Electrochemical interfaces play a pivotal role in numerous applications, including energy storage devices such as batteries, supercapacitors, fuel cells, and electrolyzers. Recent advances in computational electrochemistry, particularly in the era of artificial intelligence (AI), have significantly enhanced our understanding of these interfaces at atomic and molecular levels, paving the way for optimized materials and improved device performance^[1].

In the field of modeling electrochemical interfaces, there are two large sub-communities. The first involves the use of classical molecular simulations, particularly classical molecular dynamics (MD), to investigate the formation and structure of the double layer at electrode-electrolyte interfaces, primarily in the context of supercapacitors^[2]. Customized methods such as constant potential^[3] and its derivatives have been developed to accurately describe polarizable electrodes, enabling simulation of complex electrodes such as porous carbon, and layered materials under potential control. For electrolytes, advancements like polarizable force fields^[4] and reactive force fields (ReaxFF)^[5] allow detailed modeling of complex electrolytic environments. Key challenges in this domain include accurately modeling pseudocapacitive interfaces and flexible electrodes^[6].

The second sub-community focuses more on metal-electrolyte interfaces, which serve as prototypical systems in electrochemistry and play crucial roles in electrocatalysis and energy storage applications^[7–9]. Quantum mechanical simulations, particularly those based on density functional theory (DFT), are widely used to study electronic structures, charge distributions, and interfacial interactions at electrode surfaces. Ab initio molecular dynamics (AIMD) further enables the study of dynamic interfacial processes, including the adsorption of reactive intermediates, electron and proton transfer events^[10]. A critical aspect of such simulations is the treatment of solvation. While implicit solvation models are still used in static DFT for efficiency^[11–13], explicit solvation—now standard in AIMD—offers a more realistic representation of solvent structure and electrolyte effects at interfaces^[14,15]. Despite significant progress, the high computational cost and limited system size remain major constraints.

The rapid development of artificial intelligence has introduced new paradigms and opportunities into the modeling of electrochemical interfaces, opening novel avenues for overcoming long-standing challenges in accuracy, scalability, and complexity. In this context, the emergence of machine-learning force fields (MLFFs) offers transformative opportunities across both communities^[16]. For classical molecular simulation practitioners, MLFFs introduce a universal modeling paradigm capable of capturing complex, chemically diverse interactions in emerging electrode and electrolyte materials with near ab initio accuracy. For electrochemical systems, proper treatment of long-range interaction is essential for the application of MLFFs^[17–20]. Properly trained MLFFs have the potential to describe both non-Faradaic and Faradaic processes^[21]. For the quantum-focused sub-community, MLFFs can accelerate AIMD simulations by orders of magnitude while preserving first-principles accuracy, thus enabling the exploration of larger system sizes, longer timescales, and more realistic electrochemical conditions^[22].

Other opportunities in the AI era include scaling microscopic simulations to macroscopic device-level analyses and fostering synergy between high-throughput computational simulations and automated high-throughput experimental techniques, potentially facilitated by AI-driven agents^[23]. Also, emerging approaches such as End-to-End differentiable MD show great promise in optimizing classical force field parameters based on experimental or ab-initio data.