The Exascale Computing Project (ECP) is delivering two exascale systems, Frontier and Aurora, and the applications and software technologies to run on them. Trends in scientific applications and computer architecture during the course of the ECP and the rise of AI/ML point toward an increasingly complex future. Architectures are relying upon accelerators of different types for performance. Scientific applications are adding physics, coupling models, and encountering new scientific workflow challenges. During this presentation, I will focus on the implications of these trends on how we will perform computational and data science and strategies that embrace the complexity with the goal of sustainably growing a computational and data science program.