Water Rights, Not Megawatts, Are Becoming the Gating Item on New Load

Leonard S. Hyman and William I. Tilles, both former heads of utility equity research at major brokerage houses, argue in a piece carried by nakedcapitalism and originally published at OilPrice that water scarcity has become an energy-security threat spanning hydropower, thermal plants, grids and fast-growing AI data centers [s1c1][s1c2][s1c3]. The consequence for anyone underwriting new load is unglamorous: water is starting to behave like a hard constraint on where a plant or a campus can go, not a paragraph in a sustainability report [s1c3].

Start with the scale. Power generation already accounts for roughly one third of water use in the United States, according to the authors [s1c4]. That is not a rounding error competing with agriculture and municipal demand; it is a claim on the same finite allocation, and the authors ask directly whether power producers, power users and everyone else will end up bidding against each other for a static or declining supply [s1c13].

The Colorado is their worked example. River flow has diminished and Lake Mead has dropped to record low levels [s1c5]. Hoover Dam sits at the foot of that reservoir and is a large power producer, and one of its biggest customers is the Metropolitan Water District of California, which the authors note stands to lose twice over, on water and on power [s1c6]. That is the shape of the risk operators keep missing: water and generation exposures are correlated, so a single hydrological event hits both sides of a balance sheet.

Europe shows the thermal version. Power producers there have faced too little water in rivers to cool their stations or to move fuel by barge [s1c7]. Add the grid: extremely hot weather affects the carrying capacity and operations of transmission and distribution facilities [s1c8]. So the bad day arrives as a bundle, with cooling limits, fuel logistics and line ratings all degrading in the same weather.

Data centers inherit the whole chain. They need water for their own operations and for the power plants serving them [s1c9]. Site diligence that stops at megawatts and fiber is measuring one leg of a three-legged dependency.

Policy is not obviously helping. The authors argue the Trump administration promotes still more water use by killing offshore wind projects, which have no freshwater needs, and encouraging nuclear and coal plants as replacements, which are big water users, while downplaying water cleanup efforts that would otherwise increase usable supply [s1c10]. On their characterisation, the favored replacement mix is more water-intensive than what it displaces [s1c17].

Nor should anyone assume price clears this. Higher prices will not increase precipitation where and when it is needed, the authors write, though they may discourage consumption, push loss reduction and reuse, and improve desalination economics [s1c11]. And because water has no substitute for the general population, raising prices or reallocating to big users carries political consequences [s1c12]. They suggest coordinated inter-industry planning instead [s1c16]. Site nominator, note the asymmetry: your permit is the discretionary one.

Two engineering points deserve a place in the model. In an editorial note, nakedcapitalism's Yves Smith, who says she is not an expert, writes that water can be reused around seven times but is potable only for the first two treatment cycles [s1c14]. She also notes that water used in thermal generation returns hot and cannot be discharged until it has cooled enough not to harm river and lake organisms [s1c15]. That is a dispatch constraint in warm weather, not a compliance footnote.

Watch whether new nuclear and coal announcements disclose water source and volume alongside capacity, and whether reuse and desalination costs start showing up in siting decisions rather than press releases [s1c10][s1c11].