Surveys: Most Finance Chiefs Say AI Costs Ran Over Budget, and Most Still Plan to Spend More

Corporate finance chiefs were promised that AI would cut costs. A batch of surveys released between September and early October 2026 says the bills are running past forecasts instead.

The common thread is pricing. Many AI vendors now charge by usage, so a bill depends on how much text, code or data the models process. That makes a budget set in January a guess by March.

The numbers

Pigment's Q3 CFO Index, released Oct. 7, surveyed 2,000 CFOs and finance executives in the U.S., UK, France and Germany. It found 83% said consumption-based AI costs pushed spending above expectations. Just 13% stayed within budget.

U.S. Bank's September survey found 51% of U.S. finance leaders said AI spending over the prior year exceeded budget. A Mavvrik report found 25% of firms delayed or canceled AI projects because of unexpected costs.

SpendHound, a YipitData subsidiary, surveyed 172 finance and procurement leaders and drew on spending data from more than 1,300 companies. It found 46% exceeded AI budgets in 2025, versus 37% for traditional finance and accounting software.

It also found 57% of respondents are not confident they are paying a fair price for AI tools, or don't know. Nineteen percent believe they are overpaying.

Spending more anyway

CloudZero's FY2027 survey shows the contradiction. 87% of finance leaders plan to raise AI budgets next fiscal year, and 45% plan increases above 20%. In the same survey, 61% admitted current AI spending exceeds what they can justify by outcomes.

Deloitte's September research found 60% of finance leaders expect AI costs and complexity to rise substantially through 2027. SpendHound found 81% expect spending on general-purpose AI to increase in 2026.

Gartner estimates worldwide AI spending will hit about $2.7 trillion in 2026, up 49.5% from a year earlier. Its forecast puts AI software at $461.6 billion and AI agents and assistants at $29.2 billion.

The savings have not shown up either. Only 7% of SpendHound's respondents said general-purpose AI reduced their traditional finance software spending. Sixty-two percent saw no change. Seventeen percent said that spending went up.

Finance takes the wheel

Bloomberg reported that finance executives are cutting AI spending and pulling budget authority into a central function. That means individual departments get less freedom to sign up for tools on their own.

Companies are also adopting stage-gate funding, which releases money in tranches only after a project hits set milestones. Finance teams are demanding measurable return on investment before approving more.

The rationing is reportedly reaching big names. Uber is said to have used its entire 2026 AI budget by April. Reuters reported in June that Amazon and other companies had put spending caps in place, and Cisco is said to have moved to curb non-essential AI tools. None of the companies named is quoted on those points in the surveys themselves.

The confidence side

Not everyone is alarmed. Pigment's respondents rated their confidence in AI's value at an average of 8.2 out of 10. U.S. Bank found 69% of finance leaders reported commercial benefits from their AI investments.

The most experienced users sound the least worried. Pigment says 31% of AI-mature companies report no significant cost-management problems, nearly double the average. Yet 35% of those same AI-mature companies are more than 50% over budget.

Pigment's CFO Susan Phan framed the findings as a planning problem. "When nearly half of leaders say elections are derailing their forecasts, and AI costs are blowing past expectations, a static budget becomes obsolete on day one," she said. Her answer is continuous scenario planning. Pigment sells finance planning software, so it has a commercial stake in that conclusion.

The same caution applies to the rest of the data. These are survey results and vendor reports, not audited company spending. Self-reported ROI is only as good as the measurement behind it. Only 51% of SpendHound's respondents said their AI investments are producing measurable ROI.

What happens next

Pigment's data points to a squeeze. It found the average amount companies are exceeding AI budgets is greater than the average increase respondents plan to add to AI budgets next year. In Pigment's reading, that suggests companies are hitting a ceiling on AI spending growth.

Whether the planned increases cover the overruns is the open question for 2027. Deloitte's respondents expect costs to keep climbing, and 61% of CloudZero's say they already cannot fully justify what they spend.