Companies are ramping up AI investment even as most of their projects fail to reach production, according to Gartner's 2026 1H CIO Report. The survey of 11,000 chief information officers found that 83% of CEOs are boosting AI budgets, yet 59% of AI initiatives never make it to live deployment. The report highlights a growing disconnect between executive enthusiasm for artificial intelligence and the industry's ability to demonstrate concrete returns on those investments.
The struggle extends beyond deployment rates. Among the CIOs surveyed, 71% said they have difficulty choosing AI use cases that will produce measurable results. Token-based pricing structures mean expenses grow with usage, prompting vendors to introduce new controls — in early July, Anthropic rolled out spend alerts, model defaults, and analytics dashboards for Claude Enterprise to help organizations track consumption alongside output. Most companies still report only deployment statistics, with a rare few measuring actual AI outcomes and a handful tracking ROI, according to veteran technology executive Ashwin Rangan, who has held senior roles at Walmart and ICANN.
The challenge stems from treating AI like conventional software when it functions more as a general-purpose technology, similar to electricity, the report explains. The new cost dashboards act as electricity meters — they show how much power was consumed from the grid, but what's lacking is a companion tool to measure business outcomes, Rangan stated. Organizations gravitate toward the wrong metrics because the useful ones are difficult to implement, said Ara Kharazian, lead economist at spend-management platform Ramp. Tokens consumed, employees using AI, and lines of code written are available measures that reveal nothing about work quality. Mohan Sankararaman, CIO of First Horizon, a regional financial services firm based in Memphis, noted that the biggest risk for technology leaders is advancing without a clear objective. "Doing more AI experiments doesn't equate to more AI value," he said in the report.
Finance chiefs are already framing AI differently than technologists do, viewing it less as a software purchase and more as a capital-allocation decision, according to Hemant Kapadia, CFO at planning-software maker Anaplan. Boards are demanding evidence that AI is delivering return on investment as the technology accelerates, with board members growing increasingly frustrated that ROI isn't being demonstrated, Rangan warned in the report. In 2026, boards and c-suite executives want proof that AI is shifting from pilots to producing ROI, with results showing business value and proper governance for risk management, Sankararaman stated. Experts in the report advise picking two or three high-impact use cases, implementing them with alignment to business outcomes, and knowing when to walk away — in certain cases, the decision not to use AI may be the most appropriate one. IT leaders who can demonstrate what their token costs buy will continue earning investment, while those who can only report expenditures may find their AI budgets facing the same scrutiny cloud spending once did. The tolerance for experimentation without accountability is narrowing as finance teams demand the same rigor for AI that they apply to every other capital decision, and organizations that built their AI strategy around hype rather than discipline will face difficult conversations when budget reviews arrive.