The past two years have seen AI move from boardroom promise to enterprise deployment at scale. Post-Enterprise Connect 2026, the conversation had clearly moved from adopting AI to whether AI is actually delivering. Vendors have flooded the market with productivity calculators quantifying the gains: ten minutes saved per meeting, two hours reclaimed per week, entire workflows stripped of manual effort. On paper, the math looks compelling. In practice, CFOs are beginning to question their ROI.
But it’s not the tools themselves that are in question. AI can draft emails in seconds, summarize hour-long calls into three bullet points, handle customer inquiries without human intervention, and automate the administrative tasks that once consumed entire mornings. The technology is genuinely capable. But capability and ROI are not the same thing. Saving time and creating value are only equivalent when the time saved is put back to work in a meaningful way.
That is the central problem facing enterprise AI deployments right now. If an employee saves an hour a week through AI assistance but spends it in an additional meeting, doom-scrolling, or clearing a backlog of low-priority emails, the organizational return on investment is exactly zero. The real question for CIOs and CFOs heading into the second half of the decade is not how much time AI is saving; it is where that time is going.
The Efficiency Trap: Why Time Saved Is Not the Same as Value Created
The assumption that saved time automatically converts to business value is what researchers are calling the "Time Saved Fallacy." Its roots go back further than generative AI. Economist Erik Brynjolfsson identified the Productivity Paradox in 1993, the observation that despite massive IT investment, aggregate productivity growth remained stubbornly flat. The same dynamic is reasserting itself now. Enterprise AI investments are projected to reach $644 billion by 2025, according to Gartner. Larridin’s report released in the same year said many organizations are struggling to show measurable business outcomes beyond initial efficiency metrics.
Part of the explanation lies in what economists call the Jevons Paradox. Named after William Stanley Jevons, who in 1865 observed that more efficient steam engines led to increased coal consumption rather than less, the paradox applies directly to AI-enabled workplaces. When AI makes it easier to produce a meeting summary, the friction of scheduling a meeting disappears. The result is not fewer meetings; it is more of them. Data from the Microsoft Work Trend Index 2024 puts the average knowledge worker between 11.3 and 14.8 hours per week in meetings, representing up to 35% of a standard workweek, even as AI summarization tools have multiplied. Estimates suggest unnecessary meetings cost U.S. companies between $37 billion and $259 billion annually in lost productivity.
Dippu Singh, Leader of Emerging Technologies at Fujitsu, sees this pattern playing out at scale:
"We are seeing a classic Jevons Paradox."
Going on to explain, Dippu says "As the cost of a resource, in this case the effort required to document and summarize a meeting, decreases, the consumption of that resource increases. Because AI makes meetings easier to digest, the organizational friction to schedule them disappears. This leads to AI agent sprawl, where the volume of low-value interactions actually inflates because the pain of the meeting has been artificially dulled."
The same dynamic appears in written communication. When an AI writing assistant reduces the time to draft an email from twenty minutes to thirty seconds, the result is rarely fewer emails; it is a dramatic increase in volume, with each message still requiring attention, judgment, and a response on the receiving end. The bottleneck does not disappear; it migrates. Organizations that deploy productivity tools without redesigning the workflows around them often find themselves generating more work, not less. The efficiency gain gets absorbed by what researchers call "organizational slack," the low-value administrative void that expands to fill available capacity.
Closing the Reinvestment Gap: A Framework That Actually Works
Identifying the problem is the easier half. The harder question is how to close the Reinvestment Gap, the space between time being saved and that time generating measurable business value. Rob Loake, Country Manager UK and Ireland at Wildix, argues the answer starts with asking a fundamentally different question:
"True return on investment from AI doesn’t come from simply doing the same things faster."
Instead, Loake argues "It comes from fundamentally changing what your teams have the capacity to achieve. The C-suite should be wary of vanity metrics and instead focus on a more strategic question: Where is the saved time being reinvested?"
