We propose to leverage the recent advances in hybrid tensor-quantum computing in thecontext of QAOA. Since the interchangeability between QC and MPS models, we expect thatpreparing product states variationally should help analyze the structure and limitations ofQAOA more deeply. The method offers a systematic Tensor Network based view of QAOA,helping identify improvements via bond dimension and entanglement control or through deepercircuits.