This post was originally published on this site.
Chief financial officers are moving rapidly from tentatively exploring generative artificial intelligence to managing it as a core element of finance operations. By the end of 2025, gen AI was no longer confined to pilot programs or isolated analytics use cases. Instead, CFOs report that they are increasingly embedding it across core finance functions, including financial reporting, capital and working capital management, and risk-related activities, all while delivering strong effectiveness across a wide range of tasks.
As adoption has deepened, CFO perceptions of gen AI risk have shifted meaningfully. The average number of reported drawbacks declined sharply over the past six months, indicating that firms are resolving many early operational and technical concerns through hands-on experience. At the same time, the remaining risks are becoming increasingly concentrated around higher-order challenges such as talent readiness, vendor reliance and data security. Instead of worrying about getting gen AI working, CFOs are increasingly focused on using it safely, consistently and responsibly.
Looking ahead, CFOs are recalibrating expectations around how the technology delivers value at scale. Rather than viewing the success of using the technology as a single milestone, finance leaders are approaching it as a phased journey that requires disciplined execution, governance and sustained organizational change.
These are some of the findings in “What Happens When CFOs Get Serious About Gen AI,” a PYMNTS Intelligence exclusive report.
The Role of Gen AI
CFOs are embedding the technology across core finance operations.
CFOs now report that gen AI plays an important role across core finance functions, including financial reporting, capital and working capital management, and strategic decision support. Notably, importance ratings rose most sharply in traditionally control-heavy areas such as risk, treasury and compliance, signaling growing confidence in governance and oversight. The expansion into control-heavy functions suggests that CFOs are increasingly confident in their ability to govern and oversee gen AI at scale.
CFO Views
Gen AI effectiveness is rising rapidly across CFO use cases.
CFOs now regard gen AI as highly effective across almost every task they use it for, with satisfaction levels having jumped into the 80–100% range by December 2025. Compared with July 2025, ratings of effectiveness have especially strengthened in areas such as code generation, automated workflows, product innovation and real-time customer responses, from roughly two-thirds to more than 90% of respondents. Previously mixed areas such as content creation, employee feedback and cybersecurity now have the vast majority of CFOs saying gen AI works well, while pockets of neutrality or dissatisfaction have largely disappeared. The collapse of neutral and dissatisfied responses suggests that CFOs have gained enough experience to deploy gen AI with confidence across high-impact use cases.
The perceived drawbacks of gen AI are shrinking as CFOs gain experience.
CFOs report that the average number of perceived drawbacks from using gen AI has fallen sharply over the last six months, dropping from 6.92 in July 2025 to 4.23 in December 2025 across the full sample. The decline is broad-based: Every industry segment and revenue band shows fewer drawbacks, with particularly large improvements among the biggest companies, where reported drawbacks during that time period fell from 7.52 to 3.44 on average. This pattern suggests that as organizations gain experience, they are resolving early concerns around gen AI governance, risk and change management rather than discovering new problems. Rather than revealing new risks, increased exposure appears to be helping CFOs manage and mitigate early concerns about the technology.
CFOs report fewer operational frictions from gen AI over time.
CFOs report a sharp decline in several operational pain points related to gen AI as programs mature. The share citing errors and output issues fell from 80.0% in July to 35.0% in December, while integration and implementation challenges dropped from 70.0% to 45.0%. Costs and maintenance concerns also declined markedly, falling from 78.3% to 23.3%. Broader constraints, including organizational issues, regulatory compliance challenges, perceived technology limitations and limited ROI, also eased meaningfully, suggesting that many early operational and technical barriers are being worked through in practice.
Some higher-order concerns persist, including talent readiness and data security and privacy, reinforcing the thesis that the next phase of adoption will depend less on resolving implementation friction and more on scaling governance and capability building. This pattern suggests that CFOs are increasingly managing gen AI as an operational capability as opposed to troubleshooting the rollout of a new technology.
Integration Timelines
CFOs are recalibrating timelines for full gen AI integration.
By December 2025, CFOs expected gen AI to take significantly longer to be fully embedded across their organizations than they did earlier in the year. Average timelines nearly doubled over last July–September, reflecting a more realistic understanding of the complexity involved in scaling gen AI across systems, workflows and controls. This shift is consistent across revenue bands, with mid- and large-revenue firms anticipating the longest integration timelines.
The ROI of Gen AI
CFOs measure gen AI ROI across multiple value levers, not as a single financial outcome.
CFOs evaluate the ROI on gen AI through a portfolio of metrics that span both financial and operational outcomes. By December 2025, improved customer experience and improved margins were the most cited measures, while a growing share of executives also tracked cost reduction, revenue expansion and lower capital expenditures. At the same time, headcount reduction is less frequently used as an ROI yardstick, suggesting CFOs are increasingly focused on productivity and performance rather than workforce cuts. This shift implies that proving the value of gen AI will require measurement frameworks that connect operational improvements to financial results, rather than a single ROI figure.
CFO expectations for ROI are becoming more differentiated over time.
By December 2025, CFO expectations for when gen AI will deliver very positive returns became more dispersed than earlier in the year. While a growing share of CFOs now anticipate near-term ROI within one to two years, others increasingly recognize that a full economic payoff will take longer. Compared with last July, expectations have shifted away from a single mid-term timeline toward a split between early value realization and longer-term transformation.
This pattern suggests that CFOs are distinguishing between targeted use cases that can produce quick returns and the slower, more complex work of enterprise-wide integration. CFOs increasingly view gen AI’s ROI as a long-term journey that involves company-wide integration, requiring disciplined execution and governance.
Read More
PYMNTS Intelligence is the leading provider of information on the trends driving AI. To stay up to date, subscribe to our newsletters and read our in-depth reports.
Methodology
“What Happens When CFOs Get Serious About Gen AI” is based on a survey of 60 CFOs working at U.S. firms generating at least $1 billion in revenue in the last year. The survey was conducted from Dec. 15, 2025, to Dec. 19, 2025. This edition examines CFO views and usage of gen AI within their company.