The big AI New Year’s resolution for businesses in 2026: ROI

After three years of experimenting and spending, and as talk of an AI bubble looms, enterprises are starting to demand results. According to Kyndryl’s recent Readiness Report, drawing on insights from 3,700 business executives, 61% of CEOs say they are under increasing pressure to show returns on their AI investments compared with a year ago. 

This is putting company leaders to the test in terms of balancing long-term innovation with the need to prove outcomes now, all while AI development continues to move at breakneck speed. It’s also creating risks of misalignment in the C-suite, with tech and business leaders looking out for their firm’s innovation while financial leaders look out for the balance sheet.  

“The last year was a lot about experimental budgets, like, ‘I’m just going to give the budget to every department [and] experiment with whatever tools they think are useful,’” said Lexi Reese, a former strategy executive at firms including Google and the current CEO of AI observability platform Lanai. “Now, it’s accountable acceleration, because the price tag on this is very expensive.”

The AI spending spree

The unprecedented amount of money being spent to develop and deploy AI has been grabbing headlines all year. Much of this surrounds infrastructure spending by frontier AI labs and eye-popping startup investments, but enterprises are heavily investing, too. Garner expects spending on AI application software to more than triple from last year to almost $270 billion in 2026. Over the past year, Reese said she’s had conversations with over 300 customers about their AI tool costs and found they are spending between $590 and $1,400 per employee annually, according to internal data shared with Fortune.

For many executives, this is causing flashbacks to digital transformations of the past—in particular, the transition to cloud, which left some with a bad taste in their mouth. Michael Bradshaw, global practice leader for applications and data at Kyndryl, told Fortune he sees this consistently in his role, where he works with executives on bridging the gap between their business strategy and what they are actually trying to do with their technology. 

“Almost two-thirds said, ‘We got to our cloud strategy by accident,’ and 95% of them said, ‘If we could do it over again, we would redo our cloud strategy,’” he noted, referring to data in Kyndryl’s report. “That is shocking to me as a practitioner, because we’re about to see the same thing happening. I think we’re seeing it with AI today.”

For example, he pointed to one customer that spent over a billion dollars just to implement its initial ERP (enterprise resource planning) software: “As they’re looking at the next wave, their frame of reference is, ‘Oh, my gosh, I’m going to have to spend that same amount of money to do the next wave of transformation, on top of the challenged business environment. I can’t afford that!’”  

Manisha Khanna, senior product manager for AI and generative AI at analytics firm SAS, told Fortune this is now the number one challenge she’s hearing from customers, who are asking: How much should I invest in these technologies so that it gives me ROI? 

“Our customers have seen some very flashy demos from various vendors. They have high expectations. But then they start asking questions as to, ‘Do you really understand how much this is going to cost in production?’ There is compute cost, data infrastructure requirements, talent cost,” she said. “How do you budget for all of that?”

The challenges hindering ROI

When Dan Rogers stepped into the CEO role at Asana just a few months ago, AI ROI quickly became top of mind. Understanding the return on AI is a top priority, he told Fortune, and he is using a formalized approach that includes both financial ROI and “human-centered” ROI metrics, such as reduced administrative burden and improved decision-making. Every functional leader owns outcomes from AI in their area and must report specific metrics, he said, adding that the company is “setting ambitious efficiency targets as part of a top-down strategy.”

At the same time, this poses a challenge around how to weigh long-term bets against short-term value. 

“Ask for financial ROI too early and you kill experimentation. Wait too long and you’re in pilot purgatory,” he said. 

Rogers noted he’s cautious about forcing every AI initiative to clear a narrow, short-term hurdle, as benefits sometimes accrue gradually. Some capabilities are infrastructure layers, for example, and you have to look at what that layer enables over time. On the flip side, the rapid innovation cycle of AI can make wait-and-see especially challenging. 

“The pace of AI change has effectively broken traditional planning cycles. We’ve moved from a 12-month ABR [annual business review] rhythm to quarterly checkpoints and continuous reprioritization, because you simply can’t wait a year to course-correct in a domain evolving month to month,” Rogers said.

Kyndryl’s Bradshaw echoed this, describing moving into “very, very rapid cycles where we’re having conversations with customers about delivering value in four to six months.” Another factor Bradshaw points to is that AI ROI can be difficult to measure, in particular with so much of it currently revolving around personal productivity technologies.

“I have agents that are helping go through my mail. How do I translate that to my personal productivity, and what business outcomes am I driving? That’s why businesses have really struggled with being able to articulate ROI,” he said.

Together, all of this is starting to create cracks in the C-suite. According to the Kyndryl report, nearly three in four CEOs said short-term ROI pressure undermines long-term innovation, and 65% said they aren’t aligned with their CFO on long-term value. 

“A CFO might want to look at it from a balance sheet perspective. The business leader wants to make sure the business model changes. The technology leader—like a CTO—wants to make sure I’m innovating, I’m applying all the latest technologies, that I have the talent and the skills in my team so I’m really able to realize value,” said Khanna. “And these three [C-suite leaders] are on completely different pages right now with regard to the expectation.”

A new year in AI ROI

Asana’s Rogers believes the larger industry conversation around spending has definitely intensified the focus on ROI. “And honestly, that’s overdue,” he said. “When reports say that 95% of AI pilots generate zero return, boards and executives are understandably asking tougher questions.” He added he thinks the push for clear ROI is a positive step, and that it’s time to shift the conversation “from novelty to outcomes.”

The growing desire to prove ROI could cause a boom for firms like Reese’s Lanai, which promise to help customers get a clearer picture of how their teams are actually using AI, what’s driving impact, and what isn’t. Lanai in particular analyzes all of the prompts users are putting into their AI tools to get granular views of their work. Reese said the key to driving ROI is to determine what your “highest-value workflows” are and adjust from there.

“If you can see the work and the workflows, you can start to standardize operations and say, ‘Sales is finding a lot of value in these workflows, so let’s double down on that. These workflows are getting a lot of adoption, but they’re super risky. Let’s nudge people in the right direction,’” she said. 

There will certainly be pressure to prove ROI in 2026, but Khanna and Bradshaw believe most firms aren’t positioned to get there. 

“We’ll see it within what we characterize as the ‘pacesetters.’ So where you have that alignment within the C-suite, where you have that outreach with your employee base, and you get that alignment of the technology strategy,” Bradshaw said. “You will see some [companies that achieve ROI in 2026]. It will not be widespread. That’d be my prediction, because most firms are not positioned to be able to do it.”