Title 15 Amendment Analysis for Spanish Fork

1. Executive summary

The choice is not this warehouse site versus an empty field, it is the proposed data center versus the distribution warehouse the building was originally built for. Against that alternative, the proposed data center build is better on every measurable dimension for the city.

No drawback to the community was found in the facility as proposed and as permitted. Sound and heat are the only effects that travel, and the nearest existing home is 1,298 ft away.⁵³ Against that alternative, truck traffic goes from thousands of diesel trips a month to none, water use drops, there is no diesel on site, and the municipal grid carries none of it. §6 sets out the alternative use side by side, §10 the air permit, §20 sound and heat.

The tax base benefits the whole community, not just the city. Roughly 90% of the levy goes to entities serving these same residents: Nebo School District takes 75.2%, about $30.1M in year one against the city's own $4.0M.¹⁵⁵⁹ Volition has separately committed to more than $500,000 a year in direct community contributions, itself 15× the city's property tax take from its largest commercial taxpayer.⁵⁷⁷³ §13 carries the arithmetic, the depreciation curve, and the caveat that no development agreement is signed yet, though Volition has so far accepted every substantive suggestion from the community and from staff.

The performance standards do their job. §6 caps noise at 65 dBA at any property line and 55 dBA at any residential zone boundary, at any hour, covering generation, storage and cooling equipment, measured under Chapter 9.32 and without displacing Chapter 9.32 enforcement.⁵¹ Evaporative cooling is banned outright, diesel is excluded by definition, and the permitted emission rates are a zoning condition the city can enforce. §8 and §20 carry the detail.

One question is worth putting to the Council on Tuesday, and it is a point of clarification: The amendment binds every applicant in the I-1 zone at up to 500 MW per facility.¹⁵¹ Its redline struck "approved as a conditional use" and "conditionally permitted" from every operative clause, while leaving the use heading pointing at "(see §15.3.08.060)", the Conditional Use Permit chapter that carries a public hearing and five findings.⁵¹ So the enacting text reads as administrative site plan review, and one stray cross-reference is the only thing suggesting otherwise. §9 sets out the redline in full.

2. Scope and sourcing

This report prioritizes primary sources across official municipal, city, county, state, and federal records, plus an adversarial red team review of each claim and individual verification of every cited source. The binding ordinance text comes from the 18 AUG Council packet with page-level citations and color-verified redlines.⁵¹ The city's official 04 AUG City Council meeting, 05 AUG Planning Commission meeting, and the 15 AUG Volition community meeting supply the record. The applicant's own materials from that meeting, its printed two-sided handout⁶¹ and the live project page its QR code resolves to,⁴ are treated as primary sources and audited in full. Parcel, tax, and receptor geometry come from the Utah County Assessor and Recorder and from computation against the state's parcel polygons. Field research was carried out directly: on-site visits to comparable operating facilities, in-person interviews with residents, employees, and neighboring businesses, and uncalibrated sound level readings, all reported below with their limits stated.

Everything material here is fully sourced unless otherwise marked. Where a claim could not be confirmed against a primary document it is marked UNVERIFIED. Where a physical observation was taken without professionally-calibrated instruments it is marked UNCALIBRATED and treated as corroboration only. Where the applicant has committed to something on the record but no signed instrument exists it is marked COMMITTED, NOT EXECUTED: the commitment is taken as made in good faith, and nothing is final until a written agreement signed and dated by both the City and Volition exists.

3. Bottom line

Volition describes a 250 MW compute load; the air permit it has applied for would authorize 393.25 MW of nameplate, 1,210 fuel cells at 325 kW each.¹³ The ordinance would authorize 500 MW per generation facility, and its two-site quota expressly does not apply to standalone generation facilities.¹¹⁵¹

The real risk is that this zoning decision is the only discretionary review any facility built under this text will ever receive. Utah's 2025 S.B. 132 created a "closed private generation system" category describing this project exactly and exempting it from Public Service Commission regulation; no interconnection means no utility review.¹² DAQ's minor-source order is ministerial and weighs no siting question. Spanish Fork land use is the last stop in the system for oversight of any kind. Which makes one drafting question worth putting on the record Tuesday: the amendment's redline struck conditional-use status from every operative clause but left the use heading citing the Conditional Use Permit chapter, so the text reads as administrative site plan review with one stray cross-reference against it. §9 sets out the redline.

4. What the amendment authorizes, at its own maximum

§15.3.24.090 permits "not more than 500 megawatts of nameplate generating capacity, measured per Onsite Power Generation Facility," and no less than 50.⁵¹ 500 MW is the number the Council is actually voting on, and it is 27% larger than the 393.25 MW Volition has applied for.¹⁵¹

The figures below use Utah DAQ's own method. DAQE-IN163550001-26 derives every permitted limit as the manufacturer's published rate times full nameplate times 8,760 hours.⁴³ Applying that arithmetic at the ordinance's own cap is not a hypothetical worst case; it is the state's method at the city's number.

TABLE 01

Two sites at the cap would reject more heat than Lake Side Power Plant, the 1,315 MW-thermal gas plant in Vineyard, and would burn 49 to 60 billion cubic feet of natural gas a year.³³

The two-site quota caps data center self-generation power levels

§13, verbatim:

The citywide ceiling is therefore not 1,000 MW. It is 1,000 MW of data-center-paired generation plus an uncapped number of standalone Onsite Power Generation Facilities, each permitted up to 500 MW.¹¹ This is just a noted observation for future consideration.

Where this could go, and where it could not

Any parcel in the I-1 Light Industrial zone qualifies. The amendment is citywide in application, not site-specific: the applicant requested a text amendment, not a rezone of one parcel. The next applicant need not be Volition, need not use Volition's equipment, and need not carry any of the six covenants Volition volunteered. It needs only to meet the objective conditions in §15.3.24, and it is approved administratively.¹¹

How far that opens is itself a drafting question. The qualifying test is a performance standard, not a technology: equipment must be CARB-certified or emit NOx at no more than 0.5 lb/MWh, and reciprocating internal combustion engines are excluded outright.⁵¹ Fuel cells are named as equipment that meets the standard rather than required by it. But the redline struck the words "without limitation" from that list, and struck "recuperated microturbines" out of it, which is the standard way an open list becomes a closed one.

What the ordinance does constrain: generation must be Ultra-Low Emission Technology, permanently islanded, non-evaporatively cooled, held to 65 dBA at the property line and 55 dBA at any residential boundary at all hours, water-capped by permit condition, and subject to battery fire-safety review.¹¹ No other Utah jurisdiction examined imposes any of those.

5. The local yardstick

what 500 MW looks like against everything already here

Every figure here comes from Utah DAQ's reported emissions inventory: 256 facility-year records covering 71 permitted Utah County facilities, 2017 through 2025.⁹ These are reported actual numbers.

Every permitted air source inside Spanish Fork

There are three. Grouping by DAQ facility ID rather than company name collapses two pairs that are the same plant under successive owners: DAQ_ID 10847, Sapa Extrusions and Hydro Extrusion USA, one facility whose street address moved from 1550 N Kirby Lane to 1550 Hydro Way after Norsk Hydro renamed the road; and DAQ_ID 11789, Longview Fibre and WestRock Longview, one facility at 2200 North Main Street with identical coordinates in every year.⁹

TABLE 02

Hydro Extrusion is 81% of every permitted ton of NOx in Spanish Fork, and nothing else is close.^{43 35}

Comparison of NOx

TABLE 03

A 393 MW data center is permitted for about the same annual NOx as the region's main hospital, a 395-bed facility. On NOx specifically this technology is clean, and the state's own inventory proves it. A facility at the full cap would rank below the aluminum plant Spanish Fork already hosts.⁴¹¹ Two sites at the cap would emit 73% of Hydro Extrusion's total and become the city's second-largest NOx source, four times WestRock.⁴³ (Spanish Fork's own 33-bed hospital holds a DAQ Approval Order but reports no emissions yet, so Provo's 395-bed facility is the meaningful hospital comparator.)⁹⁵⁴

Everything Spanish Fork generates today is wind, solar, or water. It has never had a thermal power plant. Its neighbors have. UAMPS Nebo, 20 minutes down the road, already reports 8× the NOx a 500 MW facility would be permitted for, and PacifiCorp's Lake Side in Vineyard, 1,204 MW combined, reports 176 tons, 27× the 500 MW figure.¹¹³³³⁴ Lehi, Springville and Payson each run their own plants, at 7.39, 3.37 and 0.57 tons respectively in 2025.⁹

Every generator that already exists inside Spanish Fork

From the Utah Geological Survey electricity series, reproducing EIA Form EIA-860 filings. All capacities are AC, and EIA lists only solar and hydro units above 1 MW.

TABLE 04

The wind park runs nine Suzlon S88 2.1 MW machines at a 33.52% capacity factor, selling under a PSC-approved power purchase agreement; it is the only utility-scale wind generation in Utah County.⁸¹¹ The solar park is UMPA's, sold to residents through the city's SharedSolar program and built on a closed landfill.⁷²¹²² The hydro plant is on Powerhouse Road, and it is the reason the road has that name: three units under the Strawberry Water Users Association, the oldest in service since 1937.¹⁰⁵⁵ Powerhouse Road serves the current warehouse site, and the city straightened it for this building's truck traffic.^6112343

Only one of those three actually supplies Spanish Fork customers. The wind park sells to PacifiCorp under a 20-year PPA and the hydro plant to the South Utah Valley Electric Service District; only the 4.0 MW solar park serves the city.⁶⁷⁸²¹ Spanish Fork is not a Rocky Mountain Power city, it runs a municipal utility supplied wholesale by UMPA, whose largest resource is market purchases, 52% of all energy delivered.⁵

Filed nameplate against filed nameplate, with no assumptions: Volition as permitted is 14.8× all 26.5 MW of existing city generation, one facility at the 500 MW cap is 18.9×, and two sites at the cap are 37.7×.¹⁸¹¹⁴³

Who actually uses the power, and what the amendment would add to it

UMPA files Spanish Fork's retail book by customer class (FY2022 Customer Profile, IRP p. xiii). This is the city's electrical demand:

TABLE 05

UMPA does not name customers in thier paperwork, but the filing establishes the ceiling. Hydro Extrusion and Mountain Country Foods are both certainly inside the five-account Large Commercial class, which consumes 48.02 GWh a year in total, so even if one plant were that entire class it would use 48 GWh a year.⁵²⁸

TABLE 06

One facility built under this amendment would consume eleven times the annual electricity of every home, business, farm, and streetlight in Spanish Fork combined, roughly 68x the entire class its largest industrial neighbors sit in, and 19x every generator in the city. Two sites would be 22x the city.⁴³ It also explains why the off-grid design is an engineering necessity rather than a marketing choice: there is no version of this facility that the city's distribution system could carry. The building's own leasing listing puts the point beyond argument: 4,000 A at 480 V with roughly 6 to 8 MW of additional power advertised as available, against a permitted 393.25 MW.⁵⁵

Additional context

NOx is the dimension where this technology wins, and it is the only one. None of these comparators carries anything close to the facility's CO2e, rejected heat, or gas throughput at the cap; Lake Side, the one Utah County facility at comparable thermal scale, would be exceeded by two sites at the cap.³³

6. What is actually being proposed

Volition Industries, Inc. applied on 10 JUL 2026 for a Spanish Fork Title 15 text amendment permitting data centers with on-site generation in the I-1 Light Industrial zone. The target is the existing 1,054,000 sq ft former Fingerhut building at 4000 E US Highway 6, marketed since as Western Distribution Center and now as Canyon Logistics Center, at the mouth of Spanish Fork Canyon.⁵⁵¹⁰

The warehouse data

TABLE 07

First, the vacancy. 451,792 of 1,054,000 sq ft, about 43% of the building, is being actively marketed, which is what "partly occupied" means in this report and is the closest thing to a measured baseline that exists.⁵⁵

Second, the listing quantifies the grid constraint from the property's own side. The building is served at 4,000 A, 480 V, 3-phase, roughly 3.3 MVA, and its broker advertises 6 to 8 MW of additional power as available.⁵⁵ Volition's permitted 393.25 MW is ~50 to 65 times that headroom, and the ordinance's 500 MW cap is 60 to 80 times it.¹¹¹⁴³

And the 36 dock doors are the truck-traffic argument. A building with 36 dock-high positions and one grade door was designed for continuous trailer cycling. The data center uses none of them.⁵¹

The alternative use, side by side

The counterfactual is this building operating as designed, not an empty field.

