Ninety-Five Percent of Fund Managers Use Generative AI
In 2024 ninety-five percent of fund managers worldwide had integrated generative artificial intelligence tools into their workflows. Adoption reached this level in less than three years, a pace without precedent in the history of technological diffusion in financial markets. Algorithmic trading, by contrast, required more than a decade to become dominant in US equity markets.
The pace matters, but the deeper transformation concerns the nature of what is being automated. Earlier technologies automated order execution, rebalancing, and quantitative screening. Generative AI automates something structurally different: the formulation of problems, the synthesis of information, and the production of analyses and recommendations. The reasoning that precedes execution is now part of the automation process.
This shift alters the cognitive structure of the sector. When most analysts rely on the same models to process the same sources, outputs converge. A shared tool produces similar results from similar inputs without requiring explicit coordination. Interpretive diversity, which in efficient markets is the condition for price discovery, contracts. The market becomes more cognitively homogeneous at the moment it believes itself more informed.
Shiller (2019) showed that shared narratives move markets regardless of their factual accuracy. When those narratives are generated by the same models on a global scale, the risk is not only that they are wrong, but that they are wrong in the same way for everyone simultaneously, with no actor positioned to correct the common trajectory.
A second effect is less visible. Generative AI lowers the marginal cost of producing analysis. An analyst who once produced five reports a month can now produce fifty with comparable effort. Volume increases, differentiation decreases. The market for financial analysis is behaving like any market transformed by a technology that lowers production costs: consolidation, commoditisation, and selection based on distribution rather than content.
Ninety-five percent signals saturation. The phase in which generative AI offered a competitive advantage to early adopters has ended. Those who do not use it fall outside the market; those who do simply meet the baseline. Competitive advantage shifts, as always, to those who manage to produce something different with the same tools. Understanding what that difference consists of is precisely the problem this series of essays addresses.