Worldwide spending on AI is forecast to total $2.59 trillion this year, a 47% increase year-over-year, according to recent estimates from Gartner. Much of it has been by vendors and hyperscalers, but enterprises are gearing up their budgets for AI as well. Gartner expects AI spending to increase 35% by next year, reaching an astounding $3.5 trillion.
The question becomes: is it, or will it soon be, delivering at least $3 trillion in value? So far, all this AI investment is not meeting its promise, at least at this time, a recent survey of 951 companies out of Bain and Company suggests.
Nearly 40% of companies that measured AI cost savings landed below 10% savings, despite their initial targets of gains of at least 11% to 20%, the consultancy finds in its research. Still, yet 90% are ramping up their budgets again, this time to build and deploy agents “that will operate with even greater autonomy, complexity, and consequence.”
It’s an annual ritual to see boards expand their automation budgets, the study’s authors, Michael Heric, Purna Doddapaneni, and Antoine Debarre, all with Bain, point out. “Every year, CEOs sign off on the next wave—robotic process automation, then machine learning, then generative AI, now agents. And every year, the savings fall short.”
This falling short isn’t enough to trigger alarms, they add. The gap is “not enough to kill the programs, but consistently, quietly, and by a margin that should be making executives uncomfortable.”
But hey, it’s AI, right? “The technology worked. The value didn’t arrive,” they write.
What’s at issue? Heric and his co-authors point out the following roadblocks to AI value:
AI automation isn’t autonomous – human workforces are still required. “Only 7% of companies are running fully autonomous agents in production today.”
Companies are making circular bets on automation. “When asked how they plan to fund generative AI and agentic AI investments, 44% of companies—the largest group—cited savings from prior automation programs. Self-funding the next wave from past returns sounds like discipline. In reality, it is a circular bet with a structural leak. The prior wave underdelivered. The savings pool is smaller than assumed."
Data is still the wall—and it’s not coming down. “Data access and integration remains the top barrier to AI progress, despite heavy investments in data modernization,” the researchers find. It’s the single biggest barrier to AI progress, cited by 41% of respondents.
The Bain team makes the following recommendations to get ahead of the curve with AI return on investment:
Don’t pave cowpaths with AI. “The question to ask before any AI program is approved is not ‘Where can we apply AI?’ but ‘If we were designing this process from scratch today, what would it look like?’”
Look at previous tech ROI. “CFOs should audit actual returns from prior automation programs, not projected returns. If the previous program delivered 60% of its targeted savings, size the current investment accordingly.”
Put someone in charge. Governance tends to be “split almost evenly between IT, business functions, and central teams, with no clear owner in most organizations," the Bain co-authors point out. "When an agent makes a consequential error in a production system, accountability cannot be improvised in the moment. It must be established in advance.”
Use AI itself to solve the data problem. Automate "one repeatable, high-value workflow where humans are currently pulling data manually, consolidating spreadsheets, and producing reports, and replacing that entire sequence with AI.”
Redesign employees’ roles. “In an agent-led operating model, employees are no longer moving work along a process; they are orchestrating, supervising, and making the high judgment calls that agents cannot. It requires deliberate investment in role redesign, new ways of working, and change management.”
Measure outcomes at the enterprise level, not the program level. “What matters for the enterprise is whether AI investment is producing better decisions, faster responses, and stronger customer outcomes.”
