Accurate estimation of regional-scale terrestrial carbon budgets is of great importance but remains challenging. With particular advantages, the Long Short-Term Memory (LSTM) networks method shows potential in improving regional carbon budget upscaling estimations. Here, based on LSTM, we upscale regional net ecosystem carbon exchange (NEE) with available flux tower measurements and satellite land surface observations in North America. With well-established ecosystem-specific LSTMs, we produced monthly NEE at a spatial resolution of 0.1° × 0.1° over 2001–2021 (labelled as MemoryFlux). Unlike existing upscaling estimates, our dataset properly identified the Midwest Corn Belt as a region of large seasonal carbon uptake during peak growing seasons, a feature revealed by previous top-down studies and recognized as a model benchmark. Moreover, the estimated seasonal variations of NEE by MemoryFlux coincided well with those by atmospheric inversions, i.e., the ensemble mean of Orbiting Carbon Observatory-2 Model Intercomparison Project (OCO-2 v10 MIP; <em>r</em> = 0.96, <em>p</em> < 0.001) and CarbonTracker2022 (CT2022) (<em>r </em>= 0.97, <em>p</em> < 0.001). The mean annual NEE was estimated at -1.27 ± 0.12 Pg C yr<sup>-1</sup>, aligning more closely with the inversions (-0.83 to -0.70 Pg C yr<sup>-1</sup>) than existing upscaling estimates (-3.30 to -1.68 Pg C yr<sup>-1</sup>) do. In addition, our estimate plausibly captured the NEE spatial anomalies caused by all the recent extreme drought and flood events. We further confirmed that considering memory effects was critical for better indicating interannual variability and spatial anomalies of NEE induced by climate extremes. MemoryFlux provides an improved bottom-up estimation of North American NEE, largely narrowing the gap with top-down inversions. This dataset can be downloaded at <a href="https://doi.org/10.5281/zenodo.20482274" target="_blank" rel="noopener">https://doi.org/10.5281/zenodo.20482274</a> (Huang and He, 2026).