The AI ROI question CFOs can't answer (yet)

CFOs know they need to invest in AI, but the business case their own discipline demands keeps refusing to cooperate.

A recent Forrester study, commissioned by Airwallex, surveyed 1,279 finance decision-makers across APAC, EMEA, and North America. One finding crystallised a tension that's been building for months.

40% of decision-makers can't move beyond short-term productivity gains because they can't build a clear business case for AI.

Meanwhile, 74% say integrating AI into finance workflows is a top priority for the next 12 months, right alongside strengthening financial controls and regulatory compliance. The appetite and conviction are there. But the gap between wanting to move and proving you should move has frozen some of the sharpest minds in finance.

If you're managing multiple entities across currencies and jurisdictions, the problem gets a whole lot bigger. Faster closes, real-time reforecasting, and live risk detection top most finance teams' AI wishlists, and every one of them demands proof in a format boards haven't learned to read yet.

Unless stated, all statistics referenced below come from "Building An AI-Ready Finance Function", a commissioned study conducted by Forrester Consulting on behalf of Airwallex, June 2026.

The pressure keeps building and the instincts make sense

We hear this from the companies we work with. CFOs and CEOs are fielding the same question from boards, investors, and peers on repeat. How are you using AI? Shannon Scott, our Chief Product Officer, called out the pressure on a recent episode of The Airwallex Podcast.

Your hesitation makes complete sense. You've built your career on proving returns before committing capital, and AI's costs shift with usage while its returns compound over years. The traditional business case collapses under its own assumptions.

42% of finance leaders fear falling behind competitors who are modernising their finance function faster. 45% believe slow scaling will limit their ability to respond to volatility. Moving now means spending without proof, while waiting means watching competitors make the kind of efficiency gains you'll spend years trying to close. Faster reforecasting, tighter close cycles, live anomaly detection. These advantages accumulate in the background and become hard to claw back.

Waiting for perfect ROI is its own kind of risk

More conservative organisations want a clean business case before they commit. You know the playbook. Pilot, measure, present to the board, get sign-off, scale. It works for software with predictable licensing fees and linear depreciation. AI breaks that sequence because its costs are variable, its benefits accumulate unevenly, and isolating its contribution from ERP migrations, data warehouse consolidation, and org restructures borders on impossible.

Deloitte's research complicates the waiting game. Almost half of companies see returns on AI investment within two years. Yet 69% say implementing governance alone takes more than 12 months, and fewer than a third of AI experiments scale to production within six months. The returns show up, but they arrive on a curve that quarterly evaluation cycles can't track. And every quarter without AI in the workflow is a quarter of institutional learning your competitors have banked and you haven't.

Paul Bassat, co-founder of Square Peg Capital, warned against the wait and see approach on a recent episode of The Airwallex Podcast.

He drew an analogy worth sitting with. Nobody asks "how are we going to use electricity?" It became the utility that powers everything, and AI will follow the same arc. The organisations that identify problems worth solving and point AI at those problems early will pull ahead.

The cost of delay is the business case

When Forrester asked finance leaders to name the biggest cost of delaying AI adoption, 42% pointed to the same thing. Falling behind competitors who are modernising faster. Every month of waiting compounds that distance across close cycles, cash visibility, and reforecasting speed.

The productive move right now is to start with the problems, not the tools. Every finance function has challenges that have gone unsolved because the technology to address them didn't exist until recently.

Once you've identified those problems, the foundation you build on determines whether AI accelerates good decisions or amplifies bad ones. If your financial systems are fragmented across multiple banks, billing platforms, and disconnected accounting tools, AI operates on incomplete data and produces decisions that look precise but lack context. Consolidating onto a unified platform gives AI a complete picture across entities, currencies, and jurisdictions, and gives your team one system to learn rather than seven.

53% of decision-makers cite their teams' limited experience running AI-enabled finance processes as a barrier. 29% point to resistance to change. These concerns feed on each other. The longer you wait to build experience, the harder the eventual transition becomes, and the wider the gap grows between you and competitors who started six months earlier.

Your rigour pulls one way. The shifting competitive ground pulls the other. Both instincts are sound. But only one of them has a ticking clock attached. The CFOs who navigate this well will be the ones who built the right infrastructure, started with the right problems, and let the measurement framework catch up to the value they're already creating.

Source: Unless stated, all statistics referenced in this article come from "Building An AI-Ready Finance Function", a commissioned study conducted by Forrester Consulting on behalf of Airwallex, June 2026.