TABLE 08

Truck traffic

Every truck figure below is computed from published ITE rates applied to the verified 1,054,000 sq ft, using rate equations reproduced by two state transportation departments.⁴¹⁴²⁴³ No traffic count for this building exists in any public record, so this describes a fully tenanted building, not present conditions of utilization, which are known to be very low.

ITE defines a high-cube warehouse as one with at least 200,000 sq ft and a ceiling height of 24 feet or more.⁴² This building is 1,054,000 sq ft with 36 ft clear height, so it is unambiguously high-cube, and Land Use Code 154 fits it better than the general Land Use Code 150 that warehouse traffic studies default to.

TABLE 09

Total weekday vehicle trip ends of 1,711 are the ITE LU 150 equation T = 1.58(X/1,000) + 45.54 at X = 1,054,000 sq ft.⁴¹ Monthly figures use 21.75 weekdays and 8.69 weekend days, with weekend volume scaled by each code's own published weekend-to-weekday ratio. ITE rates are trip ends, so one truck arriving and leaving is two trips.⁴³

The dock doors provide an independent physical check, and it corroborates the lower half of the range. At 36 dock-high positions, the three methods imply 3.2, 4.8 and 7.9 trailer loads per door per day on a 24/7 operation. Three loads per door per day is ordinary; eight is a hard-running cross-dock. Working the check the other way, 36 doors at three loads per door per day gives 6,575 truck trip ends a month, which lands on the LU 154 figure almost exactly.⁴³

TABLE 10

Scale in context: the whole city's peak electrical demand is roughly 90 MW. UMPA's Integrated Resource Plan, filed 30 NOV 2022, records Spanish Fork City's FY2021 annual peak at 74.10 MW against 297,313 MWh of annual sales for 14,468 customers, and forecasts that peak growing to 115.76 MW by FY2027 (IRP pp. 81 to 82).⁵ The ~90 MW figure used in the city record on 04 AUG sits between the two, so this report uses it as the current-era number and shows the FY2027 forecast alongside.⁵¹ Every scale multiple below survives either figure. At 250 MW the facility generates ~2.8× the whole city's peak consumption; at the 500 MW cap, ~5.6×, or 4.3× against the FY2027 forecast.¹¹⁵¹ It also explains why the off-grid design answers a real constraint rather than a marketing one: city staff put a grid-connected 50 MW load at "two new substations… 30 to 40 million dollars," and a 250 MW grid-tied load in Spanish Fork is not possible at all.⁵¹

7. Where the procedural record actually stands

TABLE 11

The Planning Commission did not endorse this. It deadlocked. Text-amendment review is advisory, so it advances regardless.

Two councilmembers asked for exactly this discretion on 04 AUG

The city's own approved-form minutes (packet pp. 59–71) record this exchange during the large-load work session, before the Volition presentation:⁵¹

The amendment reaching the Council on 18 AUG does not contain that review, and nothing in the public record shows that examination happening. Two councilmembers and the Planning Commission have now asked for discretion the ordinance does not contain, and no one in the public record argued for removing it. Staff's rationale, that pre-set objective standards are more legally defensible than ad hoc discretionary ones, is mainstream and legitimate, but it was never weighed against those asks in any document available to the public.

8. The ordinance, provision by provision

Source: https://www.utah.gov/pmn/files/1467389.pdf, pp. 61–66.

Genuinely strong:

• "Ultra-Low Emission Generation Technology", CARB-certified or NOx ≤ 0.5 lb/MWh; explicitly "does not include reciprocating internal combustion engines." Objective, numeric, verifiable, and it categorically excludes the gas engines Novva and Joule Capital actually built.¹¹ Bloom's ES 6.5 datasheet states 0.003 lb/MWh of NOx, a 167× margin.³ (0.07 lb/MWh is the CARB DG certification standard, the ceiling a unit must beat to be certified, not Bloom's rate.)³⁰
• §F6 Noise is the strongest performance standard in the document. Verbatim: sound "including generation equipment, Energy Storage Systems, and cooling equipment, shall not exceed 65 dBA at any property line of the site, and shall not exceed 55 dBA at the boundary of any residential zone, at any hour, measured in accordance with the procedures of Chapter 9.32 of this Code," and "compliance with this subsection does not excuse compliance with Chapter 9.32."⁵¹ So it is a continuing limit, not a design target; it names cooling equipment, the source that generates complaints nationally; it adopts a measurement procedure; and it preserves the city's standing enforcement. Against the existing I-1 limit of 80 dBA day / 65 dBA night, a facility here must meet the city's nighttime standard around the clock. Volition wrote this. The one omission is a scheduled test after construction, leaving enforcement to complaint under Chapter 9.32.
• §I9, closed-loop, air-cooled, or dry cooling required; evaporative and once-through banned. This converts the water promise into law and is the most valuable protection in the document.
• §E5, permanent electrical isolation, no interconnection, parallel operation, or export.
• §9 Water Use was strengthened: applications must include "an annual water use estimate, by category of use", explicitly covering landscaping, reviewed by the Public Works Director, and "the approved estimate shall be a condition of the permit." This closes the landscaping gap that went unregulated at Novva.
• §7 Air Quality requires the facility to "at all times operate within the emission rates demonstrated under subsection C", making the emission rates a zoning condition the city can enforce, not merely a DAQ matter.
• §4 Energy Storage requires UL 9540 listing, UL 9540A large-scale fire test data, NFPA 855 compliance, and Fire Marshal review with an emergency response plan before certificate of occupancy.

The text carries a capacity ceiling, and it is the number to argue about. §2 reads "…a minimum of 50 megawatts and not more than 500 megawatts of nameplate generating capacity, measured per Onsite Power Generation Facility."⁵¹ Note the floor as well as the ceiling: nothing under 50 MW qualifies at all, so every facility this text authorizes is a large one. The section above works through what that permits at the cap.

Still materially weak in the text going to a vote:

• No heat or thermal standard of any kind. The Commission's third recommendation, a deeper discussion of heat impacts, did not make it into the ordinance. At 605–742 MW-thermal this remains the clearest substantive gap, and the Commission itself asked for it.⁴³
• §K11 decommissioning: unchanged. Exhibit A still reads "The City may require a decommissioning plan" within 12 months. No bond, no surety, no escrow.⁵¹
• §J.10, sewer discharge is NOT prohibited, only subject to pretreatment. Staff confirmed this on the record. The Volition representative testified "there is no scenario where we are offloading water." The ordinance and the testimony cannot both be the operative commitment.
• No GHG provision, and no regulator in this process will supply one. Utah DAQ quantifies CO2e in approval orders but does not cap it, and GHG BACT does not reach minor sources.
• The by-right mechanism, the one item on this list worth a question on Tuesday. Exhibit A contains zero occurrences of "Conditional Use Permit", "conditional use" or "Planning Commission may impose", against three of "site plan application", and the applicant's own §L discretionary-conditions clause does not exist in the enacting instrument.⁵¹ The redline shows the removal was deliberate, striking conditional-use language from four separate clauses while leaving the heading's "(see §15.3.08.060)" cross-reference in place. §9 sets that out in full, and it is the single clarification this report recommends asking for.

Of the six volunteered covenants, five are substantially in the text: no grid interconnection (§5), the noise limit (§6), no diesel engines (via the Ultra-Low Emission definition), decommissioning (§11, but without financial assurance), and the water use cap, which §9 makes a binding permit condition. Only annual public reporting is missing. No draft development agreement is publicly posted anywhere.

9. The best question on the agenda

Exhibit A still points to the hearing chapter

This is the strongest finding in the review, and no keyword search will surface it. It was found by following a cross-reference rather than searching for a phrase.

What Exhibit A contains, and what it does not

Exhibit A is §15.3.24.090, subsection O, items 1 through 13. Read in full, two things are true at once.⁵¹

TABLE 12

So the enacting text routes these facilities through site plan review, contains no Planning Commission conditioning authority anywhere, and still carries a heading that points at the Conditional Use Permit chapter, where the public hearing and the five mandatory findings live.¹¹ Whether that cross-reference imports those procedures or is inert decides whether any facility built under this text ever gets a hearing.

The packet contains two documents that describe different ordinances

The justification statement travels in the same packet the Council votes from, and on four points it does not describe Exhibit A:⁵¹

TABLE 13

One more feature of the current text. The definition of Ultra-Low Emission Generation Technology reads "includes, fuel cells, and linear generators meeting the foregoing standard." There is no "without limitation" and no other technology named, so as written the list is arguably closed and would exclude a future ultra-low technology without a further amendment.⁵¹

The cross-reference itself

Exhibit A, the clean enacting text, reads "B. Uses Subject to Conditions (see §15.3.08.060)." §15.3.08.060 is Title 15's Conditional Use Permits chapter. It makes the Planning Commission the land use authority, provides that it "shall hold at least one public hearing," and requires five mandatory findings before approval.²⁸⁵¹

Title 15 uses two established forms, and Exhibit A matches neither. An independent sweep of the code found 17 headings reading "Uses Subject to Conditional Use Permit (see §15.3.08.060)", where the cross-reference carries the hearing and findings machinery, and 16 separate bare "Uses Subject to Conditions" headings, none of which carries it.²⁸⁵¹ Exhibit A creates a third form that exists nowhere else in the code: the "Conditions" label attached to the Conditional Use Permit cross-reference. That is the anomaly, and it is a residue of the heading it replaced.

The redline settles what a string search could only suggest, and it is the strongest evidence in this report. Packet page 113 carries the amendment as tracked changes. Rendered and read at the pixel level, because strikethrough does not survive text extraction, the drafters struck "conditional use" in four separate places and inserted "conditions" in its place:⁵¹

TABLE 14

That is not drafting ambiguity. It is a deliberate, systematic removal of conditional-use status from every operative clause, and Exhibit A confirms the result.

The surviving "(see §15.3.08.060)" is therefore best read as residue of the heading struck around it, not as deliberate incorporation. The by-right reading is the likely one, and the Council should know that is what it is voting for. Deleting the stray cross-reference is a one-line amendment that would settle it.

Two more dangling references point the same way. §7 requires operating within emission rates "demonstrated under subsection C", but Exhibit A has no lettered subsections: the single biggest credit this report gave staff, making emission rates enforceable as a zoning condition, hangs on a reference to a subsection that does not exist. And §13's two-site quota identifies uses "as identified in §15.3.16.120(C)", after the same ordinance renumbers that subsection to (B). Three unresolved cross-references, all inside the text being enacted.⁵¹

Why this is the right question and not a gotcha. There are two readings and they lead to opposite outcomes. Under surplusage, the cross-reference is a leftover, courts disregard it, the use is purely administrative, and the entitlement runs with the land with no future hearing. Under incorporation by reference, "(see §15.3.08.060)" imports the CUP chapter's procedures, every future applicant gets a Planning Commission hearing and findings, and the ordinance is far stronger than it appears.²⁸

Nobody can object to that question, and either answer is worth having. "Surplusage" tells the Council on the record that it is granting a permanent by-right entitlement, which is the single fact this whole review exists to surface. "Incorporation" means the protections are real and the ordinance is considerably stronger than its critics allow. This ask rests on the enacting text itself, which is why it holds regardless of what anyone intended.

10. The air permit is not issued, and the comment window closes after the vote

Volition's live website states "Permit Status: Approved." Its August handout states "Utah DEQ issued the permit in 2026."⁶¹ The 05 AUG deck told the Planning Commission a permit "has been approved." All three are wrong, and the state's own document says so in terms.

DAQE-IN163550001-26 is an Intent to Approve, issued 4 AUG 2026 by the Utah Division of Air Quality, signed by the Division's New Source Review Section Manager. Its cover letter: "The attached document is the Intent to Approve (ITA)… The ITA is subject to public review. Any comments received shall be considered before an Approval Order (AO) is issued." And its Public Notice Statement: "Air pollution producing sources and/or their air control facilities may not be constructed, installed, established, or modified prior to the issuance of an AO by the Director… The proposed conditions of the AO may be changed as a result of the comments received."

TABLE 15

The Council is being asked to grant a permanent by-right entitlement eighteen days before the air permit that governs the facility can become final, and three days before the public can still demand a hearing on it.²⁸

What the permit actually contains, and it is worth stating fairly. The facility is permitted as 1,210 Bloom Energy Server 6.5 units, 393.25 MW combined nameplate, SIC 4911 Electric Services, classified New Minor Source, with no applicable federal standards.¹

TABLE 16

Every criteria-pollutant limit is the manufacturer's datasheet rate applied to the full nameplate for all 8,760 hours of the year. That is conservative on utilization and carries no margin at all above the vendor's specification. There is no degradation allowance and no measurement uncertainty.⁴³ If the units do not hold their published rates, the facility is out of compliance immediately.

