The integration of Artificial Intelligence (AI) into Product Lifecycle Management (PLM) is increasingly recognized as a strategic enabler of product innovation, operational excellence, and sustainable manufacturing. While recent advances have demonstrated the potential of AI in supporting specific PLM tasks, most implementations remain functionally isolated, relying on data-centric models deployed within individual lifecycle phases, with limited support for semantic integration and knowledge reuse. This paper introduces a formal hybrid AI framework that combines statistical and symbolic AI to support lifecycle-wide product knowledge modeling and explainable decision support. The architecture comprises three interoperable layers: a statistical inference layer, a symbolic reasoning layer, and a Large Language Model (LLM)-based interface. Rather than reporting empirical results, this paper proposes a conceptual framework and articulates a research agenda that addresses several open challenges in hybrid AI for PLM. These include the semantic enrichment of statistical outputs within domain knowledge, the incremental integration and validation of emergent insights, and the design of human-in-the-loop interaction mechanisms tailored to industrial decision-making environments.