Chapter 15.4
In this chapter · 5 sections
Water Stewardship
Evaporative cooling buys PUE at a permanent water-consumption line; dry heat rejection buys near-zero site water at a hot-weather energy penalty. Pick per basin, before the campus is plumbed and permitted.
What you'll decide here
- Whether this site evaporates water (low-PUE, water-exposed) or runs non-evaporative heat rejection (low process-water demand, weather-dependent energy penalty) — the irreversible fork that sets your basin risk and your social license, with its annualized PUE consequence calculated from the site’s energy bins rather than a portable penalty figure.
- Which water source you contract for — potable, reclaimed/non-potable, or harvested/grey — because reclaimed water can improve the permit case in a stressed basin only if seasonal volume, chemistry and permitted discharge also close, even when it raises capex.
- What you commit to disclose: source-split, on-site (Scope-1) vs total/embodied (Scope-3, including the water behind your electricity) water footprint — and whether 'water-positive' is a credible volumetric-replenishment program or a marketing claim a regulator will pierce.
- Your blowdown, cycles-of-concentration, biocide, and ZLD/discharge strategy — the operating decisions that determine whether you have a Clean Water Act permit problem, a Legionella liability, or a salt-disposal cost nobody budgeted.
- Whether a basin-level water-risk screen gates the site at all — run it before, not after, you contract power (the siting gate lives in Chapter 3.7).
For two decades water was a footnote in data-center design: cheap, abundant near the cheap-power and cheap-fiber sites operators already wanted, and invisible to the public. The AI build-out ended that. A single hyperscale AI campus can consume up to ~5 million gallons a day — the municipal draw of a town of 10,000–50,000 people — and it does so in exactly the arid, power-rich basins (Texas, the US Southwest, the Gulf, inland Spain) that the power-bound era pushed operators toward. The result is that water has hardened from a siting footnote into a top-three permitting pillar, alongside power and zoning, and into the most visible front in the community fight over whether a campus gets built at all.
Stewardship — the operating, accounting, and social-license discipline an operator owns for the life of the asset — is the subject here. The upstream siting gate (is there water, of what seniority, in what basin, under what regulator?) is run in Chapter 3.7, and the cooling-plant and cooling-tower engineering lives in Chapter 5.8. Choosing wrong costs you in PUE, in permit risk, in capex, and in the social license that is now as binding a constraint as the interconnection queue. Underneath everything sits the energy–water nexus: the cooling design that saves water spends energy, and the one that saves energy spends water. Stewardship is the practice of choosing which one to spend, per site, and accounting for it honestly.
The energy–water nexus: what each basin makes you spend
Heat rejection is the seam where energy and water trade against each other, and the trade depends on the ambient condition and the selected loop temperature. Evaporative (open-loop) cooling — cooling towers and adiabatic assist — exploits the latent heat of vaporization: evaporation carries latent heat away at the water’s operating temperature, an enormous thermal lever that lets a tower reject heat at a wet-bulb temperature far below the dry-bulb air. A tower’s access to that lower wet-bulb sink can reduce compressor duty when the selected chiller and weather bins support it, and roughly 85% of the water an open-tower plant withdraws is consumed — it leaves as vapor, not as discharge; the site’s exact fraction follows its blowdown, drift and cycles of concentration. Dry, non-evaporative heat rejection sends heat to outdoor air without consuming process water — at the cost of a higher approach temperature, more compressor hours in hot weather, and an annual electrical penalty determined by the named dry cooler, chiller and weather bins, concentrated in the summer peak that also stresses the grid.
This choice defines the water-stewardship strategy, and it is irreversible in the same sense the cooling-modality decision is: you plumb and permit a campus for towers or you do not. Choose evaporative and you have bought access to the lower wet-bulb sink and a permanent water-consumption line that a regulator, a journalist, and a county board will all scrutinize for thirty years. Choose dry, non-evaporative heat rejection and you have removed the process-water draw that mobilizes opposition and unlocked arid, power-rich sites — at a standing energy tax on every hot-weather hour for the life of the building, and with the water behind your megawatts still on the ledger. Either way you buy a penalty; stewardship is picking the right one for the basin you are in. The exception that proves it is conditional: a warm-water loop rejecting to dry coolers in a temperate climate lands low PUE and near-zero site WUE together, so model the hot-weather power draw and the upstream footprint before treating an arid site as settled.
