Companies turn to AI tokenomics to track AI costs and returns

In short: More companies are using “AI tokenomics” to measure how much AI use costs, and whether that spending is paying off.

What's going on

“AI tokenomics” is an emerging discipline that treats tokens as the basic unit for measuring AI usage. A token is a small piece of text that an AI system reads or writes. It might be a whole word, part of a word, punctuation, or even a space, depending on the system.

This matters because many AI services now charge by tokens, not by the number of employees using the tool or by time. It is similar to paying for electricity by the kilowatt-hour instead of paying a flat monthly fee. The more text you send into the AI and the more it writes back, the more tokens you use, and the higher the bill.

Companies are starting to use token counts to model and predict costs across common tasks, like customer support replies, writing marketing copy, or helping programmers write code. They also use token tracking to compare different AI models, for example using a cheaper model for simple, high-volume work and a more expensive one for harder tasks. Groups such as the FinOps Foundation describe this as “FinOps for AI,” meaning a way to bring budgeting and accountability practices to AI usage.

What to watch

Tokens are a good way to measure what AI costs, but they do not automatically show what AI is worth. The next step for many organizations will be tying token spending to outcomes they care about, like faster response times, fewer support tickets, higher sales, or lower labor costs, and then cutting back on AI uses that do not show clear benefits.

Source: NYTimes