An accurate description of materials properties often requires atomic-scale models as well as the electronic structure of the materials under consideration. First-principles molecular dynamics (FPMD) provides a seamlessly coupled, consistent picture of both atomic structure and electronic structure. This approach is becoming increasingly popular for simulations of materials having complex structure, such as nanoparticles or interfaces. FPMD simulations are limited by the high computational cost of the underlying electronic structure computations, which are most often carried out using Density Functional Theory (DFT).  In recent applications, the need for higher accuracy has motivated the use of hybrid-DFT approximations, which further increases the computational cost of simulations. We discuss the current limits and tradeoffs of feasible FPMD simulation, using examples of nanoparticles and solid-liquid interfaces relevant to solar energy conversion.