Predictions have been revised to contemplate an ice-free Arctic Ocean in September by 2037 or 2040, or even earlier. The future development of the seasonal ice zone (SIZ) and concomitant ecosystem changes are by far the greatest challenge for the future Arctic Ocean. The greatest relative changes in gross primary production (GPP) are found in the low productive SIZ, i.e. at the rim of the deep Arctic Ocean. Here our knowledge about plankton is insignificant, let alone the regulation of processes such as primary production. The SINMOD C flux model predicts that GPP increases with increased air temperature (0–6°C; decreased ice cover) from 5 to 35 g C m^–2 a^–1. However, the existing algorithms are not sufficient for predicting the future. As the upper Arctic Ocean receives more heat and fresh water, nutrients for phytoplankton growth ceases, increased light will increase GPP, but not the harvestable production. Bacterial production may increase by twofold and respiration by eightfold. Models of GPP in the Arctic Ocean must be improved to include realistic temperature-dependent respiration, metabolism and a negative ice-melt effect. Current GPP projections are thus too high. The role of Arctic plankton communities as CO₂ sinks weakens and reverts to become CO₂ sources to the atmosphere along the range of warming predicted for the 21st century.