Satellite-based column-averaged dry air CO₂ mole fraction (XCO₂) retrievals are frequently used to improve the estimates of terrestrial net carbon exchanges (NEE). The Orbiting Carbon Observatory 3 (OCO-3) satellite, launched in May 2019, was designed to address important questions about the distribution of carbon fluxes on Earth, but its role in estimating global terrestrial NEE remains unclear. Here, using the Global Carbon Assimilation System, version 2, we investigate the impact of OCO-3 XCO₂ on the estimation of global NEE by assimilating the OCO-3 XCO₂ retrievals alone and in combination with the OCO-2 XCO₂ retrievals. The results show that when only the OCO-3 XCO₂ is assimilated (Exp_OCO3), the estimated global land sink is significantly lower than that from the OCO-2 experiment (Exp_OCO2). The estimate from the joint assimilation of OCO-3 and OCO-2 (Exp_OCO3&2) is comparable on a global scale to that of Exp_OCO2. However, there are significant regional differences. Compared to the observed global annual CO₂ growth rate, Exp_OCO3 has the largest bias, and Exp_OCO3&2 shows the best performance. Furthermore, validation with independent CO₂ observations shows that the biases of the Exp_OCO3 are significantly larger than those of Exp_OCO2 and Exp_OCO3&2 at mid and high latitudes, probably due to the fact that OCO-3 only has observations from 52° S to 52° N. Our study indicates that assimilating OCO-3 XCO₂ retrievals alone leads to an underestimation of land sinks at high latitudes, and that a joint assimilation of OCO-2 and OCO-3 XCO₂ retrievals is required for a better estimation of global terrestrial NEE.