million of lines still probably fit in memory.. Note that $#{$aoa[0]} assumes all lines are of the same length as you said.

use strict;
use warnings;

my @aoa;
while (<DATA>) {
    chomp;
    push @aoa,[split '',$_];
    
}

foreach my $col(0..$#{$aoa[0]}){
  print "Column $col: ",
        (join ' ',map { $aoa[$_][$col] } 0..$#aoa),
        "\n";

}

__DATA__
ACATCACCTC
ACATCACCTC
ACATCACCTC
ACATCACCTC

# out 

Column 0: A A A A
Column 1: C C C C
Column 2: A A A A
Column 3: T T T T
Column 4: C C C C
Column 5: A A A A
Column 6: C C C C
Column 7: C C C C
Column 8: T T T T
Column 9: C C C C
[download]

L*

UPDATE if really care memory you can try the following (*untested*)approach:

# pseudocode!!

# analize first line
my $line = <$fh>;
chomp $line;
# compute last index of the future array (or future string? be aware o
+f possible off one errors!!);
my last = length $line - 1;
# rewind the filehandle
seek $fh,0,0;

sub get_column{
  my $col = shift;
  my $line = shift;
  if($col==0){$line=~/^(.)/}
  elsif($col==$last){$line=~/(.)$/}
  else{ $line=~/.{$col-1}(.)/} # or $last - $col? 
  return $1;
} 
while (<$fh>){
   chomp;
   print get_column(3,$_)

}
[download]