Companies need AI 'nutrition labels' to curb spending

Technology companies have consumed AI services at an unprecedented scale over the past year. Google alone now processes over 3.2 quadrillion tokens per month, a sevenfold increase compared to a year earlier. Uber, for instance, exhausted its entire 2026 AI budget by April.

The effort to curb excessive AI usage is justified, as companies lacking visibility into their consumption see poorer returns. Studies indicate that organizations with full insight into their AI operating costs are five times more likely to report established return on investment (ROI).

The core problem is that most users do not know how much energy and computing resources even the smallest prompt consumes. AI became easy to use before its cost became easy to understand. The proposal suggests that AI tools should provide employees with clear information about the true cost of each prompt before it is executed – a so-called 'nutrition label'.

While improvements in AI infrastructure and token limits may help control overconsumption, they do not solve the fundamental issue: employees cannot assess whether a prompt's computational effort is appropriate for its purpose. A single inefficient workflow can cost a company hundreds or even thousands of euros per month per employee, considering repeated attempts and unclear requests. Furthermore, energy consumption is a significant concern.

In the future, companies will demand greater transparency from AI models. AI service providers are likely to begin offering information on the number of tokens, costs, and computational intensity for prompts. This would help employees assess a prompt's cost in relation to its value and, at the same time, enable companies to track their AI expenditures more effectively.