WUE drivers and the source-strategy decision
Water Usage Effectiveness (WUE) — liters of water per kWh of IT energy, with the boundary deciding whether that is total input or net consumption — is the headline metric, defined in ISO/IEC 30134-9 and set against the EU's total-water-input reporting boundary alongside the rest of the metric stack in Chapter 15.1. The industry average for evaporatively-cooled facilities sits around ~1.8–1.9 L/kWh; best-in-class evaporative designs reach ~0.3–0.7 L/kWh; and designs with dry, non-evaporative heat rejection trend to ~0 WUE-site. Microsoft’s FY2025 cooling-and-humidification WUE was ~0.27 L/kWh for its fully owned and controlled datacenters operating for the full year. Separately, its December 9, 2024 zero-evaporation design announcement estimated avoiding more than 125 million liters per datacenter annually, with Phoenix and Mt. Pleasant pilots scheduled for 2026 and new sites to begin operating in late 2027. The headline number, though, hides three drivers an operator actually controls.
Driver one — cooling modality. The evaporative-vs-non-evaporative heat-rejection choice above is the dominant lever; nothing else moves WUE as far. Driver two — cycles of concentration. In an evaporative plant, the same makeup water can be recirculated several times before dissolved-solids buildup forces a blowdown. Running more cycles of concentration cuts makeup and blowdown but concentrates salts; scaling and corrosion depend on chemistry, while Legionella control also depends on temperature, residence time and the water-management program — a direct water-vs-chemistry trade managed in Chapter 5.8. Driver three — water source. WUE counts liters on its declared input or consumption boundary, regardless of quality; the stewardship question is not just how much, but which water — and that is where the source strategy becomes the most defensible lever you have.
Source is a decision in its own right. Potable water is the path of least engineering resistance and the path of greatest political resistance — Loudoun County data centers drew ~899 million gallons of potable water in 2023 (a ~250% jump), and ~57% of US data-center water still comes from potable sources, the fact that most reliably mobilizes community opposition and ratepayer-cost arguments. Reclaimed / non-potable water (treated municipal effluent, industrial grey water) costs more in treatment, conveyance, and sometimes a dedicated purple-pipe agreement, but it is increasingly the condition of the permit itself: water-stressed jurisdictions now mandate reclaimed sourcing, and a reclaimed-water commitment is often what converts a contested rezoning into an approved one. Harvested / on-site sources (rainwater capture, condensate recovery, treated process water) are marginal in volume but valuable in the disclosure narrative. The downstream cost is asymmetric: potable is cheap to build and expensive to defend; reclaimed is expensive to build and cheap to defend.
| Strategy | Annualized PUE | WUE (L/kWh) | Water source fit | Basin / siting fit | Primary downstream cost |
|---|---|---|---|---|---|
| Open evaporative (towers + adiabatic) | ~1.08–1.15 in hyperscale evaporative and free-air fleets (Meta 1.08, Google 1.10, Microsoft/AWS ~1.15) | ~1.8–1.9 typical; 0.3–0.7 best-in-class; above 2 for tower-fed chillers | Reclaimed strongly preferred; potable is the political liability | Water-abundant, cool, power-constrained basins | Permanent consumption line; blowdown/ZLD; Legionella program; social-license exposure |
| Hybrid (dry + adiabatic trim on peak days) | ~1.2 (CIBSE hybrid example) | Water only in wet-trim hours; mixed-design fleet references 0.20 (Meta) and 0.27 (Microsoft FY2025) | Reclaimed for the trim; small makeup volume | Hot-but-not-extreme; the pragmatic 2026 default | More plant complexity; still permits as a water user |
| Liquid cooling + dry, non-evaporative heat rejection | ~1.05–1.15 temperate (design illustration); hot climates add a summer compressor penalty | ~0 process water | Potable fine — tiny one-time loop fill, not ongoing draw | Water-stressed, hot, or politically exposed basins | Standing energy/PUE tax on hot-weather hours; higher heat-rejection capex |
| Full air-cooled / dry (no liquid) | ~1.4–1.6 (Uptime DX example 1.40; ~1.6 industry average) | ~0 process water | Potable fine; negligible | Arid sites; air-qualified racks only — Supermicro’s 60 kW HGX B300 layout is the named ceiling, while frontier training and inference run liquid-cooled NVL72-class racks | Cannot reach NVL72-class rack densities; worst energy per unit work where compressor hours are high |
The table is the same kind of cascade the workload archetype drives in Chapter 1.1: the leftmost column is the input you control, and everything to the right is a consequence you inherit for the life of the asset. The crucial 2026 shift is in the third row. Direct-to-chip liquid cooling — required by named dense NVL72 designs (Chapter 5.4) — is frequently confused with water consumption because it is 'water cooling,' though it consumes almost none. The technology-cooling loop is a sealed, recirculating circuit; its water is filled once and topped up rarely. What consumes water is the heat-rejection stage at the back of that loop — and a liquid plant with dry final heat rejection consumes essentially none. This is why the dense-liquid future is, if anything, easier to make water-neutral than the air-cooled past: the heat is already in a contained loop, and you get to choose how to reject it.
