-climate WUE can reach 2.5+ L/kWh; evaporative cooling tower analysis
  5. KUT Radio, “Hays County residents aren't happy about a data center moving in,” June 4, 2025
  6. KUT Radio, “With 5 data centers on the horizon, Hays County water advocates see the fight as just beginning,” February 26, 2026
  7. Edwards Aquifer Authority, J-17 Index Well Data and Drought Management Rules
  8. Texas Water Development Board, 2022 State Water Plan — Water for Texas
  9. NRDC, “Thirsty Data Centers,” 2022
  10. Siddik, M.A., Shehabi, A., & Marston, L., “The environmental footprint of data centers in the United States,” Environmental Research Letters, Vol. 16, No. 6, 2021
  11. Crystal Clear Special Utility District, Service area and infrastructure records
  12. City of San Marcos, Council meeting agendas and recaps