Quick Answer
The office that owns the AI budget usually has the least visibility into it. In CloudZero's survey of 252 senior finance leaders, 65% of CFO-led AI programs sit at the two weakest allocation tiers, against 43% of engineering-led ones. Forecasting and outcome measurement are level between the two. Allocation is the only capability that diverges, because it depends on tagging inside systems finance does not run.
The office most likely to own the AI budget is the one least able to see inside it. In CloudZero’s survey of 252 senior finance leaders, 65% of CFO-led AI programs sit at the two weakest allocation tiers, against 43% of engineering-led ones.
CFO-led programs forecast just as well as engineering-led ones, measure outcomes just as well, and see their spend data just as quickly. Where they fall behind is allocation, which depends on systems finance does not run.
What is AI cost visibility?
AI cost visibility is the ability to see what AI spending consists of below a single line on a bill. This post asks a narrower question, which function can actually see it, and the survey measured that as three separable capabilities:
- allocation: the spend that can be attributed to a team, product, feature, or customer;
- forecasting: the total that can be predicted;
- outcome measurement: the return that can be shown.
An organization can be strong at two of these and weak at the third, which is why a single AI maturity score can bury the problem rather than surface it. Across the survey, allocation is the weak third, and it is weakest where the budget sits.
2026 State of AI Spend Report
Why Finance Caps What It Can’t See
Learn how finance leaders are managing AI spend, and how they’re approaching FY2027
Why the CFO’s office has the weakest AI cost visibility
Finance can own the number, forecast it, and build a board story around it from the top down. It cannot tag a workload it does not run.
On every other capability besides allocation, finance and engineering are roughly comparable: 52% of CFO-led programs forecast strongly against 43% of engineering-led ones, and 59% measure outcomes strongly against 64%. Neither of those differences is statistically significant. Allocation is the only capability where the two genuinely diverge.
The mechanism shows up in who holds the data.
In CFO-led programs, finance tracks the AI spend 53% of the time and engineering does 16% of the time.
In engineering-led programs, that reverses: engineering tracks it 48% of the time against finance at 24%.
Allocation depends on tagging and cost-center mapping inside the systems engineering runs, so whoever runs those systems ends up able to see the spend, regardless of who owns the budget.
Most CFO-led programs have committed to the initiative without the means to see what it costs.
What poor AI cost visibility costs the people who own it
The function most likely to own AI spend is also the one that finds it hardest to defend.
62% of CFO-office-driven programs agree that AI spend is harder to defend than any other line in the budget, against 40% of everyone else. Agreement runs at 49% overall, so the complaint is not unique to finance, just concentrated there.
That is the same group already carrying the weakest allocation. They own the budget and they cannot see inside it. They also carry the heaviest share of the defense burden.
How much AI spend companies cannot account for
Most finance teams cannot break AI spend past a rough category.
54% of finance teams sit on the bottom two rungs of the allocation ladder, seeing a total or a rough split and nothing underneath it. Only 12% can break AI spend down finely enough to act on it inside the quarter.
It is worth knowing what that means in money. Cross-referencing reported AI and cloud spend against the share attributable to AI, the 135 companies at the two weakest allocation tiers are carrying at least $385 million in combined annual AI spend as a number they cannot decompose. These are not small companies. Their reported spend sits close to that of the companies at the two strongest tiers.
Allocation is also the only one of the three capabilities where FY2027 demands more than finance currently has. Forecasting is already slightly ahead of where teams say they will need to be. Outcome measurement is further ahead still. AI cost allocation is the gap.
What actually closes the AI cost visibility gap
Reassigning the budget does not close it. The survey is clear that moving ownership between finance and engineering does not move the capability, because the capability lives in the systems rather than in the org chart.
What closes it is a shared view. Finance owns the number and engineering owns the systems that could explain it. Both are looking at AI spend, and they are not looking at the same version of it. The organizations that escape this are the ones where one view of the same spend is readable by whoever owns the budget and whoever runs the infrastructure, as opposed to a reporting layer added afterwards.
Independent research points the same direction. KPMG’s Global AI Pulse Q2 2026, covering 2,145 senior leaders across 20 countries and territories, found that only 35% have full visibility into their AI operating costs, and that those who do see their costs are five times likelier to report established AI ROI, 15% against 3%.
Frequently asked questions about AI cost visibility
Building AI cost visibility where the budget actually sits
Owning the AI budget and being able to see inside it are two different things, and most finance teams have bought only the first.
The organizations that have both spend what their peers spend. They know which part of it is working, so they can put more behind that part.
CloudZero’s full research covers all 13 findings from 252 senior finance leaders, with the complete methodology and every response distribution published alongside it.
Read the full report: Why Finance Caps What It Can’t See
