Five Questions Every CFO Should Ask About AI Spend

In 2026, only 20% of finance teams can forecast their AI spend within ±10% accuracy. That means 80% of CFOs are approving technology budgets with a material line item they cannot predict.

This is not a technology problem. It is a visibility problem. And it starts with five questions that most AI deployments cannot currently answer.

1. WHAT AI SYSTEMS ARE WE RUNNING?

Not at the vendor level — at the seat level. Which employees have active licences? Which tools are they using? Which ones have been provisioned and forgotten? The AI Register is the starting document. Without it, the answer to every subsequent question is a guess.

2. WHAT IS EACH SEAT ACTUALLY COSTING US VS. WHAT IT COULD COST?

The gap between what you pay for a seat's plan tier and what the seat's actual consumption would cost on a lower tier is recoverable spend. Industry average: 35% of AI budget is recoverable through rightsizing without any capability reduction. On a $500,000 annual AI budget that is $175,000 available for reallocation — or for funding the next AI initiative.

3. CAN WE FORECAST NEXT MONTH'S SPEND WITHIN ±10%?

If the answer is no, the budgeting process is broken. Accurate forecasting requires real-time burn rate data per seat, per vendor, with trend analysis and planned initiative adjustments. It is achievable — but not from a vendor invoice that arrives at month-end.

4. WHICH TEAMS ARE GENERATING ROI FROM AI, AND WHICH ARE NOT?

AI spend without attribution is infrastructure spend with extra steps. If you cannot answer which department, project, or workflow each dollar of AI spend is supporting, you cannot build an ROI narrative for the board. Chargeback — allocating AI costs to the business units consuming them — is the governance maturity milestone that turns AI from a cost centre into an accountable investment.

5. WHAT IS OUR REGULATORY EXPOSURE?

As of June 2026, the EU AI Act, the US AI Executive Order, and Canadian sector guidance all require organizations to maintain documented AI inventories with cost data. The CFO's AI spend report is also, increasingly, a compliance document. Finance teams that have built spend visibility are ahead of the regulatory curve. Finance teams that haven't are building a liability.

These five questions are the AI FinOps maturity audit. If your organization can answer all five with current data — not last quarter's estimates, not spreadsheet approximations, but real-time per-seat visibility — your AI governance practice is ahead of 80% of the market.

If it can't, the gap between where you are and where you need to be is exactly the size of the problem.