Mathematical potential is to a large degree innate but mathematical attainment requires tuition - even the best mathematicians needed teachers: Newton famously referred to standing on the shoulders of giants (though this may have been a dig at Robert Hooke who was on the short side). Euler had 1:1 tuition from Bernoulli. Even Ramanujan, though very isolated, had some good textbooks to learn from. All of us, no matter how brilliant, need to be taught what has gone before. But some need only the faintest sketch of a concept before they grasp it - they learn it so quickly that it seems like they're remembering something they knew before.
Another related one I don't really buy is "he taught himself to read". Really? He picked up a book with no pictures, stared at it hard for a long time, then miraculously it all fell into place and he was able to read it? Or did someone teach him the alphabet, perhaps put magnetic letters on the fridge, read stories with him every day, probably pointing to the words as they read, let him watch alphablocks or other children's programmes which are loaded with phonics knowledge, etc.? Genuinely "teaching oneself to read" is a feat on par with deciphering the Rosetta stone. Being immersed in a phonics-rich environment and filling in some gaps is rather different, though again they need only the faintest outline of a concept.
The point being that "He requires some degree of teaching, therefore the ability is not innate, therefore he is not gifted" is a false line of reasoning. If he loves learning, learns exceptionally quickly, and is doing KS2 maths in nursery, then the closest word we have for that in English is ... gifted. You might not like it or its connotations, but that is the word other people use, and avoiding or denying it cuts you off from the literature that would tell you how to deal with it.