How closely do you want to approximate it? Zero is a pretty good approximation, only out by at most 10^20.
Otherwise, the sum is going to be given by a fifth-power expression, and will be divisible by
61n(n+1)(2n+1) being the corresponding sum of squares; hence it is a quadratic times that thing. (It's a fact, I seem to remember, that the sum of nth powers is a multiple of the sum of n-2th powers.)
That thing is 338350, so we're looking at roughly
1002×338350=3,383,500,000. I'll knock some sig figs off that because there's no reason at all for the quadratic to have leading coefficient 1; let's go with
1×1010 as the possibly-correct order of magnitude.
Alternatively:
∑i=1ni4=nE(X4) where
X follows the uniform distribution on
{1,2,…,n}. Also
∑i=2n+1i4=nE((X+1)4), so
(n+1)4−1=nE((X+1)4−X4)=nE[4X3+6X2+4X+1].
That is,
n3+4n2+6n+4=4E(X3)+6E(X2)+4E(X)+1.
E(X)=2n+1;
E(X2)=21(n+1)(2n+1) by the usual formula for the sum of squares; hence
E(X3) can be calculated by simply rearranging.