Exactly, cognitive load is dynamic not static, and you can actually hold many more things in working memory than the oft-repeated 3-7 items (that's more if you're trying memorize and recall unrelated, novel items).

Once you commit a particular concept to long-term memory and it's not "leaky" (you have to think through the internal behavior/implementation details), then now you have more tools and ways to describe a collective bunch of lower-level concepts.

That's the same feeling programmers used to more powerful languages have to write less powerful languages — instead of using 1 concept to describe what you want, now you have to use multiple things. It's only easier if you've not grokked the concept.