The Amazon rainforest has historically functioned as a major carbon sink, but recent studies suggest this capacity is weakening under the combined pressures of deforestation and climate change. In the Vegetation Photosynthesis and Respiration Model (VPRM), the optimal temperature for photosynthesis (T<sub>opt</sub>) controls the temperature dependence of carbon uptake. To investigate how T<sub>opt</sub> influences long-term net ecosystem exchange (NEE) over the Amazon under the SSP5-8.5 scenario, we conducted multi-decadal sensitivity simulations (2030<span>–</span>2100) using fixed literature-derived and temperature-dependent T<sub>opt</sub> formulations within VPRM. VPRM runs were conducted offline, driven by meteorological fields derived from Weather Research and Forecasting Greenhouse Gas (WRF-GHG) model simulations, and evaluated for April and September to contrast Amazon seasonal extremes. Under SSP5-8.5 warming, minimal land-use change leaves climate-driven thermal stress as the primary control on long-term NEE. During the dry-season peak (September), when thermal stress is most severe, the lower fixed-T<sub>opt</sub> formulation causes the historical carbon sink to progressively weaken and transition to a net source by 2090–2100. In contrast, the temperature-dependent T<sub>opt</sub> formulation mitigates high-temperature photosynthetic inhibition, preserving daytime uptake and delaying this sink-to-source transition. By 2100, T<sub>opt</sub> formulation uncertainty introduces a regional NEE spread of 0.054 PgC month<sup><span>−</span>1</sup> in April and 0.157 PgC month<sup><span>−</span>1</sup> in September—sufficient during the dry season to reverse the sign of the regional carbon budget. These findings demonstrate that model representations of photosynthetic thermal acclimation can strongly influence the projected Amazonian carbon sink resilience under future warming.