Water Requirements for Kuna Data Centers

Hydrogeologic Setting: The data centers draw from the Treasure Valley aquifer system, part of the Western Snake River Plain (WSRP) aquiferpubs.usgs.gov 1isu.edu 2. This aquifer underlies Ada and Canyon counties, consisting of thick sequences of unconsolidated sand, gravel, and clay with interbedded basalt layers up to ~1,525 m thick. Near Kuna (south Ada County), permeable basalts and coarse alluvium yield high well outputs (often thousands of GPM), whereas finer lacustrine deposits in some areas limit yields to tens of GPM. Groundwater generally flows northwesterly toward the Snake River and Boise River; in south Ada, flow is toward the Snake River to the south/southwest.

Recharge Sources & Historical Trends: Historically, artificial recharge from irrigation has dominated aquifer replenishment. In the Treasure Valley, ~90% of aquifer recharge came from canal leakage and flood irrigation percolationpubs.usgs.gov 3research.idwr.idaho.gov 4. Surface water from the Boise River Project’s canals (e.g. New York Canal) seeped into the ground, along with deep percolation of excess irrigation water applied to fields – together accounting for an estimated 80–90% of annual aquifer input. This practice caused water tables to rise significantly in the early- to mid-20th century; under flood irrigation, shallow groundwater levels in parts of the Snake River Plain rose ~60–70 ft from 1900s to 1950sisu.edu 5. However, with conversion to sprinkler irrigation and urban development, incidental recharge has declined, leading to groundwater level declines in recent decades. In the Boise Valley, as more land urbanizes and irrigation water is tightly managed, the aquifer has shown areas of drawdown and storage loss. A local news investigation in 2023 noted that domestic wells around Boise have begun drying up “amid rapid development” as flood irrigation wanesboisedev.com 6. Nitrate contamination is also a concern: south Ada County (including Kuna) overlies a nitrate priority area where >25% of wells exceed 5 mg/L nitrate-Nwww2.deq.idaho.gov 7. Long-term monitoring by DEQ/USGS/IDWR shows significant upward trends in nitrate in parts of this aquifer, prompting its designation as a “Lower Boise/Canyon” Nitrate Priority Area (ranked 8th worst in Idaho). In sum, the Treasure Valley aquifer is productive but under stress from reduced recharge and legacy agriculture impacts. Conclusion: The Meta and Gemstone data centers are tapping a confined alluvial-basalt aquifer with a history of heavy irrigation recharge and emerging declines/nitrate issues – Yes.

Water Sources and Rights – Meta (Brisbie) and Gemstone (Diode) Projects

Meta/Brisbie (East Kuna): Meta’s data center (under shell company Brisbie, LLC) is supplied via the City of Kuna’s municipal water system, which relies on groundwater wells in the WSRP aquiferboisedev.com 8research.idwr.idaho.gov 9. Kuna holds municipal water rights (e.g. Permit 63-31741, 18 CFS, service area-wide) and has drilled new high-capacity wells and built storage to serve the Meta site. The project’s development agreement confirms city water utility service to Meta’s campus, facilitated by infrastructure that Meta funded and transferred to the citykunacity.id.gov 10. For example, in late 2021 the city approved an infrastructure agreement with Brisbie for new wells, pipelines, a booster station, and a 2.0 MG storage tank to deliver both potable and cooling make-up water. The Kuna City Council also annexed the Meta property into the Kuna Municipal Irrigation System, pooling any appurtenant surface water rights for city use. Notably, the farmland Meta occupies had canal rights from Boise–Kuna Irrigation District; these rights were surrendered or assigned to the city’s pressurized irrigation network upon development. For groundwater, Brisbie secured an IDWR transfer of existing irrigation well rights to municipal use: Application No. 87042 proposed to change irrigation rights to industrial for the Meta data centerresearch.idwr.idaho.gov 11. (IDWR’s notice states: “Brisbie LLC…proposes a water right transfer to support industrial uses at a data center…near Kuna”, but the final order was not located – see D5.) In sum, Meta’s cooling and domestic water is sourced 100% from groundwater pumped year-round by Kuna’s wells (no direct canal water), under city rights and any transferred rights from the propertyboisedev.com 12.

Gemstone/Diode (Southwest Kuna): The Gemstone Technology Park (Diode Ventures) will operate its own private well field drawing from the same WSRP aquifer, independent of city utilities. The 620-acre site came with historic groundwater irrigation rights tied to the Yamamoto farmdiodeventures.com 13. IDWR records show D. Yamamoto Development LLC held Permit 63-33415 for 640 acres irrigation (2.98 CFS, up to 2,880 AF/year combined with surface supply) in T1N R1E Sec. 5 & 8 – covering the Gemstone parcelresearch.idwr.idaho.gov 14. That groundwater right was supplemental to Boise Project canal water (condition 065 required using all surface water first). Diode has applied to transfer these irrigation rights to industrial use. In its public outreach, Diode states the transfer will “result in a 33% reduction” in allowable annual volume, reflecting the need to leave formerly recirculated irrigation water in the aquifer. In other words, if the farm was permitted ~X AF/year, the data center will be limited to ~0.67·X to account for the historical return flow (≈ 33% of irrigation diversions) that previously recharged groundwater. The instantaneous diversion rate (pump size) remains the same, but total annual pumping is capped lower than ag usage. This ensures the new year-round pumping does not exceed the former consumptive use of crops (Anchor +2 compliance pending IDWR’s final order – see D5). As of Jan 2025, Kuna P&Z reports confirm Gemstone will “work off a well for water”, with on-site production wells supplying all cooling and facility needsboisedev.com 15. No surface water will be used; indeed the City of Kuna’s staff report notes no new strain on city water since Gemstone isn’t connecting to municipal supply. Instead, Diode must get IDWR approval for its well usage and aquifer recharge plan (infiltration of wastewater) before construction permits are issued. Conclusion: Both Meta and Gemstone data centers are sourcing groundwater from the Treasure Valley aquifer year-round, via dedicated high-capacity wells (municipal and private). This answers the first question in the affirmative – Yesboisedev.com 16.