And the handout's carbon figure is honest.⁶¹ "Under 1.43M tons of CO2e a year at full build" traces exactly to the permit's 1,434,855 TPY.¹ Volition is publishing its own permit's potential-to-emit number, which is larger than this report's operating estimate of roughly 800,000 to 980,000 metric tons at the 75% utilization the Volition representative stated. Both are correct at their stated basis. Quoting either without the basis is not.

11. Why the by-right question matters far more here than it normally would

Utah enacted S.B. 132 (2025), "Electric Utility Amendments", Laws of Utah 2025, ch. 318, effective 7 May 2025, creating Utah Code Title 54, Chapter 26. It defines a new regulated category at 54-26-101(1):¹²⁴⁴

That is a verbatim description of this facility, at 2.5× the statutory threshold. And 54-26-504 provides that such systems "are exempt from commission oversight or regulation as a public utility."⁴⁴ A conventional 250 MW power plant in Utah passes through several independent public-interest gates.⁵¹ This one passes through almost none:

TABLE 17

Title 54 Chapter 26 is a Public Utilities framework, silent on municipal zoning; it neither preempts local land-use authority nor supplies a substitute for it.⁴⁴ So the entire public-interest review of a 250 MW industrial power plant reduces to one thing: the Spanish Fork zoning decision. And §15.3.24 as drafted converts that decision into an administrative permit issued over a counter, with no hearing and no findings.¹¹²⁸⁵¹

This is not an argument that Volition is dangerous. It is an argument about where the only remaining valve is. The off-grid design that makes this project genuinely attractive is the same design that removes it from every regulator except the city. Islanding is a benefit and a jurisdictional exit at once, and only one of those is in the deck. It also disposes of the most common rebuttal at hearings, "the state and DAQ will regulate this anyway": for a closed private generation system, the state has affirmatively legislated that it will not.¹²²⁸

And the exemption is not incidental to the design. It is contingent on it. 54-26-101(1) defines a closed private generation system as one that "is not connected to and operates independently from the transmission system of a qualified electric utility."¹² Interconnection is physically available: a 345 kV hub sits under a mile due north of the parcel.⁴⁰⁴³ Tying into it would satisfy the one condition that currently removes this facility from Public Service Commission jurisdiction, and would add a FERC-jurisdictional interconnection queue and years of system-impact study on top.⁴⁴

And the reporting backstop does not exist. S.B. 282 (2026) would have required an annual electricity- and water-consumption report from the generation provider to the host municipality; it failed, confirmed two ways from primary sources.⁴⁴⁴⁹ H.B. 585 (2026), which would have required data centers to disclose "the percentage self-generated" and "the type of generation facility and fuel source," also failed.⁵⁰ So there is no state-law disclosure obligation covering this facility's energy, emissions, or water. The "annual public reporting" covenant Volition volunteered, and which §15.3.24 omits, has no statutory backstop whatsoever. If it is not in the ordinance or the development agreement, it does not exist.¹¹

Is by-right an outlier in Utah? No. It is the norm.

TABLE 18

Two conclusions follow, and they point in opposite directions.

1. On the mechanism, Spanish Fork is the Utah norm. Eagle Mountain and Bluffdale permit data centers administratively by a plain "P" in a use table; Lehi permits them by silence; Eagle Mountain moved toward this posture recently, Ordinance O-40-2024, 3 SEP 2024, titled in part "REMOVING CONDITIONAL USES." The criticism that Spanish Fork is doing something procedurally unusual does not survive.²⁸

2. On the substance, Spanish Fork's ordinance is the most regulated of any Utah jurisdiction examined, by a wide margin. It is the only one with a megawatt cap, a citywide quota, a data-center-specific noise standard, a cooling-method restriction, an emissions requirement, an electrical-isolation mandate, a water condition, or a decommissioning provision. Eagle Mountain and Bluffdale host operating data centers under codes that impose none of those.

That is a genuine credit to Spanish Fork staff: they wrote more protection into this ordinance than any Utah city that already has data centers, before the first one arrived. The criticism that remains is not that §15.3.24 is weak by comparison. It is that its specific gaps, no heat standard, no post-construction noise verification, no decommissioning security, no potential-to-emit cap, are real, and three of them are gaps the applicant's own filing describes as already filled.¹¹

One point of context. Among jurisdictions that considered this question recently, Saratoga Springs, Box Elder County, Loudoun County VA, every one moved toward more discretion, while Eagle Mountain's and Bluffdale's permissive treatment predates the current wave.⁵¹ That distinction is real, but it is weaker than the cross-reference question and should not lead.

12. Four state-law facts nobody in the Spanish Fork record has mentioned

All four come from enrolled bill text read in full.

1. The fuel cells are already sales-tax exempt, and the rest may be too. Utah Code 59-12-104(80) exempts "sales of a fuel cell as defined in Section 54-15-102," unconditionally and independent of anything about data centers.⁴⁵ The Bloom array, plausibly the largest single equipment line in the project, is exempt on its own terms. Separately 59-12-104(84) exempts equipment bought by a "qualifying data center," which 59-12-102(109) defines as a facility that "(c) is a new operation constructed on or after July 1, 2016; (d) consists of one or more buildings that total 150,000 or more square feet." The 1,054,000 sq ft clears (d); prong (c) is genuinely uncertain, since this is adaptive reuse of a building the leasing listing dates to 1998.⁵⁵ The councilmember weighed "$50 million existing assessed value" against "$4 billion in equipment investment" (04 AUG minutes p. 69) with no offsetting entry, and sales and use tax on a $4B purchase is not a rounding error.⁴⁵⁵¹

None of this contradicts "no incentives requested." These exemptions are automatic, and H.B. 507 (2026) provides at 11-41-202(2)(e) that "a sales and use tax exemption described in Section 59-12-104 does not constitute an incentive."⁴⁵⁴⁸

2. Utah's only data-center disclosure law is keyed to water, and this design sits under it. H.B. 76 (2026) created Utah Code 73-5-8.3, requiring that "before a land use authority… approves a land use application for a new large data center, the land use authority shall notify" the Divisions of Water Rights and Water Quality and the water provider.⁴⁷ "Large data center" is defined by annual withdrawal of 75 acre-feet or more. Volition's 250,000 gal/yr is 0.77 acre-feet. Two orders of magnitude below the trigger.⁵¹

Read both ways: it is real evidence for the closed-loop design, and it means the state's only data-center notification regime does not reach this project. 73-5-8.3 expressly contemplates "generation by the operator of the large data center of energy to power a specific facility," so the statute knows self-generation exists and still gates on water. Combined with the failure of S.B. 282 and H.B. 585: no Utah statute requires this facility to report its energy use, emissions, self-generation, or water.⁴⁷⁴⁹⁵⁰

3. An NDA with the city would now be a protected record. S.B. 318 (2026) would have prohibited municipal NDAs with prospective data centers and voided existing ones. It failed.⁴⁴ H.B. 507 did the opposite, amending 63G-2-305(2)(b) to protect "confidential economic development information… that takes the form of a nondisclosure agreement between a person and a governmental entity exploring an economic development opportunity." The Volition representative told Council conversations began "around October or November" (p. 64), roughly nine months before the application surfaced.⁴⁶⁴⁸ Any such NDA is now statutorily shielded from GRAMA, a live constraint on the records strategy in open-questions.md.

4. A mechanism now exists to divert the tax base being promised. H.B. 507 also enacted Title 63N, Chapter 3a, Part 4, creating "Regionally Significant Zones with Energy Implications." Under 63N-3a-402(3)(b) a zone proposal may "(i) capture up to 60% of the property tax increment… and (ii) divert up to 100% of personal property tax revenue" generated within it.⁴⁸ The entire $40M/yr pitch rests on personal property tax on equipment. From 6 MAY 2027 a municipality may generally "not offer an incentive for a large load data center that is not located within a zone,"⁵¹ pushing large projects toward the mechanism. Nothing suggests anyone has proposed a zone here, but the fiscal case Council was shown treats the $40M as settled municipal revenue while state law contains a documented pathway to route most of it elsewhere in a later proceeding.

13. The $40M/year does not mean what the room heard

The year-one arithmetic is sound, and almost none of the number survives contact with who receives it and how fast it depreciates. The figure is gross, and gross is what an applicant quotes. Neither correction below challenges Volition's honesty, the slide names "the largest share flowing to Nebo School District" and its footnote states the rate is "declining as personal property depreciates."³⁴ They are corrections to what the room is likely to have heard.

Correction 1: 75% of it is not the city's

Utah County Assessor, parcel 67:228:0003, Tax Area 150.¹⁰ The combined certified rate for tax year 2026 is 0.009563:¹⁵

TABLE 19

A gross $40M/year is roughly $4.0M/year to Spanish Fork City.⁴³ The rest is real public revenue and Nebo School District is a legitimate beneficiary, but the number was presented in a Spanish Fork Council chamber, weighing a Spanish Fork land-use decision, without that breakdown.³⁴

One commitment worth pinning down, from the 23 JUL marketing video: "A share of every site's revenue flows straight back into the community that hosts it." A revenue-share is materially stronger than a flat ">$500,000," because a percentage scales with the facility and a flat figure does not. Ask which form the development agreement takes.

The tax base is a community benefit, not a city line item

Nebo's share alone is more than five times Spanish Fork City's entire annual property tax revenue from every parcel in the city, $5,885,340.⁵⁹ At recent Utah costs that is a new elementary school every year, or a new high school every three to four years, against Utah high school builds of $77.5M to $145M.⁷⁴ Nebo School District and Utah County have a direct and substantial interest here, and neither has appeared in the record.

The direct community contribution is larger than it sounds

Volition has committed to more than $500,000 a year in direct community contributions, described in the 23 JUL video as "a share of every site's revenue."¹⁶⁵¹ Against what the city collects today:

TABLE 20

The proposed contribution by itself is 15× the city's property tax take from its single largest commercial taxpayer, and twice what the city receives from the eleven largest combined.⁵⁷⁷³ That is before a dollar of property tax.

Correction 2: the $40M is a floor, not a ceiling, and it depreciates slowly

The $4B figure does not include the servers. Volition's own project page states the investment covers "retrofit, fuel cells, electrical, fire, security, and engineering."⁴ Compute hardware is not in that enumeration. Volition builds and powers the facility to the rack; the tenant installs the servers, and that equipment is additional taxable value on top of the $40M, in an amount nobody can state. So $40M/year is the applicant's expected minimum, not its total.

That changes which depreciation curve governs, and it runs in the proposal's favor. Utah Admin Code R884-24P-33 sets percent-good schedules by class, and two classes behave completely differently:¹⁴

TABLE 21

(Each column models the full $4B in that class, at the 2026 combined rate of 0.009563.)¹⁴¹⁵

Volition's $4B sits overwhelmingly on the durable side. Fuel cells, electrical, fire and security are generation equipment, and the retrofit is real property, reassessed at market rather than run down a percent-good schedule. None of it is Class 12, because the compute is not Volition's, and the base holds 74% of its value at year ten.

Two features of the plan push the same direction. The Volition representative told Council the Bloom equipment would undergo "a phased upgrade approximately seven or eight years after installation", paid monthly in advance for a full "swap out and retrofit."⁵¹ Replacement equipment is assessed as new, resetting the curve. And tenant compute refreshes on its own cycle, each refresh re-entering the roll at 62% good. The fast-depreciating asset is also the one that gets replaced most often.

One thing not to do: do not net the sales-tax exemption against the property tax. 59-12-104(84) is a sales and use tax exemption only. It does not touch property tax, and conflating the two produces a wrong number in both directions.⁴⁵

Correction 3: what $40M/year means against what these businesses actually pay

This is public record, and it should be quoted rather than computed. Utah County publishes the tax actually billed on every parcel, by year; 2025 is the most recent complete year.⁵⁷

TABLE 22

The eleven largest commercial taxpayers in Spanish Fork paid $2,485,898 between them in 2025. Volition's claimed $40M is sixteen times all of them combined.⁵⁷

The hospital is not the benchmark it looks like: the city's most valuable property pays less than five other businesses, an effective 0.001734 against the 0.009632 everyone else pays, the standard Utah charitable exemption.⁵⁷⁵⁸ The correct benchmark is WestRock at $333,850, the largest property tax bill in Spanish Fork.