Scope & caveats
The near-zero endpoint requires dry, non-evaporative heat rejection; loop closure alone does not determine WUE-site.
Scope & caveats
Fully owned and controlled datacenters operational for the full fiscal year. Numerator: cooling and humidification water; not EU all-functions total input. The zero-evaporation next-generation design is a separate claim.
Scope & caveats
Expected avoided cooling-water consumption per datacenter from the announced zero-evaporation design, not a metered FY2025 fleet saving.
Scope & caveats
Typical open-tower evaporative share of withdrawn water that leaves as vapor; the site’s fraction follows its blowdown, drift and cycles of concentration.
Scope & caveats
Company-reported fleet figure for a mixed evaporative-assist fleet; a single hot-basin site posts far higher.
Water quality, blowdown, Legionella & discharge compliance
An evaporative plant is not a closed system; it is a continuous flow of water in (makeup) and water out (evaporation plus blowdown). As pure water evaporates, dissolved minerals concentrate in the recirculating loop, and at some threshold — the cycles of concentration — the loop must dump a fraction to drain (the blowdown) and replace it with fresh makeup. That blowdown is a regulated discharge: it carries elevated total dissolved solids, conditioning chemicals, biocides, and corrosion inhibitors, and in the US it falls under the Clean Water Act NPDES (Section 402) permitting regime if discharged to surface water, or a local pretreatment/sewer-use agreement if sent to a POTW. Underestimating blowdown chemistry is how a campus that cleared its withdrawal permit gets blindsided on the discharge side.
The other side of cycling water harder is biology. A cooling tower is a warm, aerated, nutrient-bearing aerosol generator — textbook habitat for Legionella pneumophila, the pathogen behind Legionnaires' disease, and an outbreak traced to a data-center tower is a reputational and legal event of a different order than a permit dispute. ASHRAE Standard 188 supplies the building-water risk framework; when adopted by law or contract, it requires a written Water Management Program built on a risk-assessment of every device where water contacts air, with defined control limits (temperature, biocide residual, conductivity), routine monitoring, validation, and documented corrective action. ASHRAE Guideline 12 supplies the supporting risk practice. In many jurisdictions an ASHRAE-188-conformant program is the de-facto standard of care, and its absence is what plaintiffs and regulators point to after an incident.
Deep dive: ZLD, blowdown minimization, and the salt-disposal cost nobody budgets
In a water-stressed basin with no surface-water outfall and a POTW that will not take saline blowdown, the operator is pushed toward Zero Liquid Discharge (ZLD) — treating the blowdown stream (brine concentrators, evaporators, crystallizers) until the only output is distilled water for reuse and a solid salt cake for landfill. ZLD removes the designed liquid blowdown outfall and recovers water; residual liquid streams, stormwater and salt handling still need their own discharge or waste authorization, and it is increasingly the price of admission in arid jurisdictions. But it is energy- and capex-intensive — evaporators and crystallizers are among the most energy-hungry equipment on a site — and it converts a water problem into a solid-waste problem: someone has to truck and dispose of the salt cake, a recurring opex line and a regulatory exposure of its own that rarely appears in the headline water narrative.
