Gartner's CIOs say AI risk is outrunning value, and AI vendors should read the numbers

The largest annual gathering of corporate technology buyers opened in Orlando on October 19 with a message that sits awkwardly beside the revenue forecasts of the companies selling them artificial intelligence. Gartner analysts Daryl Plummer and Kristin Moyer told the IT Symposium/Xpo keynote audience that 86% of chief information officers report that AI risk in their organisations is growing faster than AI value. Gartner's research, as presented on stage, puts enterprise spending on AI at roughly 18 times its level of two years ago, yet only 13% of CIOs in its annual survey of more than 2,000 technology executives say their company has realised significant or transformational value from it.

These are not the views of sceptics on the sidelines. They are the self-assessment of the people who sign the purchase orders on which much of the AI software industry's valuation rests. For founders and investors who have spent two years modelling enterprise adoption as a steadily rising curve, the Gartner data offers something more useful than enthusiasm: a set of numbers on how that adoption is actually going inside large organisations, and what buyers intend to do next.

Adoption is wide, returns are narrow

The figures describe a market that has bought broadly and harvested selectively. According to the keynote, the average organisation increased its number of AI use cases by more than 400% in a single year. Over the same period the success rate has not improved: Gartner said about one in five AI projects succeeds, the same proportion as a year earlier. A separate Gartner survey of more than 1,000 functional leaders, conducted in the second quarter, found that 58% see some positive value from their AI initiatives, but the analysts stressed that most of the value comes from a minority of successful projects rather than from the long tail.

Productivity, the benefit most often cited in vendor pitches, is the thinnest result of all. CIO Dive reported that only 16% of enterprises say they have seen productivity gains. Gartner also put a price on a cost that rarely appears in business cases: about four in ten workers said they had received low-quality AI output in the previous month, with each instance taking nearly two hours to correct. Ms Moyer estimated that this rework costs an average Fortune 500 company about $52m a year.

None of this means enterprises are retreating. It means the second phase of corporate AI adoption looks different from the first. The first phase was about access: licences, pilots and internal experiments, often bought on a per-seat basis. The second, on Gartner's evidence, will be about pruning.

From careless consumption to outcomes

The analysts' advice to CIOs was unusually blunt about the vendors in the room. Mr Plummer told the audience that vendors want companies to use AI agents everywhere in order to build momentum for their own products, and urged CIOs to stop what he called careless consumption and move to outcome-based development. His recommendation was to reduce waste before cutting investment, and to concentrate spending where outcomes matter most. Ms Moyer, according to InformationWeek, told CIOs to demonstrate measurable business value before increasing AI investment, adding that the biggest drag on AI momentum is using it when it is not needed.

Translated into procurement behaviour, that guidance points to three changes. Use cases will be audited and consolidated rather than multiplied. Successful deployments will be scaled at the expense of new pilots. And ownership of AI spending will be shared more formally between the CIO and the chief financial officer, with security and legal teams involved, which in practice lengthens sales cycles and raises the bar for renewal.

For software companies whose revenue growth has been driven by expanding seat counts or token consumption inside existing customers, this is the more important signal. Gartner is not telling buyers to spend less on AI in aggregate. It is telling them to spend it on fewer things, with evidence attached.

The agent governance gap

The sharpest number in the keynote concerned autonomous agents, the category into which venture capital has poured money all year. A Gartner survey of more than 2,000 CIOs and technology executives in the second quarter found that 71% of organisations expect to deploy agentic AI within 12 months, but only 21% currently have governance capabilities for it in place. A third-quarter survey of 303 CIOs and IT risk leaders found that only 25% align technology risk with business risk. Ms Moyer's verdict, as reported by InformationWeek, was that in such conditions it is not possible to create value faster than AI can create risk.

Gartner's proposed remedy has three parts: a "value-risk flywheel" to judge which deployments compound value and which compound risk, an internal "AI central bank" acting as a shared control plane for accountability and economics, and a deliberate effort to cultivate employees who use AI creatively. InformationWeek cited IKEA as an example of the last point, noting that the retailer reskilled 8,500 call centre workers to sell interior design services after chatbots took over routine calls, generating more than $200m in additional revenue in fiscal 2025.

The governance gap is also a market. Gartner's data suggest that only 1% of CIOs have purchased AI insurance and 34% are considering it. Tools for monitoring agent behaviour, detecting shadow AI and recording AI decisions as a system of record are the kind of products that become budget lines once a board decides the risk is real.

What the numbers mean for AI revenue

The disconnect between spending and realised value matters for private valuations because so much AI revenue is still early, usage-based and concentrated in pilots. A buyer that has increased its use cases by more than 400% in a year and now intends to cut the weakest of them will not churn uniformly. It will keep the vendors that can show a measurable result and drop those that cannot. That favours companies selling into a specific workflow with a clear cost or revenue line attached, and penalises horizontal assistants whose value is diffuse.

It also explains why pricing models are under pressure. Seat-based pricing assumes value scales with the number of users. Gartner's evidence says it scales with the number of successful deployments, which is a much smaller figure. Vendors that can tie their fees to outcomes, or at least report outcomes credibly, will find renewal conversations easier than those relying on adoption metrics.

Our view

Gartner's keynote is the most useful corrective the enterprise AI market has received this year, precisely because it comes from the buyers rather than from critics. Spending is up 18-fold, use cases are up more than 400% in a year, and still only 13% of CIOs report transformational value. That is not a bubble bursting. It is a market moving from experimentation to selection, and selection is harsher on suppliers than experimentation ever was.

For founders, the implication is to sell evidence, not access: instrument every deployment, report the business result in the customer's own numbers, and expect the CFO in the room. For investors, net revenue retention in 2027 is the metric to watch, because the pruning Gartner describes will show up there first. The clearest beneficiaries are likely to be vendors whose products make AI governable, measurable and insurable. The most exposed are those whose growth depends on customers continuing to add pilots they cannot yet justify.

Sources

• Business Wire via Finviz: Gartner Says AI Risk is Rising Faster Than AI Value
• CIO Dive: AI risks, value remain unaligned for CIOs
• InformationWeek: Gartner: CIOs must balance AI value with risk, and fast
• Techstrong.ai: CIOs must compound AI value faster than accelerating risk, Gartner warns

Filed under AI. Written by Invest Times Editorial under our editorial standards. Commentary and analysis for information only, not investment advice.