And the city's own revenue is the number that matters most

Spanish Fork City's audited FY2025 statements report total property tax revenue of $5,885,340, against $5,741,117 the year before.⁵⁹ That is the whole city: every home, every business, every parcel.

TABLE 23

Two true statements, and the Council needs both. Volition's gross claim is 6.8× the city's entire property tax revenue, which is why the number sounds unreal in a Council chamber. But the realistic municipal benefit is roughly $4.0M a year, which would raise the city's own property tax revenue by about 68%, a very large benefit, and smaller than the headline by a factor of ten. For scale, $4.0M is roughly 120× what the city currently collects in property tax from its largest commercial taxpayer, which bills $33,444.¹⁵⁵⁹ Setting $40M beside $5.9M compares a total against a share.

Two rates are in play and are not interchangeable: the 2026 certified rate of 0.009563 for projections, and the 0.009632 actually applied to 2025 bills for comparisons against real billed dollars.¹⁵⁵⁷⁷³ At the certified rate $4 billion produces $38.3M, so $40M is sound as a gross year-one figure.

The assessed-value baseline checks out, the acreage in testimony does not

The councilmember's "approximately $50 million existing assessed value" is confirmed: $50,898,400 total assessed for 2026, owner of record WDC Industrial Utah Owner LP.⁵¹ An $81,000,000 figure in the parcel abstract is a trust deed, not a sale price: Utah is a non-disclosure state.

The Volition representative told Council the building "sat on approximately 25 acres."⁵¹ The Assessor record shows 69.478641 acres, corroborated at 69.09 by computing the area of the state's parcel polygon.¹⁰ ~23–25 acres is the single-story footprint, not the parcel, leaving roughly 44 acres of open ground for the fuel-cell field, transformers, switchgear and cooling plant. The site is physically ample. It was the testimony that was wrong, not the plan.⁵⁵ But note what else 44 acres means: the ordinance's 500 MW cap is not constrained by land. At Bloom's published density 500 MW needs roughly 14–18 acres of yard, so the parcel can hold the full entitlement twice over.¹¹⁴³

14. How Utah County property taxes are actually calculated

Every argument above about the $40M, and most of the neighborhood argument about property values, runs through a mechanism almost nobody in the public record has described. It is entirely statutory, it is not what most people assume, and getting it right changes what the Council should say about revenue.

The chain, step by step

TABLE 24

That disposes of the most common objection to the system, which is that assessors cannot possibly be driving every street in the county. They are not, and the statute never asks them to. Values move through mass appraisal and Commission-ordered factoring applied to geographic areas, and human inspection of an individual property is a once every five years event.⁶⁷⁶⁸⁷²

The certified tax rate is the part almost nobody knows, and it is the whole answer

Utah has been revenue-based rather than rate-based since 1985. The statute is blunt:

Revenue is fixed first. The rate is whatever number produces it. A taxing entity may not levy above its certified rate, except during a truth-in-taxation exemption period, unless it advertises, mails notice, and holds a public hearing.⁷⁰

When values across the county go up, the rate mechanically comes down, and that is directly observable in Spanish Fork's own records. Four large commercial parcels, four owners, five consecutive years, all in Tax Area 150 carry an identical effective rate to six decimals in every year, because the rate belongs to the tax area and not the parcel; a fifth parcel in a different area returns a different rate, which confirms the reading:⁷³

TABLE 25

The rate fell 14.2% in four years. Nobody voted for a tax cut. That is Truth in Taxation operating exactly as written.

So why did an announcement next door not raise your taxes

Because rising value is not the thing that raises your bill. What raises your bill is your value rising faster than the average inside your taxing entity, or the entity voting an increase after a noticed hearing. The same parcels show it:⁷³

TABLE 26

Every one gained a third to a half in assessed value and none saw its bill rise anywhere near that much. The gap is the certified rate falling underneath them.

Three consequences to carry to the podium: a new landmark, temple, park, school or data center does not raise your taxes by existing, only by making your property appreciate faster than the county average; an announcement is not evidence of value, since the annual update runs on current market data and factoring on assessment studies, so a change appears only once transactions demonstrate it;⁶⁷⁶⁸ and the reverse attribution is the common error, people see the bill move, look at what changed nearby, and connect the two, when the bill moved because of a countywide ratio study and a divisor.

The finding that matters for this amendment

New growth is deliberately excluded from the certified rate denominator. The statute subtracts eligible new growth from the value the prior year's revenue is divided by, and "locally assessed new growth" expressly excludes any change in value "as a result of factoring in accordance with Section 59-2-704, reappraisal, or another adjustment."⁶⁹ It captures new construction, not market drift. Two consequences follow, and the room needs both:

One carve-out worth an ask. Section 59-2-924 already excludes, from both revenue and aggregate taxable value, personal property that is semiconductor manufacturing equipment.⁶⁹ The Legislature has already removed one class of high-value technology equipment from this machinery, and the great majority of Volition's projected $4B is personal property of a broadly similar character.

15. Claims that fail

TABLE 27

The carbon numbers these claims are measured against

Bloom's February 2026 datasheet for the ES 6.5 publishes 679–833 lb CO2/MWh, and that printed figure reconciles to 0.1% with the datasheet's own heat rate through EPA's standard natural-gas carbon factor.³²⁵ Utah DAQ's Intent to Approve puts the facility's potential CO2e at 1,434,855 tons per year (DAQE-IN163550001-26, 4 AUG 2026).¹⁴³ At the 75% utilization the Volition representative stated, the realistic figure is ~796,000–977,000 metric tons/yr. Methane in a solid-oxide fuel cell yields CO2 as its primary carbon product: suppressing combustion eliminates thermal NOx, not carbon.⁵¹ Upstream methane leakage adds 76,000–125,000 t CO2e.⁴³

The estimate is more likely low than high. In the one case where a state regulator measured Bloom units in the field, Delaware found a 24-month average of 823 lb CO2/MWh against Bloom's advertised 773 (NBC Bay Area, 25 JUN 2015), and a separate installation, AC Transit, held the advertised rate only for the first three of 22 months.⁵² That is a decade old and predates this platform, but it means the datasheet figure is a floor. Even at 823 lb/MWh Bloom still beats the Utah grid, which EIA puts at 0.601 metric tons CO2/MWh for 2024, about 1,325 lb/MWh.²⁷ The criticism is the gap between marketing and measurement, not the absolute number.

1. CO2 "the same amount as three or four diesel trucks"

The answer depends entirely on scope, and no scope is stated anywhere. On total facility CO2 the permit's 1,434,855 t/yr against 3–4 trucks at 171.5 t each fails by 1,900–2,500×.¹⁴³ On criteria pollutants converted to CO2-equivalent, the basis Volition actually described, short-lived climate forcers carry published indirect GWP100 values, several negative (NOx nets between −11 and −108 per kg, SO2 around −40, while CO is +1.9 and VOC +4.5), and applied to the five permitted pollutants the sum lands between roughly −400 and +110 t CO2e against 3–4 trucks at 515–686 t. On that basis the claim is defensible, even conservative.

One carbon claim holds up cleanly. "The lowest-available carbon at this scale" is correct, and the honest framing is the one Volition should have used:

TABLE 28

Roughly 29% cleaner than on-site gas engines and 43% cleaner than the Utah grid, per unit of compute. Both things are true at once: this adds roughly 890,000 tons of CO2 to Utah County that does not exist today, and it is the lowest-carbon way to run that compute given what actually gets built. Telling a council it equals four trucks throws away a real advantage.

2. The emissions comparison table

Slide P-08 puts a "natural gas reciprocating engine" at ~59,700 tons NOx/yr with no capacity, capacity factor, or basis stated. Utah's own permit for 235.6 MW of exactly that equipment (Novva, DAQE-AN160660003-24) is 51.28 tons/yr, which scales to ~86 tons at this facility's 393.25 MW.¹³¹ The "Utah grid average ~2,600" row has no stated basis at all.

The defensible claim is ~10× cleaner than West Jordan and ~45× cleaner than the Millard County facility, with no HAPs and no purchased offsets.³² Inflating that to ~11,500× replaces a verifiable advantage with an indefensible one and hands critics the easiest attack in the record.

3. "Equal to 4–5 semi trucks per year"

The slide states no basis, and the two available fleet baselines differ by roughly 20×.⁴³ What can be said uses only the deck's own numbers: the 04 AUG Council deck puts its stated truck fleet at ≈2 metric tons NOx and ≈50 kg PM2.5/yr (420,000 truck-miles, 65,000 gallons of diesel, internally consistent at 4.76 g NOx/mi and 0.119 g PM2.5/mi).⁴³ Against the facility's 5.17 short tons NOx and 0.12 short tons PM2.5:¹

NOx: 4,690 kg ÷ 2,000 kg = 2.3× PM2.5: 109 kg ÷ 50 kg = 2.2×

Two pollutants, computed independently, agree: the facility is roughly 2.2–2.3× the truck fleet the applicant's own deck describes, not four or five trucks out of it. That is a finding about the slide, not about the site. The deck never states what its fleet figure represents, and it cannot be read as the fully tenanted warehouse this report compares against: the building is mostly vacant today and its present truck traffic is very low, so any baseline anchored to current operations answers a question nobody is voting on.⁵¹ The comparison that matters is the proposed data center against the fully tenanted distribution warehouse the building was built for, and that one is computed from ITE rates in §6 rather than taken from the applicant. On that basis the warehouse runs 6,396 to 12,849 diesel trip ends a month and the data center none.⁴³

4. "~13 W/m²" of heat

Over the verified 69.5-acre parcel, 605–742 MW-thermal gives ~2,150–2,640 W/m².¹⁰⁴³ To get 13 the denominator must be 47–57 km², well beyond Spanish Fork's entire municipal area. The figure is facility heat ÷ city area, printed beside "Max of +5°F on-site," differing by a factor of ~170–210 in basis. As a city-scale comparison it is coherent; as printed it invites the wrong inference.

5. The Novva proof point

The Volition representative cited Bloom running 50 MW at Novva with Oracle as tenant, the largest single-site Bloom deployment in existence. Refuted across four independent layers: Utah DAQ's approval order (72 gas engines, 37 diesel generators, zero fuel cells), Novva's own 16-release press archive, Novva's Utah marketing page, and Bloom's newsroom.³¹ The real project is Oracle's "Project Jupiter," up to 2.45 GW of Bloom fuel cells in Doña Ana County, New Mexico, announced 27 APR 2026.²⁴

But the correction runs in two directions. The largest documented single-site Bloom deployment in North America is 73 MW (AWS/Cologix, Hilliard, Ohio, PUCO-approved JUN 2025, under legal challenge by the City of Hilliard); the largest anywhere is the 78.96 MW Shinincheon Bitdream plant in South Korea, operating since 2021.²⁴ At the permitted 393.25 MW, Spanish Fork would be 5.0× the largest fuel cell installation on earth, and 6.3× at the ordinance's 500 MW cap. Volition described this accurately at Council: the closest comparable "was not a Volition facility," only "the same general contracting and engineering team."¹¹¹

16. Property values, which the evidence addresses better than either side admits

Four studies bear directly on this and none says what either camp claims. The strongest evidence finds no negative effect and often a positive one, while the study conducted nearest a proponent's interest has the weakest numbers of the four.

The best study says proximity is associated with higher prices, not lower

George Mason University's Center for Regional Analysis modeled every 2023 BrightMLS home sale in Northern Virginia, the densest data center cluster on earth, using a hedonic regression that explains 86.7% of price variance.⁶² Its data center result:

TABLE 29

The sign is what matters here: homes farther from a data center are worth less, not more. The authors state it plainly: "The analysis fails to demonstrate statistical evidence that proximity to a data center negatively impacts housing values."⁶² And they anticipated the obvious objection. Because data centers usually sit on industrial land, controlling for industrial proximity might absorb the penalty, so they re-ran the model with the industrial variable removed. The data center coefficient stayed negative at −$9,745, R² essentially unchanged at 0.864. The effect survives the objection.

Two more studies point the same direction, with real caveats

Priest, Not In My Back Yard! The Effects of Data Centers on Housing Prices matches Virginia air-permit issuances to ZIP-level house price indices in a difference-in-differences design and reports effects "economically small and slightly positive," with confidence intervals that rule out substantial declines.⁶³ The full text sits behind a CAPTCHA that has failed three times across this review, so only the abstract is verified. Treat it as supporting, not load-bearing.

