Human experience and physical systems are fundamentally grounded in the 3D world. Consequently, 3D understanding, interaction, and generation are essential capabilities for advanced AI systems. However, available 3D data remains far scarcer than data for other modalities, such as text and images. The objective of this project is to investigate how combining different modalities can enhance diverse 3D tasks. For example, multimodal data fusion can increase the reliability of 3D perception systems; data-rich modalities like text or images can serve as control signals for 3D content generation and interaction; and existing foundation models can act as strong priors across various 3D tasks. Ultimately, we hope that the results of this research lead to advancements in diverse applications such as augmented reality, autonomous driving, remote sensing, and robotics.