Cooling Methods and Water Demand – Efficiency vs Consumptive Loss

Data Center Cooling Technologies: The cooling design dictates water use. Most large data centers use evaporative cooling towers or adiabatic coolers to reject heat efficiently, at the cost of significant water consumptioneesi.org 17. Industry-wide, the average Water Usage Effectiveness (WUE) is ~1.9 L per kWh of IT energy, meaning nearly 2 liters of water evaporated for every kWh of compute. High-efficiency facilities like NREL’s data center have achieved WUE ~0.7 L/kWh by aggressive economization and minimizing tower useenergy.gov 18. In contrast, air-cooled or dry-cooled designs can attain WUE ≈ 0 (no direct water use) but usually at a power efficiency penalty (higher PUE). Table T-1 compares cooling modes relevant to Idaho’s climate:

• Conventional Air + Water (Cooling Tower): WUE ~1–2 L/kWh; achieves low PUE (~1.1–1.2) by utilizing evaporative heat rejection in hot weather. Suitable for power-limited sites, but consumptive water use is high (80%+ of water evaporates). Annual water for a ~200 MW IT load can reach 100–200 million gallonsboisedev.com 19 (Meta’s largest 2020 centers used 170 Mgal). Idaho benefit: ample cool months for economizer use, but dry summers require heavy tower usage.
• Hybrid Air (Adiabatic/Fluid Cooler): WUE ~0.2–1.0 L/kWh (sprays water only during hottest periods)eesi.org 20. PUE slightly higher (~1.2–1.3) because dry cooling is used in milder conditions. Water use is lower and seasonally skewed (peaks on very hot days). Idaho: feasible given moderate summer wet-bulb temperatures; reduces annual water demand ~50% compared to full tower coolingenergy.gov 21.
• Direct Liquid (Warm Water) Cooling: WUE ~0 (closed loops, no evaporative loss)eesi.org 22. Heat is removed via liquid-to-liquid heat exchangers and cooled by towers or dry coolers. Can yield ultra-low PUE (~1.1) since cooling water is delivered close to chip heat source, improving efficiencyenergy.gov 23. However, waste heat still ultimately dumped – if via cooling towers, water use shifts to that stage; if via radiator, power usage rises. Idaho: could reuse warm water for secondary heating needs. Requires water quality management to prevent corrosion.
• Immersion Two-Phase Cooling: WUE ~0 on-site; uses specialty dielectric fluids (e.g. 3M Novec) that boil off chip heatmotivaircorp.com 24. Yields excellent PUE (~1.03–1.1) and near-zero water use, but introduces PFAS-based “forever chemicals” risksdatacenterdynamics.com 25. EPA now classifies PFAS coolants as hazardous substances, imposing strict leak containment and reporting rules. 3M’s exit from the PFAS coolant market by 2025 underscores environmental concerns. Likely not chosen in Kuna due to these liabilities.

Meta has not publicly detailed its cooling system, but its historical water disclosures (only ~1% from groundwater, 17% recycled water, rest municipalboisedev.com 26) suggest evaporative cooling with municipal make-up. Gemstone, however, alludes to using hybrid cooling: Diode’s FAQ mentions “cooling water outflow will be treated to drinking water standards”diodeventures.com 27 and sent to infiltration basins, implying a cooling tower with blowdown (outflow) rather than 100% dry cooling. During summer smoke or wildfire events, both sites may temporarily disable air-side economizers to avoid smoke ingestion, forcing full mechanical cooling and spiking water consumption (tower operation 24/7). Uptime Institute warns that ash-laden air clogs filters and cooling coils, necessitating recirculation mode and potentially reducing cooling capacityjournal.uptimeinstitute.com 28. In practice, sustained smoke could drive water use even higher than design peaks (as free cooling is unavailable), a risk factor for any facility relying on evaporative towers.

Consumptive Use and Cycles: In evaporative systems, ~80% of water withdrawn is consumed via evaporation and drift losseesi.org 29. Only the remaining ~20% exits as blowdown (high-mineral wastewater) or is reused elsewhere. High cycles of concentration in cooling towers can improve water efficiency: going from 3 to 6 cycles cuts makeup water ~20% and halves blowdown volumeenergy.gov 30. Both data centers are likely to run towers at 4–6 cycles if water quality allows, to minimize waste. Formula check: Evaporating 1 liter of water dissipates ~2,430 kJ of heat; for example, a 1 MW IT load (at PUE 1.2) rejects ~0.2 MW of heat to cooling (Q_reject = 1 MW × 0.2 = 0.2 MW). That requires ~0.03 L/s evaporation (Q_reject/h_fg ≈ 200 kW / 2,430 kJ/L) or ~0.11 m³/hour (~29 gal/min). Over a year that 1 MW would consume ~95 acre-feet if fully evaporative – illustrating the immense water footprint at scale. By maximizing economizer hours (cooling with outside air or water-side free cooling), the facilities can slash this usage. Idaho’s cool season offers many economizer hours, but July–August and smoky periods will demand heavy water use. Conclusion: Meta and Gemstone have substantial withdrawal caps (exact volumes pending final permits) and will see summer peaks in demand. The consumptive fraction is very high (~80%) under evaporative cooling, with only ~20% returned as treated water. Under hybrid/dry modes, consumptive fraction remains >> irrigation’s, just spread over more hours. These facts confirm the data centers’ water use is largely consumptive – Yes.

