Our algorithmic media landscape is driven by the miniaturization of smartness, a process centered on embedding artificial intelligence into increasingly compact and efficient devices. At the core of this transformation lies the graphical processing unit (GPU), a hardware originally designed for graphics acceleration but now integral to AI and computational parallelization. My PhD examines the political economy of GPU miniaturization by bridging two communities situated at opposite ends of its lifecycle: liquid-nitrogen overclockers in Taipei, Taiwan, who push computational limits through extreme cooling practices, and urban miners in Agbogbloshie, Ghana, who recycle and repurpose discarded GPUs. By tracing their material, environmental, and knowledge infrastructures, my PhD challenges dominant narratives of AI immateriality and universality, revealing smartness as deeply embedded in geological, thermal, and cultural processes. As a research-through-design PhD, it also addresses these elemental practices of labour by combining academic research and fieldwork with hands-on critical and speculative design projects. Disentangling smartness and its miniaturization through situated practices and embodiments placed in dialogue with qualitative data and analysis, the research fosters alternative modes of inquiry. By doing so, it invites us to speculate in more-ended ways about the past, present and near-future of computing power and its entanglements.