NBER Working Paper 35194 (Alvarez, Argente, Chow and Van Patten, May 2026) uses a shift-share instrumental-variable design across roughly 3,130 US counties and finds a house price index coefficient of +0.177 (SE 0.028), falling to +0.136 with state fixed effects but staying significant.¹⁸ Two limits matter. It is an NBER working paper and explicitly not peer-reviewed, and it is county-level, so it describes what happens to a county's housing market, not what happens to the house across the street.

The study that complicates the picture is the industry's own

Integra Realty Resources examined homes within 1.5 miles of four Indiana data centers from 2021 to 2026, and presented the analysis in support of a zoning variance a developer was seeking.⁶⁴ The headline is that nearby values rose 42% against 41% countywide. The detail is less flattering: values rose near all four sites, but at Microsoft LaPorte the 1.5-mile increase of +42% lagged the county's +48%, and the favorable average is carried by Amazon New Carlisle alone at +49% against +32%. The method is also thin: 20 Zillow estimated values per location. This is weak evidence in either direction, and it is worth citing mainly because it is the proponent-side study and it does not clearly support the proponent-side claim.

What none of this settles

Every study above measures conventional data centers. None measures one with a 393 MW on-site power plant, because essentially none exists. Heat, gas throughput and continuous generation noise appear in none of these samples.

The argument specific to this site, and it is a fair one

The counterfactual here is not farmland. It is an operating one-million-square-foot distribution warehouse, and that matters on these very coefficients.⁵⁵ GMU's industrial-distance penalty is $47,523, three times the data center coefficient and opposite in sign.⁶² The industrial penalty is already priced into every home near this site, because the site is already industrial.

One finding worth carrying to the podium for a different reason

The same NBER paper that finds house prices rise also finds electricity prices rise, coefficient +0.009, significant at the 1% level across every horizon tested.¹⁸ A competing study reaches the opposite conclusion: Watten, Bistline and Blanford estimate that data centers caused average US retail rates to fall modestly between 2015 and 2024, though they caution that "future supply constraints could reverse the effect."⁶⁵ Two of the three authors are at the Electric Power Research Institute, an industry-funded body, which the Council should weigh.

The literature genuinely disagrees on ratepayer impact, and this is the one dimension where the Spanish Fork proposal is unusually well positioned. A permanently islanded facility that draws nothing from the municipal grid cannot raise Spanish Fork ratepayers' bills through the mechanism either study is measuring.¹¹ That is a real advantage and the applicant has not made enough of it.

17. Claims that hold up

• NOx 5.17 tpy is the permitted limit, and it is exactly Bloom's datasheet rate at full output. 0.003 lb/MWh × 393.25 MW × 8,760 h = 5.17 tpy, matching DAQE-IN163550001-26 to 0.05%.¹³⁴³ The same is true of CO (22.39) and VOC (17.22). The permit assumes 100% capacity factor, which is conservative, but it carries no margin whatsoever above the vendor's published specification.¹ VOC exceeding NOx by ~3× is correct SOFC behavior, not an error.
• "<65 dBA at 10 ft per unit", verbatim from Bloom's datasheet including the measurement distance, and Volition disclosed the limitation itself, to the Council, in print. The footnote on that slide reads: "PER-UNIT SPEC FROM DATASHEET; ARRAY-WIDE ACOUSTIC MODELING AND AN ENFORCEABLE NOT-TO-EXCEED LIMIT ARE BEING COMMISSIONED BEFORE FINAL APPROVAL." The ordinance requires neither the array-wide modeling nor any post-construction check.³¹¹
• Water arithmetic, 500,000 gal fill ≈ 2.6 households and 250,000 gal/yr ≈ 1.3, sourced to the Central Utah Water Conservancy District's 155 gal/person/day filing and Census household size, correctly applied.⁵¹
• "+5°F max on-site" and "<1°F beyond a quarter mile", both plausible under light wind.
• "Less than half of what Lake Side Power Plant produces" held at the 250 MW rack load but fails at the permitted 393.25 MW: 605–742 MW-thermal is 46–56% of Lake Side's ~1,315.^1334351 That slide is still the only one in the deck that discloses its own methodological limitation.
• ~$40M/yr in year one, arithmetically sound. $4B × 0.95% = $38M.⁵¹
• "We can't say that we are net zero" and "we don't make a claim to being a huge employment benefit." Volition volunteered both when it would have been easy not to.
• Minor-source permitting is legally correct. Utah County is Marginal ozone nonattainment, so NNSR is 100 tpy, and the claimed figures clear both PSD and NNSR.¹³ The offset question under R307-421 is genuinely open rather than cleared: that rule turns on a modeled ambient concentration, not a tonnage threshold, and no modeling is in the public record.²⁹
• Zero operational water for the ultra-low class is sourced, not asserted, to the Bloom Energy Server 6.5 and Mainspring Linear Generator 3 MW datasheets, and §9 makes the annual estimate a binding permit condition verified before approval.
• "More protective than the current code, which has no standard for this use at all" is fair. Eagle Mountain and Bluffdale host data centers under codes with no data-center standards.

The strongest fact in the proposal is one Volition never put on a slide

The city manager, answering the Mayor on 04 AUG, supplied a number the applicant did not:

250,000 gal/yr against ~5,000,000 gal/yr is a 95% reduction in site water use, it comes from city staff rather than the applicant, and it appears in neither slide deck.⁴³⁵¹ Two related items deserve the same credit:

• Contaminated coolant is not going to the sewer. The Volition representative, on the record: the water would be "removed into tankers, treated, and transported to an authorized waste facility rather than discharged into the local wastewater system" (p. 66). The ordinance as drafted permits sewer discharge, which is looser than what the applicant committed to, the same pattern that runs through this whole review.
• A third councilmember built the right kind of condition. If the facility ever needed hundreds of millions of gallons it would require water infrastructure the site does not have, so prohibiting infrastructure expansion makes the water cap physically self-enforcing rather than trust-dependent (p. 69). That is the best piece of reasoning by any official in the record and a template for every other condition in this ordinance.

One caveat: the ~5M gal/yr figure is a staff statement in minutes, not a metered record. Ask for the parcel's municipal water billing history, a routine GRAMA request that either confirms the best argument in the proposal or undercuts it.

18. The company

As of 20 FEB 2025, volition.eco described a residential and commercial solar, HVAC, air-filtration, battery, and microgrid installation business. No mention of data centers anywhere. The Internet Archive holds no content capture after that date. The data center identity is roughly 12–18 months old.

TABLE 30

Ratio: $2.8M documented capital against a ~$4B claim, roughly 1:1,430. The gas lateral alone, at $14M, is five times the company's entire documented raise.⁵¹

Two claims need correcting. "500+ MW of team experience" appears in company material only as "over 500MW in new development in the works", a forward pipeline figure, not a track record; and the Novva team-overlap claim could not be located in any company material and shows no overlap with any named Novva executive, unverified and unlocated, not disproven. Pipeline claims mostly evaporate: San Jose and Austin, nothing, and "Stockman Flats" is a real place with Public Infrastructure Districts created at owner petition but no MW figure, no data center reference, and no Volition mention in the primary record.

The counter-case. Bloom Energy is named as sponsor and partner in the city record, a NYSE company with $2.45B cash that runs exactly the third-party-ownership structures this project needs. Volition's four years deploying third-party-owned generation, batteries and microgrids under PPA structures is adjacent experience for the hardest part; the data center half is what is new. Adaptive reuse removes execution risk, the build is phased, and the company volunteered enforceability and offered CUP review that staff declined.

The problem is not bad faith. It is capitalization and track record, against a permanent entitlement.

19. The tenant chain

Spanish Fork City → Volition Industries, no operating facility, $2.8M documented capital, no SPV named → Nscale (alleged), $14.6B valuation, ~$450M run rate, $51B backlog, heavily levered → undisclosed "large brand"

Nscale has no public mention of Utah, Spanish Fork, or Volition anywhere, not its site list, newsroom, investor materials, or trade press. Pre-signing NDAs are standard, so this is not evidence of fabrication. It means the city has no independent corroboration of the tenant on a project whose entire economic case depends on one, and Volition's own timeline is expressly conditional on "a complete contract" that is not documented.

Nscale's two largest anchors are Microsoft and OpenAI, the exact two the Volition representative said are "not in play." That does not identify this site's sub-tenant, and both statements can be true. The closest structural precedent is public: Ionic Digital Inc. → Nscale → Microsoft at Ward County, Texas, a 126-month triple-net lease for 234 MW, where Ionic Digital's SEC filing states "We currently have only one customer" and warns explicitly about Nscale nonperformance. Same chain shape, with the difference that Ionic is public and discloses the risk.

The "no government, no surveillance" commitment is almost certainly sincere and almost certainly unenforceable. The city has no privity with the sub-tenant's sub-tenant, GPU workload content is not externally verifiable, and no municipality has ever enforced such a restriction. Better to ask for an audit right than accept a promise.

20. Sound and heat, the two things that reach a neighbor

The nearest existing occupied home is 1,298 ft from the property boundary, due west in Oakridge Cove.⁵³⁴³ "The closest homes are a mile away" is not accurate. And because the amendment applies citywide, the receptor that matters is the nearest home to whichever I-1 parcel the next applicant selects.

Heat is not perceptible at a mile and unregulated at any distance; sound level is unpredictable from public information; sound character is the real exposure and the one nothing in the code addresses.

Heat

605 to 742 MW-thermal at the permitted 393.25 MW is a real quantity that disperses fast: on-site temperature rise falls from 3.0 °F at 5 m/s to about 1.1 °F at 30 mph, and at a mile it is zero.¹⁴³ Volition's "+5 °F max on-site" and "under 1 °F beyond a quarter mile" are both plausible. Nobody will feel this at a mile. The gap is regulatory, not physical: §15.3.24.090 contains no thermal standard at all, no modeling requirement and no measurement, and the Planning Commission asked for one.⁵¹⁴³

Sound level

1,210 modules attenuated over 1,609 m gives 15 to 40 dBA against a 30 to 40 dBA rural night ambient. That ambient floor is not quiet in practice. A rail crossing sits directly across US-6 from the nearest residential receptors and its horn sounds through the day and night, which is the loudest recurring sound most of these homes already live with.⁴³ The 25 dB spread is what the public record supports, not either endpoint, and it covers everything from inaudible to clearly present. Treat every figure as a lower bound: inverting a "<65 dBA @ 10 ft" spec as a point source is wrong for a cabinet 29 feet long, and finite-line and ISO 3744 treatments give 84.5 and 90.6 dBA sound power against the 82.7 used.³⁴³

The cooling plant, not the fuel cells, is the likely dominant source. In every documented complaint case nationally the driver is HVAC, chillers and blowers. Volition has no diesel, which removes the loudest source outright, but cooling is what people sue over. Microsoft is currently defending a suit from residents of Sturtevant, Wisconsin over "excessive and pervasive noise" from data center HVAC.

The binding number is stricter than the criticism suggests. §6 sets 65 dBA at the property line and 55 dBA at any residential zone boundary, at all hours, against the existing I-1 limit of 80 dBA daytime.⁵¹ That is a 15 dB reduction from what the site could lawfully do today, roughly a halving of perceived loudness, around the clock. The number is good, and the enforcement is better than this report first stated. §6 binds "at any hour", expressly covers cooling equipment, adopts the Chapter 9.32 measurement procedures, and adds that "compliance with this subsection does not excuse compliance with Chapter 9.32."⁵¹ The limit is continuing and independently enforceable, not a design target. What §6 omits is only a scheduled measurement after the plant is running, so enforcement waits on a complaint.

Field observations at an operating comparable, 14 AUG 2026 UNCALIBRATED

On 14 AUG 2026, between roughly 16:00 and 18:00, the author visited the Novva Data Center in West Jordan, interviewed six people and took uncalibrated phone-based sound readings.⁶⁰

Method: perimeter roads driven twice, clockwise, at a constant 18 MPH on cruise control, only the passenger window open, microphone pointed at the property. Constant speed holds the car's own noise steady, which is what makes the relative comparisons meaningful even though the absolute level is not. Readings approximate property-line distance, though the road-to-fence setback was not recorded.

What was found. Readings ran approximately 65 dB, but wind, truck traffic, construction and neighboring commercial activity were audibly louder than any data center equipment at every point. All six interviewees reported no noise issues, all six volunteered that neighboring activity is louder than the site, and all six confirmed the diesel generators have never operated, the site having had no outage.

