For some years now, the data-centric digital circuits of global finance have been pushing financial markets towards a comprehensive algorithmic mode of (infrastructural) management. Artificial Intelligence (AI) is somehow enthusiastically said to be the new frontier of this process. Building on the Science and Technology Studies (STS) definition of infrastructures – as evolving sociotechnical systems that combine human and non-human elements – and in the political economy of digitalized finance, this work addresses AI in and as an (emergent) financial infrastructure. Clearing the field, it first presents a solid definition of currently in motion, existing AI. It then addresses the state-of-the-art of the most salient AI applications in financial markets in different areas (retail, accounting, credit, risk assessment, etc.). Once paved the way for a multistakeholder empirically grounded understanding of actual AI penetration in finance, it finally elaborates on some concrete evolving trends in this field vis-à-vis the forecasts (sociotechnical imaginaries) provided by relevant actors (major players and regulators).