The unprecedented expansion in machine learning (ML) and artificial intelligence (AI) has put unwonted pressures on conventional computing architectures, demanding particularly brain-like computations for tasks. While conventional von Neumann-based computing architectures have demonstrated immense flexibility and computational power across a wide array of algorithmic workloads, their inefficiency and scalabilities have emerged as prime concerns in the presence of increasingly complex modern AI workloads, such as transformers and large language models (LLMs). In addition, the inherent bottleneck imposed by the sequential, clock-based architecture of von Neumann systems restricts their capacity to model highly parallel and event-driven biological processes common to natural neural systems.