Abstract
<jats:p>U.S. data center operators face growing scrutiny over water and electricity use. Dry cooling has been positioned as the lower-water-impact choice on the basis that it uses no on-site water; this view excludes the upstream water and carbon associated with the additional electricity dry coolers consume. We compared dry and adiabatic cooling for a 10 MW reference facility across twelve U.S. data center hubs selected for adiabatic suitability, spanning hot-arid Southwest, high-desert Mountain West, high-plains, and Mediterranean climates, at three ASHRAE design conditions per location (0.4%, 1%, 2% exceedance); humid climates are addressed in companion work (in preparation). Annual cooling electricity, on-site and upstream water, Scope 2 carbon, and operating cost were computed for the 29 of 36 cases where both technologies are feasible. Adiabatic cooling reduced heat-rejection electricity by a median of 30.5% and maximum 41% under continuous-spray, fan-modulating operation, at a trade ratio ranging from about 8 to 56 gallons (30 to 210 liters) of on-site water per kWh saved, depending strongly on location. This exceeds the upstream water saved per kWh by about an order of magnitude, so adiabatic shifts the water burden from grid to site rather than reducing operational water. At Phoenix, Las Vegas, and Tucson 0.4%, dry cooling cannot reach the 107 °F (41.7 °C) target fluid leaving temperature (FLT) and is unavailable as a primary technology. On an energy-and-water operating-cost basis (excluding demand charges, which would further favor adiabatic), the cost ranking turns on the electricity-to-water tariff ratio more than on climate alone; El Paso, Tucson, Albuquerque, and Reno define the tariff and climate boundary where the adiabatic advantage breaks down. A four-axis framework pairing design conditions, grid carbon, water stress, and the electricity-to-water tariff ratio supports siting choices and bears on current state-level moratorium debates.</jats:p>