Tokenmaxxing: Why AI consumption needs control

The AI industry recently saw a surge in spending, with Anthropic's Claude Fable 5 model garnering attention. This model costs $10 per million input tokens and $50 per million output tokens, double the price of Anthropic's previous most expensive model, Claude Opus 4.8. As AI models evolve and businesses adopt them more widely, the issue of controlling AI consumption and avoiding wasteful spending becomes increasingly important.

While some businesses only use AI for specific tasks, others run models non-stop, leading to rapid token consumption and budget overruns. The shift toward AI in the UK software market has accelerated, with Anthropic's market share increasing significantly. However, this growth comes with financial challenges, as AI models require substantial resources.

Organizations are turning to their finance teams to gain visibility into AI usage and costs. Finance can help identify inefficiencies, such as teams using AI for peripheral tasks rather than core business activities. By tracking AI spend and performance metrics, businesses can ensure that AI investments translate into measurable productivity gains, rather than just higher software costs.

The key to effective AI control lies in understanding where AI creates value and aligning investments with business outcomes.