What is token maxxing and why are companies worried about it?

You've probably come across the term "token maxxing" in recent discussions about generative AI. The phrase describes a growing tendency among organisations to view higher AI usage as a sign of greater productivity. In practice, it refers to measuring success through the number of AI tokens consumed rather than the business value generated. Tokens are the units large language models use to process text and information. Every prompt, response and AI-powered workflow consumes tokens, making them a key factor in how AI services are priced. From AI adoption to AI returnsDuring the early phase of enterprise AI adoption, many companies focused on encouraging employees to use AI tools as widely as possible. Higher usage was often seen as evidence that AI initiatives were succeeding. As organisations move from pilot projects to large-scale deployments, however, attention is shifting to the cost implications of that approach. Industry executives say token consumption is an input cost rather than a business outcome. As AI deployments expand across thousands of employees and autonomous AI agents, enterprises are increasingly evaluating whether spending on tokens is translating into measurable gains in productivity, efficiency or revenue. Why companies are tracking tokens more closelyThe concern has become more pronounced as companies deploy agentic AI systems capable of handling complex and long-running tasks. Such systems can consume significantly more tokens than traditional chatbot interactions, creating a new layer of cost management challenges.In response, technology services firms are introducing tools that monitor token usage alongside workflow performance and business results. Some are also developing token metering and optimisation capabilities to help customers understand where AI spending is creating value and where it may be generating unnecessary costs.The discussion is also influencing commercial models. Companies are increasingly exploring outcome-based pricing structures that tie AI spending to business goals instead of pure consumption metrics.Rather than asking how much AI is being used, organisations are increasingly focused on whether that usage is delivering measurable returns. The goal is no longer maximising token consumption, but maximising business impact from AI investments