AI training and deployment consume substantial electricity, but carbon outcomes remain weakly integrated into routine model development decisions. This paper presents the Green AI Carbon Optimizer with two primary contributions:
-
Carbon Aware Cloud Region Recommendation Method: We combine regional grid carbon intensity, renewable share, and data center Power Usage Effectiveness (PUE) into a unified scoring model across 100+ regions from major cloud providers. For a reference workload (8*A100, 100h), estimated emissions in our sampled regions range from 7.74kg to 272.00kg CO₂. Selecting the best region instead of the worst corresponds to a 97.2% reduction relative to the worst case. Ablation shows that ranking by renewable share alone can select regions with higher CO₂ emissions than rankings including grid carbon intensity.
-
Global AI Energy Demand Forecasting Pipeline: We fit a power law relation between parameter count and training energy using 26 anchor models. This fit is combined with scenario assumptions on model growth, hardware efficiency, and training frequency, and we evaluate sensitivity to inference ratio and ecosystem scaling. Across scenarios, projected 2030 demand ranges from 7TWh to 1,436TWh under the stated assumptions, highlighting the importance of deployment choices, model scaling discipline, and transparent energy reporting.
Blogger's Review: This research emphasizes the importance of considering carbon emissions in AI model development, especially within cloud computing environments. By effectively selecting regions and forecasting energy demand, we can promote more sustainable AI practices and minimize environmental impacts.