Enterprise AI: Only 13% of CIOs See Significant Value — Investment Discipline Makes the Difference
The New Development
Gartner surveyed 2,000 CIOs and technology leaders between May and July 2026 for its annual study, reported on October 28 by IT for Business. The headline finding comes down to one number: 13% of respondents state that their company has already achieved significant or transformational benefits from AI. Over the same period, the number of use cases multiplied fivefold between 2024 and 2025, with a success rate of roughly one in five.
"Volume does not equal value," summarizes Alicia Mullery, VP Analyst at Gartner. The firm points to a lack of discipline: few organizations define the expected value of a use case before launching it, and even fewer set explicit criteria for walking away.
What Sets Successful Companies Apart
Gartner isolates a group of "vanguard" organizations and examines what they do differently. Two practices carry major weight:
- Organizations with explicit criteria to decide when a pilot will not move into production are 2.3 times more likely to belong to this vanguard group.
- Those with defined rules for retiring initiatives already in production that underperform are 2.6 times more likely.
These practices remain the exception: 31% of respondents have pilot-to-production gating rules, and only 17% have formalized kill criteria. In other words, most enterprises accumulate experiments without any mechanism to make hard calls.
High-performing organizations share three additional traits: tight collaboration between the CIO and the CFO, an assessment of business-unit maturity prior to any investment, and a deliberate plan to redesign roles in a workplace combining humans and AI agents.
Cost: The New Friction Point
The survey highlights a contradictory mandate. While 51% of CIOs anticipate higher total cost of ownership across their software stack due to AI—inference costs, token consumption, infrastructure—58% simultaneously face heightened expectations to cut costs through that very same AI. Traditional IT budgeting models, built around fixed software licenses, struggle to accommodate variable usage-based expenses.
Another finding runs counter to the narrative of workforce replacement: 40% of respondents expect IT headcount to grow in 2027, particularly in cybersecurity and data engineering. According to Gartner, the true differentiator is not headcount reduction but "redesigning how work gets done."
What About European SMEs?
Gartner's study is global and skews toward large enterprises. Three recent regional surveys show that SMEs and mid-caps are starting from even further back.
In France, the DGE's France Num 2026 barometer (9,655 respondents) finds that 40% of VSEs and SMEs use AI, but only 19% pay for a solution. Usage is concentrated in text generation (34%) and conversational assistants (24%). Integration spending remains marginal: 3% fund custom development and 2% pay for advisory services. Only 5% leverage AI on their own business data to steer operations.
The Coface Data & AI Barometer (900 executives, October 5) notes that companies with over €100 million in revenue are nearly three times more likely to describe themselves as advanced than those under €10 million. In Luxembourg, FEDIL's AI Survey 2026 identifies total cost of ownership and data maturity as the top barriers to scaling, ahead of user awareness.
The picture is consistent: initial adoption has happened, mostly free and individual; value creation—which requires integration, clean data, and governance—remains to be built.
Implications for Decision-Makers
Financial. An AI budget can no longer be managed like a fixed software license. Leaders must forecast variable costs, track them by use case, and compare them against measured gains—or watch bills climb without visible return.
Organizational. Without kill criteria, every pilot turns into a zombie project draining time and focus. Deciding not to industrialize a pilot must be treated as just as legitimate as launching one.
Human. Both Gartner and FEDIL put skills at the center of the equation: data analysis, AI engineering, and change management. Winning companies plan role evolution proactively rather than waiting for it to happen by default.
Five Rules for an SME or Mid-Cap
- Write down the expected value before launching: quantified time saved, cost avoided, or revenue generated, along with an accountable owner.
- Set production gating criteria and kill criteria upfront, complete with a scheduled review date.
- Limit your portfolio: three closely tracked use cases beat twenty scattered experiments.
- Involve the finance team in monitoring variable costs (tokens, APIs, infrastructure).
- Invest in data and skills before tools: clean master records, team training, and documented workflows.
Caveats and Limitations
The Gartner survey is self-reported and its sample is dominated by large organizations; the "2.3x" and "2.6x" ratios describe correlations rather than strict causation. The France Num and Coface barometers survey different populations and cannot be summed directly. Finally, a low rate of "significant" value does not mean AI delivers no utility—it means few companies know how to measure it yet.
Conclusion
Gartner's message is not a dismissal of AI, but a call for method. Value comes from a small number of well-chosen use cases that are measured rigorously and stopped when they fail to deliver. For SMEs and mid-caps, which lack the budget or bandwidth to multiply pilots, this discipline is less a constraint than a competitive advantage.
What Next?
FASEYA Consulting conducts AI feasibility studies and builds your business cases, KPIs, and use-case governance. FASEYA Academy trains your teams on practical role-specific workflows, and FASEYA Tech industrializes selected use cases by connecting them to your data. To scope your priorities in 30 minutes: request a diagnostic or email contact@faseya.com.
Frequently Asked Questions
What does "significant or transformational value" mean in the Gartner survey?
It reflects the self-assessment of surveyed CIOs: they judge whether the benefits achieved from AI go beyond marginal productivity gains to meaningfully shift organizational performance. Only 13% answer yes.
Why is stopping an AI project a sign of maturity?
Because the majority of use cases do not generate the expected return. Companies with formal kill criteria free up budget and engineering capacity for use cases that actually work, instead of keeping endless pilots alive.
Should an SME wait until its data is perfect before getting started?
No, but it should select initial use cases compatible with its current data quality while cleaning the master data repositories that future use cases will depend on.
Sources and References
Read also
60% of Companies Use AI Without Governance: The Usage Policy the AI Act Already Expects From You
Six out of ten French companies have no formal framework governing their use of artificial intelligence, according to the Coface barometer published on October 5. Yet Article 4 of the EU AI Act, which requires organizations to foster staff AI literacy, has applied since February 2025 and has fallen under market surveillance authorities since August 2, 2026. A written usage policy is the primary way to demonstrate compliance.
E-Invoicing: 30 Million Invoices, 3% Rejection Rate — What SMEs Must Fix Before September 2027
Six weeks after the September 1 launch, French authorities have released their first assessment of mandatory e-invoicing: 30 million invoices exchanged, but still a 3% rejection rate between certified platforms. For SMEs, which will be required to issue electronic invoices starting in September 2027, these figures represent less a victory lap than a checklist of projects to launch right now.
Need to address these challenges in your business?
Our practice directors are available for a 30-minute diagnostic session to assess your priorities.
Prefer a direct discussion?