When does making a microservice smaller actually cause more problems than it solves?

We've been pretty gung-ho about splitting services whenever we see a clear boundary, but I'm starting to wonder if there's a point where it stops helping, or even makes things worse. Like, if a tiny 'utility' service has just one job and only one other service ever calls it, are we just making more work for ourselves with no real upside? And what about services that always get deployed together because their changes are always linked? Are we just making deployments harder, adding more network steps, and piling on operational tasks without actually getting the separation we wanted? I'm really curious about real-world stories where a service was too small.