AI Energy Consumption: Quantifying and Reducing the Carbon Footprint

The article addresses the growing concern of AI energy consumption, highlighting that AI queries use significantly more electricity than traditional searches. A typical ChatGPT interaction consumes about 0.3 watt-hours, ten times more than a Google search. Generating an AI image consumes approximately 1.2 watt-hours, and a short AI-generated video can require anywhere from 30-950 watt-hours. Training a model like GPT-4 reportedly consumed between 50-62 gigawatt-hours of electricity. The article emphasizes that inference, or answering actual queries, accounts for 80-90% of AI's total energy consumption, surpassing the energy used for training.

The environmental cost of AI is real and growing fast, but the article suggests that individual choices can significantly reduce the AI carbon footprint. Strategies can cut the carbon footprint by 50-90% without sacrificing capability. Google disclosed in August 2025 that a median text prompt to Gemini consumes 0.24 watt-hours and produces 0.03 grams of CO2. The article highlights the need for greater transparency and accountability in AI infrastructure to address the environmental impact of AI.

The potential impact of AI's energy consumption is significant, with the rapid growth of AI potentially leading to a substantial increase in global energy demand and carbon emissions. However, the article also suggests that there are opportunities to mitigate this impact through increased efficiency, transparency, and responsible AI usage. The need for sustainable AI practices is crucial for ensuring that the benefits of AI are not outweighed by its environmental costs.

Transparency Compliance: As an AI, I have analyzed the provided text and generated the above summary. The analysis is based solely on the information provided in the source document. My purpose is to provide an objective and informative summary, and I do not endorse or promote any specific product or service.