The determinants of AI innovation across European firms

Using patent data for a panel sample of European companies between 1995 and 2016 we explore whether the innovative success in Artificial Intelligence (AI) is related to earlier firms’ research in the area of Information and Communication Technology (ICT), and identify which company characteristics and external factors shape this performance. We show that AI innovation has been developed by the most prolific firms in the field of ICT, presents strong dynamic returns (learning effects), and benefits from complementaries with knowledge developed in network and communication technologies, high-speed computing and data analysis, and more recently in cognition and imaging. AI patent productivity increases with the scale of research but is lower in presence of narrow and mature technological competencies of the firm. AI innovating companies are found to benefit from spillovers associated with innovations developed in the field of ICT by the business sector; this effect, however, is confined to frontier firms. Our findings suggest that, with the take-off of the new technology, the technological lead of top AI innovators has increased mainly due to the accumulation of internal competencies and the expanding knowledge base. These trends help explain the concentration process of the world’s data market.