Ninety-two percent of technology executives say their organization tracks the financial impact of AI-generated work. Two percent say more than half of that work is actually recorded as a business outcome.
That number comes from Lanai’s 2026 AI Labor Report, published June 9 and covered in Forbes on June 24. Two hundred U.S. technology executives at organizations with 1,000 or more employees. Not a small sample. Not an outlier industry. The gap between “we track it” and “we can prove it produced something” is ninety points wide.
This is not a technology failure. The AI works. The models produce output. Employees use the tools daily. The failure is that nobody built the system to record what that work actually produced. No attribution methodology. No financial linkage. No line on the P&L that says “AI contributed this.”
Lanai’s CEO Lexi Reese calls the pattern “AI labor orphaning.” AI does the work. The work never enters the ledger. The output is real, but the accounting system has no category for it.
And here is why that matters right now: 79% of those same executives worry their AI budgets will be cut because the spending cannot be tied to revenue or profit.
They are not wrong to worry.
The Budget Cycle That Eats Itself
Bain released its 2026 Automation and AI Pathfinder Survey the same month. Nine hundred fifty-one global companies. The finding that should stop every executive mid-sentence: nearly 40% of companies that measured AI cost savings landed below 10%, despite targeting 11% to 20%.
The technology delivered. The savings did not arrive at the scale the business case promised.
And yet 90% of those same companies are increasing their budgets again. This time for AI agents that operate with greater autonomy, greater complexity, and greater consequence.
Where is the money coming from? Forty-four percent of companies said they are funding the next wave of AI from savings generated by the previous wave. Self-funding sounds like financial discipline. In practice, it is a circular bet. The prior wave underdelivered. The savings pool is smaller than what was projected. And the investment case for the current wave was sized against projections, not actuals.
Bain’s researchers put it directly: companies that do not validate their reinvestment math against what automation actually returned are compounding risk, not managing it.
The Invisible Worker Problem
The Lanai report surfaces something more specific than a budget gap. It surfaces an attribution gap.
Eighty-seven percent of organizations credit AI-assisted output entirely to the human employee, sometimes or always. Performance reviews, bonus pools, promotion decisions — all built on work where the machine contribution is invisible. Nobody is hiding it. No system exists to record it.
Eighty-eight percent have no formal methodology for attributing business outcomes to AI. Forty-three percent assume that if AI was involved, it contributed. Thirty-eight percent rely on educated guesses. Twelve percent have a defensible answer. The rest are guessing.
That is not a measurement problem you solve with a dashboard. That is a structural gap in how the company describes itself on paper.
Ninety percent of organizations have no single function responsible for tracking AI return on investment. Nobody owns the number. Finance thinks IT has it. IT thinks the business units do. When the CFO asks whether AI spending is creating measurable value, the room goes quiet.
The Operations Problem Nobody Wants to Name
BCG’s 2025 research found just 5% of companies achieving AI value at scale. McKinsey’s State of AI survey found 88% of organizations using AI regularly, but only 39% attributing any EBIT impact to it. Six percent qualified as high performers capturing real enterprise value.
The pattern across all three studies is the same: AI adoption accelerated. The operations layer to capture value from that adoption did not.
This is the part most companies skip. They buy the technology. They roll out the tools. They approve the budget. And they never redesign the systems that track, attribute, and govern the work those tools produce. The financial systems, the performance systems, the accountability structures.
One example from Lanai’s report makes the cost concrete. A finance team measured two groups running identical workflows for 30 days. Same work, same output quality. The only variable was which AI model each group happened to use, a choice nobody had made deliberately. One group’s bill: $52,015. The other’s: $13,007. Same work product. Fixing that single default saved roughly 5% of the team’s annual spend, straight to the bottom line.
That is not an AI problem. That is an operations problem. The technology worked fine in both cases. The difference was whether anyone was watching.
What This Means This Week
If your organization is in the 90% that increased AI budgets this year, ask one question before the next cycle: can finance tie that spending to a business outcome with evidence, not estimates?
If the answer is no, you do not have an AI problem. You have an accounting problem that happens to involve AI. And that accounting problem is about to become a budget problem, because CFOs are done approving spend on faith.
Bain’s top performers did not find better technology. They treated data access, governance, and process redesign as leadership problems before they were budget problems. They built the attribution layer before anyone asked them to.
Everyone else is running a company on numbers that leave out an entire category of worker. That gap does not close on its own. It compounds.