Gartner Survey: Fewer Than a Quarter of Enterprises Have Scaled AI Across Multiple Business Units

According to survey data released by Gartner, fewer than a quarter of enterprises have successfully scaled AI across multiple business units. Gartner surveyed more than 1,300 leaders from organizations with revenues of $50 million or more between January and April of this year.

Slow rollout hasn't dampened investment plans. Eighty-five percent of tech leaders said they plan to increase AI investment next year. Meanwhile, about 11% of surveyed organizations said they don't know how much they spent on AI in 2025. Tina Nunno, Distinguished VP Analyst and Fellow at Gartner, said in the report that low visibility into AI spending increases the risk of missing out on returns, and that without disciplined measurement tied directly to business outcomes, organizations risk wasted resources and unmet expectations.

Other surveys point in the same direction

In an August report from Infosys, nearly three-quarters of senior executives said less than 25% of their AI pilots had been successfully scaled, and two-thirds said they struggled to measure the ROI generated by AI. In an August report from Deloitte, only one in five senior managers and C-suite executives said they were ready to redesign work processes so that AI agents could execute autonomously.

What the profitable companies did

In Gartner's data, companies with higher returns were the ones that continuously tracked the ROI of their AI initiatives, treated those initiatives as a single value portfolio, and evaluated performance on a regular basis. Companies that followed this framework and discontinued underperforming initiatives reported positive returns on 81% of their AI projects.

VP Nunno noted that tech leaders can get swept up in the AI hype cycle. The most widely pursued use cases—such as cybersecurity, threat detection, and IT service desk automation—don't always deliver the highest ROI. The top three use cases Gartner identified as generating positive returns were intelligent IT asset and cost optimization, synthetic data generation, and automated code generation and refactoring. Nunno said the use cases that deliver the greatest value to a company are those targeted at specific business needs and grounded in data where necessary, and that CIOs should prioritize use cases where the business case and financial case align, given their organization's circumstances, AI maturity, and business needs.

Implications for AX at Korean Companies

Because this survey targeted organizations with revenue of $50 million or more, companies in Korea below that threshold should look at the underlying structure rather than the raw numbers. The structure the survey reveals is simple: some organizations have decided to increase investment but can't count their spending, while the organizations that generated returns had two disciplines in place—measurement and the willingness to stop. For organizations where a single pilot represents a significant share of the annual budget, these two disciplines carry even more weight than they do for large enterprises.

Another point concerns the choice of use cases. Gartner's finding that the most widely pursued use cases differ from the ones that actually generate returns is a warning against choosing based on what others are doing. If you're considering adoption, the right order is to first identify which task's cost or throughput is the bottleneck, then select a use case that addresses that specific bottleneck.

There's one thing worth checking in your own organization this week: count how many of your ongoing AI initiatives have documented performance metrics and stop conditions. A project without metrics can't prove success even if it succeeds, and a project without stop conditions has no basis for being halted even if it fails.

Source: Fewer than 25% of enterprises have scaled AI successfully