The argument I have always heard is that microservices are a way to solve problems arising in large organizations with multiple teams trying to push code all in to the same repository and managing the deployments from such an organization. I get why separating the different teams' services in to separate silos helps, but the fact that the solution puts a network between the different parts where there was none feels wrong. You are trading one problem for another problem. Networks are slower and less reliable than calling a function in the same application.

Microservices may be the solution to this problem right now, but I believe someone is going to come up with some other solution (tooling etc) that allows you to get the benefits of migrating to microservices without having to add a unnecessary network layer just to solve an organizational problem.