KPMG Survey Finds AI Investment Rising as Companies Struggle to Measure Returns

Key Takeaways

• AI Investment Continues to Rise: Organizations plan to invest an average of $210 million in AI over the next 12 months, up from $186 million in the first quarter of 2026, according to KPMG's survey of 2,131 senior leaders.
• Financial Oversight Remains Uneven: Although 61% review AI costs during project approval and 59% monitor operating costs, only 12% consistently assess AI value against cost across the organization.
• Regional AI Maturity Gaps Narrow: Sixty-four percent of organizations in the Americas are scaling AI or beyond, compared with 61% in Asia Pacific and 56% in EMEA, reducing the regional maturity gap by half since the first quarter.
• Formal Governance Is More Common Among Mature Organizations: Fifty-five percent of organizations operate a formal AI harness layer, rising to 86% among those reporting established returns on investment.
• Cybersecurity and Sovereignty Gain Priority: Eighty-six percent of organizations are adapting cybersecurity operations to address AI-related threats, while 72% formally consider model sovereignty in AI decisions.

Deep Dive

Companies are preparing to spend an average of $210 million on artificial intelligence over the next 12 months, a 13% increase since the beginning of the year. Yet only 12% consistently measure the value of their AI investments against what they cost, according to a new global survey from KPMG.

The findings expose a weakness in how businesses account for increasingly substantial technology investments. Most organizations have procedures for reviewing AI expenditure before projects are approved, and a majority monitor costs once systems are operating. Far fewer have established consistent methods for determining whether those investments are delivering sufficient returns.

KPMG's Global AI Pulse survey questioned 2,131 senior leaders across 20 countries between July 23 and August 26. Average planned AI investment rose from $186 million in the first quarter to $210 million in the third, while organizations continued expanding deployment and introducing more formal management controls.

The financial oversight gap is particularly apparent when comparing organizations at different stages of AI maturity. Among those reporting established returns on investment, 48% consistently evaluate AI's value against its cost, four times the proportion across the full survey.

Even among organizations that have demonstrated returns, however, consistent financial evaluation is far from universal.

The survey also found that regional differences in AI maturity are narrowing. Sixty-four percent of organizations in the Americas have reached the stage of scaling AI or progressed beyond it, compared with 61% in Asia Pacific and 56% in Europe, the Middle East and Africa.

The gap between the leading and trailing regions has fallen from 16 percentage points in the first quarter to eight in the third. As more organizations reach advanced stages of deployment, KPMG's research suggests that the management arrangements supporting AI are becoming an increasingly important measure of maturity.

Who Is Responsible for AI?

Responsibility for AI-informed decisions is moving further into executive management, although the arrangements differ considerably between organizations. Fifty-three percent of respondents said accountability rests at the C-suite level or above. Thirty-five percent assign responsibility to a named executive, while another 18% place it with the CEO or executive committee.

The figures come as AI agents become more widely used by employees. Significant employee adoption of agents rose to 34% in the third quarter, compared with 25% at the beginning of the year.

Their expanding use presents management with questions that become harder to resolve through individual project teams. Organizations must determine how AI systems are controlled across business functions, who is responsible for their use and how decisions involving those systems are supervised.

KPMG examined the adoption of what it calls an AI harness layer, a combination of controls and tools used to manage the relationship between AI models and their application in business operations. The arrangements can include formal cross-functional or enterprise-wide management structures intended to coordinate deployment and oversight.

Fifty-five percent of surveyed organizations operate a formal harness layer. Its prevalence increases sharply with reported AI maturity.

Among organizations still experimenting with AI, 31% have established a formal cross-functional or enterprise-wide management layer. The proportion rises to 58% among those scaling AI and 86% among organizations reporting established returns on investment.

Of the latter group, 65% have deployed such a layer across the enterprise, compared with just 10% of organizations in the experimentation stage. These differences are among the clearest in the survey. They show how much more common formal management arrangements are among organizations reporting established AI returns, although the research does not demonstrate that introducing those arrangements produces better financial results.

