Net-zero by 2030 looks unlikely for US AI servers, even with best practices, model finds

Study title: Environmental impact and net-zero pathways for sustainable artificial intelligence servers in the USA

Published November 10 2025 in Nature Sustainability, the study models the sustainability implications of rapidly expanding generative AI server installations across the United States, projecting annual water and carbon footprints through 2030 and testing mitigation options.

Read the study abstract

The rapidly increasing demand for generative artificial intelligence (AI) models requires extensive server installation with sustainability implications in terms of the compound energy–water–climate impacts. Here we show that the deployment of AI servers across the United States could generate an annual water footprint ranging from 731 to 1,125 million m3 and additional annual carbon emissions from 24 to 44 Mt CO2-equivalent between 2024 and 2030, depending on the scale of expansion. Other factors, such as indus…

By the Truvace deskJuly 22, 20261 min read1 sourceEvidence: Moderate · 60/100 — Observational study · 1 independent sourcePeer-reviewedTRV-2026-0474· cite & verify

Environmental impact and net-zero pathways for sustainable artificial intelligence servers in the USA — "CAMPUS - Military Data Center" by Welcome to Switzerland backstage! is licensed under CC BY 2.0. To view a copy of this license, visit https://creativecommons.org/licenses/by/2.0/.

In brief

The findings matter because they translate AI growth into measurable resource pressures and show that meeting corporate net-zero aspirations will require more than offsets, with remaining uncertainty about how fast grids decarbonize, where servers are built, and whether best practices can be deployed at scale given current infrastructure.

Main points

Study quantifies US AI server deployment impacts as 731 to 1,125 million m3 annual water footprint and 24 to 44 Mt CO2-equivalent additional annual carbon emissions between 2024 and 2030.
Best-practice mitigation could reduce emissions by up to 73% and water use by up to 86%, but is limited by current energy infrastructure.
Authors find net-zero by 2030 unlikely without heavy reliance on uncertain carbon offsets and water restoration, and point to clean energy potential in Midwestern states.

The problem

US AI servers are projected to add 731 to 1,125 million m3 of annual water use and 24 to 44 Mt CO2-equivalent of extra annual emissions between 2024 and 2030, making net-zero by 2030 unlikely.

The rundown

The peer-reviewed analysis models US AI server growth from 2024 to 2030, estimating annual impacts of 731-1,125 million m3 water and 24-44 Mt CO2e depending on expansion scale, with variation driven by efficiency, grid decarbonization, and siting.

It evaluates net-zero pathways, finding that even with best practices that could cut emissions up to 73% and water up to 86%, infrastructure constraints and dependence on uncertain offsets and restoration make 2030 net-zero unlikely, prompting calls for coordinated private and regulatory action and leveraging Midwestern clean energy.