Make or buy your artificial intelligence? Complementarities in technology sourcing

Abstract

We investigate firm decisions to adopt artificial intelligence (AI) technology and how adoption is sourced: by purchasing commercial readymade software, by developing or customizing solutions in-house, or both. Using a cross-sectional data set of 3143 firms from across Europe, we examine the extent to which sourcing strategies exhibit complementarity or substitution. We find that adoption of AI using readymade software as a sourcing strategy is now increasingly commonplace, but differs across industrial sectors. Further, complementarities between sourcing strategies are common across sectors, though with some differences in strength and some exceptions. Our results show that sourcing strategies play an important role in shaping AI adoption decisions among firms.