This is interesting in that it so surprising to me. However, the major win I see for microservices vs monoliths is not addressed in your article.
The major problem with monoliths is how hard they can be to replace / refactor down the road. Any changes to a monolith system will be a long, complex, risky effort - likely resulting in big misses in timelines and budgets.
Refactoring / changing microservices, on the other hand, is by definition a smaller undertaking with a much better defined interface that can be maintained with less side effects.
The thought of starting monolithic and moving to microservices is interesting and you make some good points. I would say make sure you start that transition as soon as it makes sense and don't wait until you have years of development into the monolith that eventually needs to be broken out.
[Disclosure: My experience with microservices has been 100% positive but I'm with a large company - to your point.]