A well-optimized monolith is going to be faster, but a less monolithic codebase makes certain bad decisions much more painful. If you know you have to make an R

A well-optimized monolith is going to be faster, but a less monolithic codebase makes certain bad decisions much more painful. If you know you have to make an RPC call of some sort to access a given resource, you're going to be more careful in how you access it; if you don't, you're going to have unusably slow and brittle software much faster than you would in a monolith. So you end up being forced to optimize earlier, and the process of optimizing a single small component is generally simpler (you can profile just that process, for instance). That said, there's nothing keeping you from gaining many of the same benefits from a well-disciplined, modular monolith, and I think there's a pretty high risk of a fast-moving and over-stressed team (like a lot of startups) being put under more stress by the time and complexity costs of microservices, and thus making much worse decisions than they would on a monolith.