Three-quarters of finance chiefs expect AI to reshape their function — but only 9% have successfully scaled it, study finds
A global survey of more than 200 CFOs and senior finance leaders reveals a stark gap between expectations for artificial intelligence and its actual deployment, as most organisations remain trapped in small-scale pilots that never reach enterprise-wide adoption.
Text by Martti Asikainen, 25.5.2026 | Photo by Adobe Stock Photos
Three in four chief financial officers expect artificial intelligence to have a high or very high impact on the finance function by 2030, yet only 9% report having scaled AI in line with their expectations, according to research published on 21 May by management consultancy BearingPoint.
The study, based on a global survey conducted in 2025 and more than 30 in-depth interviews with CFOs and senior finance leaders across Europe and the United States, found that 73% of respondents described their current level of AI adoption as minimal or basic. More than 80% anticipated significant or moderate changes to finance roles within five years.
Stuck at the pilot stage
The research identifies what it calls a “pilot trap” — a pattern in which individual AI initiatives produce encouraging results but fail to spread across organisations. Promising use cases in areas such as financial forecasting, process automation, and decision support remain isolated, delivering value in narrow contexts without reshaping how finance functions as a whole.
Data quality emerged as the most significant barrier to scaling. Three in four CFOs cited it as a major obstacle to adoption. Fragmented technology systems, unclear governance, and limited in-house capabilities were also widely cited as constraints slowing progress.
“The barriers to scaling AI in finance are not technological, but rather structural,” said Olivier Beugnet, Partner at BearingPoint. “Data quality, governance, and process design are what determine whether a pilot becomes a scalable capability or stays an experiment.” Beugnet did not elaborate on how organisations that have successfully scaled AI differ in their approach to data infrastructure.
Redesigning, not just deploying
The report argues that closing the gap between ambition and execution requires organisations to redesign their operating models — restructuring processes, data foundations, governance frameworks, and job roles — rather than simply adding AI tools to existing structures.
According to the study, organisations that are making the most progress have begun aligning AI initiatives with broader transformation programmes, standardising processes before attempting to automate them, and building the governance structures needed to sustain adoption at scale.
The role of finance professionals is shifting as a result. As AI takes on more of the routine analytical work, the study suggests that finance staff are increasingly expected to interpret and validate AI-generated outputs rather than produce the underlying numbers themselves. Organisations investing in hybrid finance and data skills, the report argues, are better positioned to sustain adoption.
“CFOs have a unique opportunity to shape the next generation of the finance function,” said Kornel Malysch, Partner at BearingPoint. “Those who close the gap between ambition and execution will define how finance creates value in the age of AI.” The report did not provide benchmarks for how long that transition is expected to take or at what cost.
Who is pulling ahead
The study finds that a small minority of organisations have moved beyond isolated pilots to treat AI as a core operational capability, embedding it into systems, decision-making processes, and continuous performance monitoring rather than managing it as a series of discrete projects. The research does not specify what proportion of respondents fall into this category beyond the headline figure of 9%.
The report outlines four areas requiring deliberate action for CFOs seeking to close the gap: operating model design, data and system transformation, governance, and workforce development.
The BearingPoint study does not specify how survey respondents were recruited or whether the sample is representative of CFOs across industries and geographies. As with all research published by a consultancy that offers services to address the problems it identifies, the findings should be read alongside independent analysis.
