The rapid advancement of artificial intelligence (AI) has led to the development of cutting-edge hardware and software tailored to meet the demands of AI-related workloads. This global effort, driven by millions of users and an enormous number of researchers and engineers, has resulted in groundbreaking innovations in GPUs specifically designed for AI tasks and software ecosystems like JAX, which enable efficient utilization of the available computing power. Previous publications have shown that these tools can also be utilized for conventional high performance computing workloads: [XueEtAl23] provided a general FEM framework for fast GPU-accelerated forward solving and automatic differentiation (AD) based inverse design, [DuHe25] employed automatic differentiation to solve inverse problems using a Material Point Method (MPM) implementation while [GaoHarms25] showed how to implement reduced precision calculations to speed up transient wave simulations. However, an application to more complex problems like orthogonal cutting (as in [AfrasiabiEtAl20]), with concrete quantitative analysis of the impact of the said tools, has yet to be presented. We investigate the potential of applying the aforementioned developments to inverse problems in solid mechanics through the example of a JAX-based MPM Johnson-Cook plasticity simulation. By leveraging automatic differentiation, mixed-precision computations, and multi-GPU computing (all cornerstones of state-of-the-art AI computation), we provide a more efficient means for solving inverse problems and conducting sensitivity analyses in a classical computational mechanics context. We analyze the effect of employing the mentioned innovations, and describe what challenges arise within the considered scope. A material parameter identification problem in orthogonal cutting is presented to demonstrate applicability to real-world engineering problems.