You can use the "only 10% of users use it" argument on so many features. That doesn't mean Linux shouldn't have it. That argument might fly with a commercial OS like Windows or Mac OS X, but some OS needs to provide a long tail of useful features for power users.
You also have to weigh the utility of features. It may be that 90% of Linux desktop users would rather have fancy composting and a "cube effect" when doing fast user switching[1], but shouldn't network transparency win because it actually does something useful?
Much as you can run X within Wayland there's absolutely no reason why it's shouldn't be possible to build networking above Wayland. Even better, there's room for competing implementations - one could implement a framework which ships some sort of serialized OpenGL commands (client-side hardware rendering), fully rasterized windows (relying on the server's video hardware), or abstract primitives (like X).
By providing a level of abstraction lower than that of X, Wayland provides a more flexible framework over which to implement whatever "long tail" of useful features the Linux community desires.
You also have to weigh the utility of features. It may be that 90% of Linux desktop users would rather have fancy composting and a "cube effect" when doing fast user switching[1], but shouldn't network transparency win because it actually does something useful?
[1] See "Is Wayland replacing the X Server?" on http://wayland.freedesktop.org/faq.html#heading_toc_j_4