Texas CFOs reckon with the rise of ‘shadow finance’
Like finance leaders around the globe, CFOs across Texas are trying to get their arms around a new problem created by the rapid adoption of artificial intelligence: “shadow finance.”
The term was a recurring theme during a recent series of CFO Alliance roundtables in Houston, Dallas and Austin last week as Texas continues to grow as a corporate hub and the newly launched Texas Stock Exchange begins trading.
According to CFO Alliance CEO and founder Nick Araco Jr., finance chiefs across all three cities raised concerns about employees signing up for their own artificial intelligence subscriptions and using the tools to generate their own metrics. Some of Araco’s CFO members shared that they’re dealing with more than 100 systems, in some cases, that haven’t been vetted.
One example discussed during the Texas meetings showed how quickly issues can pop up. An AI tool interpreted “+31 headcount” as “+31 vehicles” in a board deck. The error made it through multiple rounds of human review before someone caught it.
“The deeper concern isn’t the error itself, it’s what it implies about scale,” Araco said. “If a mistake that obvious got three rounds of review before catching, how many smaller, less obvious ones aren’t getting caught at all?”
AI adoption is moving faster than trust
Though the conversations took different turns as the CFO Alliance moved across Texas, the concerns about how technology affects the integrity and functionality of finance were consistent.
In Houston, Araco said CFOs spent significant time on the gap between AI adoption and the returns finance teams can measure. The group discussed Deloitte data showing 63% global AI adoption, while only 21% of finance leaders report measurable ROI clarity.
“That gap is the real story right now, not the adoption number,” Araco said.
In Dallas, cost was a major part of the discussion, too. Some CFOs said OpenAI-related costs had up to a 50-fold increase, pushing some toward internal or private large language models as they look for more control over costs and company data.
Austin brought a totally different tone, Araco explained. He described CFO sentiment there as “dizzyingly optimistic,” with finance leaders focused on getting their organizations “game ready” for AI even as governance continues to catch up.
Across the three cities, CFOs are also beginning to draw clearer lines around what they will allow AI to do inside finance. One of those lines, Araco said, is the ERP system.
AI agents can draft journal entries, for example, but CFOs said a person should still review and load them rather than allowing AI to write directly into the system.
Finance chiefs are also separating work that requires an exact answer from areas where AI can be useful as more of an analyst. Code can handle work that has to be precise, such as margin attribution, while AI can be used to stress-test assumptions.
“Systems cannot regulate their own compliance,” Araco said.
CFO Alliance’s own maturity tracking puts AI governance around the same 21% level as ROI clarity. Araco said the two problems are closely connected.
“Governance and measurement are lagging adoption in lockstep, not independently,” he said. “Both problems are the same underlying problem.”
CFOs become ‘guardians of narrative integrity’
Araco said the conversations across The Lone Star state focused on the CFO’s role as more employees gain the ability to produce their own analysis via AI.
He described CFOs as becoming “guardians of narrative integrity,” responsible for keeping the company’s story “coherent, defensible and grounded in the ERP” as AI accelerates decision-making and spreads data across more tools.
Dallas offered one example of how quickly that responsibility is changing. CFOs there discussed how compressed decision cycles have reduced the effective shelf life of a strategic plan to roughly 30 days.
“When your plan goes stale in a month, the CFO’s narrative has to be rebuilt and re-defended constantly, not just delivered once a quarter,” Araco said.
When participants were asked to identify the top skill for the CFO of 2030, Araco said finance leaders pointed to the ability to translate complex data into clear, connected strategic narratives, a shift they described as moving “from crunchers to storytellers.”
The group also discussed CFOs taking on the truth-teller role more formally and owning more of the strategic planning process. Technical fluency with AI still matters, but Araco said empathetic communication was viewed as an equally important skill rather than a soft add-on.
“Data has to travel with narrative, not replace it,” Araco said.
