Carbon capture, utilization, and storage (CCUS) is a suite of technologies designed to separate CO 2 from industrial sources or the atmosphere, convert it into value-added products, or permanently sequester it in geological formations. Although essential for achieving global carbon neutrality, the gigatonne-scale deployment of CCUS is currently constrained by high energy penalties in capture, thermodynamic limitations in utilization, and rigorous monitoring requirements to ensure storage security. Artificial intelligence (AI), which encompasses machine learning (ML), deep learning (DL), and generative algorithms, offers a data-driven approach to addressing multiscale physicochemical challenges by identifying complex correlations in high-dimensional datasets. This review synthesizes AI applications across the CCUS value chain. In the capture domain, we discuss how generative models and high-throughput screening accelerate the discovery of high-performance sorbents and solvents, while surrogate models optimize process dynamics to improve energy efficiency. For utilization, we highlight AI’s role in navigating the vast chemical space of catalysts to overcome thermodynamic scaling relations and optimize biochemical pathways. In geological storage, we analyze how DL architectures, from computer vision for seismic interpretation to Fourier neural operators for rapid plume forecasting, automate site characterization and enhance real-time leakage detection. Furthermore, this review elucidates AI-facilitated system-level integration, enabling techno-economic optimization of capture-transport-storage networks under policy and market uncertainties. Finally, we address critical challenges regarding data standardization, model interpretability, and generalizability, and project a future in which large language models and digital twins drive autonomous, self-optimizing CCUS operations.
Yang et al. (Mon,) studied this question.