Preventing the failure of structures and components is at the heart of materials science. Failure can take different forms, including corrosion, brittle fracture and ductile failure as well as combinations thereof, and strongly depends on the loading regime (e.g., creep, fatigue) and the surrounding media as evidenced e.g., by liquid metal or hydrogen embrittlement. We utilize and combine various methods such as ab initio based modelling (DFT, ab initio MD, ab initio thermodynamics) and semi-empirical as well as machine learning potentials for large-scale atomistic simulation methods (e.g., MD, kMC, NEB) to study the fundamental failure mechanisms at the atomic and microstructure level. The so-gained understanding not only enables a targeted design for improved materials performance and weight reduction, but also contributes to better material models and design guidelines, thereby laying the foundations for a more sustainable use of materials.