In this paper, we combine recent progress in neuro-morphic computation and neuromorphic mixed-signal hardwareto present the first step towards an implementation of a neu-romorphic spike sorting algorithm, that has been proven ableto extract and decode spikes, in real time. This implementationis based on TSMC 180nm technology. Combined with a neu-ral recording system, we anticipate this approach to leverageefficient neuromorphic brain-machine interfaces for embeddedrehabilitation prosthetic control