<p>Agriculture is commonly portrayed as a major source of greenhouse gas emissions, yet it also represents one of the largest human-managed biological systems regulating carbon exchanges between the atmosphere and the biosphere. This study reassesses the role of global agriculture within the terrestrial carbon cycle by quantifying gross photosynthetic CO₂ uptake from crops, pastures, and managed forests and by evaluating alternative agricultural development pathways through 2050. Using FAOSTAT data for 156 crops integrated with FAO estimates for pastures and managed forests, we estimate that agricultural systems assimilated approximately 47.64 Gt CO₂ in 2023, including 21.87 Gt CO₂ from crops alone. This gross uptake exceeds current annual anthropogenic CO₂ emissions and approaches total anthropogenic greenhouse gas emissions expressed as CO₂ equivalents. However, most of the assimilated carbon is subsequently returned to the atmosphere through respiration, decomposition, livestock metabolism, biomass utilization, and food consumption. Gross uptake should therefore be interpreted as a measure of managed biogenic carbon cycling rather than permanent carbon sequestration. Historical analysis indicates that crop CO₂ assimilation increased from approximately 5.4 Gt CO₂ in 1961 to 21.9 Gt CO₂ in 2023, reflecting the combined effects of technological progress, yield improvements, agricultural intensification, and expansion of photosynthetically active biomass. Over the same period, agricultural productivity increased much faster than cropland area, reducing the land required to satisfy growing food demand and thereby limiting the conversion of natural ecosystems. To explore future trajectories, we developed the Emission Scenarios Simulation Dynamic Model (ESSDM), a scenario-based accounting framework that evaluates alternative production pathways under the constraint of meeting projected global food demand. Three scenarios were examined: Sustainable Intensification (SI), Moderate Expansion with Sustainable Intensification (MESI), and Organic Farming with substantial cropland expansion (OF). The simulations reveal substantial divergence among scenarios. By 2050, cumulative emissions are projected to reach 163.51 Gt CO₂e under SI, 241.35 Gt CO₂e under MESI, and 493.99 Gt CO₂e under OF. The markedly higher emissions associated with OF are primarily driven by lower average yields and the resulting expansion of cropland area (+ 52.45% relative to 2023), which generates substantial land-use change emissions through ecosystem conversion. Overall, the results indicate that the climate performance of agricultural systems depends not only on direct greenhouse gas emissions but also on productivity, land-use efficiency, and their influence on future land demand. The analysis highlights that protecting forests and grasslands from conversion remains a central climate objective and that sustainable intensification provides the most effective pathway for reconciling food security with climate mitigation under the assumptions considered. More broadly, the study suggests that agricultural assessments may benefit from integrating emission inventories with information on managed carbon fluxes and land-use dynamics when evaluating alternative development pathways.</p>