We extend the Ax-Katz theorem for a single polynomial from finite fields to the rings Z_m with m composite. This extension not only yields the analogous result, but gives significantly higher divisibility bounds. We conjecture what computer runs suggest is the optimal result for any m, and prove a special case of it. The special case is for m = 2^r and polynomials of degree 2. Our results also yield further properties of the solution spaces. Polynomials modulo composites are the focus of some computational complexity lower bound frontiers, while those modulo 2^r arise in the simulation of quantum circuits. We give some prospective applications of this research.