In the p dimensional space formed by p asset returns, PCA finds the most important k directions that capture the most important variations in the given returns of p assets. Usually, k is less than p. Therefore, by using PCA you can decompose the p asset returns into the k factors, which greatly redu