Physical Address
23,24,25 & 26, 2nd Floor, Software Technology Park India, Opp: Garware Stadium,MIDC, Chikalthana, Aurangabad, Maharashtra – 431001 India
Artificial intelligence has moved from a niche technology to a central pillar of the global digital economy, but new research suggests its environmental cost is growing quietly and rapidly. A study finds that the AI boom in 2025 is driving a sharp rise in carbon emissions and water consumption, largely due to the energy demands of data centres powering large language models and other AI systems. As companies race to deploy increasingly complex tools, the research highlights how electricity use, cooling needs, and limited transparency are combining to create environmental pressures that remain poorly regulated. The findings raise fresh questions about how AI growth fits into global climate and water security goals at a time when both resources are already under strain.
Rising Emissions Linked to AI Computing Demand
The study estimates that AI related activity in 2025 could be responsible for around 80 million tonnes of carbon dioxide emissions. That figure places AI among major global emitters, comparable to the annual emissions of large cities and a noticeable share of aviation related pollution. The study focuses specifically on AI workloads, separating them from broader data centre activity to better understand how generative tools are reshaping energy demand.
This surge is driven by the enormous computing power required to train and run advanced AI models. Tasks such as real time language processing, image generation, and continuous user interaction require servers to operate around the clock. Much of this electricity still comes from fossil fuel based grids, meaning that every increase in computing demand translates directly into higher emissions unless energy systems shift more rapidly to cleaner sources.
Water Use Surges as Data Centres Expand
Alongside carbon emissions, the study points to a dramatic rise in water use linked to AI operations. Researchers estimate that AI related systems consumed around 765 billion litres of water in 2025. This figure exceeds the amount of water used globally for bottled drinking water over the same period, highlighting the scale of demand created by digital infrastructure.
Water is primarily used to cool data centre servers that generate intense heat during operation. Additional water is consumed indirectly through electricity generation, particularly where power plants rely on water intensive cooling processes. In regions already facing water stress, the expansion of data centres has begun to attract public concern, especially where local supplies compete with industrial demand driven by global tech firms.
Transparency Gaps and Growing Policy Concerns
One of the study’s central findings is the lack of clear reporting from major technology companies. While many firms publish sustainability reports, few break down energy use, emissions, or water consumption linked specifically to AI systems. This makes it difficult for regulators and the public to assess the true environmental cost of the technology’s rapid expansion.
The research argues that this gap shifts environmental burdens away from companies and onto society. Communities hosting data centres often face higher water demand and pressure on power grids, while climate impacts extend far beyond national borders. As AI becomes embedded across industries, the study calls for stronger disclosure requirements and clearer policy frameworks to ensure that innovation does not move faster than environmental accountability.
References:
https://doi.org/10.1016/j.patter.2025.101430
Banner image: Photo by Dhahi Alsaeedi on Unsplash