Apples-to-Apples: Irrigation Recharge vs. Data Center Return (Seasonal Ledger)

To test “water-positive” claims, we compare historical irrigation return flows to data center aquifer returns in the same aquifer, same season, same location. Before development: The farmland in question (Meta and Gemstone sites) received Boise River Project canal water April–October, flood-irrigating crops. Only a portion of applied water was used by plants; the rest percolated downward or ran off. For traditional flood irrigation in the Treasure Valley, on-farm efficiency can be ~50–70%, meaning 30–50% of water diverted ultimately recharged groundwater or surface drainspubs.usgs.gov 31diodeventures.com 32. IDWR’s transfer analysis for Gemstone essentially assumed ~33% of groundwater irrigation was returning to the aquifer (hence the 33% volume cut). Monthly pattern: In spring (Apr–Jun), canal deliveries flood fields that were dry over winter – initial deep percolation is high as soils saturate, providing a large spring pulse of recharge to the aquifer. Through mid-summer (Jul–Aug), farmers irrigate regularly; though evaporation and crop uptake peak, some water still percolates past root zones, maintaining aquifer levels. By fall (Sep–Oct), irrigation slows and canals are shut off by mid-Octoberkunacity.id.gov 33. Virtually no recharge occurs in winter (Nov–Mar) from these lands, as precipitation (~11″/yr in Kunawww2.deq.idaho.gov 34) mostly runs off or is held in the soil, and canals are dry. Overall, the historic aquifer recharge was highly seasonal**, concentrated in the irrigation season (perhaps ~2–3 AF/acre over Apr–Oct for flood irrigation on this parcel).

Under data center use: Groundwater pumping occurs year-round to meet cooling loads and other needs. However, only a small fraction of the pumped water is returned to the aquifer on-site. Gemstone plans to infiltrate its cooling tower blowdown via basins year-roundboisedev.com 35. Blowdown is perhaps 15–25% of makeup water (assuming 4–6 cycles of concentration). Thus, if Gemstone pumps (for example) 100 AF in July, ~80 AF evaporates and ~20 AF is released to infiltration basins as cooled, treated effluent. That 20 AF will seep into the same aquifer (lag time dependent on soil percolation). In contrast, the former farm in July might have received, say, 120 AF of surface water and 20 AF of well water, of which 30–50 AF percolated back (a higher absolute return). Crucially, timing differs: In summer, the farm likely recharged more water locally than the data center will, because the data center’s return (blowdown) is a smaller share of a smaller total withdrawal. In winter, the farm recharged nothing, whereas the data center will still send some water to ground (e.g. if Gemstone pumps 40 AF in January with low loads, maybe ~8–10 AF of blowdown infiltrates). This winter return is beneficial to aquifer levels, but the volume is relatively small and spatially concentrated in basins.

To illustrate, Table T-3 (conceptual monthly water ledger):

(Numbers are hypothetical for pattern illustration.) This shows the farm delivered much more water annually (and thus more absolute recharge) than the data center will pump. Even though a farm’s consumptive use was high in summer, the non-consumed portion (return flow) over the season could rival or exceed the data center’s managed returns. Only in winter does the data center provide a new source of aquifer recharge (albeit small), since the farm provided none then. But adding water in winter when the aquifer is naturally rebounding from fall rains is not equivalent in value to the lost summer recharge – summer is when aquifer levels drop and pumping demand from all users peaks.

Same-location caveat: The irrigation return water spread broadly under fields and via shallow drains, seeping to the aquifer over a wide area. The data centers’ returns are point-discharged in limited basins or reuse fields. For Gemstone, 30 acres of infiltration beds will receive all cooling effluentwww2.deq.idaho.gov 36; Meta will send its treated wastewater to 30 acres of alfalfa/corn irrigation nearby. This spatial concentration means localized mounding of groundwater beneath basins, but reduced recharge distribution compared to water spreading over hundreds of acres of farmland. In effect, the aquifer volume may be similar on an annual basis if the 33% cut is accurate, but the timing and distribution are not. Summer net withdrawals will be larger and broader in impact (drawdown) without the diffuse flood recharge. Therefore, any claims that these data centers are “water-positive” or use “less water than agriculture” are NON-PROBATIVE in an aquifer-season context – they do not compare like with like. Meta’s pledge to be “water positive by 2030” relies on basin-level offsets (e.g. funding remote conservation projects)boisedev.com 37, not on replenishing the same aquifer it draws from at the same time. Gemstone’s claim of using less water than prior irrigation is based on annual volume limits, not on matching the crucial summer recharge that flood irrigation provided (no evidence shows July–Aug aquifer balance parity) – thus NON-PROBATIVE without further hydrologic proof. Conclusion: Neither facility has demonstrated same-aquifer, same-season neutrality; offset claims fail the apples-to-apples test – Nodiodeventures.com 38. (NON-PROBATIVE)

