1. There's a hard limit to the rate of computation you can do within a single instance. Anything beyond that will require distribution, and you as an engineer have to be able to handle that.
2. That limit is very high nowadays, and distributed systems have very high overheads. Most systems that are distributed don't actually need to be, and are paying the overhead costs unnecessarily.
Comments
I think two things are true
1. There's a hard limit to the rate of computation you can do within a single instance. Anything beyond that will require distribution, and you as an engineer have to be able to handle that.
2. That limit is very high nowadays, and distributed systems have very high overheads. Most systems that are distributed don't actually need to be, and are paying the overhead costs unnecessarily.