The currency is dead in the water if an ASIC comes out before more than half of the currency is mined IMO.  Litecoin would be an example of this.  It only has 30 of 80 million mined, and now the entire revenue stream of the network basically goes straight to the pockets of ASIC manufacturers.  There's no reason for the average human to want to become involved in the network under that situation, and it has hampered the uptake of BTC severely.  CPU coins, on the other hand, are born with this problem from day one due to botnets, and the fact that very few people own large CPU mining rigs to compete with them.  A long GPU mining period seems essential to establishing a currency.

As for the monopolization of XMR due to ASIC, such as what happened to Litecoin, XMR has a rather short mining period, a little too short in my opinion, so there's less risk of a situation like mentioned above occurring.  My idea of soft forking to have multiple PoW algorithms should the first one become ASIC monopolized too soon is probably completely unnecessary given how short XMR mining period is.  Dealing with continuous weighting upkeep in that situation is also a developer hassle and kind of defies the idea of decentralization having someone babysit the code.