In basic kinetic theory we often use cubes. For example, the derivation of the ideal gas law is done with a cubical volume. Is a cube necessary for it to work? If not, wouldn't a sphere be more symmetric and thus simpler?
Using a cube is an extension of the choice of coordinates used in the mathematical modeling. $dV=dxdydz$ is a small cube after all. $dV$ is more complicated in radial coordinates