The Energy and Water Footprint of Generative AI

Interactive analysis of resource demands and environmental impact

Energy Consumption by Use Case
Source: Research data from "The Energy and Water Footprint of Generative AI" (Režun & Kapusta, 2025)
AI vs Bitcoin Energy Consumption
Source: de Vries (2024); International Monetary Fund (2024)
LLM Energy Efficiency Comparison
Source: OpenAI (2025), Anthropic (2025), Replicate (2025), TogetherAI (2025)
Key Energy Metrics
GPT-3 Training Energy
1,287 MWh
Equivalent to annual energy consumption of 123 U.S. households
Single ChatGPT Query
~0.3 Wh
10 times more energy than a Google search
Projected Global AI Energy (2027)
85-134 TWh
Comparable to the Netherlands' annual electricity consumption
Source: de Vries (2023); Ren & Chien (2024); IEA (2024)
Water Consumption by Use Case
Source: Research data from "Making AI Less Thirsty" (Li et al., 2025)
Key Water Metrics
GPT-3 Training Water Consumption
~5.4 million liters
In Microsoft's U.S. data centers (includes 700,000 liters on-site)
Water per ChatGPT Response
~10-50 mL
A 500ml bottle of water for 10-50 medium-length responses
Google Data Centers Water (2022)
5 billion gallons
20% increase from the previous year
Source: Li et al. (2025); Yale Environment 360 (2024)
Regional Data Center & AI Energy Consumption
Source: IEA (2024); TechInsights (2024); World Economic Forum (2024)
Global AI Energy Consumption Forecast
Source: IEA (2024); de Vries (2024); IMF (2024); Axios Generate (2025)
Task-Specific Resource Usage
TaskModelEnergy (Wh)Water (mL)
Email (500 words)GPT-4o0.310
Email (500 words)Llama 3.10.210
Essay (1,500 words)GPT-4o0.930
Essay (1,500 words)Llama 3.10.620
Saying "Please" to ChatGPTChatGPT0.010.3
Source: Režun & Kapusta (2025); Futurism (2025); Washington Post (2024)
Water Efficiency by Location
Source: Li et al. (2025); Microsoft Sustainability Report (2024)
Sources and References
  • Režun, T., & Kapusta, D. (2025). The Energy and Water Footprint of Generative AI: A Vanguard Leadership Perspective. International Leadership Journal.
  • Li, P., Yang, J., Islam, M. A., & Ren, S. (2025). Making AI Less "Thirsty": Uncovering and Addressing the Secret Water Footprint of AI Models.
  • de Vries, A. (2024). AI vs Bitcoin mining: Which consumes more energy? WIRED Middle East.
  • International Energy Agency (2024). Electricity 2024.
  • World Economic Forum (2024). AI and energy: Will AI reduce emissions or increase power demand?
  • Axios Generate (2025). Another case for worry — but not panic — over AI's climate toll.
  • The Washington Post (2024). A bottle of water per email: the hidden environmental costs of using AI chatbots.
  • Futurism (2025). Sam Altman Admits That Saying "Please" and "Thank You" to ChatGPT Is Wasting Millions of Dollars in Computing Power.
  • TechRadar (2025). Are you polite to ChatGPT? Here's where you rank among AI chatbot users.
© 2025 AI Energy Impact Dashboard | Created as part of research project on The Energy and Water Footprint of Generative AI