Enterprise AI investment is growing, but substantial business value remains elusive. Only 13% of CIOs say their organizations have realized significant or transformational AI value, according to findings Gartner released October 19 at its IT Symposium/Xpo in Orlando.
The survey of 2,002 CIOs and technology executives found that just 31% have rules for stopping unsuccessful AI pilots. Only 17% have criteria for abandoning low-value initiatives already in production. That leaves many organizations without formal guidelines for deciding when continued AI investment no longer makes business sense.
Gartner finds few CIOs have AI stopping rules
Gartner's 2027 CIO and Technology Executive Agenda survey, conducted from May through July 2026, identified a high-performing group of CIOs called “vanguards.” They reported stronger business outcomes and high CEO-perceived AI value.
Organizations with rules for stopping pilots before production were 2.3 times more likely to qualify as vanguards. Those with criteria for ending low-value production deployments were 2.6 times more likely. The findings show an association, not proof that stopping rules independently improve financial returns.
“Establishing and adhering to these criteria is a team sport,” said Gartner Vice President Analyst Alicia Mullery. She also emphasized the need for business partners willing to abandon investments that deliver insufficient value.
The 13% figure does not mean other enterprises received no AI benefits. It measures those reporting significant or transformational value, rather than every improvement achieved through AI.
An independent BearingPoint study of 1,050 executives published October 1 found that nine in 10 surveyed organizations would continue investing in AI despite limited expected returns. BearingPoint also found that nearly three-quarters of organizations implementing AI reported measurable financial benefits, although its survey used different measures from Gartner's research.
Rising AI costs complicate investment decisions
Financial pressure is increasing alongside deployment. Gartner found that 58% of CIOs face greater pressure to deliver cost improvements through AI, while 51% expect AI to increase their technology solutions' total cost of ownership.
AI operating costs can fluctuate with consumption, complicating conventional software budgeting. Google's Gemini Enterprise pay-as-you-go pricing illustrates the trade-off: lower upfront commitments can introduce variable expenses as workloads expand.
Model inference, cloud infrastructure, monitoring and human oversight all contribute to long-term costs. Evaluating those expenses against business outcomes becomes essential before renewing or expanding deployments.
Visibility creates another challenge. Dataiku's cross-platform AI agent monitoring plans illustrate a potential limitation: the information available to IT teams depends on what each connected platform exposes. IT teams need reliable operational data to assess whether deployed systems justify continued investment.
What eWeek Found: Production AI exit rules lag pilot controls
Gartner's survey reveals a 14-percentage-point gap between CIOs reporting pilot stopping criteria (31%) and those with production exit rules (17%). Fewer CIOs report rules for ending deployed AI projects than for stopping pilots, although shutting down production systems can involve more operational dependencies.
Ending a pilot may prevent further integration costs. Retiring production AI can require changes to cloud services, data pipelines and business workflows. Separate research on enterprise AI governance gaps also found shortcomings in visibility and traceability as organizations expanded AI deployments.
Gartner's published findings do not explain why production exit rules are less common or prescribe financial thresholds for ending deployments. They also cannot establish whether adopting such rules would independently improve returns.
Before deployment, CIOs should establish measurable success targets, cost-review schedules and clear authority to end funding. Exit plans should also account for system dependencies and service continuity, particularly where AI supports critical business operations.
Want to learn more AI tips, tricks, and prompting techniques? Let us teach you How to Talk to AI for free! Try our six-minute course at The Neuron Academy, our practical learning platform designed to help professionals use AI more confidently at work.
Learn a few simple ways to write better prompts and get more useful results from AI, or browse our other AI course for free for seven days. eWeek readers get free 7-day access. Check out all the lessons here →