Two years into the enterprise AI boom, the boardroom question has shifted from “Are we doing AI?” to “What are we getting for it?” New research from Gartner suggests the honest answer, for most organisations, is: not enough.
Only 13% of CIOs report that their enterprise has realised significant or transformational value from AI so far. Use cases have multiplied rapidly over the past two years, yet the likelihood that any given one succeeds remains stubbornly at one in five.
Speaking at Gartner IT Symposium/Xpo in Orlando, Alicia Mullery, VP Analyst at Gartner, put it bluntly: “Volume is not value. AI is creating a reckoning for organisations that lack fundamental disciplines, such as defining use-case value ahead of time and establishing criteria for when to walk away.”
The research behind the headline
The findings come from Gartner’s 2027 CIO and Technology Executive Agenda survey, which gathered responses from 2,002 CIOs and technology executives between May and July 2026. Gartner uses the term “vanguards” for CIOs who are delivering stronger business outcomes while also reporting high CEO-perceived AI value. They are the benchmark against which the rest of the field is being measured.
What sets them apart is less about technology than about management habits. Three practices stand out.
1. Knowing when to stop
Vanguards are more likely to set explicit stopping rules at two points in the AI lifecycle: before a pilot reaches production, and after a system is live but failing to justify its cost.
• Organisations with guidelines for deciding when a pilot should not progress to production are 2.3 times more likely to be vanguards.
• Those with guidelines for abandoning low-value AI initiatives already in production are 2.6 times more likely.
The gap between knowing and doing is wide. Only 31% of CIOs have the first practice in place, and just 17% have the second. As organisations move quickly from experimentation to live deployment across GenAI, traditional AI and AI agents, that omission looks increasingly costly.
Mullery stresses that this is not something the technology function can achieve alone. “Establishing and adhering to these criteria is a team sport,” she said. “CIOs need business partners who are willing and ready to walk away from low value investments.”
2. Treating AI economics as a lifecycle responsibility
The financial pressure is building from both directions. Some 58% of CIOs say they are under greater pressure to deliver cost improvement from AI, while 51% expect AI to increase the total cost of ownership of their technology solutions.
That tension exposes a structural problem. Many organisations still rely on funding and governance models built for cloud and SaaS, where costs are largely predictable. AI behaves differently. “Unlike traditional technology costs, which are largely predictable, AI costs are probabilistic and determined at run time,” Mullery explained.
As technology budgets shift towards GenAI and agentic AI, and spending on legacy infrastructure contracts, the CIO’s remit has to widen. Approving the initial purchase is no longer enough; CIOs will need new approaches to funding, governance and cost accountability across the full lifecycle.
3. Redesigning work, not just cutting it
Perhaps the starkest gap concerns people. Organisations with explicit plans to redesign roles for a human-agent workforce are 2.4 times more likely to be vanguards. Yet only 4% of CIOs say they have such plans.
The popular assumption is that AI will mainly be a tool for trimming technology headcount. Gartner’s data points the other way. “While many leaders assume AI will primarily serve to reduce technology headcount, the true vanguard differentiator lies in workforce redesign,” said Mullery.
The logic is that as AI lowers the cost and effort of producing technology, overall demand for technology work rises. Some 40% of respondents expect to increase IT headcount in 2027, particularly in cybersecurity and in data and analytics. At the same time, roles are compressing and expanding.
“Rather than maintaining static job definitions, vanguards recognise that as AI takes on specialised tasks, the boundaries between established technology jobs begin to blur,” Mullery said. “For example, designers can increasingly perform testing activities, while testers participate in design.” The consequence: fewer distinct role types, each with a broader span of responsibility.
What CIOs should take from this
The message from the data is uncomfortable but actionable. The leaders extracting real value from AI are not necessarily spending more or moving faster. They are:
• Defining value before building, so success has a measurable meaning.
• Setting stopping rules for both pilots and production systems, and enforcing them with business partners.
• Governing AI economics as a continuing cost, not a one-off purchase.
• Redesigning roles for a workforce in which humans and agents work side by side.
With only one in five use cases succeeding, the ability to end the other four early may prove to be the most valuable AI capability a CIO can build.