A ~7 GW power plant and data center is proposed for Hubbard, Texas.

Seven gigawatts on 800 acres — what that actually means

By generating capacity, this single site would be larger than every existing power plant in Texas — twice the size of the W.A. Parish complex outside Houston, and roughly three times Comanche Peak nuclear.

The cards below put the proposed facility in context against the largest existing power plants in Texas, the peak demand on the ERCOT grid, the footprint of Hubbard itself, and standard industry benchmarks. The capacity figure used throughout is the continuous-operation total of roughly 7,230 MW, summed from the per-unit ratings in the applicant's own equipment inventory; it excludes the ~19.5 MW of emergency-standby diesel generators. Reference figures come from publicly reported industry data and the U.S. Census.

Sources: capacities and ERCOT peak figures are from publicly reported utility-industry data; population figures are from the U.S. Census. Permit-specific numbers (acreage, unit counts, turbine models) are from the air-permit application itself, summarized in the next subsection.

Project overview

31 combustion turbines, 24 reciprocating engines, 6 emergency diesels, and a hyperscale data center — all on one 800-acre parcel southeast of the City of Hubbard.

Site plan — buildings

Site plan — emission sources

On day one, the plant would be roughly fourteen times larger than the data center it's built to serve

For most of the buildout, the majority of permitted generating capacity would have no contracted on-site customer. The air permit does not constrain what happens to that surplus.

Two very different capacity figures appear in the public record for this campus. The ~7.2 GW on this page is the continuous generating capacity summed from the TCEQ air-permit application. Nexus's own public statements describe the data-center side of the campus with figures that are much smaller in the near term and grow over a multi-year horizon. Both are accurate; they describe different sides of the same project.

Two numbers, two sides of the same campus

On the generation side — the TCEQ permit sets the ceiling for what the plant would be authorized to operate. Every combustion turbine and reciprocating engine in the equipment inventory is permitted and environmentally reviewed up front, regardless of whether the corresponding data-center load yet exists to consume it.

On the data-center side — Nexus has described a phased buildout in statements to industry press and in public project filings:

• Roughly 500 MW of data-center capacity online in the first phase (late 2026), per company statements reported by industry media (Baxtel, Data Center View, Data Center Map).
• 612 MW (30 buildings at about 20 MW each) described in the December 2025 filing that preceded the current expanded application.
• ~7.7 GW of total facility power as the eventual full-buildout target that Nexus has publicly stated, per reporting by industry trackers and Global Energy Monitor.

What the math implies at each phase

The middle row is an interpolation between the company's first-phase and full-buildout figures; Nexus has not published a detailed phase-by-phase schedule. The generation figure in every row is the permitted capacity, not an operating load factor. Units that are installed but not yet needed would simply sit idle.

In the early years, the majority of the permitted generating fleet would have no contracted on-site load. Three possibilities exist for how that unused capacity is handled:

1. Excess units remain installed but off, consistent with Nexus's stated behind-the-meter strategy and with how permitted capacity commonly ramps at phased plants.
2. Nexus later applies for partial ERCOT interconnection to sell surplus generation as merchant power. The air permit does not prohibit this; it regulates emissions, not the business model. Grid interconnection at any later phase would be handled separately by ERCOT and the Public Utility Commission of Texas.
3. The data-center buildout accelerates faster than publicly scheduled, closing the gap sooner than the multi-year horizon.

Why Nexus chose behind-the-meter — and why that matters

"Behind-the-meter" means the power plant is not connected to the ERCOT grid; electrons move from the turbines directly to the on-site data center through private infrastructure. Nexus has publicly cited speed-to-market as the reason: bypassing ERCOT's large-load interconnection queue (backlogged at roughly 233 GW of pending requests) and the rate-base oversight that comes with traditional utility generation.

Crucially, behind-the-meter is a commercial and interconnection choice, not a regulatory category imposed by TCEQ. The air permit regulates emissions; it does not determine whether the plant can eventually sell power to the grid. A shift toward merchant operation at any later phase would not require re-opening the air permit.

