There have been vociferous calls for ‘tree-centered’ vegetation models to refine predictions of forest carbon (C) cycling. Unfortunately, our global survey at flux-tower sites indicates insufficient empirical data support for this much-needed model development. We urge for a new generation of studies across large environmental gradients that strategically pair long-term ecosystem monitoring with manipulative experiments on mature trees. For this, we outline a versatile experimental framework to build cross-scale data archives of C uptake and allocation to structural, non-structural, and respiratory sinks. Community-wide efforts and discussions are needed to implement this framework, especially in hitherto underrepresented tropical forests. Global coordination and realistic priorities for data collection will thereby be key to achieve and maintain adequate empirical support for tree-centered vegetation modeling.