Most succinctly put, standard deviation is about where the data are clustered in one sample data, while SEM is where the means would be clustered around many samples taken of a given set of things.