Gartner: Only 13% of CIOs Report Significant AI Value, 31% Have Exit Rules

Gartner said on 19 October 2026 that only 13% of CIOs report their enterprise has realised significant or transformational value from AI, in a press release based on its 2027 CIO and Technology Executive Agenda survey of 2,002 CIOs and technology executives, fielded between May and July 2026. The finding landed the same day Gartner opened its IT Symposium/Xpo in Orlando, where it said AI spending had increased 18-fold over the past two years. The interesting part of the survey is less the 13% than what it says about how few organisations have rules for stopping AI work that is not paying off.

What happened

According to XenoSpectrum's write-up of the release, which reproduces its figures, only 31% of CIOs have criteria for stopping pilots from moving into production, and only 17% have criteria for ending live AI initiatives that deliver low value. Gartner groups its top-performing respondents as "Vanguard" CIOs and says those with pilot exit criteria were 2.3 times more likely to be in that group, and those with criteria for ending low-value production work 2.6 times more likely. Gartner VP analyst Alicia Mullery's summary was that "volume is not value."

Cost is now part of the conversation. Fifty-eight percent of CIOs said pressure to deliver cost improvements through AI had increased, and 51% expect AI adoption to raise their total cost of ownership. Only 4% said they have a clear plan to redesign roles for humans and AI agents working together.

On stage in Orlando, Gartner's Kristin Moyer gave the agentic numbers, as reported by InformationWeek: 71% of CIOs expect to deploy agentic AI within 12 months, but only 21% have agentic AI governance in place today. "That is out of alignment," Moyer said. She added: "Use AI agents when you need autonomy, not just automation." Only 1% of CIOs said they had bought AI insurance, while 34% are considering it.

The numbers

Why it matters

Gartner's numbers line up with what other surveys found this autumn, from different angles. The Wharton and GBK Collective study published a week earlier found 80% of senior respondents report positive AI returns but only 28% report strongly positive ones. Gartner's 13% uses a stricter bar ("significant or transformational") and a different population (CIOs, not business managers), so the two are not in conflict. Both say the same thing: almost everyone gets something, few get a lot.

The exit-criteria finding is the more actionable one. A portfolio with no kill criteria cannot reallocate money from failures to successes, so spend grows and value does not. That is an old lesson from IT portfolio management, and it is striking that seven in ten CIOs have not applied it to the fastest-growing line in their budget.

What sceptics say

There are reasons not to over-read the 13%. It is self-reported, and "transformational" is a high bar for technology that most enterprises have been deploying seriously for under three years. Gartner also has an interest in telling CIOs they need more governance advice. And the correlation between exit criteria and Vanguard status is just that: well-run IT organisations probably do both, without one causing the other.

There is also a case that the comparison is moving. In Gartner's previous CIO survey, released in October 2025 with 2,501 respondents, 64% planned to deploy agentic AI within 24 months. This year 71% say within 12. Intent is accelerating even as reported value stays thin, which is either a sign of conviction or the definition of a hype cycle, depending on your priors.

As opinion: the gap that worries me is 71% deploying agents against 21% with agent governance. In my world that ratio usually shows up later as an incident review.

What it means for engineering and DevOps teams

Platform and DevOps teams are usually the ones asked to "productionise" an AI pilot. That puts you in a position to insist on the controls Gartner says are missing.

• Write the exit criteria into the pilot ticket. Define the metric, the threshold and the date before the pilot starts. If the pilot misses, it does not get a production environment. This is the same discipline as an SLO, and the evals-as-unit-tests approach makes it measurable.
• Price the TCO honestly. Half of CIOs expect AI to raise total cost of ownership. Include inference, evaluation runs, observability, on-call and the human review time in the estimate, not just the vendor licence.
• Ship governance with the agent. Identity, scoped permissions, audit logs and an off switch should be in the first deploy, not added after something goes wrong. Agent failure modes in production shows why.
• Measure value with your own telemetry. Self-reported value is how we got to 13%. Instrument the workflow before and after.

FAQ

What did Gartner's 2026 CIO survey find about AI value?

Released on 19 October 2026, it found only 13% of 2,002 CIOs and technology executives say their enterprise has realised significant or transformational value from AI. Only 31% have criteria to stop pilots and 17% have criteria to end low-value live initiatives.

How many CIOs plan to deploy agentic AI?

Gartner analyst Kristin Moyer said 71% of CIOs expect to deploy agentic AI within 12 months, while only 21% have agentic AI governance in place today, according to InformationWeek's report from Gartner IT Symposium/Xpo in Orlando.

What are AI pilot exit criteria?

They are predefined conditions, such as a target metric and deadline, under which an AI pilot is stopped instead of moving to production, or a live initiative is shut down for delivering low value. Gartner found CIOs with such criteria were more likely to be among its top performers.

Related: AI Agent Failures in Production: 5 Incidents, 14 Failure Modes (2026), Policy as Code for AI Agents in 2026 and DevOps Still Matters in 2026.