This capstone project addresses the critical data gap in estimating Scope 3 greenhouse gas (GHG) emissions from cloud-computing at MIT, where existing spend-based frameworks lack the granularity and transparency required for effective emissions management. The rapid growth of cloud services, AI workloads, and data centers has increased electricity consumption and carbon emissions, making accurate cloud-emissions accounting an urgent sustainability challenge. The project aims both to highlight and quantify the shortcomings of the current approach and to move beyond these limitations by integrating financial data with granular supplier usage-based metrics to identify emissions seasonality and forecast emission portfolio until 2030. Although the study faced significant challenges due to data availability and siloed departments, it successfully establishes a theoretical framework for more credible carbon accounting. Recommendations include a phased implementation strategy, beginning with the standardization of procurement processes and progressing toward a full usage-based methodology that utilizes region-specific carbon intensity factors to align with industry’s best practices.
