Why...

does white castle kingside when against the indian defense? It seems like suicide.

1.d4 2. Nf6

Well, the positon after d4 Nf6 is still very undefined. Not in all variations B tryes a kingside attack. W more often than not castles kingside because after moving c4 and d4 his queenside is a bit open.

Probably you meant why W castles kingside in the King's indian defense (2.c4 g6 3. Nc3 Bg7 e4 and so on) where, indeed, B often goes for the king.

I'm not an expert,but generally speaking W (often, not always) castle kingside even here because he wants to push his queenside pawns and create weakness in B queenside. His reasoning is that B treaths are bigger (a mating attack!) but also slower and he hopes to inflict such damages in black queenside to force B on the defensive. Both sides plans are quite suicidal indeed ("mate or die" VS "i don't care about my king")  and this makes most variaton(there are also quiet lines) of the KID very sharp!

There are exceptions, for example in the saemish W can consider O-O-O and a pawnstorm on B king. This has been played in countless GM games. But yes, it's more normal to play O-O and use the queenside space advantage.

Some situations are comical in that after castling queenside white can't exactly pursue play on the queenside without weakening his king while black can continue seeking his natural kingside play.  Like it was said, these situations are pretty sharp.

If white is piece heavy on the queenside, it is possible to develop a queenside initiative after opening lines... even if the white king is castled queenside.

Such positions are much too sharp for my taste!