To facilitate the discovery and design of new materials, both “at-scale” and “multi-scale” models and simulations are used to predict the properties and behavior of heterogeneous, imperfect, real materials. To account for the most important properties and, models and simulations cover a wide range of space and time scales, starting with the nucleus and the atomic electronic structure (nm) all the way to the real parts (meters), and from point-defect formation (pico-seconds) to the operating characteristic times (months, years). We present examples of using atomistic results to directly predict properties and phenomena of real materials or to inform models and simulations at higher time and space scales. The examples include point defect formation in cerium oxides, formation of precipitates in aluminum-copper alloys, microstructure evolution of uranium oxides, and phase stability of plutonium-gallium alloys. A common theme for these case studies is modeling the free energy of multi-component systems at finite temperature. We extensively discuss the challenges associated with the feedback between atomistic and higher scales and focus on the question “With the advances in theory and computation, is it possible and/or desirable to predict continuum properties of real materials from atomistic models and simulations?” Our answer is “Yes, but No”.