With well formatted input, this sort of task is fairly simple. However, as toolic points out, your lack of formatting makes this unnecessarily difficult. If I assume your files are:

file1.txt:

SNDK 80004C101 AT
XLNX 983919101 BB
NETL 64118B100 BS
AMD 007903107 CC
KLAC 482480100 DC
TER 880770102 KATS
ATHR 04743P108 KATS
RBCN 78112T107 JT
TXN 882508104 KATS
STM 861012102 KATS
[download]

file2.txt:

AT AU
AU AU
AV AT
BB BE
BS BR
BSE HU
BZ BR
CC CL
CD CZ
CG CN
[download]

and I assume the mappings of 1st to 2nd column in file 2 are many to one(so I can use a hash), then you'll likely want something like this:

#!/usr/bin/perl
use strict;
use warnings;

my %mapping;
open my $key_handle, '<', 'file2.txt' or die "Open fail: $!";
while (<$key_handle>) {
    my ($key, $value) = split;
    $mapping{$key} = $value;
}

open my $data_handle, '<', 'file1.txt' or die "Open fail: $!";
open my $out_handle, '>', 'file3.txt' or die "Open fail: $!";

while (<$data_handle>) {
    my @columns = split;
    print $out_handle "$columns[0]\t$columns[1]\t$mapping{$columns[2]}
+\n"
}
[download]

However, my guess at your file 1 contains the string 'KATS' multiple times in what I guess was column 3 and never appears in your file 2. Given the distribution of 'KATS' in your specified file 1, this mapping is impossible unless file 1 contains only one line.