As businesses commit significant budgets to new tools, boards increasingly expect finance leaders to explain whether the returns justify further spending. 

That places the CFO in an influential position. It also creates a governance challenge: the same function helps allocate AI investment, assesses financial returns and provides assurance on the reliability of its outputs.  

Deloitte’s 2027 Finance Trends research, based on 1,434 finance leaders at businesses with more than $1bn in revenue, found:  

  • 54% lead cross-enterprise AI and technology capital allocation 
  • 48% own AI trust, including the reliability and explainability of outputs 
  • 48% are responsible for AI spending and cost control 

The implication is clear: evaluating AI requires more than a calculation of investment returns. Finance teams need a framework for allocating capital, monitoring costs, assessing outcomes and sharing accountability. 

When finance approves an AI budget and reports its returns to the board, a conflict emerges. The person (or function) evaluating performance has played a role in deciding whether to proceed with the investment. 

Auditors call this a self-review risk. People and teams find it harder to reassess decisions objectively when they were responsible for making them. This becomes particularly important when the return on AI takes time to emerge.  

With 84% of CFOs yet to see a return on AI in finance, boards and finance teams may be assessing investments before their full impact is clear. 

Robert Checchia, CFO at Benzinga, described that longer-term accountability in Soldo’s CFO Playbook: “You don’t just have to make the recommendation [for AI], you have to, a year from now, explain what went well and what did not.” 

A stronger governance model separates advocacy from assurance wherever practical and brings other functions into the assessment process. 

Build shared accountability for AI decisions 

While finance offers capital allocation, cost management and evidence of return, AI governance shouldn’t rest with finance alone.  

  • Technology teams understand infrastructure and implementation.  
  • Business functions understand operational value.  
  • Risk and governance teams pose challenges across controls and compliance. 

Combining those perspectives gives boards a more complete basis for deciding where AI should be used. 

Deloitte found that 77% of finance leaders are comfortable allowing AI to make some decisions, while only 14% are comfortable allowing it to make important ones. This stark difference demonstrates the importance of controls. 

Checchia describes a similar model at Benzinga: “We empower everyone to start and build their own tools in AI. We also keep the guardrails so that nobody can go rogue with the tools.” 

Make AI costs visible before measuring returns 

An assessment of return depends on understanding the cost. AI makes that more difficult because pricing can vary by seat and usage, while some tools overlap without a central owner. 

Rohan Hewavisenti, CFO at Amnesty International, says: “One of the challenges with using generative AI is not knowing what the costs are, because you can’t pay by token. So, it’s really like a taxi meter without knowing how many miles you’re going to use, or without people seeing that taxi meter running.” 

Even when visible, usage can be difficult to allocate to products, teams or commercial outcomes. For finance, that creates a measurement problem. 

A practical AI cost-control framework should make it easier to: 

  • Identify AI subscriptions, licences and usage-based spend 
  • Assign ownership to tools and budgets 
  • Detect duplicate or overlapping services 
  • Allocate costs to the teams and projects using them 
  • Review spending against the budget 

Give boards a clearer answer on whether AI is worth it 

The board is right to keep the CFO close to AI investment. Finance plays an important role in determining affordability, scrutinising value and managing costs. Effective governance distributes responsibility across finance, technology, business teams, risk and governance to ensure success and impartiality. 

The central question remains whether the total investment is creating sufficient value.  

Soldo gives finance teams real-time visibility and controls over business spending, helping them understand where money is being spent and by whom. For businesses reviewing the cost of AI and other technology investments, clearer spend data can provide a stronger foundation for governance discussions.