Artificial Neural Networks (ANN) are reasonably well served today by von Neumann CP architectures and GPU variants, especially when assisted by co-processors optimized for streaming matrix-vector math. We are seeing steady progress in chips like the Google TPUS, Nvidia GPUS, Baidu XPUs and many application-specific hardware that continue to push performance and energy efficiency of ANNs. Still, they fall far short, by many orders of magnitude, from the computational power and energy efficiency of the human brain. Value of ANNs was not completely appreciated till the advent of fast CPUs, GPUs and loads of fast, high bandwidth memory like DRAM. Similarly, we believe the same is the case is for Spike based Neural Networks, that may provide the needed boost. The Brain is a massively parallel architecture supported by extremely high 3D connectivity, co-located logic, and memory, suitable for life-long learning and decision making in a dynamic environment, and relies on stochastic decision processes and shows resilience to instantaneous errors. In this workshop, we will discuss recent progress in smart interconnect, spin and ferroelectric based spiking neurons and stochastic synapses being pursued in NEW LIMITs, CAPSL, ASCENT, and C-BRIC centers.