Dive Brief:
- Nearly seven in 10 companies (68%) say their artificial intelligence initiatives exceeded budgets for at least part of the past year, with one-third (33%) reporting that overruns occurred mostly or always, AI security firm WitnessAI said in a report released Wednesday.
- The survey of 300 U.S. business executives also found that companies are struggling when it comes to proving that AI investments are paying off. Only 9% of respondents said that more than three-quarters of their AI initiatives delivered a measurable financial return.
- “AI initiatives are exceeding budgets, underperforming against ROI expectations, and running into governance bottlenecks that slow deployment without reducing risk,” the report said.
Dive Insight:
The findings highlight the growing challenge companies face in forecasting and managing AI costs, as they navigate unforeseen expenses, limited visibility into spending and rapidly evolving innovations and pricing models.
The challenge has only intensified as many organizations have begun deploying AI at scale.
The percentage of organizations orchestrating multiple AI agents across workflows doubled from 9% to 18% in the second quarter, according to a June report from KPMG.
At the same time, some companies are experimenting with what KPMG described as “unconventional and often counterproductive” ways to drive AI usage.
Forty-one percent of respondents said they would consider using a practice known as “token-maxxing,” which gamifies token consumption through incentives and leaderboards, according to the KPMG report. Twenty-two percent were opposed to the practice, and another 37% were neutral on the topic, the report found.
Meanwhile, the rise of “shadow AI,” where employees use AI applications outside formal IT and procurement processes, also contributes to the problem, according to WitnessAI.
IT and infrastructure was found to be the largest single source of shadow AI activity, at 47%, ahead of sales and business development (34%) and marketing and communications (33%).
“According to the findings, the department most responsible for governing AI use within the organization is also the department most likely to be operating outside its own policies,” WitnessAI said in its report.
Thirty percent of respondents reported that unmanaged or poorly governed AI usage has led to cost overruns, and 27% said it has resulted in delayed or canceled AI initiatives.
“The findings reveal a gap within financial risk management: organizations are deploying AI agents faster than they are building the financial visibility, ownership structures, and risk management practices needed to manage them responsibly,” WitnessAI said.
All of this comes as finance leaders also face increasing pressure to show that AI investments are delivering measurable returns.
A CloudZero survey found that 87% of finance leaders feel pressure to connect AI spending to business outcomes within the next year, but only 22% have already achieved that goal.
Part of the difficulty in showing ROI from AI adoption is that it rarely appears in a single metric, according to WitnessAI. The impact “surfaces instead as fragments scattered across budgets, productivity reports and individual business units, making a single ROI figure hard to pin down,” the report said.
