A KPMG survey of 2,145 senior leaders across 20 countries found that 29% of corporate executives have difficulty understanding and controlling AI operating costs as they scale deployments, a challenge that has grown alongside a shift to usage‑based pricing models.

The survey, released in July 2026, indicates that many organizations are still building the capabilities needed to forecast, monitor, and manage AI spending effectively.

Nearly half of the respondents said they have rephased AI projects when costs exceeded expected value, and lower‑cost, high‑fidelity models have become the fastest‑growing influence on AI strategy, up seven percentage points from the first quarter.

Amazon plans to spend about $200 billion this year on AI capacity in its AWS data centers, a 50 % increase over the prior year, while Microsoft expects total capital expenditure of $190 billion, a 61 % rise.

Both companies are investing heavily in forward‑deployed engineering: Amazon has committed $1 billion to an AWS Forward Deployed Engineering team to help customers adopt AI agents and shorten deployment timelines, and Microsoft is providing $2.5 billion to establish Microsoft Frontier Company, which aims to let customers amplify their intelligence with AI.

KPMG notes that governance remains a concern, with questions about who is responsible for decisions made by statistical models that can produce erroneous outputs, commonly referred to as "hallucinations".

Executive accountability is important, but the consultancy says governance ultimately depends on day‑to‑day operating practices, including clear rules for employee intervention, cost ownership, output review, and incident handling; few organizations describe these practices as fully embedded.

Additionally, GPTZero claimed that a forensic review of KPMG's October 2025 report "Total Experience: Redefining Excellence in the Age of Agentic AI" found only five of 45 citations accurately referenced the intended source, with the remainder containing misleading or vague information; KPMG subsequently removed the report from some websites and issued a statement reaffirming its commitment to content accuracy and responsible AI use.

As AI spending accelerates, the ability of companies to manage costs and establish clear governance frameworks will determine the broader success of enterprise AI initiatives.