Quantum ESPRESSO is an open-source software package for quantum-mechanical materials modeling based on density-functional theory, pseudopotentials, and plane waves. It is known for its performance on a wide range of hardware architectures, from laptops to supercomputers. This paper discusses the ongoing effort to port Quantum ESPRESSO to heterogeneous architectures with hardware accelerators, which is essential for achieving exascale computing. Quantum ESPRESSO has evolved significantly since its inception in 2002, incorporating advanced theoretical methods and algorithmic improvements. It now supports a wide range of property calculations, including activation energies, superconducting transition temperatures, phonon linewidths, and nuclear magnetic resonance chemical shifts. The software is also used in conjunction with other codes for further processing, such as Quantum Monte Carlo and many-body perturbation theory. The paper discusses the challenges and opportunities of new heterogeneous architectures, which require significant changes to scientific software. The goal is to achieve performance portability, allowing high performance on different architectures with minimal hardware-specific code. This involves refactoring Quantum ESPRESSO into multiple layers, enabling independent maintenance and enhancement of components. The paper also describes the current status of Quantum ESPRESSO on NVIDIA GPUs, highlighting the need for further optimization and the potential for future improvements. The software has been ported to GPU architectures, with the latest version, QE-GPU, providing significant speedups for certain calculations. The paper presents benchmarks showing the performance of Quantum ESPRESSO on both CPU and GPU architectures, demonstrating the potential for exascale computing. The paper concludes with an outlook on future developments, emphasizing the importance of performance portability and the need for sustainable development models. The work described aims to ensure that Quantum ESPRESSO remains a powerful and flexible tool for quantum-mechanical materials modeling in the era of exascale computing.Quantum ESPRESSO is an open-source software package for quantum-mechanical materials modeling based on density-functional theory, pseudopotentials, and plane waves. It is known for its performance on a wide range of hardware architectures, from laptops to supercomputers. This paper discusses the ongoing effort to port Quantum ESPRESSO to heterogeneous architectures with hardware accelerators, which is essential for achieving exascale computing. Quantum ESPRESSO has evolved significantly since its inception in 2002, incorporating advanced theoretical methods and algorithmic improvements. It now supports a wide range of property calculations, including activation energies, superconducting transition temperatures, phonon linewidths, and nuclear magnetic resonance chemical shifts. The software is also used in conjunction with other codes for further processing, such as Quantum Monte Carlo and many-body perturbation theory. The paper discusses the challenges and opportunities of new heterogeneous architectures, which require significant changes to scientific software. The goal is to achieve performance portability, allowing high performance on different architectures with minimal hardware-specific code. This involves refactoring Quantum ESPRESSO into multiple layers, enabling independent maintenance and enhancement of components. The paper also describes the current status of Quantum ESPRESSO on NVIDIA GPUs, highlighting the need for further optimization and the potential for future improvements. The software has been ported to GPU architectures, with the latest version, QE-GPU, providing significant speedups for certain calculations. The paper presents benchmarks showing the performance of Quantum ESPRESSO on both CPU and GPU architectures, demonstrating the potential for exascale computing. The paper concludes with an outlook on future developments, emphasizing the importance of performance portability and the need for sustainable development models. The work described aims to ensure that Quantum ESPRESSO remains a powerful and flexible tool for quantum-mechanical materials modeling in the era of exascale computing.