Key message Bi-axis training increased vegetative shoot light interception and modeled photosynthetic rate, and reduced fruiting shoot mutual shading and vegetative growth by optimizing canopy structure compared to single-axis in young apple trees. Improving light interception and distribution within canopy are constant objectives of training through manipulating tree architecture. A bi-axis training system with two primary scaffolds has been proposed to improve flowering and dry matter production in apple trees. In this study, 'Fuji' apple trees trained using a bi-axis or single-axis training system were compared. Twelve three-dimensional (3D) virtual apple trees were reconstructed by combining 3D digitizing and allometric relationships for three shoot types (vegetative long shoot, VL; fruiting shoot, FS; vegetative short shoot, VS), to evaluate canopy structure impacts on light interception. Light interception efficiency was evaluated by silhouette to total area ratio (STAR). The potential canopy photosynthetic rate was evaluated by the eco-physiological RATP model. The leaf area of VL in bi-axis trained trees was approximately 40% lower than that in single-axis trees. Lower leaf area and more uniform spatial distribution were noted in VL in bi-axis than in single-axis trees. This led to more even spatial light distribution and more shoots having higher STAR in bi-axis than in single-axis trees, regardless of the whole tree or shoot type. In the virtual orchard, bi-axis trees had a 25% and 10% STAR increase in VS and VL, respectively, but a similar STAR for FS compared to single-axis trees. Mutual shading between neighboring trees in the virtual orchard made of bi-axis trees was significantly lower than that with single-axis trees for FS. The modeled canopy net photosynthetic rate was 26% higher in bi-axis than in single-axis trees. With the increase in tree age, the projected leaf area increased, but the porosity of the canopy decreased, and mutual shading increased for the whole canopy and all shoot types, irrespective of training systems. These results indicated that manipulating scaffolds in apple trees can regulate canopy structure, light interception, and vegetative growth during the early developmental stages.