Infrastructure, Permits, and Public Health Safeguards

Water & Wastewater Infrastructure: Both projects constructed extensive new water infrastructure under regulatory oversight. Meta/Brisbie funded a ~$8–10 million well field, reservoir, and pipeline network handed over to the City of Kunakunacity.id.gov 39. This includes high-capacity production wells (depth ~300–400 ft into the lower Boise aquifer sands/basaltresearch.idwr.idaho.gov 40), a 2 million gallon elevated storage tank, and distribution mains connecting to Meta’s campus. On the wastewater side, Meta is building a $50 million industrial wastewater treatment plant for its campusboisedev.com 41. Per its DEQ Reuse Permit application, this plant consists of three facultative lagoons (ponds) and chlorination, producing Class C recycled waterwww2.deq.idaho.gov 42. The plant will treat up to 0.5 MG/month initially (limited since cooling tower blowdown is apparently not entering this plant). The treated effluent will be stored in two lagoons in winter and land-applied on 30 acres of city-owned fields growing alfalfa/corn during April–Oct. DEQ’s staff analysis confirms this reuse system (Permit M-269-01) is solely for domestic sewage and some equipment washdown water from Meta, not the cooling tower blowdown. Meta’s cooling water strategy remains unclear; given no mention in the reuse permit, Meta may be disposing of its tower blowdown via a separate method (e.g. discharge to the city sewer system upstream of the new plant, or perhaps a deep injection well if ever approved). No UIC (underground injection) permit has surfaced, so likely Meta’s blowdown is blended with other wastewater and ends up in the reuse lagoons anyway. Gemstone/Diode, in contrast, will have on-site wastewater treatment for both domestic and cooling flows completely off-gridboisedev.com 43. Plans include septic systems for sanitary sewage (to be approved by Health District) and infiltration basins for cooling tower blowdown. Engineering details haven’t been publicized (the developer must submit designs to DEQ for approval), but Diode commits to treating cooling effluent “to drinking water standards” before infiltrationdiodeventures.com 44. This suggests a filtration and RO (reverse osmosis) or similar treatment skid to remove scaling minerals and additives, ensuring the water percolating is potable-grade. Chemical use: Both sites will use corrosion inhibitors and biocides in cooling water. Those must be disclosed in reuse permit plans. The reuse permit for Kuna (Meta) indicates chlorine disinfection and likely use of common tower chemicals (e.g. sodium hypochlorite, anti-scalants) – all of which must meet Idaho’s reuse water standards to protect soil and groundwater. Any hazardous chemicals (like PFAS) are not expected in use; Meta’s cooling is standard water-based, and Gemstone explicitly claims no chemicals of concern in outflow. (If immersion cooling had been used, PFAS-based Novec fluid could pose a spill risk, but neither project opted for that. Given EPA’s new PFAS rules, use of such fluids would trigger stringent spill reporting and waste management requirementsdatacenterdynamics.com 45.)

Legionella & Health Protections: Evaporative cooling towers carry a known risk of Legionella bacteria growth and aerosol transmission. Both operators are expected to implement rigorous water management plans per ASHRAE Standard 188ashrae.org 46. This includes maintaining biocide residuals, regular cleaning, and drift eliminators on towers to 0.001% drift or better (minimizing any offsite exposure)journal.uptimeinstitute.com 47. Idaho has no specific state Legionella regulations, but general duty applies; a major Legionella outbreak would expose the operators to liability. As a preventive measure, Meta and Gemstone will likely continuously treat cooling water (e.g. chlorination or bromination systems) and monitor for bacteria. Their DEQ permits require routine monitoring of recycled water quality and soil/groundwater near reuse siteswww2.deq.idaho.gov 48 – this indirectly ensures proper biocide use (excessive bacteria in lagoons or drain fields would be a red flag). For example, any E. coli detected in reuse water would violate Class C standards, prompting corrective action. Likewise, nitrate in groundwater near reuse fields will be sampled to ensure agronomic uptake (the permit limits loading so that nitrate in underlying water does not exceed 5 mg/L over background) – important given Kuna’s nitrate-priority status.

Governance & Enforcement: The key enforceable instruments are: IDWR water rights/transfer conditions, DEQ reuse permits, and local development agreements. IDWR’s permit for the Yamamoto well (63-33415) had robust conditions like mandatory flow metering and monthly water level measurements in the wellresearch.idwr.idaho.gov 49, and a trigger that if pumping “is causing material injury” to seniors, the Director can order curtailed use or mitigation. We expect similar or stronger monitoring requirements in the new industrial transfer approval (Anchor sources to be obtained – see D5). DEQ’s reuse permit M-269-01 (Kuna) imposes extensive monitoring and reporting: the city must sample lagoon effluent, soil nutrients, and shallow groundwater at the reuse site regularly and submit annual reportswww2.deq.idaho.gov 50. Any violation (e.g. rising nitrate or pathogens) could lead to permit modification or revocation. For Gemstone, DEQ will likely issue a reuse permit or a recycled water memorandum of understanding covering its infiltration basins, with requirements to sample the percolate and perhaps nearby well water. Until then, the city attached conditions to zoning approval: Gemstone cannot start operations until DEQ signs off on its water plans and Central District Health approves septic systemsboisedev.com 51. The City of Kuna itself has no ordinances limiting industrial groundwater usage beyond what IDWR allows, but its development agreement can include penalties for overuse or require emergency curtailment plans. (No such provisions are visible publicly for Brisbie’s agreement – likely because city rights suffice for now. This may be a governance gap: see Risk Register.)

