I really hope that people are slowly starting to understand that using kafka and turning it in a single point of failure (yes, it fails) of your architecture is not a good idea.
This pattern has it's uses, but if you are using it everywhere, every time you have some sort of notifications because "it's easy" or whatever, you are likely doing it wrong and you will understand this at some point and it will not be pleasant.
I've yet to see the numbers you list live, but in addition to pure uptime, I've seen a lot of instances where it has been used as the default communication channel instead of proper RPCs, with devastating effects on the long run from the point of view of issues, debugging, predictability, etc...
It's being overused where it has no reason to be, good and simple on paper, awful in the way many are using it. Most of the times, you don't need it.
This pattern has it's uses, but if you are using it everywhere, every time you have some sort of notifications because "it's easy" or whatever, you are likely doing it wrong and you will understand this at some point and it will not be pleasant.