How the physical infrastructure compares to a standalone power plant

Setting the "data center" framing aside, the physical reality of the site is this:

• By permitted generation capacity, 7.2 GW would place this site among the top five largest power stations in the United States and larger than every existing power station in Texas, including W.A. Parish (3.65 GW) and Comanche Peak (2.4 GW).
• By emissions inventory and permit burden, this is a utility-scale natural-gas generating facility, requiring the same federal PSD, greenhouse-gas PSD, and state NSR analyses any major gas plant of this size would require — regardless of where the electrons eventually go.
• By footprint, the 55 combustion units along the northern boundary and in the detached southeast cluster dominate the site plan. The data halls ("DTC") occupy the interior.

Texas needs more generation. The question is where, and on what terms.

ERCOT has been transparent about the state's need for additional generating capacity — summer peak demand keeps rising, the large-load queue (driven largely by data centers and industrial electrification) has grown dramatically year over year, and reserve margins remain tight. New gas generation is part of the near-term answer most industry analysts expect.

What is less settled — and, for neighbors, what can feel less settling — is where that generation should sit, how emissions are controlled, how closely the performance is monitored, and how the communities that will live alongside it are involved in the decision. The TCEQ permit process is designed to surface exactly those questions. On a site of this scale, the answers matter for the residents of Hubbard, Dawson, Mt Calm, Malone, Mertens, Penelope, and surrounding communities — regardless of which label (data center or power plant) appears in public-facing materials.

Sources: Nexus Hubbard Power, LLC air-permit application (Section 3 — Project Description & Emission Estimates); company statements reported by Baxtel, Data Center Map, Data Center View, and Data Center Dynamics; project tracking by Global Energy Monitor and Cleanview; ERCOT large-load queue context from Utility Dive.

Equipment inventory — every turbine and engine, unit by unit

The 7.2 GW headline figure isn't marketing — it is straight arithmetic on the per-unit ratings Nexus itself filed with TCEQ. The table below shows every line of that arithmetic.

The application identifies 55 combustion turbines and reciprocating engines designed for continuous operation, plus 6 emergency-standby diesel generators. The MW ratings below are taken directly from Section 3 of the application; adding them up yields the 7.2 GW headline figure used elsewhere on this page.

Capacity summary: approximately 7,230 MW (~7.2 GW) of continuous generating capacity at full buildout, plus ~19.5 MW of emergency-standby capacity limited by permit to a maximum of 100 hours per year of non-emergency operation. Per-unit MW figures are verbatim from Section 3 of the air-permit application. The "~7 GW" headline elsewhere on this site is the rounded continuous-operation subtotal; it does not include the emergency diesels.

Source: Nexus Hubbard Power, LLC air-permit application, Section 3 (Project Description & Emission Estimates) and the Site Plan – Emission Sources figure.

Associated data-center built environment: the modeling protocol's building-downwash input (Table 3-1 of Appendix E) depicts 23 data-center buildings on the site, each with a footprint of roughly 122 m × 396 m (~400 × 1,300 ft) and a height of 10.67 m (~35 ft). A separate battery-energy-storage building measures roughly 472 m × 101 m (~1,550 × 330 ft). For scale, a single data-center building at these dimensions covers about the floor area of ten NFL playing fields, and the 23-building cluster alone accounts for roughly 275 acres of covered footprint — about a third of the 800-acre parcel, not counting the power block, substation, engine halls, battery building, and internal roads.

Location and surrounding receptors

A church, a medical clinic, a public park, and a school/daycare all sit within one mile of the facility boundary. Direct mailed notice reaches only a small subset of households inside that ring.

The proposed site is located at 880 State Highway 171 in Hubbard, Hill County, Texas 76648, immediately southeast of the City of Hubbard and near the Hill County / Navarro County line. It lies approximately 20 miles east of Interstate 35W at Hillsboro, 30 miles north of Waco, and 80 miles south of the Dallas–Fort Worth Metroplex. Nearby communities include Dawson (Navarro County, ~8 mi southeast), Mt Calm (Hill County, ~6 mi southwest), Malone (Hill County, ~7 mi west), Mertens (Hill County, ~12 mi northwest), and Penelope (Hill County, ~10 mi northwest).

The applicant's own Area Map identifies four sensitive receptors within one mile of the facility boundary — a church, a medical clinic, a public park, and a school/daycare — all clustered in or near the City of Hubbard to the northwest.

