Data centers' carbon footprint higher than earlier estimates, study finds

Data centers expanding to meet rapid growth in artificial intelligence leave a carbon footprint significantly larger than previously calculated, according to a new study. The research warned that annual energy consumption by data centers could double between 2020 and 2030, potentially adding substantially to global greenhouse gas emissions. The authors found that the facilities release carbon through both direct electricity use and indirect sources, including cooling systems, server manufacturing and water consumption. The energy required to train AI models at large-scale data centers has grown at a compounding rate compared with traditional cloud services, the study said. Researchers called on cloud providers to disclose energy data more transparently and to accelerate the adoption of renewable energy at their facilities. No further details on the study's methodology or publication outlet were immediately available.