CIOs earn AI reprieve, but ROI pressure is surging

2026 arrived as the year AI ROI would need to get real. After years of experiments and pilots that largely failed to scale, CEOs’ No. 1 priority for CIOs was to achieve demonstrable benefits from AI investments.

Under that pressure, CIOs started to feel the heat, with 71% of IT leaders in February saying they believed they had until midyear to prove AI value or face budget or job fallout, according to a survey published by AI platform provider Dataiku. For many, experience may have informed that anxiety, as three-quarters of CIOs surveyed then also said they had remorse over at least one major AI vendor or platform selection made in the past 18 months..

Six months later, and passed that midyear mark, CIOs who have set their course for AI ROI are finding that destination remains elusive. Still, there hasn’t been a spate of CIO firings, and AI spending continues to grow. About 71% of organizations plan to increase AI spending this year, but only 27% expect near-term ROI, according to recent research by IT solutions provider TEKsystems.

That metric aligns with findings from CIO.com’s State of the CIO survey from earlier this year, when 40% of IT leaders said some AI initiatives (between 30% and 70%) were meeting ROI goals. While progress remains the same, the drumbeat to prove value goes on.

“CIOs are feeling pressure to demonstrate that AI is delivering measurable business value, not just experimentation,” says Jed Dougherty, SVP of AI and platform at Dataiku. That’s because, while CIOs’ worst fears haven’t been realized, organizations are putting greater scrutiny on AI investments, he adds.

“The CIOs who are succeeding aren’t deploying AI everywhere,” he says. “They’re building the governance, data, and operational foundation that lets the business scale AI responsibly and demonstrate real outcomes.”

A pronounced focus on AI spending and costs

Bob Hutchins, CEO at AI advisory firm Human Voice Media, believes CIOs’ early year anxiety wasn’t likely based on any formal deadlines.

And if any CIOs have been fired since February because they missed AI targets, those changes are likely hidden from public scrutiny either in reorganization efforts or other leadership changes, he says.

“Midyear came and went and there was no apparent bloodletting,” he adds. “I haven’t seen credible evidence of mass firings of CIOs due solely to missing return on investment targets for artificial intelligence.”

But organizations do seem more focused on spending their AI budgets wisely, Hutchins notes. Nearly half of all organizations surveyed recently by KPMG have delayed, stopped, or scaled back AI projects due to budgetary constraints, he says.

“Companies are stopping poorly performing projects, scaling back pilot programs, decreasing the number of vendors they use, creating cheaper models of products and services, and giving more control over AI approval to the financial department,” he adds.

Ryan Ries, chief AI and data scientist at AI and cloud consulting firm Mission Cloud, also sees IT leaders still under pressure to improve AI results.

While firing a CIO midyear looks bad on an earnings call, IT leaders now face budget triage efforts related to AI, he says.

“Money still flows to projects with a hard number attached,” he adds. “Pilots without one get quietly starved but not killed outright. The AI landscape is constantly changing, and companies are trying to figure out all the new tools like coworking and coding solutions.”

Moreover, facing increased uncertainty over AI pricing, CIOs are re-examining AI adoption metrics and becoming more aware of the hidden costs of AI.

Cost control is also receiving greater emphasis as some early agentic forays have shown how an AI agent can cost more than an employee without limits in place.

CIOs still on the hot seat

CIOs are also finding that it is taking their organizations more time to figure out how to use AI tools to their full advantage, and that they are constantly reacting to errors, Ries notes.

As a result, IT leaders appear to be putting in more effort to find AI value than they were earlier this year, he says. “Fewer are succeeding than leadership wants to admit,” he adds.

While most organizations were experimenting with the so-called “art of the possible,” IT leaders seeing success are focused on attaching a metric to every AI project before launch, not after, he says.

Many IT leaders still aren’t taking that approach, however. “The CIOs still stuck are the ones running pilots that never graduate to production, usually because nobody can explain what the model is doing under the hood, and teams are trying to answer the wrong questions with AI,” he says.

It’s possible that CIOs have gotten a reprieve due to the ongoing complexity of the AI ROI mandate, Ries adds.

“Boards don’t fire on a spreadsheet’s calendar, but the underlying pressure was real, and it hasn’t eased,” he says. “Instead of a hard cutoff, CIOs now face constant reporting. Monthly board briefings on AI performance are becoming standard, not optional.”

CIOs should remain on their toes and focus on driving AI value, Ries says. “The fear of a July guillotine was overstated,” he adds. “The fear of ongoing, permanent scrutiny was not, if anything it has increased, due to how quickly cost overruns can happen.”

Budget pressure is real

Like other observers, Mridul Nagpal, CTO and co-founder of AI software development company Krazimo, sees a growing focus on AI budgets and untargeted spending.

“The pressure is real, but it’s reshaping spend more than cutting it,” he says. “What’s actually at risk is the undifferentiated AI budget — the ‘we’re doing AI’ line item with no outcome attached.”

Many CIOs are now starting to show AI value, but by narrowing their approaches, not expanding them, he adds.

“CIOs who funded broad experimentation are the ones sweating; CIOs who tied spend to a specific, measured workflow are defending, and often growing, their budgets,” he says. “The fallout is landing on unaccountable AI spend, not AI spend per se.”

The CIOs showing returns have quietly killed sprawling AI pilot portfolios and doubled down on a handful of use cases that reached production, Nagpal adds.

Like Ries, Nagpal believes that earlier CIO fears were a bit overblown and, at the same time, they’ve gotten more time to prove AI value.

“Boards softened the ‘or else’ because the whole market discovered the pilot-to-production gap is real and hard, so the deadline quietly moved,” he says. “But the underlying expectation didn’t disappear — it matured from ‘show me AI’ to ‘show me AI that pays for itself.’”