The decision tree is therefore: (1) can you discharge blowdown to a POTW or surface water under an achievable permit? If yes, optimize cycles and chemistry and discharge. (2) If no — saline basin, no outfall, or a no-net-discharge mandate — you are in ZLD territory, and the right move upstream is often to avoid the evaporative plant entirely (the non-evaporative/dry fork), because eliminating evaporation eliminates the concentration problem at the root. ZLD is the expensive way to keep evaporating in a place that does not want you to; non-evaporative final heat rejection is the way to avoid the fight. → cooling-tower water chemistry in Chapter 5.8.
Water-positive accounting: the honest footprint vs the marketing claim
'Water-positive' has become the headline sustainability commitment of the hyperscaler era — Microsoft reports that in FY2025 its replenishment exceeded its consumption for the year — a milestone toward its 2030 water-positive commitment, which requires sustaining that balance; Google has committed to replenish 120% of the freshwater it consumes by 2030 — actual replenishment reached >7 billion gallons in 2025 (4.5B in 2024), with contracted projects projected to deliver 19B+ gallons/year once fully implemented; Meta committed in August 2021 to be water positive by 2030 — a strategy built on dry cooling and watershed restoration, updated in December 2025 with watershed-restoration targets of 200% of consumption in high-stress basins and 100% in medium-stress. The claims are real programs, and they are also the part of the water story most vulnerable to being pierced by a regulator or a skeptical community — because the accounting hides three distinctions that determine whether the claim is meaningful.
Distinction one — on-site vs total footprint. The water you withdraw and consume at the campus — direct/on-site water — is the visible number. But the electricity that powers the campus is itself produced with water — thermoelectric and hydroelectric generation evaporate enormous volumes — so the total footprint includes a large indirect electricity-supply and value-chain water term behind every kWh. A site that runs zero process water on-site can still carry a substantial water footprint through its grid mix. Honest disclosure reports both; a water-positive claim that quietly ignores the energy-embodied term is incomplete by construction.
Distinction two — replenishment vs reduction. 'Water-positive' is usually a volumetric claim: the operator funds watershed projects (wetland restoration, irrigation-efficiency, leak repair, aquifer recharge) that return more water to the basin than the campus consumed. That is a legitimate and valuable thing — but it is offsetting, not reduction, and the offsets are only as good as their additionality, permanence, and — critically — their locality. Replenishing a watershed in a wet basin does nothing for a stressed basin where the campus actually draws. The credible programs explicitly target same-basin replenishment; the weak ones average across geographies in a way that lets a thirsty site hide behind a wet one.
Distinction three — basin-level risk and social license. Water is the most local of all the sustainability variables. Carbon is global and fungible; a ton avoided anywhere counts. Water is not: the only basin that matters for a community's drought, its wells, and its ratepayers is their basin. This is why water dominates the social-license fight even when the absolute volumes are modest against agriculture or municipal use — and why source/total footprint disclosure, same-basin replenishment, and a reclaimed-water commitment are worth more to a project's license than any global metric. The disclosure frameworks that formalize this — CSRD/ESRS, ISSB, and the ISO 30134 WUE standard — are treated in Chapter 15.7.
Deep dive: the indirect electricity-supply water term, or why 'zero-water on-site' is not zero-water
The cleanest illustration of why total-footprint accounting matters is the closed-loop, dry-cooled campus that reports ~0 L/kWh on-site WUE and then draws its power from a grid heavy in thermoelectric generation. Conventional thermoelectric plants (coal, gas, nuclear) reject waste heat through cooling towers or once-through cooling and consume water in the process — on the order of ~1–3 L/kWh of generation depending on technology, with evaporative-cooled thermal plants at the high end. A data center that has driven its on-site water to zero but consumes that grid power has simply moved its water footprint upstream to the power plant; the basin impact is real even though it is invisible on the campus meter.
