This dissertation studies inter-firm network formation and performance implication. Different from current network formation literature that focuses on the actor or dyad level, this dissertation examines network formation and its performance implication at the group level. Specifically, I examine: 1) How do inter-firm groups with different levels of density form? 2) When is a firm more likely to participate in a group with mostly unfamiliar firms? and 3) How do group internal and external network structures influence task performances? Using Venture Capital (VC) investments as the research context, I develop novel empirical designs to quantitatively test my theory. In Essay I, to investigate how groups with different levels of density form, I emphasize the path-dependence effect of previous ties among all potential group members and simultaneously examine the formation of all ties in a group. I find that both anticipated environmental adaptation and anticipated internal cooperation are important considerations in a group formation. Taking a firm-focused group perspective, Essay II studies when a firm participates in a group with mostly unfamiliar firms. The empirical results show that the group participation of an unfamiliar firm depends not only on the uncertainty it brings in value creation but also on the uncertainty in value appropriation. Essay III examines the impact of syndicate density and structural holes and finds that both have impacts on the startup company performance. This dissertation enhances our understanding of network formation by bringing in a brand-new perspective, by uncovering group-level antecedents of network formation, by illustrating the impact of concerns in value appropriation, by exploring group dynamics, and by linking network formation behaviors with task performance at the group level.