The application also depicts a 3,000-foot ring and a one-mile ring around the facility boundary as the primary public-notification radii — the distances within which the applicant identified landowners whose land falls inside the required notice zone. In practice, TCEQ's "direct notice" requirement for a major-source air permit is largely satisfied by a sign posted at the site entrance and a legal notice published in a local newspaper of general circulation. Individual certified-mail letters are required only for narrow categories (for example, elected officials, school districts, and specific adjacent landowners); the broader population inside the one-mile ring often learns of the application only by encountering the posted sign or the newspaper notice. Several parcels of privately owned land share fence lines with the proposed boundary, and a few are nearly surrounded by it on multiple sides.

Pollutants requested in the permit

Nine separate categories of regulated air pollution — including nitrogen oxides, fine particulate, formaldehyde, and ammonia slip — are requested across the four source classes on the site.

Pollutants by source category

Source categories and control technologies are as listed in the applicant's BACT analysis and Table 2 emissions summary. Pollutants shown are those quantified for each source category in the application. Tan chips in the emergency-generator row indicate pollutants that diesel combustion produces but the gas-fired units do not.

Why "major source" matters

The application explicitly states that the project will be a "major source" under both the federal Prevention of Significant Deterioration (PSD) program and the federal greenhouse-gas PSD program. Major-source designation is not a formality — it triggers a stricter review, including a Best Available Control Technology (BACT) analysis and an air-quality impact analysis using AERMOD, the EPA's regulatory-standard atmospheric dispersion model — a computer simulation that predicts how emissions from the stacks travel through the air and what pollutant concentrations reach the ground at the property line and at nearby receptors. The application contains a BACT analysis, an AERMOD modeling protocol prepared on Nexus's behalf by SLR International Corporation of Tomball, Texas (SLR Project No. 120.021834.00001), and tables comparing the proposed emission limits to entries in the EPA RACT/BACT/LAER Clearinghouse (the national registry of control-technology determinations regulators consult to decide what "best available" looks like for a given pollutant and source type).

What the modeling shows — headroom, not abundance

The local airshed has almost nothing in it today. The nearest ozone monitor sits 3 ppb under the federal standard, and TCEQ's own approved methodology already pushes the project past that line.

A major-source air permit is not approved on emission totals alone. The applicant also has to show, through AERMOD dispersion modeling, that the concentrations the project causes at nearby receptors stay inside the federal National Ambient Air Quality Standards (NAAQS) and, for PSD pollutants, inside the PSD Increments. Nexus's Air Quality Analysis Protocol, prepared by SLR International, walks through this exercise pollutant by pollutant. The numbers in it are revealing: for most pollutants the project would stay below the NAAQS, but the margin for ozone is thin enough that the answer depends on a modeling input choice TCEQ has already flagged.

The airshed around Hubbard is essentially empty today

One reason the scale of this project stands out is that it is being proposed into an airshed that currently has almost nothing in it. TCEQ's own 2023 State Summary inventory — cross-checked in a spreadsheet TCEQ attached to the public-information response — identifies only two permitted sources within ten kilometers of the proposed site, and both of them report zero tons per year of every criteria pollutant. The entire Hill County inventory for 2023 totals roughly 9.6 tons of NO_X, 44.1 tons of CO, 27.1 tons of PM₁₀, 15.9 tons of PM_2.5, and 2.4 tons of SO₂ across every permitted emitter in the county combined. Nexus's Air Quality Analysis uses a worst-case NO_X figure of 1,599.66 tons per year (slightly higher than the 1,549.50 tpy permit ceiling because the model uses conservative potential-to-emit assumptions). That single project, on a single parcel, would add roughly 166 times the entire existing NO_X inventory of Hill County — into a ten-kilometer radius that today has effectively no other emitters.

Ozone — a three-parts-per-billion margin

Ozone is not emitted directly by the facility; it forms downwind when the NO_X and VOC emissions from the plant react in sunlight with other pollutants already in the air. TCEQ evaluates that secondary formation using a methodology called MERPs (Modeled Emission Rates for Precursors), which converts the project's NO_X tonnage into an estimated ppb contribution to 8-hour ozone at the nearest regulatory monitor.

The nearest ozone monitor on the regulatory network is the Italy site in Ellis County (AQS 48-139-1044), whose 2024 design value is 67 ppb. The 8-hour ozone NAAQS is 70 ppb. That leaves only 3 ppb of headroom before a new source would push the regulatory design value to the standard.