This produces two non-obvious stewardship moves. First, clean-power procurement and water stewardship are coupled: shifting to wind and solar (which consume negligible operational water) reduces the indirect electricity-supply water footprint at the same time it cuts carbon — one of the rare places the energy–water nexus points the same direction. Second, an operator that wants a defensible water-positive claim has to be explicit about scope: reporting direct/on-site WUE alongside indirect electricity-supply and total value-chain water footprints. A claim that conflates the two — counting only the on-site number while running on water-intensive thermal power — is the kind of thing a CSRD auditor or an investigative journalist will surface. Honest scope discipline is the difference between a credible program and a claim that ages badly. → metric definitions in Chapter 15.1; clean-power procurement in Chapter 15.3.
Building a defensible water-stewardship program
Trace the hot-day boundary. At six cycles, blowdown = 300/(6 − 1) = 60 m³; tower makeup = 300 + 60 = 360 m³. Total input = 360 reclaimed + 15 potable = 375 m³/day. Discharge = 60 + 15 = 75 m³/day; net consumption = 375 − 75 = 300 m³/day. Input-WUE = 375/240 ≈ 1.6 L/kWh; consumption intensity = 300/240 ≈ 1.3 L/kWh. Normal operation fits the assumed reclaimed entitlement with 15 m³/day spare. It still requires accepted chemistry, source and discharge evidence before operation.
Flip. At three cycles, blowdown = 300/(3 − 1) = 150 m³ and makeup = 450 m³: the source is short by 75 m³/day. With the original source limit, the crossover is cycles = 375/(375 − 300) = 5; below five, reject the full wet duty. In drought at six cycles, maximum evaporation = 300 × (6 − 1)/6 = 250 m³/day, about 17% below the requested duty. Reduce wet heat rejection, use a qualified dry backup or curtail load; replenishment credits cannot supply the missing water. Chapter 5.8 converts the permitted evaporation into heat-rejection duty.
Do not average the WUEs. Annual total input = 300 × 90 + 65 × 375 = 51,375 m³; annual IT energy = 300 × 120 + 65 × 240 = 51,600 MWh. Their ratio is about 1.0 L/kWh; the day-weighted mean, [300 × (90/120) + 65 × (375/240)]/365, is only 0.89 L/kWh because the hot days carry more IT energy. Method: DOE cooling-tower mass balance and Chapter 15.1 for metric boundaries. Choose the source for its drought entitlement and chemistry, then carry its energy penalty into the annual bins.
Pulling the threads together, a water-stewardship program that survives a contested rezoning, a CSRD audit, and a drought year is built from a small number of decisions made early and documented honestly:
- Run the basin-risk screen before contracting power. Water-stress mapping, water-rights seniority, and basin-security under drought are siting gates — US drought (D1–D4) peaked near 37% of the country in late 2025, and a groundwater-certificate denial (the Arizona model) can strand a campus regardless of how good the power deal was. → Chapter 3.7.
- Make the evaporate-vs-design-out call explicitly, per basin. Screen dry, non-evaporative heat rejection against summer electrical headroom in stressed basins; retain evaporative cooling only where drought supply, treatment and discharge pass and the energy saving relieves the binding constraint.
- Commit to the most defensible source you can engineer. Reclaimed/non-potable where the drought delivery, treatment, conveyance and fallback case passes; treat a potable-only design in a stressed basin as a permitting liability to eliminate.
- Own the ASHRAE-188 Water Management Program and the blowdown/discharge chain as one system — cycles, chemistry, Legionella control, and NPDES/ZLD strategy optimized jointly, under one owner.
- Disclose source-split, on-site WUE, and total/embodied footprint — and, if you claim water-positive, back it with same-basin, additional, permanent replenishment whose volumetric benefit a third party can verify.
Cite this chapter
Fehn, J. (2026). Water Stewardship (Chapter 15.4). The Definitive Guide to AI Data Centers. https://aidatacenterguide.com/part-15-sustainability-and-efficiency/15-4-water-stewardship (accessed 2026-09-29).
@misc{aidc-15-4,
author = {Fehn, Jacob},
title = {Water Stewardship (Chapter 15.4)},
howpublished = {The Definitive Guide to AI Data Centers},
year = {2026},
url = {https://aidatacenterguide.com/part-15-sustainability-and-efficiency/15-4-water-stewardship},
note = {Accessed 2026-09-29}
}