Neural simulations at a large scale often suffer from computational inefficiencies that limit their scalability and their ability to improve neural structures for specific applications. Oak Ridge National Laboratory’s SuperNeuro is a Python-based neural simulator that addresses these limitations and provides AI practitioners with a fast and scalable solution. It also allows users to simulate their own spiking mechanisms in a human-interpretable manner. SuperNeuro provides superior computing performance compared to existing simulation platforms by leveraging GPU computing. This simulator is versatile enough to handle different types of workloads, including neuroscience, deep learning (SNN), and general-purpose computing workloads such as graph algorithms, neuromorphic data structures, and μ-recursive functions. SuperNeuro easily integrates with other learning tools for SNN optimization and facilitates large-scale AI experimentation on accelerated computing infrastructures.