Lisa Heneghan, global chief digital officer at KPMG International, said organizations need senior leadership accountability and coordinated controls to manage AI across the business.

“What scale demands is not simply more AI, but the accountability and coordination to run it well and treat it as an enterprise-wide priority,” Heneghan said.

Security and Control Become Operational Concerns

The demands of operating AI at scale are also changing how organizations approach cybersecurity. Eighty-six percent of respondents said their organizations are adapting cybersecurity operating models and strengthening policies, processes and controls to address AI-accelerated threats.

Security receives particular attention among organizations already reporting established returns on AI investment. Seventy-one percent of that group include cybersecurity and data security among their AI budget priorities, compared with 36% of organizations still experimenting with the technology.

The survey measures reported priorities rather than the effectiveness of existing security arrangements. Nevertheless, the difference indicates that cybersecurity commands greater budget attention among organizations with more mature AI programs.

Control over the underlying technology is another growing concern.

Seventy-two percent of respondents said their organizations formally consider model sovereignty when making AI decisions. That includes 21% with an enterprise-wide strategy subject to regular review. Among organizations reporting established returns, 53% have such a strategy.

Model sovereignty encompasses questions about where AI models, data and intellectual property are hosted, who controls them and the extent to which organizations depend on outside providers.

Those considerations have practical consequences for businesses incorporating externally supplied models into their operations. Changes in access to a model, interruptions to a provider's service or restrictions affecting data can disrupt processes that have come to rely on the technology.

The survey places these dependencies alongside cybersecurity as matters of operational resilience. Organizations with established AI returns are more likely to have formal strategies for managing them, suggesting that questions of provider control and continuity become more prominent as deployment matures.

The Difficulty of Accounting for AI Spending

Financial management remains one of the least developed areas identified in KPMG's research. Sixty-one percent of organizations review AI costs during project approval, and 59% monitor costs during operation. Only 12% consistently assess the value of AI against its cost across the organization.

A company may know how much it spends on models, infrastructure and implementation without having a reliable measure of the business value those systems produce. That becomes a more consequential omission when AI is deployed across multiple departments, each with different operating costs and performance expectations.

The survey found a considerably higher rate of consistent value assessment among organizations reporting established returns, at 48%. The relationship is noteworthy, although it leaves open whether more disciplined measurement contributes to better returns or whether organizations with established AI programs are better equipped to measure them.

There is also a distinction between reporting measurable benefits and demonstrating that those benefits justify the investment. KPMG's research found that 89% of organizations report measurable value from AI, despite the much smaller proportion consistently evaluating that value against costs across the enterprise.

For boards and executive committees, the figures raise questions about the evidence supporting AI investment decisions. A project can meet its implementation timetable, achieve widespread employee adoption and produce measurable improvements without necessarily delivering the financial return anticipated when the expenditure was approved.

The growing use of AI agents may complicate those assessments further. Benefits can appear in the form of employee time saved, increased operational capacity or improvements in the quality of decisions, each of which requires different methods of evaluation. Measuring the cost of deployment is generally more straightforward than establishing the financial contribution of those changes.

KPMG's findings offer no evidence that higher AI expenditure guarantees stronger returns. Nor do they establish a particular management structure as the cause of successful deployment.

They do show that organizations reporting established returns are much more likely to have formal AI management layers, defined sovereignty strategies and consistent approaches to evaluating costs against benefits.

The narrowing regional maturity gap suggests that more businesses are acquiring the ability to deploy AI extensively. Whether their financial oversight can keep pace is a separate question, and one that is becoming harder for corporate leadership to leave unanswered as investment increases.

The GRC Report is your premier destination for the latest in governance, risk, and compliance news. As your reliable source for comprehensive coverage, we ensure you stay informed and ready to navigate the dynamic landscape of GRC. Beyond being a news source, the GRC Report represents a thriving community of professionals who, like you, are dedicated to GRC excellence. Explore our insightful articles and breaking news, and actively participate in the conversation to enhance your GRC journey