While answering this question on Mathematica.SE, I started wondering about the properties of the sequence asked about.
Consider $a_1=12$, and perform
$a_1 = 12 \rightarrow (1^3 + 2^3) = 9 = a_2$.
Then again
$a_2 = 9 \rightarrow 9^3 = 729 = a_3$,
and
$a_3 = 729 \rightarrow (7^3 + 2^3 + 9^3) = 1080 = a_4$,
and so on. We arrive at the sequence $(12, 9, 729, 1080, 513, 153, 153, 153, 153, ...)$, with the number $153$ repeating on and on. Let's examine how the sequence looks like for different starting values $a_1$:
It turns out, that for $a_1\in\{1, 2, 3, \ldots, 99, 100\}$ the sequence either ends up with a repeating single number, or with two/three alternating values.
So let's check these multiplicities for $a_1\leq 10^4$:
Multiplicity $1$ occurs 8543 times, $2$ --- 501 times, and $3$ --- 956 times.
I haven't noticed any obvious pattern in the $a_1$s that lead to different multiplicities of the limitting sequence. Questions that arise:
- Are the numbers $1,2,3$ the only multiplicities possible for all $a_1\in\mathbb{N}$?
- What governs the value of the multiplicity in dependence on $a_1$?
- How universal this behaviour is when sums of squares, quartics, etc. are considered?
- Do other types of functions of the
IntegerDigitshave such interesting/universal properties? - What are other interesting questions to ask about?

