Artificial intelligence may have become a core part of business operations, but many companies still do not fully understand how much they are spending on it.
A new survey by consulting firm KPMG has found that only 26 per cent of organisations have a comprehensive view of their AI costs, highlighting a growing challenge as businesses rapidly increase their reliance on AI-powered tools and services.
The findings come at a time when companies across industries are reassessing their AI spending amid concerns that costs are rising faster than anticipated.
Most Companies Lack Visibility Into AI Spending
According to the KPMG survey, only about one in four organisations has a complete understanding of how AI usage is being billed and managed.
The report found that 50 per cent of companies have only partial visibility into their AI spending, while 22 per cent said they have little to no understanding of costs until invoices arrive.
The lack of transparency has become increasingly problematic as AI providers shift towards usage-based pricing models.
“It’s a new resource that needs to be managed that didn’t exist quite that way, and we’re seeing exponential growth,” KPMG Global Head of AI Steve Chase was quoted as saying by The Wall Street Journal.
The findings suggest that while companies have moved quickly to adopt AI technologies, many have not yet developed the systems needed to effectively monitor and control costs.
Token-Based Pricing Driving Up Costs
A major reason behind the growing expense is the industry’s move towards token-based billing.
Companies such as OpenAI and Anthropic increasingly charge customers based on token consumption, with tokens serving as the basic unit used to process AI prompts and responses.
Under this model, businesses purchase access for employees but face additional charges once usage exceeds allocated limits.
As employees use AI tools more frequently and for increasingly complex tasks, token consumption can rise dramatically.
According to Chase, KPMG has worked with organisations that exhausted annual cloud and token budgets within just a few months. One client reportedly saw token consumption increase sixfold in a short period.
The challenge is particularly acute as AI agents become more capable and are tasked with handling longer workflows and larger volumes of data.
Businesses Begin To Pull Back
The report comes amid signs that some companies are beginning to moderate their AI usage due to mounting costs.
Ride-hailing giant Uber reportedly exhausted its annual AI budget within months and subsequently imposed a company-wide cap of approximately $1,500 per employee for AI-related spending.
Other organisations are also reassessing how AI tools are deployed internally as they seek clearer evidence of productivity gains and return on investment.
Even AI industry leaders acknowledge that cost management has become a growing concern.
OpenAI CEO Sam Altman recently remarked that companies are increasingly discovering that AI expenses can quickly exceed expectations.
The debate over AI spending is unfolding as major technology firms continue investing heavily in infrastructure. Google recently announced plans to raise billions of dollars to support AI development, while Anthropic is reportedly preparing for a public listing. OpenAI is also widely expected to pursue an IPO in the future.
The Next Phase Of AI Adoption
The findings point to a broader shift in how businesses view artificial intelligence.
For much of the past two years, the focus was on adoption and experimentation. Now, organisations are entering a phase where efficiency, governance and cost control are becoming just as important as innovation.
As AI usage continues to expand across software development, customer service, research and everyday workplace tasks, companies may need to develop more sophisticated methods for monitoring consumption and managing budgets.
The challenge for businesses is no longer simply deciding whether to use AI. It is figuring out how to use it effectively without allowing costs to spiral out of control.
