This talk will provide an overview of recent advances in GPU-accelerated molecular simulations and the computing architectures that enable them. With the end of Moore’s law, performance improvements now depend on algorithmic innovations and specialized hardware. I will first discuss modern GPU architectures, supercomputers, and programming approaches that leverage massive parallelism for scientific computing. I will then highlight developments in Amber, a leading biomolecular simulation package used for computational drug discovery. Its molecular dynamics engine features a highly optimized CUDA implementation for Nvidia GPUs, recently extended to AMD (HIP/ROCm) and Intel (SYCL) devices, enabling simulations of systems with millions of atoms. Finally, I will present QUICK, a massively parallel quantum chemistry code for Hartree–Fock and DFT calculations with Gaussian basis functions on GPUs. QUICK performs all major computational steps—including Fock matrix construction and gradient evaluation—entirely on single or multiple GPUs, and couples efficiently with Amber for QM/MM molecular dynamics. Together, these efforts illustrate how advances in software and GPU technology are transforming molecular simulations, enabling faster and more accurate modeling of complex chemical and biological systems.