New Delhi [India], September 10 (ANI): The rapid adoption of agentic artificial intelligence (AI) is creating a new and difficult-to-predict cost burden for companies, forcing chief financial officers (CFOs) to rethink how AI spending is budgeted and monitored, according to a report by Gorilla Logic, a tech company.
The report said AI token costs, which were largely treated as a minor technology expense until recently, are increasingly emerging as a significant corporate budget issue as companies expand the use of AI agents for coding, software development and other business functions.
Unlike traditional software, where costs generally increase with the number of employees or seats, agentic AI costs are linked to usage.
It stated, “AI token costs have gone from a minor technology detail to a board-level budget problem in under a year. Twelve months ago, “token cost” wasn’t a line item on most CFOs’ radar -now it’s showing up”.
The report said this makes spending difficult to forecast because a single employee or autonomous AI agent can consume large amounts of tokens during complex or prolonged tasks.
AI tokens are the basic units of text that AI models use to read, process and generate information. A token can be a whole word, part of a word, or punctuation.
Every time a user asks an AI model a question or receives a response, tokens are consumed. For companies running AI at scale, these tokens are essentially the “units of usage” they pay for — meaning that higher token costs can directly increase their AI-related expenses.
The report estimated that allowing employees to use two agentic coding tools could take the potential cost to USD 3,000 per engineer per month, or USD 36,000 annually for one employee. With approximately 4,000 engineers, the report estimated that this could amount to about USD 144 million in AI spending.
Meta has also warned internally that its AI token costs could reach billions of dollars in 2026, according to the report.
The problem is particularly pronounced with agentic systems because AI agents can process thousands of tokens of background information with every prompt. Some models also use additional tokens for their reasoning processes, further increasing consumption.
According to Bob Graham, Chief Growth Officer at Gorilla Logic, “There is a lesson to be drawn here. If the biggest technology companies in the world can be surprised by this expense, so can anyone. Token consumption has become a genuinely new expense category, one that did not exist on any P&L two years ago. It does not behave like the software costs finance teams have spent decades learning to model and factor into traditional ROI calculations. Understanding why AI token costs behave so differently from traditional software spend is the first step toward getting them under control.”
The report said AI spending risks are increasingly divided into three broad categories. Business users in areas such as sales, marketing and operations generally use seat-based tools costing around USD 20 to USD 30 per user per month, making them relatively easy for finance teams to budget.
The report also highlighted the growing use of AI cost-management tools, including agent gateways and orchestration systems. Such systems can route tasks to cheaper models when appropriate, impose spending limits and provide companies with greater visibility into AI usage.
Another approach is the use of hybrid small language model (SLM) and large language model (LLM) architectures. The report said companies can use smaller, specialised models for narrow corporate tasks instead of relying on expensive frontier models for every request.
The report said the growing challenge for CFOs is therefore not simply controlling AI usage, but understanding which teams are driving spending, what budgets are being used and whether the AI expenditure is supporting a measurable business outcome. (ANI)