Conclusion: Both data centers have built modern infrastructure to pump, store, use, treat, and return water with regulatory oversight. They employ standard safety measures (chemical treatment, redundant containment, monitoring) to protect public health and the aquifer. However, no evidence of any novel enforced condition to ensure aquifer neutrality was found – e.g. neither project is required to replace in-summer recharge losses gallon-for-gallon. The only enforceable limits are the annual volumetric caps and water quality standards in their permits. This points to a reliance on existing water law frameworks rather than project-specific mitigation mandates – a potential governance gap flagged below.

Risk Register – Cooling & Water Impacts

Drought & Aquifer Depletion: Risk: Continuation of the ongoing decline in aquifer levels due to reduced incidental recharge (from canal urbanization) coupled with new high-volume pumping. In dry years, canal allocations to farmers drop, so even less surface water seeps into ground, while data centers will still withdraw their full groundwater rights. Mechanism: Year-round pumping without equivalent recharge leads to net overdraft; over time, water tables fall. Likelihood: High (regional trend already observedisu.edu 52). Magnitude: Moderate to high – could lower local water table by several feet per year near the well fields. Shallower domestic wells nearby may go dry, and future municipal wells might need drilling deeper. Mitigation: None currently mandated beyond IDWR’s general “material injury” processresearch.idwr.idaho.gov 53. Aquifer recharge projects (e.g. winter injection of canal water) are being studied by IDWRidwr.idaho.gov 54, but none are in operation in Kuna. Residual Risk: Significant. No specific condition requires Meta or Gemstone to halt pumping if regional levels fall – a gap unless IDWR intervenes under a formal Ground Water Management Area or similar.

Seasonal Water Shortage (Peak Summer Conditions): Risk: During extreme heat or wildfire smoke events, water demand spikes (due to 24×7 tower use) and could temporarily exceed pumping capacity or rights limits. If a severe drought coincides, the city might enact water restrictions. Mechanism: On very hot days, the data centers’ cooling water flow will max out. Gemstone is designating capacity for this, but if multiple hot, smoky weeks occur, its annual allotment (cut by 33%) could be exhausted faster than anticipated. Meta’s supply depends on city wells that also serve other users; summer is peak use for all, so pressure drops or emergency curtailments are possible. Likelihood: Moderate. Boise area sees 100°F+ heat waves and wildfire smoke each summer (recent summers have had weeks of heavy smoke). Magnitude: Low to moderate – short-term need to rent temporary chillers or dial back IT load if water is insufficient. Mitigation: Both facilities likely have some on-site water storage (e.g. Meta’s 2 MG tank doubles as fire and buffer supplykunacity.id.gov 55). Contracts with the city might allow interruptible supply reduction to other users to favor Meta (unknown). Residual Risk: Manageable operationally, but if climate change worsens (longer smoke periods), water-heavy cooling could become a liability. Long-term, transitioning to more water-efficient cooling (liquid or dry coolers) is a mitigation strategy being considered industry-wideeesi.org 56.

Interference with Other Water Users: Risk: Large well pumping can draw down local groundwater levels, affecting neighboring wells or connected surface water. Mechanism: Cone of depression around the city’s East Kuna well field and Gemstone’s wells could overlap with domestic/ag wells in the vicinity. Also, the aquifer here may feed springs or drains; lowering it could reduce those flows (though none significant identified in immediate area). Likelihood: Moderate on well interference – IDWR permit 63-33415 explicitly anticipated this, requiring monitoring and enabling “show cause” proceedings if senior rights are harmedresearch.idwr.idaho.gov 57. Complaints can trigger mitigation (e.g. deepening a neighbor’s well or providing alternate water). Magnitude: Localized – a few domestic wells might need lowering of pumps or new casing if water levels drop several feet. The Boise River or major canals are far enough that impacts should be negligible to surface flows (plus mitigation Condition 121 on the rights compels offset releases if Snake River flows for salmon are impacted – unlikely given volume, but a backstop). Mitigation: Ongoing compliance with IDWR monitoring and reporting. If interference is proven, operators would likely finance affected well improvements (common practice under Idaho water law). Residual Risk: Low if monitored – Idaho’s system protects older rights; the data centers’ 2010–2022 priority dates are junior to most irrigation rights, so they’d be curtailed first in a conflict. Notably, no Groundwater Management Area has been officially designated in south Ada (unlike Mountain Home or Boise Front), meaning no active management plan exists – issues will be handled case-by-case.

