UTF-8 vs UCS-2 (especially on ndb)

[Mirko Raner]

I'm in the process of migrating a InnoDB database to ndb, and we're frequently running into the 14000-byte row size limit imposed by ndb. There are quite a number of VARCHAR columns in our DB, and we're using the utf8 character set.
My question is: what (if any) is the benefit of using UTF-8 over UCS-2?
Since VARCHAR needs to allocate memory for the worst-case scenario (i.e., the maximum length) VARCHARs in UTF-8 require 3 times the length, whereas UCS-2 only requires 2 times the length. UTF-8 is optimized for scenarios with mainly one- and two-byte characters, but if the storage mechanism has to assume three bytes anyway, UCS-2 seems to be the better choice.
Am I overlooking something here? It seems like using UTF-8 for VARCHARs is a waste of space (especially problematic for MySQL Cluster with the smaller row memory limit of 14000 bytes).
Any insights?