The MERPs calculation depends on which hypothetical source in TCEQ's pre-approved list the applicant picks to represent the project. Nexus ran the calculation two ways and put both in the file TCEQ released:

Values taken directly from the "O3 MERPs Calculation.xlsx" workbook that TCEQ released in Public Information Request 114407. In its March 17, 2026 Modeling Protocol Review Response, TCEQ noted that the 3,000 tpy Parker source is not on TCEQ's pre-approved APDG 6232 list and asked Nexus to rerun the analysis using one of the approved sources. Under the approved input, the modeled total at the Italy monitor sits just over the NAAQS rather than just under it.

Fine particulate — secondary formation is already most of the SIL

The same MERPs methodology applies to PM_2.5 formed downwind from the project's NO_X and SO₂. TCEQ's Significant Impact Level (SIL) for 24-hour PM_2.5 is 1.2 μg/m³ — the threshold above which a project must go beyond a screening analysis into a full cumulative NAAQS and PSD Increment evaluation. From the workbook TCEQ released:

Values from the "Secondary Formation of PM2.5 (MERPs)-based on 0.13.xlsx" workbook released in PIR 114407. Under the approved-list basis, secondary PM_2.5 formation alone would consume more than four-fifths of the 24-hour SIL, before any of the project's direct PM_2.5 emissions are added. Under either basis, the project-caused secondary concentration falls short of the SIL; with direct emissions layered in, the combined result is closer to the SIL than the 49% headline suggests.

Why this matters for a public comment

The modeling is not a veto — TCEQ can accept a thin margin and issue the permit anyway, and an exceedance against a monitor's 2024 design value does not by itself mean the permit is denied. What the numbers establish, however, is that this project is not being proposed into a zone with comfortable air-quality margins. It is being proposed into an airshed that currently has almost no other emitters, into a monitor whose design value sits just 3 ppb below the ozone standard, and where the methodology TCEQ itself prefers already pushes the modeled total past that standard. Public comments that ask TCEQ to (a) require use of the approved MERPs basis, (b) require fresh AERMOD and AERSURFACE model versions, and (c) treat the Italy monitor's 3-ppb headroom as a binding constraint rather than a starting assumption are exactly the kind of technical input TCEQ has an obligation to address in its Response to Comments.

Noise and light — outside TCEQ's permit, but real for neighbors

If the applicant chooses not to mitigate, the soundscape and the night sky around Hubbard will change permanently — and TCEQ has no authority over either. Public comments are how those concerns reach the official record.

TCEQ's air-permit decision does not weigh noise, light, or property-value impacts. Those are regulated — when they are regulated at all — through other channels: county and municipal ordinances, common-law nuisance claims, and private negotiation with the applicant. For a facility operated continuously at this scale, however, both the acoustic environment and the night sky around the site will change materially. Because TCEQ will not factor these into the permit itself, raising them in formal public comments is the way to put community concern on the official record.

Noise — what 55 continuously operating combustion units sound like

Several classes of equipment on this site are commonly dominant noise sources at gas-fired power blocks and data-center campuses: the large frame and aeroderivative gas turbines with their inlet and exhaust systems; the halls housing the 24 Everllence reciprocating engines; the balance-of-plant equipment — air-coolers, radiators, and cooling towers (the application references "advanced cooling technologies such as chilled water loops"); the high-voltage transformer bays; and, during periodic testing, the six emergency diesel generators.

At the equipment itself, unmitigated near-source sound levels for this class of fleet are typically in the 80–90+ dBA range — from the louder end of busy city traffic up into levels associated with hearing loss under sustained exposure. Well-engineered installations can be driven substantially lower at the fence line and at nearby residences: modern plants often design to approximately 50–60 dBA at the facility boundary and into the low 50s at homes, particularly at night. The measures that make the difference are engineering commitments the applicant can choose to make: acoustic enclosures, inlet and exhaust silencers, barrier walls, siting and massing of buildings between sources and residences, and operational restrictions such as limits on nighttime engine testing. None of this is mandated by the TCEQ air permit; it is secured, if it is secured at all, through voluntary applicant commitments or through county and municipal requirements.