Water Quality & Contamination: Risk: Degradation of groundwater quality from concentrated return flows or chemical spills. Mechanism: Nutrients: The reuse of wastewater on limited acreage could leach nitrates if over-applied. DEQ’s permit limits loading, but if the data centers expand faster than reuse acreage, nutrient buildup could occur. Chemicals: Cooling tower water contains treatment chemicals (anti-corrosion, biocides like bromine/chlorine). The infiltration of blowdown “treated to drinking standards” suggests these will be mostly removed, but any failure in treatment could introduce biocide residuals or metals to the aquifer. Likewise, a bulk chemical spill (say 55 gal of biocide) could infiltrate via basins. For Meta, any accidental release from its site (diesel, coolant, etc.) could reach groundwater if not contained – the area is not served by storm sewers, so spills would go to ground. Likelihood: Low. Both projects have engineered controls: lined basins or at least controlled application rates, and chemicals handled per hazardous materials code (secondary containment for bulk storage, etc.). Magnitude: Potentially moderate if it occurred – e.g. a significant chlorine release could sterilize soil and create chlorinated byproducts in groundwater. Mitigation: DEQ’s oversight: monitoring wells near reuse fields will detect any creeping nitrate or contaminantswww2.deq.idaho.gov 58. If exceedances are found, permits would force reduced loading or improved treatment. Spill prevention plans (SPCC) will be in place for fuel tanks and chemicals (especially at Meta, which has large backup generators and fuel storage on-site – air permit documents not in scope here but presumably required). Residual Risk: Low with proper management. The nitrate priority area status keeps regulators attentive; Kuna’s permit will likely be scrutinized for nitrate trends in reporting periodswww2.deq.idaho.gov 59. Gemstone’s commitment to drinking-water-quality effluent sets a high bar – if achieved, aquifer quality might even improve (slightly lower TDS in percolate than typical groundwater).

Legionella/Public Health: Risk: Aerosolized Legionella from cooling towers causing illness in the community. Mechanism: Cooling towers can emit droplets carrying bacteria, leading to Legionnaires’ disease outbreaks if poorly maintained. Likelihood: Very low if best practices followed. Modern towers with drift eliminators cut droplet release by 99.9%, and Idaho’s dry climate is less conducive to long-distance plume survival. Both sites will have maintenance staff (Meta’s facility staff, Gemstone may contract out) who understand ASHRAE 188 requirementsashrae.org 60. Magnitude: High if it happened (could hospitalize or kill vulnerable individuals, significant liability). Mitigation: Ongoing water treatment (continuous biocide feed and periodic testing). The DEQ permit’s emphasis on disinfection (chlorine) for reuse water suggests pathogen control is front-of-mindwww2.deq.idaho.gov 61. Additionally, any on-site building HVAC will have filtered make-up air, etc., to protect employees from tower drift. Residual Risk: Very low – typically only arises in “set and forget” systems; here, the spotlight on water use makes it unlikely they’d neglect microbial control.

Regulatory/Governance Gaps: Risk: Lack of proactive limits or required offsets means the aquifer could be drawn down significantly before any authority steps in. Mechanism: Neither project was conditioned with an adaptive management trigger (e.g. “if monitoring wells show X ft decline, reduce pumping” – no such condition found). Without a Ground Water Management Area designation, IDWR’s only recourse is after-the-fact curtailment via injury complaints or a new administrative area. At the city level, Kuna currently has no ordinance capping industrial water use in droughts (unlike some municipalities). Likelihood: High that this gap exists (it does). Magnitude: The consequence is that preventive action is unlikely – pumping will continue until a crisis (dry wells, etc.) forces action. Mitigation: The Ada County Commissioners in 2023 funded a Regional Wells Study to consider creating a local Groundwater Management Districtadacounty.id.gov 62 – which indicates concern, but it’s only exploratory. Enhanced monitoring networks (IDWR has some in East Kuna) can provide early warning if levels plummet. Residual Risk: Medium. It’s essentially a policy risk – until rules catch up, the aquifer relies on the data centers’ voluntary stewardship or broad state law. For example, Meta’s corporate goal to replenish more water than it consumes by 2030boisedev.com 63 could drive it to fund aquifer recharge projects locally, but this is not enforceable or confirmed (and Meta’s reported offsets so far are out-of-basin). Without binding requirements, the “water-positive” claim remains PR, not an operating condition. This gap is noted for corrective action (see D5).

In conclusion, both data centers rely on a finite groundwater resource with seasonally unbalanced use. They have modern engineering and permits to mitigate contamination and waste, but no requirements to truly balance what they take vs. historically was replenished. Chapter 11 will next examine how these water constraints interplay with energy and grid reliability, and whether water-limited operations could drive the adoption of advanced cooling or storage solutions. (Spoiler: in a water-scarce future, energy storage might be easier to come by than water – making the case for investing in water-efficient IT infrastructure now, rather than later.)

T-1. Cooling Methods Comparison (Idaho Climate Focus)

Sources: ASHRAE & Green Grid (WUE)eesi.org 68; DOE FEMPenergy.gov 69; EESI; DCD industry reportsmotivaircorp.com 70datacenterdynamics.com 71.

T-2. Water Sourcing & Permit Summary

Notes: Meta’s reuse permit M-269-01 covers domestic sewage from data center, not cooling waterwww2.deq.idaho.gov 86. Cooling blowdown handling for Meta remains unclear – possibly included in overall wastewater flow to lagoons (to be confirmed, D5). Gemstone’s water rights transfer expected to mirror Meta’s 33% consumptive use reductiondiodeventures.com 87; volume shown is estimated. Enforcement of pumping limits relies on IDWR – both will need telemetry meters and annual reporting per state law.

