Water Availability, Sourcing & Climate-Driven Cooling Strategy (Siting Gate)

Power is the first gate. Water is the second — and in a growing list of markets it is becoming the gate that actually kills deals, because unlike power you frequently cannot buy your way past it at any price. You can pay for a faster interconnection, a behind-the-meter turbine, a more expensive PPA. You cannot, in an over-allocated basin in a drought year, conjure a firm water right that does not exist. That asymmetry is why this guide treats water as a top-three screen alongside power and fiber, run at the desktop-diligence stage, before land is controlled — not as a mechanical-engineering afterthought once the cooling vendor shows up.

The choice is unusually clean here because water sits at the center of a trilemma: a site optimizes for at most two of {low water consumption, low energy/PUE, low capital cost}, and the climate decides which corner you are forced toward. Evaporative cooling is cheap and energy-efficient but drinks water. Closed-loop / dry cooling consumes almost no water but burns more energy and costs more capital, and in a hot climate it caps your achievable density. Hybrid splits the difference. The chapter's job is to make you choose that corner deliberately, rather than inheriting it by accident from whatever the EPC proposes. The operational stewardship side of the same coin (replenishment, WUE targets, disclosure) lives in Chapter 15.4; the cooling hardware that implements these choices is engineered across Chapter 5.7 and Chapter 5.8.

A data center touches water in two places, and a serious diligence pass counts both. Scope 1 (direct) water is what evaporates at the cooling tower or adiabatic stage on site — the number the local community sees in the permit application. Scope 2 (indirect) water is what evaporates at the power plants feeding your load, and for a grid-tied facility it is typically the larger of the two: LBNL estimated US data-center indirect water consumption at roughly twelve times direct consumption in 2023. The consequence for siting: a site that looks water-light because it runs dry coolers can still carry a large embedded water footprint through a thermoelectric-heavy grid, and a hot, dry, solar/wind-rich grid can be the genuinely water-frugal answer once you net the two. Decide which scope your stakeholders are actually optimizing — the regulator and the community usually mean Scope 1; an honest stewardship accounting means both.

The reason water has moved up the screen is scale. US data centers directly consumed roughly 17 billion gallons in 2023; LBNL's central scenarios put 2028 direct consumption between roughly 38 and 73 billion gallons, with hyperscale sites accounting for about half. A single 100 MW evaporatively-cooled campus can withdraw 3–5 million gallons per day in peak summer — comparable to a town of tens of thousands of people, landing on the same basin, often during the same drought. That is no longer a footnote; it is the variable that triggers moratoria, disclosure laws, and the opposition coalitions covered in Chapter 3.11.

Not all water is the same water. The first sourcing decision is to design the cooling plant around reclaimed (non-potable) water wherever it is available, because it sidesteps the political fight over drinking water, often comes with incentives, and is increasingly mandated as a condition of approval. But reclaimed water is not free of constraints — it carries higher fouling, scaling, and biological load, which raises water-treatment capex and tightens the blowdown/discharge chemistry. Each source is a different gate with a different lead time; the table below is the fork.

The table is a priority stack. The default move in a water-stressed market is to engineer out of potable dependence — onto reclaimed where a utility will sell it, onto a closed loop where it will not. The expensive end of the table (desal, RO of produced water) is where operators go when water is the hard gate and power is plentiful — the Gulf and parts of the US Southwest — converting a siting blocker into a capex line. The mechanical implementation of these loops, and the makeup-water treatment train each source demands, is detailed in Chapter 5.7.

The decision that separates a durable site from a fragile one is rarely whether water is available in an average year — it is where you rank when there is not enough. In prior-appropriation regimes (the US West), seniority is everything: a junior right gets curtailed first, and a data center is almost always junior to century-old agricultural and municipal claims. In riparian and permit-allocation regimes, the regulator can simply tighten or decline to renew an industrial allocation when a basin tips into shortage. The consequence is severe and asymmetric: a facility that depends on a water right it loses in a drought year does not run at reduced capacity — if its cooling architecture cannot fall back to a dry mode, it loses goodput exactly when the grid is also stressed.

This is why curtailment is not a footnote but a design input. A site whose cooling can ride through a water-supply interruption — hybrid towers that pivot to dry operation, on-site storage sized for a defined outage, a closed-loop primary with evaporative assist only on the hottest days — converts a hard curtailment risk into a graceful energy penalty. A site whose density depends on year-round evaporation has no such fallback. Model the failure explicitly: the water-curtailment FMEA in Appendix F treats loss-of-makeup-water as a first-class failure mode with a defined detection, ride-through, and load-shed response, the same way the electrical chapters treat loss of a utility feed.

