Leveraging Climate and Time Zone: A Global Approach to Reducing AI's Carbon Footprint

This paper explores a global strategy to reduce the carbon footprint of AI data centers by exploiting climate and time zones. Based on the analysis of the existing research, we put forward “when”, “where” and “diagonal” strategies, and prove their feasibility and potential benefits by means of algorithms and experiments. The results show that the AI's carbon footprint can be significantly reduced by dynamically allocating workload in a global data centre while improving energy efficiency. Although these strategies offer promising solutions, it is important to take into account the challenges of geographic constraints when considering real-world implementation. This research contributes to the growing debates on sustainable AI development and encourages future studies on optimizing AI infrastructure for environmental efficiency.