T-3. Seasonal Water Ledger – Irrigation vs. Data Center (Kuna area aquifer impacts)

⚑ “Aquifer Recharge/Return” here means water that reaches groundwater beneath the project area in the same month (or shortly thereafter). It excludes surface runoff or water leaving the site. Bottom line: The irrigated farms provided large seasonal recharge pulses (spring and summer) that helped maintain the aquifer and down-gradient water suppliespubs.usgs.gov 92. The data centers, in contrast, return a smaller total volume and do so mostly in winter and shoulder seasons – not when the aquifer historically got its big drink. This mismatch means claims of water neutrality are not hydrologically supported (NON-PROBATIVE), as discussed above.

T-4. Infrastructure & Safety Measures

T-5. Risk Register – Water & Cooling Impacts

Overall Conclusion: The Meta and Gemstone data centers will undoubtedly stress the local aquifer more than the pre-existing farms did, particularly in summer, due to high consumptive use and loss of incidental recharge (offset claims notwithstanding). They operate within legal permits, but those permits focus on quality and usage limits, not on preserving aquifer volume. Thus, the burden falls on monitoring and future management actions. Stakeholders should push for data transparency (monthly withdrawal and return reporting) and consider establishing a groundwater management area for south Ada County to proactively handle cumulative impacts. The companies tout sustainability, so holding them to account with actual data will be key to ensuring the aquifer beneath Kuna doesn’t become another casualty of high-tech growth.

D3: Calculations & Formulas Appendix

• Heat Rejection vs. IT Load: Q_reject = P_IT × (PUE – 1). For example, at 50 MW IT load and PUE 1.2, heat to remove = 50 MW × 0.2 = 10 MW (equivalent to 34.1e6 BTU/hr). This must be carried away by cooling systemsenergy.gov 117.
• Evaporation Rate from Cooling: 1 L of water evaporation absorbs ~2.43 MJ (677 Wh). Thus ṁ_evap = Q_reject / h_fg. In above example 10 MW = 10 MJ/s; divide by 2.43 MJ/L → 4.12 L/s evaporated (≈65 gpm). Over 24 h that’s 356 m³ (~94,000 gal). WUE in this scenario = (356,000 L per 24h) / (50 MW×24h = 1.2e9 Wh) = 0.296 L/kWh for that day (a relatively efficient day with some economizer use). Industry average WUE ~1.9 L/kWh suggests many days of much higher evaporationeesi.org 118.
• Cycles of Concentration & Blowdown: CoC = ratio of recirculating water TDS to makeup water TDS. Blowdown = Evap / (CoC – 1). At 4 CoC, blowdown = 1/3 of evap. In our 10 MW example, if running 4 CoC: Evap 94k gal/day, Blowdown ~31k gal/day. Total makeup ≈ 125k gal/day. Consumptive use = Evap + drift (~Evap, drift negligible when <0.002%) ~94k gal. Consumptive fraction ~75%, return ~25% (blowdown)energy.gov 119. If improved to 6 CoC: blowdown drops to ~19k gal, consumptive ~83% (better, but still majority lost).
• Irrigation vs. Data Center Water Balance: Assume 1 acre of land. Historic farm: received ~3 AF/acre per season from April–Oct (typical for Idaho alfalfa). Consumptive use by crops ~2 AF, return flow ~1 AF (33%)diodeventures.com 120. Data center on 1 acre (notional; actual facility spans >50 acres but water use scales with IT not land): uses ~30 AF/acre-year of water (for ~5 MW IT per acre of building). Consumptive ~24 AF, returned ~6 AF (assuming 80/20 split)eesi.org 121. Timing differs: the 1 AF farm recharge mostly in summer; the 6 AF DC return spread all year. The DC uses 10× the water per acre, but that comparison is apples-to-oranges since the data center acre produces economic output, not crops. The key is aquifer impact: farm added +1 in summer, DC adds +6 annually but mostly non-summer. To truly offset, DC would need to return ~1 AF in summer on that acre, which it does not – hence not neutral seasonally.

(All figures conceptual; actual Meta/Gemstone values to be measured once operational.)

Quote Bank (Key Verbatim Receipts):

