We review the empirical literature about the implications of the computerization of the labor market to see whether it can explain observed computer adoption patterns and (long-term) changes in the wage structure. Evidence from empirical micro studies turns out to be inconsistent with macro studies that are based on CES production functions. We propose a micro foundation for the CES production function that allows for changes in the underlying structure. We adapt the macro model by incorporating computer skills, complementary skills and fixed costs for computer technology usage suggested by the micro literature. It turns out that fixed costs for computer technology usage explain different patterns of computer adoption and diffusion between several types of workers and countries; it also provides very plausible patterns of the timing of wage inequality and developments over time.