For states, the no-cloning theorem is one of the first encounters in quantum information: an arbitrary unknown pure state cannot be perfectly copied onto a waiting register. It’s simple then to infer that cloning N states onto M registers won’t work either, but by relaxing the problem to allow for imperfect (approximate) copying, one can investigate the rate and quality of N-to-M cloning. But what if we go further? Can we imagine cloning the cloner, in other words, cloning devices themselves? Questions around states have constituted decades of investigation and were, most recently, extended to cloning unitary gates. In this work we pose the question comprehensively and study general quantum channel cloning via a framework based on higher-order quantum operations. We establish necessary conditions for a family of channels to exhibit super-replication—those that can be cloned with a quadratic rate at high-quality (bounded error). Noisy phase-gate channels can be super-replicated but the full set of noisy unitaries; classical noise; and amplitude damping channels cannot. We present algorithms that search for channel cloners and study specific cloning strategies. In many cases, estimating and re-preparing is enough: for the unitary gates for which super-replication protocols were known, we construct simpler alternatives and establish a direct connection between channel cloning and Bayesian estimation.