Dear CFOs: don't stop the employees who consume the most AI budget

Many companies are currently making the same mistake. As soon as an employee suddenly spends €1,000 or more per month on AI tokens, a discussion about cost control immediately arises. But that's exactly where it goes wrong. Because the employees who use the most AI budget are often precisely the people who are building completely new ways of working.

Many companies are currently making the same mistake. As soon as an employee suddenly spends €1,000 or more per month on AI tokens, a discussion about cost control immediately arises. Finance looks at dashboards, sees usage increasing, and thinks: this needs to be more efficient. But that's exactly where it goes wrong. Because the employees who use the most AI budget are often not the ones wasting resources. They are precisely the people who are building completely new ways of working within your organisation.

These are the employees who run multi-agent workflows while they sleep. People who deploy multiple AI systems simultaneously to conduct research, generate reports, produce content, write code, perform customer analyses, and automate processes. Where three departments were once needed, now a single AI-native operator manages entire chains from prompts and smart workflows.

Many CFOs still look at AI as if it were a traditional software license. A fixed cost that needs to be kept as low as possible. But AI doesn't work like SaaS did. AI isn't a tool you pay for per seat. AI is variable output. The more strategically an employee uses AI, the greater the output he or she can generate.

This also means that high token costs are not automatically bad. In fact, the opposite is often true. The employee who spends €1,500 a month on AI but accelerates the work of five people with it, is probably not your biggest cost centre. They might just be your most valuable employee.

Many organisations still focus too much on cost reduction instead of output maximisation. They see an AI invoice rise by €500 and immediately try to implement limits. But that same €500 could, in the meantime, be responsible for tens of thousands of euros in additional productivity, faster execution, fewer operational delays, and higher margins.

The danger is that companies will slow down their most innovative employees precisely when they are starting to build a competitive advantage. Because the heaviest AI users are often the first people within an organisation to understand how work is fundamentally changing. They don't use AI as a chatbot, but as infrastructure. They build systems where AI agents collaborate, tasks are distributed automatically, and output continuously flows without human intervention.

This is where the new form of leverage within companies arises. No longer just through headcount or management layers, but through employees who use technology to exponentially increase their capacity. One person can now deliver output that previously required an entire team. Not because that person works harder, but because AI fundamentally multiplies production capacity.

Therefore, it might be time to look at AI budgets differently. Software budgets are probably no longer the right model. AI is much more like variable production capacity. The smarter you support the right people, the greater the output that comes from your organisation.

The question for CFOs should therefore not be: “How do we get those token costs down?”

The real question is: “Which employees are creating 10x more value thanks to AI, and how do we ensure they can build even faster?”

How to make AI investments measurable

Smart management of AI budgets requires data, not slamming on the brakes. We help CFOs and management boards with AI Consultancy, with practical reports via AI at Work, and with an AI keynote that aligns the board.

• Best AI tools
• GEO: discoverable in AI search engines
• Writing AI prompts