When I use pandas read_csv to read a column with a timezone aware datetime (and specify this column to be the index), pandas converts it to a timezone naive utc
You can even try :
date_parser = lambda x : pd.to_datetime(x.str[:-6])