We study a structural model of R&D alliance networks where firms are engaged in R&D collaborations that lower their production costs while competing on the product market. We provide a complete Nash equilibrium characterization, analyze welfare and determine the optimal R&D subsidy program that maximizes welfare. We then structurally estimate our model using a unique panel dataset of R&D collaborations and annual company reports. We use our estimates to study the impact of targeted vs. non-discriminatory R&D subsidy policies, and empirically rank firms according to the welfaremaximizing subsidies that they should receive by the planner.