Data-center lenders screen on-site water; power-plant water use goes unpriced

Lenders that finance data centers screen for the water drawn on site. Trellis reports that the diligence stops there, and that the larger number stays outside the file: according to an analysis by Ceres, a nonprofit, and projections from Bluefield Research, a market research firm specializing in water issues, most of a data center's water footprint is tied to electricity usage rather than cooling, and disclosure of that indirect consumption has a meaningful gap.

The mechanism is old and unglamorous. Most coal, natural gas and nuclear power plants heat water into steam to drive a turbine, then use more water to cool the steam back into liquid, so a data center buying electricity from such a plant consumes water it never sees. Janus Henderson, an asset manager, cites Lawrence Berkeley National Laboratory research putting indirect consumption at up to 75 percent of a facility's total water footprint. Meta, one of the few operators that has published such an estimate, reported an indirect figure for 2024 more than 20 times the water its data centers consumed directly.

None of that water appears on an electricity bill. Lenders assess power the way they assess other purchased services: cost, reliability, availability. Among the issues these deals hinge on, tenant credit, lease terms, power delivery and queue position, water at the generating plant is not usually a consideration, and the private credit firms now handling 70 to 90 percent of individual project-level borrowing for new construction are the ones doing that assessment. Risk that goes unpriced at the financing stage surfaces later, in the article's account, as delay, curtailment or a cost passed down the contract.

Keeping the exposure with the water utility has a reasonable defense. The utility is regulated, carries its own water-planning duties and is already scrutinized by credit rating agencies. What the defense does not produce is a number. No settled way exists to measure or report a data center's indirect water footprint, and Amazon, which tracks indirect water consumption, has not published figures, citing an absence of industry standards.

Trellis frames the water supply behind the generating fleet as an underaccounted risk sitting where AI's power demand meets economic growth, which puts the measurement problem on the same side of the table as the lenders: private credit writes 70 to 90 percent of individual project-level construction debt, so a standard for screening indirect water would likely arrive through those underwriting teams rather than through voluntary operator disclosure.