AI Data-Centre Water Tracker: an open, reproducible referee for data-centre water burden

Abstract

AI disclosure: parts of this text were artificially generated with AI assistance and reviewed by the author. The models, and the conflicts they create, are named in the Conflict of interest section. An open, reproducible referee for the water burden of AI data centres. Ren et al. (arXiv 2606.21760, June 2026) define a Water Consumption Impact index (WCI) for AI data centres and apply it to ten sites, but release no code or data and leave two channels open: the coupling between a data centre and the hydropower on its grid, and the off-site relocation of water that closed-loop cooling produces. This working note ships the open version. It rebuilds the WCI for each of the ten seed sites from public inputs (operator environmental reports, utility open-records, USGS streamflow and USBR reservoir studies) and checks each recomputed value against the published one; it adds a hydropower-coupling flag asserted on primary reservoir data for two sites; and it adds a relocation ledger that quantifies how much of a site's water footprint moves off-site into the electricity supply. Three findings. Closed-loop cooling does not remove water, it relocates the great majority of the footprint to the grid, roughly 92 to 95 per cent on the modelled inputs used here, with the direction holding across every plausible variation of them. Three of Ren et al.'s ten sites, across three different operators, are non-operational planning figures presented as measured, and are excluded from the measured set. Recomputing the same index on the operators' FY2025 reports, with capacity and peaking held constant, moves five of the six primary-verified sites up by 15.6 to 33.3 per cent in a single reporting year, while the sixth is flat on total withdrawal even though its draw on potable municipal supply rose more than sixfold. Every input is named and motive-tagged, and a dependency-free Python script (reproduce.py) recomputes both tables and self-checks them against the cited figures. Every measured claim is traced to the primary document named in the Verification note; modelled or estimated inputs are identified as such rather than presented as measurements. The contribution is the open recomputation and the source ledger, not automated data extraction. Conflict of interest: Anthropic Opus 4.8-5.0 assisted with data retrieval, calculation, literature search and drafting across v0 through v0.4. OpenAI GPT-5.6 Sol assisted with the v0.5 disclosure and bundle revision. Anthropic states that it signed an agreement to use the compute capacity at the Colossus 1 data centre (https://www.anthropic.com/news/higher-limits-spacex). OpenAI and Oracle state that they entered an agreement to develop additional Stargate data-centre capacity in the United States (https://openai.com/index/stargate-advances-with-partnership-with-oracle/). The author therefore treats neither lab as neutral on the buildout. All operators are scored the same way, the seed set is Ren et al.'s Table 6 rather than author-chosen, and every estimate is named and motive-tagged. Independent analysis and open-science documentation only, not investment advice.