The rapid growth of artificial intelligence (AI) has led to increased energy consumption in data centers, raising concerns about carbon emissions. To address these challenges and support sustainability goals, the proposed research, DecAI, aims to decarbonize AI through long-term impact assessment and optimization. DecAI develops a framework to estimate carbon emissions across AI's development, operation, and use. This framework enables stakeholders, such as AI providers and data center operators, to evaluate and compare scenarios to reduce emissions. The framework will be released as a software tool to guide emissions reduction strategies. DecAI also optimizes AI operations with algorithms that manage compute, network, and storage resources in a carbon-efficient way while supporting long-term benefits for the electricity grid. By addressing AI's entire lifecycle, DecAI offers a holistic approach to reducing carbon emissions.

2025

2029