With the advances in semiconductor process technology and the trend of "system on a chip" or "system in package," silicon intellectual property (IP), which is a reusable IC design component, has become an important approach for 1C design. Since most IP providers have different IP technology and core competences, more and more IP providers start to collaborate with one another in order to obtain complementary resources or abilities to provide complete IC design solutions to customers. However, the academic understanding of this issue is still a puzzle. This research then intends to answer the research question: Why does the inter-firm collaboration, a special type of networks, mostly occur for the transaction of silicon intellectual property in the semiconductor industry? By exploring five typical cases, we find that the most common inter-firm collaboration models and difficulties of collaboration between the design service providers and other types of third-party IP providers with different business models. Transaction cost theory is then imported to interpret why the collaboration will mostly occur and how to solve the difficulties during the collaboration process.