A recent 2024 analysis shows that every ChatGPT query consumes roughly the same amount of water as a 500 ml plastic bottle. The study links the high energy demand of large‑language‑model inference to water‑intensive cooling systems, highlighting a tangible sustainability challenge as billions of daily prompts strain global water resources for future generations.
Why Each ChatGPT Prompt Consumes a Half‑Liter of Water
Large‑language‑model inference requires massive computational power, which in turn generates heat that must be dissipated. Data‑centre cooling systems often rely on fresh water to absorb this heat, converting electrical energy use into measurable water consumption.
Data‑Centre Cooling Drives Water Use
High‑performance GPUs can draw up to 700 watts each, and clusters of these chips operate like industrial ovens. To prevent overheating, operators circulate large volumes of water through cooling loops, directly linking electricity draw to water intake.
Scale of Daily Queries Amplifies Impact
ChatGPT serves over 800 million weekly users, handling approximately 2.5 billion prompts per day—about 30 000 queries per second. At the estimated 500 ml per query, daily water consumption reaches the equivalent of 2.5 billion half‑liter bottles.
Environmental and Public‑Health Implications
The hidden water footprint adds a new dimension to AI’s environmental impact, especially in regions where water scarcity already threatens ecosystems and public health.
Potential Bottlenecks for AI Growth
As AI services expand, the demand for fresh water could become a limiting factor for new data‑centre construction, particularly in arid zones where cooling efficiency is already challenged.
Community Health Concerns
Intensive water withdrawal near data‑centre sites may reduce local water availability, increasing the risk of sanitation issues and related health problems for nearby communities.
Industry Strategies for Reducing Water Consumption
Data‑centre operators are exploring alternative cooling methods and water‑recycling technologies to lower fresh‑water intake while maintaining performance.
Air‑Side and Climate‑Based Cooling
Some facilities relocate to cooler climates or adopt air‑side cooling, which relies on ambient air rather than water, significantly cutting water use.
Closed‑Loop Water‑Recycling Systems
Closed‑loop systems capture and treat used cooling water, allowing it to be recirculated multiple times and reducing the need for new water sources.
What Users Can Do
While a single prompt has a negligible effect, awareness of the aggregate impact encourages responsible AI usage and supports demand for greener infrastructure.
- Stay informed: Understand the resource intensity behind AI services.
- Advocate: Support companies that adopt sustainable cooling practices.
- Optimize usage: Consolidate queries when possible to reduce overall demand.
