Exploitative learning and exploratory learning are of great significance to the survival and long-term development of an organization. The former refers to the improvement of existing knowledge, skills and paradigms. The latter refers to the search for new knowledge, technology and experience. Therefore, the dual learning equilibrium (balance between exploitative learning and exploratory learning) is very important. However, for the engineering project organization, due to the pressure of cost and time cycle assessment, the engineering project organization is often more inclined to exploitative learning rather than exploratory learning. Therefore, how to realize the dual learning equilibrium (balance between exploitative learning and exploratory learning) of engineering project organization has become an urgent problem to be solved. Through formal control structures (outcome control vs. behaviour control) and guanxi (affect vs. face) can get the information needed by organizational dual learning (exploratory learning vs. exploitative learning), but whether the balance (formal control structures vs. guanxi) is still an effective strategy to promote dual learning equilibrium (balance between exploratory learning and exploitative learning) is still lack of theoretical and empirical support. It was examined assumptions related to how dual learning equilibrium is influenced by Guanxi and formal control structures equilibrium in project partnerships. Practically, this paper investigates the extent to which distinct components of Guanxi (i.e., affect and face) are differentially and interactively associated with outcome and behavioural controls on influencing ambidextrous learning equilibrium. Survey data were collected from 221 construction project partnerships in China. The results showed that both of the affect and behavioural controls have a positive impact on the ambidextrous learning equilibrium. Besides, the equilibrium between formal control structures and Guanxi can promote the ambidextrous learning equilibrium of the engineering project team. Compared with behavioural controls, the equilibrium between the affect and outcome controls has a stronger positive correlation with the ambidextrous learning equilibrium of the engineering project team. Compared with the outcome controls, the equilibrium of face and behavioural controls has a stronger positive correlation with the ambidextrous learning equilibrium of the engineering project team. However, the equilibrium between the affect and behavioural controls has no significant impact on the ambidextrous learning of the project team. On the whole, the two-pronged and balanced strategy is still effective for dual learning equilibrium, but it should avoid the balanced use of affect and behavioural controls. In addition, the affect plays a negative moderating role in the process of behavioural controls affecting ambidextrous learning equilibrium. Face plays a positive moderating role in the influence of formal control structures on ambidextrous learning equilibrium. These findings have important implications for firms hoping to facilitate the dual learning equilibrium in China and other countries.