The six were two residents living directly across the street, two employees of the neighboring construction supplier, a security guard at that business, and Novva's own guard. Five of the six have no interest in defending the site, and the two most probative are the residents, the closest available analogue to the receptor this ordinance protects.

Novva achieves this with no acoustic barrier at all: wrought iron fence around the entire property, which provides no meaningful attenuation.⁶⁰ No one can claim a barrier is doing the work, and this report's own finding that barriers could buy back 5 to 15 dB describes headroom that has not been needed.

On sound character, the field evidence cuts against this report

Asked specifically about tonality, the answer was that nothing was annoying about the quality of the sound. It simply sounded like fans running.⁶⁰ Broadband fan noise is the benign case. A narrow-band whine or hum is what generates complaints, and none was reported.

That matters because tonality is one of the strongest arguments in this report, and a field observation at an operating facility does not support it. Two limits keep it from settling the question: no spectral measurement was taken, and a moving vehicle with an open window masks subtle tonality. It says nothing about night-time audibility at distance. But it is evidence, it points against the concern, and it belongs in the record even though it weakens a point made elsewhere here.

What would actually decide whether it is noticed

A decibel limit cannot capture the failure mode that generates complaints. A narrow-band tone is picked out of a quiet background far below the level at which broadband noise registers. Two measured anchors bound the propagation question. Refractive focusing tops out around +6 dB (DOE/SERI MOD-1, Kelley et al. 1985). And the distance match is nearly exact: van den Berg's 2006 Groningen measurements found a steady source audible above background at 1,500 m in 38% of night hours against 0.7% of day hours, and a receptor at one mile sits at 1,609 m.

The wind is mostly protective, and the geometry matters more than this report first stated. Five years of KSPK ASOS observations put the canyon drainage flow moving toward 308 to 314° at night.⁴³ The nearest homes bear roughly 263° from the site, due west, which is 45 to 50° off that axis, so they are not squarely downwind after dark.⁵³⁴³ Homes to the north-northwest sit closer to the drainage axis than the nearest homes do, which is where night-time exposure would concentrate if it occurs at all.

At 30 mph, wind through trees, fences and eaves puts local ambient at 50 to 60 dBA, 10 to 20 dB above the facility's estimated 35 to 45.⁴³ That is complete masking. The worst case is not the windy nights but the opposite: wind-farm acoustics has known for decades that complaints cluster where the ground is calm while the source region is windy. The condition that could make this facility audible is narrow: light drainage flow around 2 to 5 m/s under a shallow stable layer, fast enough to carry sound and benefit from refraction, too slow to mask it.

21. What would have to be true, in both directions

TABLE 31

Note the pattern: every adverse outcome is about the ordinance, not the applicant.

Two of them already happened once, in this exact building

Fingerhut built this structure, and the record is contemporaneous press rather than inference. The Volition representative identified the site to Council as "the approximately one-million-square-foot former Fingerhut building."⁵¹

TABLE 32

Deseret News, 20 OCT 2003 carries the 520/800 employment figures and the 163 acres; 30 OCT 2001 carries the $70M, "never fully opened," and the 100-of-250 layoffs.¹⁶¹⁷ Only the $220,000 road figure appears in both; cite them separately.

One conflict, now settled: the 2003 article places the former plant on 163 acres against the assessor's 69.478.¹⁷ Three independent records agree on the smaller figure, the Assessor at 69.478641, the leasing listing at 69.95, and this review's own computation of the state parcel polygon at 69.09.¹⁰⁵⁵⁴³ The 163-acre figure describes something other than this parcel, most likely a larger original assemblage, and should not be used for buffer or expansion arguments.

Where the national record actually lands

The closest structural analogue is xAI "Colossus," Memphis, per NAACP v. X.AI Corp., No. 3:26-cv-00074 (N.D. Miss., filed 14 APR 2026): "Neither xAI nor CTC Property LLC applied for or received any air permits prior to constructing and beginning to operate these sixteen gas turbines" (¶197), reaching "thirty-five turbines for a total generating capacity of 421.4 MW" by 31 MAR 2025 (¶198).

Why it transfers: an off-grid, self-generating AI data center, the configuration Spanish Fork is being asked to authorize by right, where the failure was process rather than technology. Why it does not: xAI ran combustion turbines against Volition's CARB-certified fuel cells at roughly 1/70th the NOx per MWh. These are allegations in a filed complaint, not adjudicated findings. The transferable point is narrower. In Memphis, when local process failed, the correction came through the Clean Air Act citizen-suit provision because a federal statute supplied a hook. In Spanish Fork, air is DAQ's and siting is the city's, and a by-right ordinance with no post-construction verification gives the city no trigger to pull.

The Utah record cuts the other way. KSL measured ~70 dB about one block from Aligned's West Jordan campus (10 JUL 2026), residents describing constant generator blowers.²⁶ No enforcement action, notice of violation, or citation is on the record against Aligned, and for Novva no documented complaint of any kind was found. Residents complain; nobody has been cited. Expect Volition to make that point, and it is fair.

The "broken siting promises" narrative does not survive verification

A public commenter alleged a documented national pattern of data centers breaking siting commitments. A dedicated retrieval pass across Northern Virginia, the densest data center region on earth, found no agency finding and no adjudicated violation against any named operator on any of those grounds. Every noise case traced there ends one of three ways: the county measured and found no exceedance, the conduct was legally exempt, or the facility held a valid permit.

At Vantage VA2, Sterling, the closest Virginia analogue with eight gas turbines running 24/7, county staff readings "have never exceeded 55 dBA" at the adjacent residential property.¹¹ At AWS / Great Oak, a 1989 HVAC exemption meant "the County has no legal ability to control noise at any level from data center cooling equipment", so resident readings of 67.4 dB were real but a violation was legally impossible to find. And "Chesterton" is not a case: withdrawn JUN 2024 before any approval.

The one fully adjudicated case points somewhere else entirely. The Prince William Digital Gateway rezonings, ~2,000 acres and ~22 million sq ft, were declared void ab initio because the County Board violated Virginia's public-notice statute: a legal ad was drafted, never confirmed by the newspaper's deadline, and never ran.¹⁹

The project is dead, finality coming from QTS withdrawing its Supreme Court filing on 2 JUL 2026 rather than from the opinion. The largest data center approval in American history was undone by a newspaper ad that did not run.

Two conclusions follow, and they pull against each other. Both are true.

1. The adversarial case against operators is weaker than the public conversation suggests. Where noise is measured by an authority with jurisdiction, exceedances mostly are not found.
2. The failures that actually get remedied are failures of local process. Digital Gateway died over notice. Great Oak went unremedied for years over a definitional exemption. Vantage's turbines entered without additional review because Loudoun's zoning definition swept "associated utility infrastructure" inside the term "data center", structurally similar to what §15.3.24 does in combining data center and generation into one administratively-approved use.¹¹

The lesson is not "watch the applicant." It is that the ordinance is the only thing that will still be here in ten years.

22. Four arguments not to make, because the evidence refutes them

An adversarial review earns its keep as much by disarming bad arguments as by finding good ones. Each of these circulates in data center opposition, and each would be corrected in the room.

1. "Data centers lower nearby property values." The verifiable literature does not support it, and the strongest study finds the opposite.⁶² The legitimate version of this concern is a question, not a claim: no study has measured a data center paired with a 393 MW on-site power plant. Ask which coefficient this facility fits. Do not assert the answer.

2. "$120 billion in off-balance-sheet AI financing, per the Bank for International Settlements." Misattributed. The figure is from the Financial Times, 24 DEC 2025; both relevant BIS publications return zero matches for it. BIS economists did call these structures "shadow borrowing", under the usual disclaimer that the views are the authors' own. Moody's $662 billion is the better number and is properly attributable. The concern is real; the citation is not, and attributing it to BIS invites a correction that discredits the point.

3. "The project will be abandoned and stick the city with stranded infrastructure." No documented case of a US municipality left with stranded infrastructure by a data center bankruptcy was found. Kearney, Nebraska's city manager stated Compute North's payments "are paid in full." Cyxtera's facilities were bought and kept running.

4. "Ratepayers will subsidize this." The critique applies to grid-connected large loads whose costs get socialized. The islanded design is the answer to this criticism rather than an instance of it.

5. "The fuel cells and the compute load will expose neighbors to electromagnetic fields." The distances defeat this before the technology matters. NIEHS, reporting WHO measurements, puts a 230 kV transmission line at 57.5 mG immediately beside the conductors, 7.1 mG at 100 ft, and 1.8 mG at 200 ft, and states that beyond about 300 ft the field is indistinguishable from ordinary household background of roughly 0.5 to 4 mG.³⁶ The nearest existing home is 1,298 ft away, and ICNIRP's public reference level for 60 Hz magnetic fields is 200 µT, or 2,000 mG.⁵³³⁷ For scale, the same federal booklet measures a hair dryer at 300 mG at six inches.³⁶

And this facility has no transmission line at all. §5 requires permanent electrical isolation with no interconnection, parallel operation, or export, so the grid-tie infrastructure that produces the fields in every one of these studies is absent by design.⁵¹

The area's existing 60 Hz environment is set by transmission the facility has nothing to do with. Four 345 kV circuits, two 138 kV circuits and roughly a dozen 115 kV lines converge at the Spanish Fork substation under a mile north of the site.⁴⁰ A residence across the street from any energized transmission line sits in a measurably higher power-frequency field than one 1,298 ft from a facility with no line at all.³⁶⁴³ State that carefully if at all. "You are already exposed to more" reads as dismissive and invites the reply "so you concede there is exposure." The affirmative version needs no comparison: there is no transmission line, and beyond 300 ft the field is household background. And note a technical point that a knowledgeable objector would raise: magnetic field scales with current, electric field with voltage, so a line's kV rating does not by itself establish its magnetic field, and a heavily loaded 115 kV circuit can exceed a lightly loaded 345 kV one.³⁶

The professional literature on data centers and EMF runs in the opposite direction from the concern. CIGRE's 2025 review of the subject is about transmission lines interfering with data centers, not data centers emitting toward neighbors, and the design reference it cites, TIA-942-C Annex F, recommends keeping the data hall itself below 37.5 mG to protect the equipment.³⁸ That is an immunity specification for servers, not an emission limit for residents.

And one argument the record hands the Council, fully drafted

Columbia Law School's Sabin Center for Climate Change Law recommends decommissioning bonds for data centers, and names two local ordinances operating at comparable scale: Susquehanna County and Smithfield Township, Pennsylvania. §15.3.24's decommissioning provision is the weakest in the ordinance, and Volition's documented lifetime capital is $2.8 million against a $4 billion build.¹¹⁵¹

This is the one gap where the Council can be handed a solution instead of a problem, and it survives every optimistic assumption about Volition, because it only matters if Volition is no longer the owner.

23. Where this report is incomplete

Three national cases were run down and each is narrower than its reputation. Cheyenne is real but is not about Meta's operations: the Board of Public Utilities issued a Notice of Violation on 2 JUL 2026 naming Goat Systems LLC over Cupriavidus gilardii contamination tied to construction-phase fill-and-flush discharge, proposed a $10,000 fine, and Meta appealed on 14 JUL disputing which entity held the permit. Unresolved, and resting on regional press rather than the agency filing. Chandler and Mesa are confirmed negatives: noise complaints never escalated past voluntary mitigation, and both cities responded with prospective ordinances rather than citations. The Dalles was a public-records fight, not a water-use violation: the city sued to keep Google's water data secret, lost, and settled. No water violation was ever found against Google there. Do not cite The Dalles as an operator violation.

Unverified and must not be stated at the hearing: the public commenter's broken-promise claims beyond the Northern Virginia set, which was retrieved and largely refuted; and a second public commenter's BIS / $120B claim, traced and misattributed, the figure is Financial Times.²⁸

Known limits on the numbers here:

• The noise model's source term is the weakest link. The independent check against KSL's measured ~70 dB at Aligned produces 63.6 dBA, evidence against the source term, not for it. Treat every noise figure here as a lower bound and the property-line study as the only authority.²⁶
• The per-truck comparison depends on an assumed 100,000 mi/truck-year that appears in no document. On the deck's own basis, 3,500 trips/month at a 10-mile round trip, the comparison is far worse for the claim. The claim cannot be evaluated without a stated basis, which is why the fleet-level figure (2.2–2.3×) is the one to rely on.⁴³
• No traffic count for this building exists in the public record, by the city, the applicant, or this review. Every truck figure in this report is calculated here from published ITE rates against the verified 1,054,000 sq ft, not measured and not taken from the applicant.⁴¹⁴²⁴³ The building is mostly vacant today and its present truck volume is very low, unmeasured, and deliberately not asserted anywhere in this report: the comparison throughout is the proposed data center against the fully tenanted warehouse, not against current conditions. Two limits on the calculation: ITE rates are national averages whose observed spread at real sites is wide, PennDOT measured 0.04 to 0.38 truck trip ends per 1,000 sq ft against ITE's predicted 0.54 for buildings of this size class; and the split between LU 150 and LU 154 depends on an operating characteristic no one has specified, since the tenant is unknown.⁴²
• The depreciation curve applies R884-24P-33, a personal property schedule. A building retrofit is real property and depreciates differently.¹⁴ The pro forma request stands; the curve is illustrative.
• The CO2 method's agreement with Bloom's own datasheet is circular, same stoichiometry, same efficiency range.²⁵ The genuine validation is the 1% agreement with Novva's regulator-issued Utah permit, derived independently.

