PROTACs are bifunctional therapeutic agents that achieve selective protein degradation by recruiting a target protein to an E3 ligase, with the linker connecting the two ligands serving as a critical determinant of ternary complex formation, cellular activity, and degradation efficacy. Current linker design is largely empirical, relying on iterative synthesis of chemically familiar linkers, which constrains the exploration of broader chemical space and slows the discovery of PROTACs with optimal geometry and degradation profiles. To address this, we propose a generative chemistry framework for PROTAC linker design that leverages ternary complex structural information and degradation activity annotations to produce linkers compatible with both molecular geometry and pharmacological intent. Conditional generative models, such as diffusion models or variational autoencoders, can be trained on PROTAC structures paired with degradation metrics like DC50 and Dmax, and conditioned during generation on the protein of interest, E3 ligase, warhead structures, ternary complex geometry, and desired degradation profile. The resulting linkers are expected to be chemically valid, compatible with the two warheads, and consistent with stereochemical constraints derived from ternary complex models, preserving key binding interactions while enabling productive POI–E3 proximity. By moving beyond empirical variation, structure-informed and degradation-aware generative design could accelerate PROTAC development and enable systematic exploration of linker space for targeted protein degradation.