• Diode (Gemstone) FAQ – “The land the project is on has historical groundwater rights for irrigation that will undergo a nature-of-use transfer to industrial use, resulting in a 33% reduction in the maximum allowable usage. The diversion rate will remain the same as it is today, but the total volume of water used will decrease.” (Diode Ventures, Gemstone Project Update, 2025)diodeventures.com 122. Claim: Data center will use less water annually than the farm did. Analysis: Based on IDWR transfer requirements, yes annual volume is cut by ~33%; however, this doesn’t account for seasonal differences. Probative? Partially – it’s true in a paper rights sense, but NON-PROBATIVE for aquifer health without seasonal context (they omit that the farm returned that 33% to aquifer).
• BoiseDev (Margaret Carmel) on Meta’s water pledge – “There are still a lot of unanswered questions about Meta’s promise to ‘add more water than we consume’ to Boise’s watershed… DEQ’s regional administrator…said the department has not been approached by Meta… Meta pointed to a blog post… listing projects like helping restore water to Lake Mead, groundwater recharge in New Mexico, and a wetlands project in Texas. The page did not offer any details about plans in Kuna.” (BoiseDev, Feb 18 2022)boisedev.com 123. Claim: Meta’s “water positive” promise lacks local detail. Probative? Yes – it shows Meta’s offsets are out-of-basin and no local aquifer recharge project exists (thus NON-PROBATIVE for Kuna aquifer).
• Kuna P&Z Staff (report on Gemstone) – “The data center would not need to tap into the city’s water and wastewater systems… it will work off a well for water. As part of the water plan, it will use infiltration basins, and whatever is not evaporated will go back into the local aquifer.” (Kuna Planning & Zoning staff report, cited by BoiseDev, Jan 29 2025)boisedev.com 124. Significance: Confirms Gemstone’s use of groundwater and intent to infiltrate returns on-site. Probative: Yes – an official summary requiring Diode to follow through with DEQ/IDWR approvals for those basins.
• IDWR Permit 63-33415 (Yamamoto farm) – “Right holder shall make full beneficial use of all surface water rights… before using ground water, and limit ground water diversions when surface supply is available… If surface right is sold, transferred or used elsewhere, this ground water right shall not be used without an approved transfer.” (IDWR Permit 63-33415, Cond 065, 2013)research.idwr.idaho.gov 125. Significance: Illustrates how historically groundwater was supplemental and tied to surface irrigation – once the farm is converted (surface water gone), the ground water right must be transferred (which is exactly happening). Probative: Yes – shows legal basis for transfer and ties to return flow considerations.
• EESI Report (Data Center water use) – “On average, a medium-sized data center can consume up to roughly 110 million gallons of water per year… Larger data centers can each ‘drink’ up to 5 million gallons per day… Together, US data centers consume 449 million gallons per day (163.7 billion annually as of 2021). The average WUE across data centers is 1.9 L/kWh… ‘0’ is ideal WUE (air-cooled only), which most cannot meet due to climate. Approximately 80% of the water withdrawn by data centers evaporates… the remaining water is discharged to wastewater.” (EESI, Data Centers and Water Consumption, 2023)eesi.org 126. Probative: Yes – authoritative stats on water use and consumptive fraction (used above to calibrate analysis).
• DOE FEMP on Cooling Towers: “Increasing cycles from 3 to 6 reduces cooling tower makeup water requirements by 20% and blowdown by 50%.” (DOE FEMP Best Practices)energy.gov 127. Probative: Yes – technical reference backing our note on CoC and efficiency.
• Uptime Institute on Wildfire Smoke: “The cooling system of almost all facilities will be affected… ash will either contaminate the water in a cooling tower or clog condenser coils. This will result in less cooling capacity, and backup units may be similarly affected. Direct air-cooled data centers need to monitor air quality and adjust filter maintenance… All data centers near a wildfire are at some risk from smoke and ash infiltration and should take action to mitigate contamination.” (Uptime Institute Journal, Sep 2020)journal.uptimeinstitute.com 128. Probative: Yes – expert warning that smoke events reduce cooling efficiency (meaning more water use to compensate or risk of capacity loss).

Each quote underscores aspects of our analysis, with A-class verifications. Where documents were missing (e.g. final IDWR transfer orders), we have flagged these in D5 below.

D5: Data Needed & FOIA Docket

1. IDWR Transfer Orders – Meta & Gemstone: Obtain the full Transfer 87042 order (Brisbie LLC to City of Kuna) and the Gemstone transfer application and order (likely filed 2023, Brisbie/Gemstone Yamamoto rights). These will detail the exact water rights numbers, volumes, conditions (consumptive use calc, mitigation). Rationale: To confirm the 33% reduction basis and any monitoring/mitigation requirements. Source: IDWR Western Region – likely via a public records request or the online water right files (not fully accessible via web). (FOIA Template to IDWR provided above can be used – request all documents on those transfers.)
2. DEQ Reuse Permit Engineering Reports: Specifically, City of Kuna Reuse Permit M-269-01 Engineering Report (likely by Parametrix, 2022) which will detail the design of Meta’s wastewater system (including any mention of cooling water handling), and any Gemstone/Diode reuse or UIC applications. As of 2025, Gemstone might still be in design; DEQ may have a draft if submitted. Rationale: These reports give monthly flow projections, acreages, and groundwater protection measures. For example, the Kuna report would clarify if Meta’s cooling blowdown is indeed excluded or combined. Source: Idaho DEQ Boise Regional Office – reuse program (the memo we cited is summary; the full app file would have more detail). (FOIA to DEQ as drafted above covers this.)
3. Irrigation District Data – Boise Project Board of Control: Need the canal delivery schedules and return flow estimates for the Kuna area. Specifically: on what date does Boise-Kuna Irrigation District typically start and stop deliveries each year, and what percentage of farm deliveries typically percolate to groundwater in that locale (south Ada, near New York Canal terminus)? The USGS/IDWR studies from early 2000s (Hutchings & Petrich 2002) might have it, but easier would be Boise Project’s own water budget. Rationale: To quantitatively support the seasonal recharge volumes we estimated. Source: Boise Project Board of Control or Nampa & Meridian Irrigation District (operating New York Canal) – likely an ops memo or an engineer’s report. (This could be a simple inquiry or request for any “return flow coefficient” data for that district.)
4. Treasure Valley Groundwater Model Runs: Specifically seek any scenario modeling showing the impact of cessation of irrigation recharge in East/South Ada County and addition of new pumping like Kuna’s industrial use. The new USGS model SIR 2023-5096 might include a scenario of full urbanization. Also, IDWR’s 2016 Eastern Treasure Valley model had predictive scenarios. Rationale: To get a quantitative handle on drawdown magnitude over time due to land use change. Source: IDWR Modeling Section or USGS Idaho – relevant publications or internal memos. (FOIA to IDWR could include “any model run results 2018–2023 for East Ada groundwater impacts.”)