The control of materials properties and their relation to the manufacturing processes are central in achieving better performance and lower manufacturing costs of microelectronic devices. To this end, ab initio methods can contribute unique information about materials properties and processes where experimental techniques are too blunt. This includes band gap engineering for optoelectronic materials, the thermochemistry and kinetics for vapor deposition processes, and the prediction of thermomechanical properties in the context of device reliability. Ab initio methods have proven to be especially useful in predicting the geometric arrangement of atoms of bulk systems, surfaces, and interfaces and the corresponding electronic, optical, and magnetic properties. This capability is particularly useful in the study of defects. The accuracy of current ab initio methods in predicting structural properties is fairly satisfying whereas the calculation of energy-related properties with an accuracy beyond 0.1 eV remains a tremendous challenge. Density functional methods are the main theoretical approach for ab initio calculations on solids, surfaces, and interfaces. These methods are evolving, but face severe difficulties as one strives for systematic improvements in accuracy. Quantum Monte Carlo methods offer a promising alternative, but are still early in their development. Thus, a combination of different ab initio methods in conjunction with reliable experimental databases remains the most appropriate practical approach at present.