Also if you use Dagger with Kotlin, modularization will yield 10x performance improvements than incremental annotation processing (through granular local caching, higher hits on remote caching, compilation avoidance, stronger encapsulation, and parallelization)