Why are solutions to the Laplace equation with Neumann boundary conditions not unique (FEM setting)?

The bilinear form is the same as for the Dirichlet problem, which has a unique solution. Why doesn't the Lax-Milgram theorem apply?