It's well-known that every natural number can be written as a sum of 4 squares of integers.

Has there been any recent progress about the similar problem for the cubes, 4-th powers and so on? I believe this was proven to be representable using some N — that depends on the power — and what's the story about it?