Final remarks
The discussion of the relational view is born from questions regarding how firms achieve average returns. According to the relational view, this is only possible because of existing idiosyncratic connections between businesses that earn relational rents, i.e., a supernormal profit that is generated in an exchange and that cannot be earned by a single firm – instead it can only be created through alliances with specific partners. One of the hypotheses developed by Dyer and Singh (1998) is that the design of specific collaborative practices in knowledge creation would facilitate exchanges between alliance partners and the creation of new knowledge. However, as already shown in the theoretical framework, no solid contributions explain the dynamics of these collaborative practices that lead to supernormal profits. This lack of explanation of the origin of collaborative practices meets with notes by Lounsbury and Crumley (2007) and Labatut et al. (2014), who emphasized at different times that one of the most intriguing aspects of recent developments in organizational studies consists of the concentrated efforts to expand the understanding of “how” and “why” organizations change and “what” the origin is of these newly adopted practices. Thus, we seek to identify the main factors that guarantee the dynamics of collaborative practices in an ambidextrous innovation process during the lifespan of the FD-SOI 28 nm collaborative project. Three main types of collaborative practices were found: assessment (1 st phase), fractalization (2nd phase) and crystallization (3rd phase). In the first phase, collaborative assessment practices of knowledge emerge to interpret reality as it occurs and the different forms in which it appears.