Quote - Umm, what ?

A cube has eight vertices.

If I subdivide it, it has 28, but only the original 8 can be manipulated in Poser, as the 20 new ones are just "virtual".

If I "freeze" the subdivision I get a high res cube with 28 "real" vertices.

Morphing the SubD cube limits me to moving those 8 vertices around, as the others will be re-generated constantly.

The simplest object contains a single vertex. How many morphs could you make for this single vertex? There are only 3 coordinates you could modify, x, y, and z. A single morph can modify these x,y, and z in some arbitrary fashion. One of the possible morphs, for example can be x=1, y=2.78, and z=0. Every different combination of x, y, and z gives a different morph. How many different combinations are there? Poser uses a precision of approximately 10 digits. Because there are 3 values, there are in total 30 digits which makes a 10^(310) = 10^30 possibilities for a single morph.
In Poser each number uses approximately 4 bytes, so one morph for a single vertex uses 12 bytes. To store all of the single vertex morphs you would need 12 * 10^30 bytes. Even if you used new 500GB DIMMs (weight ca. 20g), you would still need (12 * 1030) / (500 * 230) = 210^19 of them or in other words 400 billion gigatons of RAM, which exceeds the specification of most computers, especially 3 year old laptops, and that is only for a single vertex! This number will not get smaller if you used more vertices, so the number of possible morphs is mainly determined by the size of a single morph, not by the number of vertices. So in practice: the smaller the morph, the more of them you can have. Of course, this does not take into account the usefulness of morphs, only their number.

Quote - With the high res cube I can move all 28 vertices completely independently from each other.

So of course I can have much more morphs with the high res cube.