There’s a noticeable and pertinent shift in Environmental, Social, and Governance focus regarding the environmental footprint of Artificial Intelligence. Earlier, the primary concern was carbon emissions, but with AI guzzling significant water resources, the focus has shifted to its massive water consumption.
Perhaps the answer lies in the accelerating demand for data centers due to the rapid growth of Artificial Intelligence workloads. This development has led to a massive increase in corporate water use, particularly in water-stressed regions.
As a result, water scarcity is emerging as a material financial risk, especially in already water-stressed or drought-prone regions, where the growth of AI data centers is doubling the strain on scarce water resources and directly competing with agricultural and household water demand.
Due to the mounting demand and to ensure water equity, ESG scrutiny of digital infrastructure is expanding beyond scope 1-2-3 carbon emissions. While accounting for carbon emissions remains crucial, water is considered the most critical risk due to local shortages and the consumption of water by AI infrastructure.
Projections as per a new study by Global Water Intelligence and Xylem suggest that the rapid boom of AI-infrastructure will drive a 129% increase in water demand across the AI value chain by 2050, adding 30 trillion liters of annual water demand.1
While AI’s water footprint is not as mammoth as that of some legacy sectors such as agriculture and livestock, electric power utilities, etc., it is nevertheless attracting significant attention for reasons, primarily:
(a) Air cooling, this technology relies on computer room air conditioning (CRAC) or computer room air handling (CRAH) units. It uses little or no water directly, but becomes less energy-efficient at higher server densities;
(b) Evaporative cooling, employs cooling towers that withdraw and consume water through evaporation. Evaporative cooling requires 1—3 liters of water per kWh of heat rejected.
(c) Adiabatic cooling, which pre-cools incoming air using a fine water mist before it passes through air-side economizers, thereby reducing cooling energy while consuming less water than conventional evaporative systems; and
(d) Direct liquid cooling (DLC) and immersion cooling, in this technology, coolant is circulated directly to server components or servers are immersed in dielectric fluid, significantly reducing reliance on air-side cooling. Although some configurations may still use water-cooled rear-door heat exchangers.
As per the International Energy Agency (IEA), evaporative cooling remains the predominant cooling technology in most hyperscale data centers due to its high energy efficiency, despite its relatively high water consumption.4
According to IEA estimates, data centers consumed approximately 20-50 billion litres of water per day globally in 2022, and this figure is projected to rise sharply with AI workload growth.5
Owing to the concerning data, corporate water stewardship is increasingly gaining centre stage in ESG reporting and is no longer being treated as a secondary metric.
This is evidenced in how the ESG scrutiny of digital infrastructure is expanding beyond scope 1-2-3 carbon emissions. Water withdrawal and local watershed impact are now explicitly evaluated in sustainability frameworks, including CSRD, ESRS E3, BRSR Core, and ISSB IFRS S2.
A study published in Nature Water introduced the Water Sustainability Index (WSI), a new standardized quantitative method that enables corporations to measure their water performance and reduce ESG greenwashing risk. The index is designed to account for different sustainability implications by integrating watershed stress scoring into performance tracking.6
The practical implications of this tectonic shift and enhanced focus on water stewardship need to be taken into account by ESG managers as well as B2B vendors supplying cooling, infrastructure, or computing services.
ESG managers must now build water into materiality assessments alongside carbon. Investor questionnaires from the CDP Water Security programme now probe digital infrastructure operators on water stewardship.
And, B2B vendors ought to address water performance metrics in sales cycles. ESG-aligned procurement frameworks increasingly include water criteria.
Microsoft and Google are good examples of how major technology companies are moving beyond reporting total water consumption to disclosing context-specific water stewardship metrics.
| Microsoft 2023 Environmental Sustainability Report disclosed water use intensity per megawatt hour. Google ESRS-aligned disclosures cite watershed-specific water stress data. |
2026 is the year in which multiple policy timelines converge. CSRD reporting obligations apply to large EU companies from FY2025 (reported 2026), ESRS E3 requires water and marine resources disclosure. SEBI BRSR Core mandatory disclosures apply to the top 1000 listed companies from FY2023-24 onwards. ISSB IFRS S2 climate disclosures are being adopted by multiple jurisdictions. CDP Water Security questionnaire adoption is growing. These obligations collectively make water performance a disclosure and liability issue, not merely a voluntary metric.
This has practical implications for investor relations teams as they must prepare for water disclosure narratives and data infrastructures. For sustainability heads, these converging timelines must align internal water accounting with external reporting frameworks before 2026 filings.
Even ESG data platforms, water accounting software, and third-party assurance providers are facing increasing demand from companies needing to comply with the converging frameworks in 2026.
Thus, the evolution of water accounting and the emergence of new measurement methodologies have significant implications for stakeholders across the ESG landscape. As water-related risks become increasingly material, companies must adopt standardized and transparent frameworks to accurately measure, disclose, and manage their water footprint in 2026 and beyond.
Browse our full library of articles, reports, and resources.
