The Arctic Ocean is undergoing rapid sea-ice loss, ocean warming, and carbonate-system restructuring, yet the future evolution of its CO2 sink remains highly uncertain. Here we use a pan-Arctic regional ocean-sea ice-biogeochemical model based on NEMO-LIM-PISCES, forced by CMIP6 SSP2-4.5 projections, to simulate Arctic Ocean carbon-cycle changes from 2015 to 2060 at 1/4° resolution. A hindcast evaluation for 2015-2025 shows that the model reasonably reproduces hydrographic structure, sea-ice conditions, and key biogeochemical variables including DIC, surface pCO2, chlorophyll, and nitrate, providing a robust basis for future CO2-flux diagnosis.

Our results indicate that the Arctic Ocean remains a net atmospheric CO2 sink, but its future uptake does not evolve as a simple monotonic increase. Instead, the sink undergoes a marked seasonal reorganization. Comparison between 2016-2025 and 2050-2060 reveals enhanced CO2 uptake during the cold season and late autumn, but weakened uptake during the warm season, particularly in late summer. Spatial diagnostics show that this seasonal redistribution is primarily contributed by Basin-like regions, whereas Opposite regions exhibit distinct or even contrasting behavior. A flux-budget decomposition further demonstrates that the warm-season weakening is mainly caused by a reduction in the air-sea ΔpCO2 gradient, driven by a rapid increase in ocean surface pCO2. This increase is dominated by nonthermal effects and is associated with rising DIC, declining alkalinity, and changes in nutrients and biological production. By contrast, the cold-season enhancement reflects the combined influence of sea-ice retreat, expanded gas-exchange windows, and altered transfer conditions.

These findings suggest that the central feature of the future Arctic CO2 sink is not simply strengthening or weakening, but a seasonal redistribution regulated jointly by sea ice, gas exchange, and carbonate-system changes. The study provides regional-scale evidence for mechanisms that may be smoothed in global climate models and improves understanding of Arctic Ocean carbon-sink stability under a mid-range emissions pathway.