Degree

Master of Science in Management

School

School of Business Studies (SBS)

Department

Department of Management

Date of Submission

Fall 2026-9-22

Supervisor

Dr. AbdulBasad Shaikh, Assistant Professor, Department of Management

Committee Member 2

Dr. Ashar Saleem

Submission Type

MS Management Research Project

Document Type

Restricted Access

Keywords

Artificial intelligence governance, AI infrastructure emissions, climate vulnerability, transboundary environmental harm, climate debt, Loss and Damage

Abstract

This thesis examines the relationship between AI infrastructure emissions in host countries and the climate impacts experienced by non-host, climate-vulnerable countries, and how this relationship is reflected in national AI governance. Drawing on existing literature and data on AI's growing carbon footprint, the transboundary nature of greenhouse gas emissions, and the documented climate exposure of Pakistan and Bangladesh, the study establishes the empirical basis for an asymmetry: the United States and China account for the majority of global AI infrastructure and its associated emissions, while Pakistan and Bangladesh, contributing under one percent of global emissions combined, are independently ranked among the world's most climate-vulnerable countries. The study then compares five jurisdictions, the United States and China as dominant AI infrastructure hosts, Pakistan and Bangladesh as climate-vulnerable non-host countries, and the European Union as a benchmark for the most advanced existing AI regulatory framework, using a structured focused comparison design. Documentary analysis of national AI policies and climate governance instruments examines what each jurisdiction's AI policy says, and does not say, about AI's environmental costs, required mitigation, and integration with climate governance. The findings identify a consistent governance gap: across all five jurisdictions, including the EU AI Act, no AI policy instrument acknowledges that AI infrastructure emissions are transboundary in nature. Where AI policy engages with climate or environment, it does so primarily by framing AI as a tool for climate action rather than as a system with environmental costs of its own. Read through environmental justice, climate debt, and postcolonial science and technology studies, this gap reflects whose experiences are recognised within AI governance and whose priorities shape it. The thesis concludes with five practical recommendations: extending emissions disclosure requirements to cover inference, linking national AI and climate policy processes, recognising AI infrastructure emissions within existing loss-and-damage frameworks, pairing AI-for-climate initiatives with acknowledgement of AI's own environmental footprint, and creating space within international discussions on AI and the Global South for the climate dimension this thesis identifies as missing, including through regional forums, coordinated ministerial positions, and sustainability-linked investment standards

Download

The full text of this document is only accessible to authorized users.

Share

COinS