Spark PCA - size of covariance matrix

Question

So I have a dataframe which has 40.107 columns and 4.827 rows and I want to perform PCA. This means I will have 40.107 features and 4.827 rows (of double values because the