PurposeThe purpose of the paper is to advance a framework that can analyze the impact of the egocentric alliance network on firm ambidextrous innovation holistically. On this purpose, the framework involves and integrates structural holes (SH), alliance functional diversity (AFD) and alliance partner geographical diversity (APGD) that measure network characteristics from structural, relational and nodal perspectives, respectively.Design/methodology/approachThe authors collected multi-source matching panel data including alliance deals, technical patents and financial information during 2000–2017 of the 106 top high-tech firms in the computer, communications, electronics and biopharmaceutical industries and conducted a three-way interaction model to uncover the complex mechanisms from a contingency perspective.FindingsEmpirical results show that SH as structural capital is positive to both exploitative and exploratory innovation. Both AFD as relational capital and APGD as cognitive capital positively moderate the SH-exploitative innovation nexus rather than the SH-exploratory innovation nexus. APGD and AFD co-moderate the relationship between SH and firm ambidextrous innovation in the way that when APGD and AFD are both high, SH has the strongest positive impacts on firm exploitative and exploratory innovation.Originality/valueThis research provides new insights into the benefit-cost paradox of the structural brokerage position (SH) by uncovering relational (AFD) and nodal (APGD) network characteristics that impact the trade-off. Also, it contributes to social capital theory by differentiating three dimensions of network capital (structural, relational and cognitive capital) and integrating them into ambidextrous innovation research. Finally, the findings give firms enlightenment to configure their egocentric alliance network for innovation ambidexterity.