Dynamical systems have recently triggered various interests for their potential as efficient and powerful information processors. Among those recent breakthroughs, one finds Reservoir Computing, a brain inspired computational paradigm proposed in the early 2000 on the basis of neural network concepts. The approach is making use of the transient motion complexity developed in the high-dimensional phase space of a nonlinear dynamical system. The computed result is obtained from a linear read-out of that transient motion in the phase space, the motion being triggered by the proper injection of the input information which is thus driving the transient dynamics. Beyond this conceptual novel paradigm, hardware proofs of principle have even been successfully proposed recently, moreover with impressive computational capabilities. We will report on recent advances in such hardware implementation of Nonlinear Transient Computing, in which an ultra-fast electro-optic nonlinear delay oscillator have been used to demonstrate a physical unit capable of processing information with multi-10 GHz bandwidth capabilities.