Potential to emit at the authorized nameplate. At the CARB DG certification standard of 0.07 lb/MWh, which is the highest rate a certified fuel cell could legally emit and 23× Bloom's published 0.003 lb/MWh, at an 85% capacity factor:³³⁰

250 MW × 8,760 h × 0.85 × 0.07 lb/MWh ÷ 2,000 = 65 tons/yr 393.25 MW × 8,760 h × 0.85 × 0.07 lb/MWh ÷ 2,000 = 103 tons/yr 500 MW × 8,760 h × 0.85 × 0.07 lb/MWh ÷ 2,000 = 130 tons/yr

for scale, at Bloom's published 0.003 lb/MWh, 500 MW gives 5.6 tons/yr

Both the permitted nameplate and the ordinance cap cross the 100 tpy NNSR major-source threshold for this Marginal ozone area.¹³ ⚠ But be careful with this one: it requires a unit emitting at the CARB certification ceiling, 23× worse than Bloom's published rate, so it is an argument about what the ordinance would permit a future applicant to build, not a prediction about this equipment.³¹¹⁴³ A potential-to-emit cap in the code, the provision the Justification Statement wrongly claims exists, is the one thing that would foreclose this permanently.

Two things the ordinance is sometimes accused of, which it does not do: it does not permit unlimited merchant power plants (§2 requires generation "scaled to serve" on-site load, and the definition bars offsite sale), and it does not lack a receptor-located noise limit (§6 sets 55 dBA at any residential zone boundary).¹¹

One fiscal item nobody has raised, and it runs in the city's favor: a 250 MW facility at 75% load consumes roughly 10 Bcf of natural gas per year. Utah municipalities levy a municipal energy sales and use tax on gas delivered within their boundaries.⁵¹ That is plausibly city-retained revenue comparable to Spanish Fork's entire 10.0% share of the property tax, and it appears nowhere in the fiscal case Volition presented or in this report's own analysis until now.

24. About the author

Sources

Every number here traces to a primary document

75 cited sources, numbered as they first appear, each linked back to the sentence that uses it. All 76 URLs were re-checked on 17 AUG 2026 and 75 return HTTP 200. One, the SSRN housing-price paper, returns 403 behind a CAPTCHA and is marked in the report as not citable. Three entries have no web address: the applicant's printed handout, the author's field observations, and the Planning Commission auto-caption transcript.

