The ability to predict, design and make the perfect material with just the right properties to do what we want, how we want, and when we want is the holy grail of materials research. Such “materials on demand” require control over thermodynamics, kinetics, nonequilibrium behavior, and structure across many length and time scales. With continuing advances in computer simulation capabilities, we have never been closer to the goal, but many challenges – and opportunities – remain. Many of those are at the boundaries of the subfields of materials research, where ideas from one area spur advances in others, and where computational tools and concepts are transferable across domains and scales. At the same time, foundational understanding at one scale can help understand new discoveries at different scales, regardless of the nature of the material and the forces holding it together. In this lecture, we show how atomic and molecular crystal structures – made possible by chemical bonds – can be realized in silico for non-interacting nanoparticles and colloids via entropic bonds. We show that similar crystallization pathways are followed by both molecular and colloidal fluids regardless of driving forces or relevant length scales. Finally, we show how colloidal crystal prediction may be amenable to modern tools used for atomic crystal prediction.