Many companies are poised to take the next step in their artificial intelligence journeys in the coming year, moving from investment in the technology — which continued at a steady pace in 2025 — to considering the return on those investments and what that means for the future of their businesses.

CFOs having a seat that table during those discussions is “super important, because we are the ones tasked with looking after the return on capital,” Harminder Sehmi, CFO of edge computing solutions provider Blaize, said.

As such, “any CFO now who is not actively engaged with their CIOs or their IT people, or their biz dev people… that want to implement AI, then they really need to get up to speed, because they have to be a key decision maker,” he told CFO Dive in an interview.

Making the Costco run

CFOs continued to pay close attention to artificial intelligence over the past year, with many eyeing the technology as a way to help mitigate talent gaps or help their finance teams fulfill a growing list of responsibilities, such as providing key insights to drive customer strategy.

“In the CFO world or the finance world, we always say, the earlier you involve us in any big business decision, the faster it goes through,” Sehmi said.

But as AI has become more commonplace both among businesses’ internal and external processes, the question facing leadership now is one of efficiency — both in terms of use case, and cost for the business.

In looking at Blaize’s customers, for example, Sehmi said that whether it’s a retail, cloud services or smart city business, “they've got to be thinking, ‘how do I get my product, my services, out to customers who want real business outcomes?”

Blaize offers AI-enabled solutions aimed at driving lower cost computing. Sehmi has served in the CFO role for the El Dorado Hills, Calif.-based company since 2023, according to his LinkedIn profile. He previously held finance roles at Truphone and Vodafone, and served a span as senior finance director for Yahoo Europe.

CFOs looking to steer an outcome-driven AI strategy must keep a close eye on both capital expenditures and operating expenses associated with the technology — which means using the right tool, at the right price point, for the right use case.

Sehmi pointed to GPUs as an example: the chips or graphics processing units are imperative to the running of many AI-enabled services, but they may not be the best fit for more mundane or day-to-day uses. He likened them to a sports or luxury vehicle car versus a family-friendly model.

One can drive a Ferrari down a fast track because it’s designed for that job, “but I want to make the Costco run every week,” he said. “Yes, I can do it with the Ferrari, but a different, smaller vehicle that's designed for that purpose is much more cost effective.”

Passing the smell test

CFOs also need to be sure they’re not slipping into bad habits when it comes to how their finance teams are using AI: the tried-and-true financial skills, as well as the know-how gleaned from a career in the space, are still critical elements finance chiefs need to hone and pass along to their teams.

“Our value is the collection of our experiences, okay, and our job is to train the next group of leaders and experts right along the way,” Sehmi said. “But today, it's so much easier to say, ‘I don't need to’.”

When it comes to hiring decisions, for example, technology has reduced the number of new employees CFOs might be searching for — rather than hiring 10 new team members, a finance chief today might only need to hire one, because “the work that the other nine are doing in the early days of their career can be done with AI,” he said.

That, however, could present challenges for leadership down the line. In five or ten years, there may not be enough skilled talent — not just in accounting, but in other industries as well — to fill the roles leadership needs them to fill.

Even among continued AI evolution, finance chiefs need to ensure they are putting the right value on the tried-and-true financial skills. When Sehmi was studying to become a chartered accountant, he learned to add numbers without the aid of machines, “and that's not because you don't have calculators and computers, [that’s] because it that gives you, as a finance professional, an ability to have a sniff test,” he said.

Sehmi said he ecourages his team to use generative AI as a tool to learn how to ask intelligent questions.

“You don't just accept the answer,” he said. “You say, ‘Okay, that's a very good start. I haven't quite thought of that. Let me go validate.’”