Most large telescope arrays (that I know of) are for radio astronomy, largely for geographical reasons (as far as I know).

Earth is a source of noise throughout the electromagnetic spectrum, so I think the benefits of building on the far side (eliminating Earth noise), will likely outweigh the costs (limited/more costly bandwidth).

I think a mixed array, consisting of "radio" and "light" telescopes could present very interesting possibilities, especially because you could dynamically allocate sparse sets of the array to different tasks.

All of that being said, I am an engineer (with an interest in array signal processing), not an astronomer.