The hunt for dark matter's true nature has led us to an unexpected cosmic signpost: the distribution of interstellar and intergalactic dust. Through high-resolution hydrodynamical simulations of Milky Way-type haloes, I'll demonstrate how dust could serve as an observational discriminator between cold dark matter (CDM) and warm dark matter (WDM). Our simulations reveal how CDM's enhanced substructure necessitates higher supernova efficiency, compared to WDM, to reproduce observed satellite galaxy populations. This creates distinct dust production rates and distributions around host galaxies. I'll present our published results examining dust mass and distribution across cosmological models, before discussing broader implications for galaxy modelling with supercomputers and new applications of AI for understanding galactic evolution.