AI has entered its “prove-it” era. The experimental frenzy of the past 18 months—PoCs, pilots, demos—has given way to tougher questions from CEOs, boards and investors: Where’s the impact? Why aren’t we seeing returns yet? Are we even looking in the right places?
According to AWS innovation strategists Syed Hoda and Glenn Holland, most companies aren’t. And the problem isn’t the technology. It’s the strategy behind it.
Hoda and Holland spoke at CFO Leadership’s recent AI in Finance Forum, in a session dedicated to tackling this exact problem. “Today is about building an AI strategy worth building,” Hoda said. “What does that mean, even? And what prevents us from doing that?”
The answer begins with a shift in mindset that CFOs, they argue, are uniquely positioned to lead.
An AI strategy that starts with AI is doomed.
The first trap is also the most pervasive: Organizations begin with the tech. They ask, How can we apply AI to this? rather than What problem is worth solving?
“The flaw is too much focus on AI,” Hoda said. “It’s not about using AI to go do something. It’s about making something better.”
Yet boardrooms often dictate the opposite. Leaders push for “gen AI-enabled” products simply because the label feels modern, not because the feature improves customer outcomes. “The No. 1 reason AI projects fail is the business case wasn’t made,” said Holland. “They have no way to ascribe what potential value they’re trying to get from these projects.”
This is where CFOs must assert themselves. Finance leaders have the clearest view of where customer friction creates financial drag, where processes are inefficient and where investment actually changes outcomes. If AI isn’t tied to those realities, it won’t scale—no matter how clever the model.
Healthy paranoia is now a leadership requirement.
The second obstacle to a durable AI strategy is a lack of urgency—a false sense of safety inside what Hoda and Holland call the corporate “castle.” Companies spend decades fortifying what made them successful. Then someone else appears with a helicopter.
“And suddenly, everything that you’ve built over those years starts slipping away because your customer of yesterday has changed,” said Holland.
The helicopter moments are familiar:
- Kodak invented the digital camera, then suppressed it.
- Nokia dominated phones until its OS collapsed overnight.
- BlackBerry believed its keypad was a moat—until it wasn’t.
As technology cycles accelerate, helicopter moments won’t announce themselves years in advance. They will appear in quarters or months.
CFOs must institutionalize that paranoia. Not fear, but vigilance. It means forcing the leadership team to answer harder questions:
- Who is tomorrow’s customer—and what will they value?
- What adjacent attacker could weaponize AI against us?
- Where are we vulnerable to speed, cost or experience disruption?
An AI strategy must be a portfolio—not a bet.
Even companies that see disruption coming often make a second mistake: investing entirely in today. Operational AI projects dominate budgets, while the initiatives that could redefine the business barely exist.
Holland and Hoda focused on a framework for innovation created by IT service management company Gartner, that dictates three different kinds of initiatives. “A proper portfolio has all three,” Hoda said:
- Defend projects, “which essentially help you maintain competitive parity.”
- Extend projects, “which help you create new differentiation. Usually, they’re focused on revenue growth, market share growth.”
- Upend projects, which are “about disruption. That’s about changing what good looks like.”
Yet in many organizations, the upend bucket—the one most likely to future-proof the company—is empty.
Holland noted that innovation portfolios resemble VC portfolios: The outliers create the outsized returns. If CFOs allow the long-term bets to shrink or disappear, the company effectively forfeits its future.
And importantly, these categories cannot be measured the same way. Hoda quoted a CTO who lamented, “My CFO wants to measure the upend projects like they’re defend projects.” But an upend project is not going to deliver an ROI this quarter or next quarter.
It is crucial that finance leadership becomes strategic leadership. CFOs must rebalance spending, rethink measurement and give disruptive ideas enough oxygen to evolve—because competitors will.
Holland joked that CFOs may need a second title: chief paranoia officer. Not because they should fear failure, but because someone must keep the organization alert to it.
“We really believe,” he said, “that the fact that you are already controlling some of the most important levers of your business make you the best person to be able to drive change, innovation, disruption, awareness in your company.”