Climate is the variable that decides which corner of the trilemma the site can occupy, and it does so through one number: free-cooling hours — the fraction of the year the ambient wet-bulb (for evaporative/adiabatic) or dry-bulb (for dry coolers) is cold enough to reject heat without mechanical refrigeration. A cold, dry site (the Nordics, the US Upper Midwest, high-altitude interior) can offer 8,000+ hours of free cooling — effectively year-round — and reach annualized PUE near 1.1 with a dry or closed-loop design, getting both the low-water and low-energy corners of the trilemma at once. That is the rare site where you do not have to choose. A hot, humid site (Singapore, the Gulf, the US Gulf Coast) offers few free-cooling hours; there, mechanical refrigeration runs much of the year, and the only way to hold PUE down is to lean on evaporation — which is precisely the water you may not have.

That is the trilemma made concrete. In a cold climate, dry cooling is nearly free in both water and energy, so you take it. In a hot-dry climate, you face the sharp version of the fork: evaporative (low energy, high water — bets against the drought), dry/closed-loop (near-zero water, higher energy and a density cap from the warmer achievable supply temperature), or hybrid/adiabatic (evaporate only on the hottest days, dry the rest of the year — the pragmatic 2026 default for water-constrained warm sites). In a hot-humid climate, dry cooling barely works and evaporation is least effective exactly when you need it most, which is why those markets increasingly turn to seawater, district cooling, or simply accept a higher PUE. The cooling-architecture fork below is the heart of the chapter.

The cross-coupling with the rest of the project is the point. A water-light design in a hot climate accepts a warmer facility-water supply temperature, which narrows the delta-T available at the cold plate, which caps the rack density you can cool — so a water decision made at siting silently constrains the compute decision made in Chapter 1.1. Conversely, the industry's move toward warm-water direct-to-chip loops (45 °C-class supply, per the Rubin-generation roadmaps) is partly a water strategy: warmer loops expand the free-cooling envelope and shrink evaporative makeup, letting hot-climate sites stay water-frugal without surrendering as much density. The warm-water loop engineering is in Chapter 5.7; heat rejection across towers, dry coolers, adiabatic, and economizers is the subject of Chapter 5.8.

Every watt the facility consumes leaves as heat, and the default is to pay — in water and energy — to throw it away. The decision worth surfacing at siting is whether a nearby heat offtaker can take it instead: a district-heating network, a greenhouse, an aquaculture or industrial process, a swimming complex. Where the offtaker exists, waste heat flips from a rejection cost into a revenue line and, more importantly for siting, into a social-license asset — a community that heats its homes from the data center is a very different political environment than one that only sees the water bill. This is why Nordic markets, with mature district-heating grids and cold climates, treat heat reuse as a primary siting advantage rather than a sustainability garnish.

The catch, and the reason it is a siting decision rather than a retrofit, is temperature. District-heating networks want supply in the 60-90 °C range; air-cooled facilities reject heat far too cold to use directly, and even warm-water DLC loops typically need a heat pump to lift the grade — which costs energy and capital. So the heat-reuse opportunity couples back to the cooling design: a site planned for a high-temperature warm-water loop near a district-heating offtaker can deliver usable heat with a small lift, while a conventional air-cooled hall far from any network cannot, at any reasonable cost. Screen for the offtaker and the loop-temperature compatibility together, at siting. The full engineering and contract economics — temperature-indexed offtake, heat-pump integration, the Stockholm-class case studies — are built out in Chapter 5.9 and the sustainability/economics framing in Chapter 15.5.

Pulling the threads together: water enters the site-scoring playbook as a hard pass/fail gate first, a weighted criterion second. The pass/fail test is binary — is there a firm, permitted, drought-durable supply (or a viable closed-loop design with the energy budget to support it) and a compliant discharge path? A site that fails that test is out, regardless of how good its power or fiber looks, because no cooling plant runs without heat rejection. Only the sites that clear the gate get scored on the softer water dimensions — WUE achievability, reclaimed availability, heat-reuse upside, regulatory durability — that distinguish a good water site from a merely adequate one.

The deepest coupling is back to power and density. A water-rich, cold-climate site lets you run evaporative or dry cooling at low PUE and push maximum density — the two-corner win. A water-poor, hot-climate site forces a choice: spend energy (and accept higher PUE and a power-side cost) to stay water-light, or cap density to live within a warmer water-frugal loop. Either way, the water decision propagates into the power bill and the compute envelope, which is why it cannot be left to the mechanical contractor after the land is bought. Water-light siting also shows up as a portfolio strategy — pairing cold, water-rich training sites with latency-driven inference sites that accept a higher cooling cost — formalized in the market-clusters and scoring playbook of Chapter 3.13.