Capgemini Research Institute’s Developing Sustainable Gen AI highlights the significant environmental impact of generative AI. GenAI consumes immense energy for training and inference, relies on GPU production involving rare earth mining, and contributes to growing e-waste—estimated at up to 5 million metric tons by 2030. Training GPT-4 uses power equal to 5,000 US homes annually. Nearly half of surveyed executives report increased GHG emissions from GenAI use, with projected carbon footprints rising from 2.6% to 4.8% of total emissions. As a result, 42% of organizations are reassessing their climate goals.

Join for free to read