Governance in practice: negotiating policy, power, and the environment in AI supply chains

The governance of Artificial Intelligence (AI) is rapidly institutionalizing across the globe, yet most debates remain strikingly dematerialized. Frameworks focus on ethics, risks, and "trustworthy AI", typically addressing data, algorithms, and uses of AI systems (Smuha 2021; Daly et al. 2019), but largely ignoring the material foundations of AI including critical minerals, semiconductor fabrication, data centers and e-waste. At the same time, AI is celebrated as a driver of "green" and digital transitions, even as its development requires energyintensive infrastructures and extractive resource supply chains that exacerbate ecological degradation and social inequality. My thesis therefore asks a seemingly simple question: What is AI made of, and how should this matter for AI governance? We argue that sustainable AI governance must be (re-)generated on material grounds. I propose the notion of material AI policy as a lens to connect AI governance with resource politics and political ecology and on Indonesia and ASEAN as a strategic yet understudied case in an environment where AI policy debates are currently dominated by software-centric perspectives from the Global North (Tironi and Albornoz 2025; Mohamed et al. 2020).