Home / Technology / AI's Growing Shadow: Skyrocketing Energy and Water Use
AI's Growing Shadow: Skyrocketing Energy and Water Use
3 Jun
Summary
- AI could consume 3% of global electricity by 2030.
- Data centers used 448 TWh electricity in 2025 alone.
- AI's environmental cost extends beyond emissions.
A comprehensive assessment reveals that artificial intelligence could consume nearly 3% of global electricity by 2030, generating emissions comparable to the UK's past annual output. Data centers, crucial for AI, consumed an estimated 448 terawatt-hours of electricity in 2025, a figure projected to climb as AI workloads grow.
Beyond electricity, AI poses significant water challenges. Data centers used approximately 9.3 trillion litres in 2025, straining aquifers. The environmental toll isn't limited to carbon emissions; researchers emphasize that focusing solely on low-carbon solutions ignores substantial water and land usage.
Training a single large AI model, like ChatGPT-5, demands vast energy and water resources. However, the daily use of AI services now surpasses training costs, with generative searches consuming ten times more electricity than conventional searches.
Prompt conciseness and user choices can mitigate AI's footprint. For instance, simpler responses can reduce energy consumption significantly. The rise of AI-generated videos is also a growing concern due to their escalating energy requirements.
Experts suggest governments must integrate AI infrastructure into water and energy planning, as its environmental burden disproportionately affects vulnerable communities. Sustainable AI development requires embedding environmental considerations from hardware design to public use.
