The goal
This project helps people understand what is currently known about AI’s environmental impact, what remains uncertain, and why positions differ.
Read our manifesto →A source-led analysisAI has no
AI has no
single footprint.
Assessing AI’s environmental impact requires looking at different parts of the system, at different scales. This guide examines energy use, water use, carbon emissions and material resources. It separates what current evidence supports from what remains uncertain. It also considers how personal judgment shapes the way we weigh those impacts.
The method
This project reviews provider disclosures and independent research, checks what each claim measures, and makes uncertainty explicit.
Published estimates can differ by orders of magnitude depending on what is counted.
Review the sources →The evidence
AI systems use electricity and water, require hardware and mineral resources, and contribute to carbon emissions. Their environmental footprint extends beyond data-centre operations to hardware production and mineral extraction.
Explore the impacts →Why people may disagree
People often approach AI’s impact through one lens, making a for-or-against position feel obvious. Looking across the evidence, scales, trade-offs and our own assumptions reveals a more nuanced picture.
Understand the nuances →