The politics of artificial intelligence supply chains

Abstract The rising demand for generative artificial intelligence (AI) is fueling the growth of extractive supply chains to build and power the infrastructures this technology demands. However, there is ambiguity within the scholarly literature about what constitutes an AI supply chain. By connecting discussions across disciplinary boundaries, this article proposes a novel theoretical framework to conceptualise the AI supply chain as consisting of four inter-connected spheres of ‘AI infrastructure’, ‘AI preparation’, ‘AI deployment’ and ‘e-waste’. It adopts the case study of OpenAI’s ChatGPT to map one such example of an AI supply chain. In so doing, it analyses emerging forms of political contestation and resistance, revealing how the development of these supply chains gives rise to political issues of supply chain opacity, an increasing concentration of actors and power and new forms of coalitional politics. The article contributes to our understanding of AI systems through the development of a more holistic approach that examines end-to-end AI production as an iterative process, providing a new perspective on the journey of material flows within these logistical networks.