The hidden cost of asking AI

Every time you use ChatGPT, generate an image, or run an AI search, energy is consumed, water is evaporated, and carbon is emitted. Most people have no idea. Five charts tell the story.
Chart 1 of 5 - The scale
Global data centres used 415 TWh in 2024 - more electricity than Australia and Argentina combined. Source: IEA (2025). Energy and AI. https://www.iea.org/reports/energy-and-ai
Chart 2 of 5 - The surge
Data centre electricity has grown ~12%/year since 2017. The IEA projects 945 TWh by 2030, equivalent to Japan's entire electricity demand. Source: IEA (2025). Energy and AI. https://www.iea.org/reports/energy-and-ai
Chart 3 of 5 - Cost per query (multivariate)
A single AI video clip uses ~3.4 million joules - enough to run a fridge for nearly ten hours. Click legend items to filter by category. Hover for exact values. Sources: Luccioni et al. (2023). https://arxiv.org/abs/2211.02001 ; AImultiple (2025). https://aimultiple.com/ai-energy-consumption
Chart 4 of 5 - Water and carbon (multivariate)
Google used 30.7 billion litres of water in 2024, a 60% rise from 2022 driven by AI cooling. Microsoft's CO2 rose 75% between 2022 and 2024 despite net-zero pledges. Sources: Google (2025) ; Microsoft (2025) ; Meta (2025) ; Amazon (2025)
Chart 5 of 5 - The renewables race (multivariate)
Renewables for data centres are growing ~22%/year but cover less than 70% of demand by 2030 even under slow AI growth. Source: IEA (2025). Energy and AI. https://www.iea.org/reports/energy-and-ai

References (APA 7th)

AImultiple Research. (2025). AI energy consumption statistics. https://aimultiple.com/ai-energy-consumption

Amazon. (2025). 2024 sustainability report. https://sustainability.aboutamazon.com/2024-amazon-sustainability-report.pdf

Ember. (2025). Yearly electricity data. https://ember-energy.org/data/yearly-electricity-data/

Google. (2025). 2024 environmental report. https://www.gstatic.com/gumdrop/sustainability/google-2024-environmental-report.pdf

International Energy Agency. (2025). Energy and AI. https://www.iea.org/reports/energy-and-ai

International Energy Agency. (2025). Renewables 2025. https://www.iea.org/reports/renewables-2025

International Energy Agency. (2026). Key questions on energy and AI. https://www.iea.org/reports/key-questions-on-energy-and-ai

Luccioni, A. S., Viguier, S., & Ligozat, A.-L. (2023). Estimating the carbon footprint of BLOOM. Journal of Machine Learning Research, 24(253), 1-15. https://arxiv.org/abs/2211.02001

Meta. (2025). 2024 sustainability report. https://sustainability.atmeta.com/wp-content/uploads/2024/08/Meta-2024-Sustainability-Report.pdf

Microsoft. (2025). 2025 environmental sustainability data fact sheet. https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/2025-Microsoft-Environmental-Data-Fact-Sheet-PDF.pdf

Acknowledgements

Generative AI (Claude, Anthropic) was used to assist with code structure and debugging. All data, design decisions, and narrative framing are the author's own. Acknowledged per RMIT Library guidelines (RMIT University, 2024).