1. ↩ Utah DAQ, Intent to Approve DAQE-IN163550001-26, Volition Industries Inc. Spanish Fork Data Center, signed 4 AUG 2026 by Alan D. Humpherys, New Source Review Section Manager daqpermitting.utah.gov
2. ↩ Utah DAQ, public notice DAQE-NN163550001-26, published The Daily Herald 6 AUG 2026 daqpermitting.utah.gov
3. ↩ Bloom Energy Server datasheet, ES 6.5, document 1016932-20260204, 4 FEB 2026 bloomenergy.com
4. ↩ Volition Industries, project page for Spanish Fork, the applicant's live public page and one of the two materials distributed at its 15 AUG 2026 community meeting, the other being the printed handout at . The handout's QR code resolves here. Captured 15 AUG 2026, HTTP 200, byte-identical on re-fetch volition.eco
5. ↩ Utah Municipal Power Agency, FY2022 Integrated Resource Plan, 30 NOV 2022 (Spanish Fork customer profile p. xiii; load tables pp. 81-82; resource table p. 30) wapa.gov
6. ↩ Utah Geological Survey, Table 5.6, Hydroelectric Power Plants in Utah, reproducing EIA Form EIA-860, data through 2024 geology.utah.gov
7. ↩ Utah Geological Survey, Table 5.7, Solar Power Plants in Utah, reproducing EIA Form EIA-860, data through 2025 geology.utah.gov
8. ↩ Utah Geological Survey, Table 5.8, Biomass, Wind, and Other Power Plants in Utah, reproducing EIA Form EIA-860, data through 2024 geology.utah.gov
9. ↩ Utah DAQ Air Emissions Inventory, live ArcGIS FeatureServer query filtered to Utah County, retrieved 15 AUG 2026 (256 facility-year records, 71 facilities, reporting years 2017-2025) services1.arcgis.com
10. ↩ Utah County Assessor, parcel 67:228:0003 property record utahcounty.gov
11. ↩ Spanish Fork City, Title 15 Land Use, archived codified text, 130 pp., last internal revision 09/2019. This document predates the amendment and does not contain it: zero occurrences of "data center," "Onsite Power Generation," "megawatt," or "55 dBA". It is cited only for provisions already codified, principally the I-1 Light Industrial zone at §15.3.16.120 and the Conditional Use Permit chapter at §15.3.08.060, whose subsection (D)(2) reads "The Commission shall hold at least one (1) public hearing on the application" and whose subsection (C) lists the five mandatory findings. The amendment text itself is in the 18 AUG Council packet, cms.spanishfork.org
12. ↩ Utah S.B. 132 (2025), enrolled, creating the statutory category closed private generation system le.utah.gov
13. ↩ EPA Green Book, Utah nonattainment and maintenance areas (Southern Wasatch Front, Marginal ozone) www3.epa.gov
14. ↩ Utah State Tax Commission, 2026 Personal Property Valuation Schedules (Utah Admin Code R884-24P-33) files.tax.utah.gov
15. ↩ Utah County Treasurer, certified property tax rates by tax area utahcounty.gov
16. ↩ Deseret News, “Fingerhut to sell Sp. Fork facility,” 30 OCT 2001 deseret.com
17. ↩ Deseret News, “Gigantic Spanish Fork center taking big steps,” 20 OCT 2003 deseret.com
18. ↩ NBER Working Paper w35194, data centers and house-price indices nber.org
19. ↩ Court of Appeals of Virginia, Prince William Digital Gateway opinion, 31 MAR 2026 vacourts.gov
20. ↩ Utah Public Meeting Notice, Spanish Fork City Council agenda, 18 AUG 2026, item 7.3 utah.gov
21. ↩ Spanish Fork City, Power Department, net metering and SharedSolar spanishfork.gov
22. ↩ Skyline Electric, Spanish Fork Solar Farm project record (4.7 MW, ~12,000 panels, 27 acres, built on a closed landfill) skyline.us
23. ↩ U.S. Bureau of Reclamation, Strawberry Valley Project history (Eric A. Stene) usbr.gov
24. ↩ DataCenterDynamics, “Oracle to deploy up to 2.45GW of Bloom fuel cells,” Doña Ana County, New Mexico datacenterdynamics.com
25. ↩ EPA, GHG Emission Factors Hub (natural gas 53.06 kg CO2/MMBtu, HHV) epa.gov
26. ↩ KSL, measured sound levels at the Aligned data center campus, West Jordan ksl.com
27. ↩ U.S. Energy Information Administration, State Electricity Emissions dataset (Utah 2024, Total Electric Power Industry, All Sources: 21,120,098 metric tons CO2; 21,168 metric tons NOx) with net generation of 35,133,906 MWh from the Utah state electricity profile eia.gov
28. ↩ Utah Admin. Code R307-401-7, Public Notice (30-day comment period; hearing request within 15 days of legal notice) law.cornell.edu
29. ↩ Utah Admin. Code R307-421, Permits: PM10 Offset Requirements in Salt Lake County and Utah County. The rule is a PM10 offset rule, and its applicability test is a modeled ambient concentration, not a tons-per-year threshold: R307-421-2 applies it to new or modified sources of sulfur dioxide or oxides of nitrogen located in or impacting Salt Lake or Utah County, where "the modeled impact is greater than 1.0 microgram/cubic meter for a one-year averaging period or 3.0 micrograms/cubic meter for a 24-hour averaging period." No 50 tons-per-year figure appears in the rule. R307-421-5 further provides that the rule takes effect in a county only upon EPA redesignation to PM10 attainment, with R307-403 controlling until then law.cornell.edu
30. ↩ California Air Resources Board, Distributed Generation Certification Program (certification standard 0.07 lb NOx/MWh; Bloom Executive Order DG-058) ww2.arb.ca.gov
31. ↩ Utah DAQ, Approval Order DAQE-AN160660003-24, Novva West Jordan, 19 DEC 2024 (72 gas reciprocating engines, 37 diesel generators, 235.6 MW, 51.28 tpy NOx, 997,261 tpy CO2e, 31.6 tpy HAPs) daqpermitting.utah.gov
32. ↩ Utah DAQ, Approval Order DAQE-AN162810001-26, Joule Capital Partners, Fillmore, Millard County, 22 JAN 2026 (~175 MW, 230.12 tpy NOx, 865,632 tpy CO2e, New Major Source) daqpermitting.utah.gov
33. ↩ PacifiCorp, thermal generation fleet (Lake Side Power Plant, Vineyard, Utah) pacificorp.com
34. ↩ UAMPS, generation projects (Nebo Power Station, Payson, 140 MW combined cycle) uamps.com
35. ↩ Mountain Country Foods, Utah operations (195 E 1600 N and 2102 N Main Street, Spanish Fork) mcfoods.com
36. ↩ National Institute of Environmental Health Sciences. (2002). EMF: Electric and Magnetic Fields Associated with the Use of Electric Power, Questions & Answers. NIEHS/NIH. Transmission-line figures therein attributed to WHO (2010) niehs.nih.gov
37. ↩ ICNIRP. (2010). Guidelines for Limiting Exposure to Time-Varying Electric and Magnetic Fields (1 Hz to 100 kHz). Health Physics 99(6):818-836. General-public reference level for 50/60 Hz magnetic fields, 200 microtesla icnirp.org
38. ↩ CIGRE. (2025). “EMF Consideration for Data Centers Near Transmission Lines: Corridor Width Recommendations,” ELECTRA No. 341, August 2025, citing TIA-942-C Annex F and IEC 61000-4-8 electra.cigre.org
39. ↩ IEEE Std 519, Recommended Practice and Requirements for Harmonic Control in Electric Power Systems (5% voltage THD at the point of common coupling); EN/IEC 61000-6-4, emission standard for industrial environments standards.ieee.org
40. ↩ U.S. Department of Homeland Security, Homeland Infrastructure Foundation-Level Data (HIFLD), Electric Power Transmission Lines feature service, queried 17 AUG 2026 for the envelope around Spanish Fork (23 circuits returned; owners, kV ratings and terminal substations as attributed in the layer). Parcel centroid and bearing computed from the Utah LIR parcel polygon for 67:228:0003 services1.arcgis.com
41. ↩ New Jersey Department of Transportation, ITE trip generation rates and equations as used by the NJDOT Highway Access Permit program (rates current as of 21 Apr 2023). Land Use 150 Warehousing weekday daily 1.58*(X/1000)+45.54, weekend daily (X/1000)0.15; Land Use 154 High-Cube Transload and Short-Term Storage weekday daily (X/1000)1.40, weekend daily (X/1000)*0.94 nj.gov
42. ↩ Pennsylvania Department of Transportation. (2024, June). Warehouse Trip Generation technical memorandum, based on ITE Trip Generation Manual, 11th Edition, with observed counts at 15 warehouse sites. Table 2 (ITE LU 150 truck rate 0.54-0.64 per 1,000 sq ft), Table 4 (ITE LU 154 truck rate 0.22), Table 8 (observed truck share of daily trips, 20% for sites of at least 1,000,000 sq ft), and the ITE LU 154 high-cube definition of at least 200,000 sq ft and 24 ft ceiling height pa.gov
43. ↩ Derived for this report. Computed with Utah DAQ's own method, the manufacturer's published rate times full nameplate times 8,760 hours, exactly as DAQE-IN163550001-26 builds its limits. Inputs are the Bloom ES 6.5 datasheet and the DAQ Intent to Approve; every step is shown in the calculations appendix alongside this report. The underlying datasheet bloomenergy.com
44. ↩ Utah Code Title 54, Chapter 26, Large-Scale Electric Service Requirements (enacted by S.B. 318, 2025), including 54-26-504 exempting closed private generation systems from Public Service Commission oversight le.utah.gov
45. ↩ Utah Code 59-12-104, sales and use tax exemptions (subsection (80), fuel cells; subsection (84), qualifying data center equipment) le.utah.gov
46. ↩ Utah Code 63G-2-305, protected records (as amended by H.B. 507, 2026, subsection (2)(b)) le.utah.gov
47. ↩ Utah Code 73-5-8.3, water-use disclosure before approval of a large data center (created by H.B. 76, 2026) le.utah.gov
48. ↩ Utah H.B. 507 (2026), enrolled, including 11-41-202(2)(e) le.utah.gov
49. ↩ Utah S.B. 282 (2026), introduced, proposed 54-26-603 annual electricity and water reporting le.utah.gov
50. ↩ Utah H.B. 585 (2026), introduced, proposed data-center generation and fuel disclosure le.utah.gov
51. ↩ Spanish Fork City Council agenda packet, 18 AUG 2026, 131 pp., published to the city's Diligent Community agenda portal. Cover page: "CITY COUNCIL Agenda Tuesday, August 18, 2026 at 6:00 PM." Contains the official 04 AUG 2026 Council minutes, the staff memo at p. 111, the applicant's redlined submittal at pp. 112-120, and the Exhibit A enacting text. The 500 MW cap appears at p. 114 (redlined) and p. 124 (clean text): "shall generate a minimum of 50 megawatts and not more than 500 megawatts of nameplate generating capacity, measured per Onsite Power Generation Facility." spanishfork.community.diligentoneplatform.com
52. ↩ Scott Pham, Vicky Nguyen and Liz Wagner, "Bloom Energy Quietly Revises Emissions Estimates," NBC Bay Area Investigative Unit, 25 JUN 2015. Reports a 24-month Delaware average of 823 lb CO2/MWh against Bloom's then-advertised 773, and the datasheet revision to a 735-849 range that followed. Separately, on an AC Transit installation the article reports the cells held the advertised rate only for the first three of 22 months, averaging 825.31 lb CO2/MWh; that is a different system from the Delaware figure and the two should not be combined. nbcbayarea.com
53. ↩ Utah AGRC Parcels_Utah_LIR statewide parcel layer, live ArcGIS FeatureServer. Used to measure boundary-to-receptor distances and, for the citywide tax base, paged in full for PARCEL_CITY='SPANISH FORK' and deduplicated by parcel: 17,120 distinct parcels, $8,048,680,000 total real-property market value, of which 370 commercial parcels carry $1,337,300,000. services1.arcgis.com
54. ↩ Intermountain Health, Spanish Fork Hospital facility page (765 East Market Place Drive; opened 5 APR 2021) intermountainhealthcare.org
55. ↩ Active leasing listing, Canyon Logistics Center, 4000 E US Highway 6, Spanish Fork (former Fingerhut building): 1,054,000 sq ft total, 46,944 sq ft office, up to 451,792 sq ft available divisible to 196,249, 36 dock-high doors plus one 12'x14' grade door, 36 ft clear, 45'x46' columns, ESFR, built 1998, 69.95 acres, 4,000 A / 480 V / 3-phase with ~6-8 MW additional available loopnet.com
56. ↩ Utah County Assessor, individual property records queried by serial number, source of the owner-of-record names and assessed values in the taxpayer table (for example serial 270150090, Hydro Extrusion USA LLC, whose owner history also records Sapa Extrusions 2016-2018 and Cressona Aluminum 2011-2015 at the same parcel) utahcounty.gov
57. ↩ Utah County, Real Property Tax Detail, the tax actually billed and paid per parcel per year. Figures above are tax year 2025, the most recent complete year; the 2026 records show $0.00 because bills have not been issued. Example: serial 26:051:0129, IHC Health Services, General Taxes $217,857.17, paid in full 02 DEC 2025 utahcounty.gov
58. ↩ Utah Code 59-2-1101, exemption from property tax for property owned by a nonprofit entity and used exclusively for charitable purposes le.utah.gov
59. ↩ Spanish Fork City, audited basic financial statements, fiscal year 2025, property tax revenue $5,885,340 (prior year $5,741,117), statement of activities and governmental funds spanishfork.gov
60. ↩ Field observations by the author, 14 AUG 2026, Novva Data Center, West Jordan, with drive-by observation of the Utah Data Center (Bluffdale) and Meta Eagle Mountain. Six interviews. Uncalibrated phone-based sound readings taken as a controlled mobile transect: perimeter roads driven twice, clockwise, at a constant 18 MPH on cruise control, passenger window only, microphone aimed at the property, so readings are approximately property-line distance. Conditions cool, ~70 °F, windy, recent rain, approximately 16:00 to 18:00. Original observation, not a web-retrievable source. Marked UNCALIBRATED throughout original observation, no web source
61. ↩ Volition Industries, two-sided printed handout distributed at the 15 AUG 2026 community meeting, photographed by the author. Both sides transcribed verbatim and audited at sources/07-handout-AUG2026-transcription-and-audit.md; source images at sources/handout-AUG2026/ (side A sha256 9bf3c49e41b19142, side B sha256 ef3d34a6fba14bda). Physical document, no web source. Its QR code resolves to original observation, no web source
62. ↩ Keith Waters and Terry Clower, Data Centers and 2023 Home Sales in Northern Virginia, Center for Regional Analysis, Schar School of Policy and Government, George Mason University, August 2025. Robust hedonic regression on BrightMLS 2023 sales, R² 0.8672; distance-to-data-center coefficient −$15,886 (t = 31.106, p < 0.001), distance-to-industrial +$47,523; reduced model without the industrial control retains a negative data center coefficient of −$9,745 (R² 0.864) cra.gmu.edu
63. ↩ Alex Priest, Not In My Back Yard! The Effects of Data Centers on Housing Prices, SSRN 6314620, 27 FEB 2026. Difference-in-differences on Virginia air-permit issuances matched to ZIP-level house price indices; effects reported as economically small and slightly positive with confidence intervals ruling out substantial declines. Abstract verified; full text CAPTCHA-blocked on three attempts. papers.ssrn.com
64. ↩ Integra Realty Resources, analysis of home values within 1.5 miles of four Indiana data centers, 2021 to 2026, based on 20 Zillow estimated values per site and presented in support of a developer's zoning variance. Per-site results as reported: Microsoft LaPorte +42% against +48% countywide; Amazon New Carlisle +49% against +32% wvpe.org
65. ↩ Asa Watten, John Bistline and Geoffrey Blanford, Have Data Centers Raised Your Electric Bill? Causal Evidence from the United States, arXiv 2606.19777, 18 JUN 2026. Instrumental-variable estimate that data centers caused average US retail electricity rates to fall modestly from 2015 to 2024, with the caution that future supply constraints could reverse the effect. Two of three authors are affiliated with the Electric Power Research Institute. arxiv.org
66. ↩ Utah Code 59-2-103, Rate of assessment of property, Residential property (effective 1 JUL 2025). Subsection (2): all tangible taxable property assessed and taxed at a uniform and equal rate on the basis of fair market value as valued on January 1. Subsection (3): residential property allowed a residential exemption equal to a 45% reduction in value. Subsection (5): no more than one acre of land per residential unit may qualify le.utah.gov
67. ↩ Utah Code 59-2-303.1, Mandatory cyclical appraisals (effective 7 MAY 2025). Subsection (2)(a): the assessor shall annually update property values based on a systematic review of current market data, using a mass appraisal system (2)(b) jointly certified with the Commission (2)(c). Subsection (1)(a)(ii)(A): the system must be able to update all parcels each year. Subsection (3)(a): the assessor shall complete a detailed review of property characteristics for each property at least once every five years le.utah.gov
68. ↩ Utah Code 59-2-704, Assessment studies, Sharing of data, Factoring assessment rates (effective 1 JAN 2026). Subsection (1)(a): the Commission annually conducts and publishes studies of the relationship between assessed value and market value in each county. Subsection (2)(a): the Commission orders each county to adjust or factor assessment rates using the most current studies. Subsection (2)(b): factoring "may include an entire county, geographical areas within a county, and separate classes of properties." le.utah.gov
69. ↩ Utah Code 59-2-924, Definitions, Report of valuation of property to county auditor and commission, Transmittal by auditor to governing bodies, Calculation of certified tax rate (effective 6 MAY 2026). Subsection (1)(i): "Certified tax rate" means a tax rate that will provide the same ad valorem property tax revenue for a taxing entity as was budgeted by that taxing entity for the prior year. Subsection (4)(a): the rate is prior-year budgeted revenue divided by the adjusted aggregate taxable value computed under (4)(b), which subtracts eligible new growth at (4)(b)(iv). Subsection (1)(t)(ii)(A): "locally assessed new growth" does not include a change in value as a result of factoring in accordance with Section 59-2-704, reappraisal, or another adjustment. Subsections (1)(b)(ii)(D) and (1)(d)(ii)(A): semiconductor manufacturing equipment is excluded from both ad valorem property tax revenue and aggregate taxable value. Subsections (2) and (3): the 1 June and 13 June transmittal deadlines le.utah.gov
70. ↩ Utah Code 59-2-919, Notice and public hearing requirements for certain tax increases, Exceptions, Audit (effective 6 MAY 2026). Subsection (2): a taxing entity may not levy a tax rate that exceeds its certified tax rate unless it meets the requirements of the section, which include a public statement of intent, mailed notice, newspaper advertisement, and a public hearing le.utah.gov
71. ↩ Utah Code 59-2-919.1, Notice of property valuation and tax changes (effective 1 JUL 2026). Subsection (1): on or before July 22 of each year the county auditor shall notify each owner of real estate on the assessment roll. Subsection (2)(a): mailed 10 or more days before the board of equalization meets and before the taxing entity holds its hearing on a proposed increase. Subsection (2)(c): the notice must state the assessor's value, the taxable value, appeal instructions, and the appeal deadline le.utah.gov
72. ↩ Utah Code 59-2-1004, Appeal to county board of equalization (subsection (3)(a): application on or before the later of September 15 or 45 days after the Section 59-2-919.1 notice) le.utah.gov
73. ↩ Utah County Clerk-Auditor and Treasurer, public parcel tax and valuation records, tax years 2021 through 2025, retrieved 16 AUG 2026 for five separate owners: four in Tax Area 150 (Spanish Fork City): Hydro Extrusion USA 27:015:0090, Nature's Sunshine Products 24:046:0067, Costco Wholesale 47:308:0001, Longview Fibre / WestRock 26:049:0009. And one control in Tax Area 151 (Spanish Fork W/SPRV Drain District), Young Living Essential Oils 56:032:0001, which returns a different rate in every year and is what establishes the area rate as an area rate. Effective rate is computed as Net Taxes divided by Total Valuation for each parcel-year; the four Tax Area 150 parcels return an identical rate to six decimals in every year, which is what establishes it as an area rate rather than a parcel rate. Example record: utahcounty.gov
74. ↩ Marjorie Cortez, "Utah high schools could cost $145M each in era of intense construction inflation," Deseret News, 1 DEC 2019. Reports Utah high school construction and rebuild costs including Farmington High at $77.5M (2018), Mountain Ridge High at $82M plus $5M furnishings (2019), and planned or rebuild costs of $145M for Cyprus High, $145M for Timpview High, over $119M for Hillcrest and more than $113M for Brighton. These are 2018-2019 figures and construction costs have risen since, so they are used here as a conservative floor deseret.com
75. ↩ Spanish Fork City Planning Commission, public meeting of 05 AUG 2026, applicant presentation and question period, auto-captioned recording transcribed in full at sources/05 AUG Planning Commission Meeting Transcript.md. A Volition representative, at 52:13: "I live in Provo and we intend to headquarter here in this building if we move forward with this project." Two limits travel with this citation. The recording is auto-captioned without diarization, which this report treats as a lead rather than as evidence, so nothing load-bearing rests on it alone. And the statement is conditional and forward-looking: it is an intention, not a commitment, it appears in no filed document, and Volition's published contact address remains 412 W Rivers Edge Drive, Provo. (transcript retained 16 AUG 2026) original observation, no web source