Abstract
Energy-system optimisation models play a central role in shaping climate policy, infrastructure investment, and the range of decarbonisation pathways considered feasible. A growing reflexive literature within the modelling community has shown that these tools are not neutral representations of future energy systems, but embed normative assumptions, epistemic values, and political choices. However, two blind spots remain. First, critiques pay limited attention to the spatial architectures of optimisation models: how decisions regarding spatial resolution, regional aggregation, and system boundaries determine which places, infrastructures, and inequalities are rendered computationally visible. Second, debates over model politics remain largely situated within Northern modelling communities, with insufficient attention to the geographies of expertise that condition who builds, adapts, and contests these models, and whose futures become computationally representable. We argue that these blind spots matter because the organisation of modelling is not incidental, but constitutive of the transition pathways that models can produce. Building on insights from critical climate geography, we conceptualise energy-system optimisation modelling as a spatial and epistemic practice that participates in producing uneven transition futures. We illustrate this argument through open-source modelling frameworks applied in lower-income countries, highlighting the distinction between additive and constitutive treatments of space. We conclude by outlining a research and capacity-building agenda for a spatially grounded, reflexive, and distributed approach to energy modelling. Specifically, an agenda that addresses two questions: whose places, knowledges, and futures are made computationally visible, and who defines the problems to which modelling is applied, and what purposes those futures are expected to serve.