Consider the Hom-Set Functor $H : C^{op} \times C \rightarrow \mathbf{Set}$.
If we let $c_1, c_2$ be categories (and assume, as is tradition, that all hom-sets are actually sets), then we have that $H(c_1, c_2)$ is a hom-set in $\mathbf{Set}$.
Question: Since $h \in H(c_1, c_2)$ is a function from $Ob(c_1)$ to $Ob(c_2)$, doesn't that imply that $Ob(c_1)$ and $Ob(c_2)$ are themselves sets? But how can we assume that?