Three decades of surveying the sky have culminated in the celebrated cosmological standard model, yet 95% of the mass-energy of the Universe is still a mystery, residing in dark matter and dark energy. To address these mysteries, major cosmological surveys are ongoing and new ones will soon start. There are tremendous modeling and simulation challenges posed by these observations in order to enable the full interpretation of the associated cosmological measurements. In this talk I will discuss recent advances in large-scale simulations on the way to the arrival of the first exascale supercomputers. I will describe an ambitious end-to-end simulation project that attempts to provide a faithful view of the Universe as seen through the Rubin Observatory's Legacy Survey of Space and Time (LSST). This resulting synthetic sky provides many opportunities for exploring new ways to optimize the science return of LSST.