It's a fair question! There are a lot of perks to splitting up your change into multiple PRs rather than one feature branch. You can parallelize getting reviewed while writing more code. If CI fails, it fails on a more narrowly scoped change and is easier to debug. If something about one of your changes is unmergable, you may still be able to merge in half of your stack, rather than your change being all-or-nothing.

A lot of the benefits remind me of other dag-based workflows, like CI, data pipelines, or build systems. Does that make sense?