ASICs require a tremendous amount of capital to go from design, tapeout, die fabrication, to hashboard production, etc. If there was a party that was interested in creating ASICs for Kaspa, it should be welcomed with open arms.
In regards to security — ask yourselves this: would a potential malicious actor attack the network with a farm of GPUs, or ASICs? Obviously GPUs — as GPUs can be repurposed to different tasks & algos after the attack. If you attack the network with ASICs, you are essentially bricking the utility of the devices & rendering them useless, as an ASIC is application specific to a single algo. It comes at a full capital loss to inflict an attack on a network in which those ASICs are only useful for.
There's an eco-friendliness argument: being (specfically optical) ASICs friendly means a larger part of the security budget comes from capital expenses. So with the same overall investment, you get the same security, but more of it goes to manufacturing rather than electricity bills.
Confirmation times assuming a limited strong attacker (< 49%) should be in the order of ~40 seconds for GHOSTDAG. The only argument for waiting 10-60 minutes (as eqv. of btc 1-6 confs) is the security budget of the attack. It is claimed that with a GPU minable coin it is cheaper to acquire 51% of the hashrate for a minute than it is to run 51% for an hour. This precisely is the argument for actual optical pow — since the security budget moves to the capital expenses, and we can be back to assuming that it is difficult to get 51% even for a minute.