We develop ACE (AI Climate Emulator) tools to emulate Earth System Models (ESMs) and apply them to the study of future scenarios and climate impacts. ACE allows us to accelerate model experiments, explore model spread, and deepen our understanding of key processes across the climate system. Calibration across models. ACE can be trained on outputs from multiple ESMs to emulate their distinct behaviors. This enables comparative analyses across models under various scenarios, helping to quantify model uncertainty and spread in climate projections. Machine learning for scenario generation. By training machine learning algorithms on large ensembles of climate model simulations, ACE can generate new, physically consistent scenarios with realistic climate variability and weather extremes. This facilitates high-resolution exploration of rare and compound events under future climate conditions. Geoengineering scenario analysis. ACE can be calibrated on climate model simulations that include both anthropogenic forcing and geoengineering interventions. This allows rapid assessment of a wide range of geoengineering strategies—such as stratospheric aerosol injection or solar radiation management—and their potential to limit global temperature rise to 1.5°C, while also evaluating their regional and systemic impacts. Research objective. The overarching goal of developing ACE is to advance our understanding of key processes along the terrestrial–freshwater–marine continuum that influence surface-atmosphere greenhouse gas (GHG) exchange. We focus on capturing the role of high-latitude systems in biogeochemical cycling and feedbacks to the climate system. ACE will enhance process-level understanding and representation in ESMs by emulating detailed dynamics of GHG fluxes in terrestrial ecosystems, freshwater bodies, marine shelves, and ocean basins. Special emphasis is placed on understanding human-induced pressures and their interactions with natural systems. Through joint analysis of observed and simulated GHG fluxes and their drivers, ACE leverages data from major infrastructures and measurement networks, including GIOS, GEM, ICOS, ACTRIS, PEEX, and SMEAR, all of which are crucial for characterizing high-latitude biomes. These data–model synergies will improve emulator accuracy and reduce uncertainty in future climate change projections, thereby informing robust mitigation and adaptation strategies.