Pricing for AI Use: Top Executives in Disarray
According to a KPMG survey, nearly one-third of business leaders find it challenging to grasp and control operational expenses when scaling up AI technologies in their companies.
Confusion Over Usage-Based Billing
Recently, Anthropic, OpenAI, and GitHub have transitioned several services from flat-rate subscriptions to usage-based billing. KPMG noted, “As these pricing models gain traction, many organizations are still developing the capabilities needed to effectively project, monitor, and manage their AI expenditures.”
Executives Face Challenges with AI Expenses
A survey involving 2,145 senior leaders from 20 nations revealed that 29 percent struggle to comprehend their operational costs as they expand their AI initiatives. Additionally, a third of senior corporate leaders pointed to a lack of understanding regarding AI expenses and economics as a barrier to deploying AI technologies.
Reassessing AI Implementations
Companies are reassessing their AI strategies amid evolving cost frameworks and increasing charges. The research indicated that nearly half of organizations have postponed AI rollouts when costs exceeded anticipated benefits.
Where’s the Value in AI?
Low-cost, high-quality models are the most rapidly expanding factor in AI strategy, rising 7 percentage points since Q1. “These measures do not indicate diminished confidence in AI. Instead, they reflect a growing readiness to assess where AI delivers significant value and where it does not. Organizations seem more focused on channeling investment into areas with the highest expected returns,” stated the report.
Amazon and Microsoft Invest Heavily in AI
Amazon intends to spend approximately $200 billion this year, primarily to enhance AI capacity in its AWS data centers, marking a 50 percent increase from the previous year. Meanwhile, Microsoft’s total capital expenditure is projected to hit $190 billion, reflecting a 61 percent rise over last year.
Challenges in AI Governance
The KPMG report highlights ongoing challenges in AI governance: specifically, the dilemma of who is accountable for decisions made by statistical models that may produce erroneous outcomes – or “hallucinations,” as technology vendors prefer to say.
Governance and Accountability
KPMG stated that while executive accountability is crucial, “governance ultimately succeeds or fails based on daily operational practices.” Organizations require clear guidelines regarding employee intervention, ownership of AI-related expenses, review processes for AI outputs, and protocols for system failures.
KPMG’s Citation Error
It seems this consultancy and service giant might be speaking from experience. Last month, research group GPTZero claimed a detailed examination of KPMG’s October 2025 report revealed that only five out of 45 citations accurately referenced the original source. The others included errors ranging from misleading or fabricated information to references that were too ambiguous to confirm.
KPMG’s Reaction
KPMG later took down the report from some of its websites and issued a statement. “KPMG International prioritizes the accuracy and integrity of its published materials. The report has been removed, and we are currently reviewing the circumstances surrounding its publication.”
Conclusion: AI Sparks Concerns and Costs
It appears the executives are just as perplexed by AI pricing as someone at a vegan festival! With all this money exchanging hands, perhaps they will figure it out before we require an AI to oversee our AI.
