I'm thinking of it in terms of "sampling error" so to speak. A single 25k waveform exists inside of one degree of rotation. You're right, in steady state, it's just a slight pitch shift.
Like, obviously we're talking about, for our purposes infinitesimally small differences, but it does remain...