AI spending is moving beyond IT budgets, forcing CIOs, CFOs and business leaders to sort out who controls the money, technology decisions and accountability for results.
A recent survey from Boston Consulting Group (BCG) of more than 1,300 CxOs and senior leaders found that corporate AI spending has doubled in a year, to 3.3% of revenue.
That shift complicates the division of responsibility because CIOs and their IT teams are still often responsible for implementing and supporting AI deployments.
The BCG survey found that 80% of AI spending now sits outside the enterprise IT budget.
AI decision-making also increasingly involves the CFO. According to a September 2026 study from the IBM Institute for Business Value, conducted with Oxford Economics, 62% of 1,500 CFOs and equivalent senior finance leaders surveyed said they had taken on enterprise technology or AI strategy leadership responsibilities in recent years. Another 56% reported greater portfolio-management and capital reallocation authority, while 54% had taken on more responsibility for business model or growth strategy design.
But those figures don't capture the full dynamic around AI decision-making.
Executives in some organizations take a collaborative approach to AI strategy, bringing together the CIO, CFO, business unit heads and other leaders in formal AI councils or ad hoc meetings to decide where to invest, how much to spend and which department pays.
BCG's report, The Formula for Agentic AI Value, points to a unified approach to AI decision-making, saying future-built companies concentrate AI efforts in a single, coherent program and make the C-suite responsible for it.
Xero centralizes AI decisions while splitting costs
Xero, which makes online accounting software for businesses, takes that collaborative approach. Xero CFO Claire Bramley said executives decided in early 2026, as they approached the April start of the company's new fiscal year, to centralize AI decision-making and oversight with company leadership.
That leadership team includes Bramley, the chief legal officer, the global chief strategy officer and other executives. Bramley also oversees Enterprise Technology, which is led by Xero CIO Fred Laury.
I'm not of the opinion that the CFO or the CIO gets to decide on their own.
Claire BramleyCFO, Xero
"We review everything together," Bramley explained. "How you think about AI investments and their connection to business value, the decisions on what AI tools to offer, those should all be enterprise-wide discussions. I'm not of the opinion that the CFO or the CIO gets to decide on their own."
As part of that centralization, company leaders decided to use companywide AI platforms rather than let every function pick its own AI platform; Bramley said executives determined that was the "most efficient and effective way to use AI." Business units can request additional AI capabilities for their unique needs.
"We're trying to balance the need for central oversight, best practices, security and governance, but not dampen the creativity and innovation in the functional teams," Bramley added.
AI-related decisions rest with that central group, but AI budgets do not. Bramley said the executive leadership team decided to divide annual AI consumption costs between two budgets: the product and technology team and IT.
The company's finance team worked with IT and engineering to estimate overall AI cost per worker. Engineers on the product and technology team had significantly higher AI use, so Xero set a higher per-worker cost for engineers and a lower one for non-engineers. The product and technology team absorbs estimated consumption costs for engineers; IT carries consumption costs for other workers, as well as AI product licensing costs.
That division isn't meant to leave IT managing AI costs on its own, Bramley said. Xero wouldn't want IT to slash programs to cover higher-than-expected AI costs. The central leadership team monitors AI costs and adjusts when needed. For example, it found non-engineers were often using more complex -- and more expensive -- AI models than necessary, so it made less-complex models the default for those workers.
Bramley said that monitoring AI use and costs and being transparent about those figures are key to the company's approach.
Company leaders discussed distributing AI consumption costs across business functions, but determined that doing so would add unnecessary complexity without providing any benefit.
Despite how the spending is budgeted, Bramley said functional teams own what they use and are accountable for ensuring their AI use provides value.
Business outcomes shape decision rights
Khalid Khan, partner and Americas lead for the Digital and Analytics practice at Kearney, a global strategy and management consulting firm, said assigning ownership -- and the AI investments themselves -- to the business is essential for success.
"If AI is going to drive meaningful P&L impact, the leaders who own the P&L must own the change," Khan said. "Organizations need AI to move from experiments into how they win and serve customers and operate the business. That requires changes to workflows, capabilities, and incentives -- changes to the operating model of the enterprise or function. The investments increasingly belong with the business leaders who can make those changes and are accountable for the results."
Khan added, "AI is no longer just a technology deployment but a value creation lever across the enterprise P&L. Ensuring the budget moves to an accountable leader with the authority to change the business is what translates this to EBITDA [earnings before interest, taxes, depreciation and amortization] impact. "
Business and functional leaders should own the intended outcome, changes to how work gets done, adoption and realized business results, Khan said.
The CFO, however, should determine capital allocation and challenge the value thesis, with finance validating benefits and informing decisions to scale, redirect or stop funding, he said.
The CIO owns architecture, integration and shared technology capabilities, working with security, data and risk leaders to set deployment requirements, Khan added.
But neither should act in isolation. "Enterprises are realizing that value from AI comes from a more connected enterprise and I think the funding model is starting to reflect that shift," Khan said.
Ashwin Ballal, CIO at software company Freshworks, agrees with that approach.
Distributed adoption can't mean distributed accountability.
Ashwin BallalCIO, Freshworks
"Business teams should own the problems they're trying to solve and the outcomes they expect, but distributed adoption can't mean distributed accountability," Ballal explained.
"Decision rights should sit close to the business outcome. Business leaders should own the problem they are trying to solve and the results they expect. The CIO should own the technology framework, including architecture, integration, data, security and governance," he said. "This allows AI to scale responsibly. And the CFO should help determine whether the investment is actually creating value."
Mary K. Pratt is an award-winning freelance journalist specializing in enterprise IT, cybersecurity strategy and data management.