There has been an explosion of innovation backed by venture capital since late 1970's.Nonetheless, a great deal of innovation still occurs within large companies. In this paper, I investigate the factors that determine when innovation is performed by venture-backed firms and when by large companies. To this end, I develop a theoretical model in which development of new technologies and products requires the collaboration of researchers, executives, and suppliers of capital. I focus on the two-tier agency problem designed to provide simultaneously the right kinds of incentives for researchers and executives. I find that if capital markets function perfectly, it is optimal for innovation to be conducted by venture-backed firms: Specialization implicit in venture form of organization mitigates two-tier agency problems. If capital markets are sufficiently imperfect, however, it is optimal for innovation to be performed by large companies: they can use cheaper internal funds to finance innovation. I finally point to the role of policy in improving capital markets and hence innovation performance.JEL Classification: D2, D82, D86, G24, O32