With AI and social robotics advancements, social robots such as Cozmo will become increasingly prevalent in homes. Previous studies have indicated that the anthropomorphic appearance of robots can elicit both favorable and unfavorable responses from consumers. However, the factors that moderate such responses are not well-understood. Therefore, the present research proposes that self-construal, an untapped topic, is a moderator. Long-term, frequent, and private interactions between consumers and social robots at home may result in consumers developing relationships with robots based on the trade-off between benefits and (psychological) costs. Through examining social support benefits and relational intrusion costs, three experiments and one text analysis demonstrate that independent (interdependent) self-construals prefer humanoid (mechanoid) over mechanoid (humanoid) SHRs due to the anticipation of net interaction benefits. This research contributes to the knowledge of the contradictory findings regarding highly anthropomorphic robots, explicitly enhancing the understanding of human-robot interaction, the “uncanny valley,” robot anthropomorphism, and social robotics. Moreover, this research offers a practical design approach to SHRs from a managerial perspective.