Yet another way -- just a plain hash:

use strict;
my %data;

open(I1,"file1") or die "file1: $!"
$_ = <I1>;   # read column headings

while (<I1>) {  # get data
    chomp;
    my ($upc,$qty) = split /,/;
    $data{$upc} = $qty;
}

open(I2,"file2") or die "file2: $!"
$_ = <I2>;   # read column headings

while (<I2>) {  # get data
    chomp;
    my ($sku,$siz,$clr,$upc) = split /,/;
    $data{$upc} .= join ',', '', $sku, $siz, $clr;

# or: $data{$upc} = join ',', $data{$upc},$sku,$siz,$clr;
}

for (sort keys %data) {
    print "$_,$data{$_}\n";
}
[download]

That's pretty brittle in a number of ways: it won't handle properly quoted fields in a CSV file (e.g. '121,MD,"Green, Kelly",87654321'); it depends rigidly on a particular ordering and quantity of columns in each file; if file2 has a upc value not found in file1, it'll print ",$sku,$siz,$clr"; if file1 has a upc value not found in file2, it'll print "$upc,$qty".

But it's simple and should do the job as you described it.