To do a complex job like that job on a MacBook, you need expertise in a lot of different areas, including surface-mount components. The biggest cost-multiplier is that even if you do everything right, sometimes the new stuff will not work -- just because. So the cost of potentially doing it all again must be partially factored in.
I thought long and hard for a way to compare that MacBook complexity to something else. And for me, that comparable complexity is rebuilding or replacing the engine in a car with one that is SUPPOSED to fit. Anyone who has attempted these projects knows they can quickly transform from "follow the directions" projects to "there is no way that will work, so now you are completely and utterly on your own" in just the turn of a page in a not-quite-correct manual.
But most folks would not take that on as a project -- they would pay someone else to do it. If you did that, a Car Engine Rebuilding Service might charge US$2000 to $3000 to do a complete swap like that in a few days time -- once they had the replacement rebuilt engine ready-to-go.
In both cases, you need to put in WAY too much money or time and expertise to go from something that works, to something a bit better that works. No one in their right mind would endure the expense or time-and-complexity to replace a car engine that works with one that is bigger, or a MacBook processor or graphics chip with one that works, but better.
But yes, it can be done.