Ah ! That's a nice challenge !!!
For a simple linear interpolation, the difficulty is to locate the best triangle surrounding the point where you want to interpolate the Z value. This is not a trivial problem, but some algorithms are quite efficient at that (one is used internally by the 3D graph).
If you have only a few X and Y (less than about 100, and homogeneously dispersed...), there is a nice vi, named "Tripolation.vi", written by Koji Ohashi. Not optimized, but enough for basic computations.
If you ever find a solution for large data sets, don't forget to tell me !
CC
Chilly Charly (aka CC)