Enterprise readiness for AI remains a growing concern, with CTO confidence in scaling the technology falling for the third year in a row, according to a report by a global digital engineering and consulting company.
In its latest "What CTOs Think" report, which is based on insights from 500 CTOs, Akkodis found that CTO confidence in their organizations' ability to implement and scale AI has slipped to 48% in 2026, down from 62% in 2025 and 82% in 2024.
"Many organizations have moved past the question of whether they can access AI," said Akkodis CEO Jo Debecker. "The biggest challenge they now face is whether they can make AI work inside the complexity of the enterprise — across legacy systems, fragmented data, risk controls, governance processes, and human workflows."
"The ability to scale AI in a meaningful way matters because that's how enterprises can see the technology's value," he told TechNewsWorld.
"Pilots can prove what is possible, but scalability is what turns AI into better decisions, faster innovation, and measurable business impact," he continued. "To get there, organizations need more than technology. They need workforce transformation, clear governance, and trust from the people expected to use AI every day."
"Organizations have spent two years running proofs of concept," added Eric Hulse, director of research at Command Zero , a cyber investigation automation company in Austin, Texas.
"The ones stuck in pilot mode are stacking up costs without capturing value," he told TechNewsWorld. "The pressure to scale is real. But the Akkodis data shows confidence in the ability to scale fell from 82% to 48% in three years. That makes sense. The more CTOs grapple with what scaling actually takes, the clearer it gets that most organizations aren't built for it."
Scaling is where many AI programs are getting stuck, observed Ryan McCurdy, vice president of marketing at Liquibase , a database-change automation company in Austin, Texas.
"Companies can get access to capable models, run demos, and show productivity gains. The harder part is turning that into work the enterprise can trust every day," he told TechNewsWorld.
When agentic AI is added to the mix, it raises the stakes, he continued. "It is not just answering questions. It can write code, generate schema changes, update pipelines, and trigger work across the business," he explained. "That requires a different operating model. Teams need to know what agents can do, where humans stay involved, and how AI-driven changes are reviewed, traced, and controlled."
"A lot of organizations have not figured that out yet," he said. "So they either keep AI boxed into experiments, or they move too fast and create risk in production. Neither path scales."
"The companies that get this right will not just buy more AI tools," he added. "They will build the structure around them, such as trusted data, governed workflows, and proof of control. That is how AI moves from interesting experiments to something the enterprise can actually run."
The report explained that as organizations move beyond pilot programs, execution complexity increases across leadership alignment, governance, and workforce trust. It found that fewer than half the CTOs (44%) believe leadership teams have sufficient AI understanding, and only 36% express satisfaction with workforce trust levels.
In addition, the CTOs said that AI progress was being limited by barriers, such as a lack of in-house technology skills (32%),…