This die is made by finding nine maximally-separated points on the surface of a sphere, and then at each point slicing off a circular facet with a fixed solid angle. All the facets are the same size (in area and solid angle), so the energy to roll off of a facet should be exactly the same for all of them. But, because they are not distributed completely regularly over the surface (since they aren't vertices of a regular polyhedron), this die is probably *not* perfectly fair, from a geometric point of view. I suspect that the bias will be fairly small (I don't know if it's small enough to be comparable to or smaller than the biases introduced by imperfect manufacturing). Should run some tests and at least get some empirical results.
Be sure to check out another d9, designed a little differently, by Magic.