Light — dark-sky standards the application does not address

A 24/7 facility with large power blocks and data halls requires continuous exterior lighting — area lighting, security lighting, and task lighting around turbine halls, engine buildings, cooling systems, tank farms, and the high-voltage yards. The application does not enumerate luminaires or propose a lighting plan, but typical industrial campus lighting is material: if fixtures are unshielded, if color temperatures skew cool (high blue content), or if illumination levels exceed what the task actually requires, the result is skyglow visible for miles and direct glare at neighboring homes, the school/daycare, the church, and the medical clinic the applicant itself identified as sensitive receptors.

The established mitigation framework for a site of this kind is a dark-sky-compliant lighting plan aligned with the joint International Dark-Sky Association / Illuminating Engineering Society Model Lighting Ordinance (MLO), using the IES BUG (Backlight–Uplight–Glare) classification system to specify no uplight (U0), low backlight, and low-glare distributions appropriate to a rural context. Warm color temperatures (2700K or below), fully shielded fixtures, time-of-night dimming, and motion-activated rather than always-on security lighting are the standard tools. As with noise, none of this is forced by the air permit — it is a set of commitments the applicant can make, or not.

Sources: equipment-noise expectations drawn from Nexus Hubbard Power application (Section 3 equipment list); noise-impact background from the Environmental and Energy Study Institute; decibel reference from Yale Environmental Health & Safety; dark-sky framework from the International Dark-Sky Association and the Illuminating Engineering Society; light-pollution baseline from lightpollutionmap.info.

Independent fence-line monitoring — going live Summer 2026

Reference-grade instruments will be installed within 10 feet of the current property line on private land adjacent to the site, capturing air, noise, and night-sky data continuously — and publishing it on this page in real time.

An independent network of reference-grade air-quality, noise, and night-sky monitoring stations is being installed within 10 feet of the current property line, on private land that is adjacent to the site. Instruments will begin capturing data as they come online this summer, with continuous public publishing on this page starting in Summer 2026. All data will be real-time, timestamped, free to access, and downloadable — establishing a baseline record before construction-phase emissions begin and continuing through operation.

Methodology & transparency

As each station comes online this summer, all raw data, calibration records, instrument metadata, and quality-assurance procedures will be published in a public archive. Data will be released under a permissive open-data license so researchers, journalists, regulators, and members of the public are free to download, analyze, and republish it. Establishing a reliable baseline before operations begin is one of the central goals of the monitoring program.

Status & timeline

The application is in expedited TCEQ technical review right now. The window for affected persons to request a contested-case hearing opens later this year — once it closes, it does not reopen.

Source: 30 TAC Chapter 39 (Public Notice) and Chapter 55 (Requests for Reconsideration and Contested Case Hearings).

Documents and source materials

Every figure on this page is sourced from the public TCEQ filing. The full record — application, modeling protocol, agency correspondence — is linked here so any reader can verify the numbers themselves.

The complete public record for this application is maintained by TCEQ. The links below are the canonical sources for everything summarized on this page — so you can read the underlying filings yourself.

Four ways to participate

A signed comment from a single resident, a single church, or a single school can trigger a TCEQ Response to Comments and a public meeting. Most permits never receive any.

Who counts as an "affected person"

"Affected person" is the legal threshold to request a contested-case hearing. It is narrower than "concerned" — and TCEQ judges it case by case.

Eligibility for contested-case-hearing standing is governed by 30 TAC §55.203. The agency considers, among other factors:

• Distance. How close the requester's interest — property, residence, business — is to the facility.
• Nature of the interest. Whether the interest is in a legal right, duty, privilege, power, or economic interest.
• Likelihood and degree of impact. Whether the requester is likely to be impacted differently than the general public.
• Distinguishability from the general public. Whether the impact on the requester is qualitatively or quantitatively different from the impact on the public at large.

Adjacent landowners, landowners whose property shares a fence line with the facility, and landowners nearly surrounded by the proposed boundary have historically been treated as having strong affected-person standing.

Frequently asked questions

Plain-language answers to the questions readers most often raise about the project — what it is, who is behind it, and what is and isn't decided yet.

Contact this site

For questions, corrections, tips, or additional public records relevant to this application, reach out by email. This address is monitored by the volunteers who maintain this site.

hubbardinfo@proton.me

This email address is for inquiries about this website only. Official comments on the air-permit application must be submitted to TCEQ through the Office of the Chief Clerk during a designated comment period; correspondence sent here is not entered into the agency's official record.