The reason is consistency.
I do not have official reference literature on this, however, I believe it is possible for us to come to a comfortable reasoning, regardless.
Let's walk through what we know from other user text-selection interactions in the same environment:
Any other selection you perform on a quantity of text places the cursor somewhere. The cursor is not ever placed ambiguously—it's always in a predictable location after every user interaction.
There is only ever one "caret" or "insertion point".
Thus, following the aforementioned established pattern, CTRL+A has to make a decision as to where the cursor shall be placed upon selection. Should the cursor be placed at the beginning of the selection, the end, or somewhere in the middle?
Let's just go ahead and agree that "somewhere in the middle" is not going to yield predictable results, and nix that option.
We're left with two options: the beginning or end.
Since the text being selected reads left to right, an assumption that most people select text with a cursor from left to right, and most active text editing happens at the end of a block of text, it seems like placing it at the end might be the best option.