In this position paper, we argue that multimodal feedback literacy for L2 writing teachers in this GenAI-ubiquitous world requires them to engage in strategic teaming with GenAI to provide feedback on both language use and multimodal synergy, and to orchestrate multiple feedback-giving modes into coherent feedback ensembles to facilitate optimal feedback uptake among diverse learners. Grounded in the social semiotic tradition of multimodality, we challenge the conventional understanding of feedback communication as language-centered and advocate an alternative conceptualization of feedback communication as multimodal. We argue against the narrow conceptualization of GenAI as a tool for generating linguistic feedback and contend that feedback automation through GenAI would dehumanize the learning potential of feedback and undermine the agency of L2 writing teachers as feedback designers. We then, in addition to the GenAI-led automation and human-led assistant models of using GenAI, present three pathways for L2 writing teachers to team up with GenAI for multimodal feedback design and multimodal feedback literacy development.