This study investigates ChatGPT-supported teacher feedback in the Chinese tertiary EFL context and explores an innovative AI-aided writing pedagogy by integrating ChatGPT into teacher writing feedback provisions to alleviate the challenges of teacher feedback in a large class, which were reported in previous research. The participants of this study were four instructors and 102 students from two undergraduate classes in the world language education program. The students completed two writing tasks: an argumentative essay and an expository essay; then, the instructors provided detailed feedback on their essays based on the ChatGPT feedback. Two prompts were provided to ChatGPT after the training: 1) corrective feedback drawing on Ferris's (2006)15 types of common errors and 2) holistic rhetorical feedback. Afterwards, the teachers adapted the ChatGPT feedback and shared the detailed individualized writing feedback with each student. We closely examined the types and features of ChatGPT-supported teacher feedback and how EFL students incorporated this feedback into their writing revisions. The findings indicate that the ChatGPT-supported teacher feedback addressed diverse error categories and included helpful comments on the overall rhetoric. Moreover, the students incorporated more of the feedback into their revisions across tasks, which reflects their deeper engagement with the feedback content. This study notes the importance of an “AI + Teacher” model that leverages the analytical strengths of AI while maintaining essential teacher‒student interactions. This new approach of ChatGPT-supported teacher feedback has great potential in L2 writing feedback provision and will shed novel light on the writing pedagogy with the aid of AI in the digital era.