The global innovation system (GIS) approach has until now focused on defining industries by a single GIS configuration at a given time. However, by analysing 739 innovation projects in the Norwegian global salmon farming industry, we find that the four ideal GIS configurations can be identified simultaneously. We argue that this is because we refer to an industry with a large diversity of innovation projects generating different resources along varied scales of interaction. Our analysis is based on text analysis of descriptive documents and interviews about innovation projects. With this information, we built four GIS configurations that illustrate how actors, networks and institutions form and are connected through different types of subsystems. The purpose is to understand why some actors have diverse priorities when participating in different types of innovation projects within the same industry and suggest relevant policy ideas.