An expert's three tips for keeping AI costs in check

Commercial artificial intelligence tools are becoming standard in many businesses and public organisations, but this widespread adoption carries a severe risk of spiralling expenses. AI usage is currently billed in tokens that are disproportionately cheap and do not reflect actual production costs, meaning prices are bound to increase. At the same time, models are growing more powerful and complex, making it difficult to predict how many tokens demanding tasks actually require. When developers and staff use AI assistance for complex tasks, or when AI agents activate one another and become multiplicative, IT budgets can be exhausted in record time.

To prevent these costs from getting out of hand, IT consultant Henrik Zangenberg offers three primary tips. First, organisations should assign a specific person to monitor AI expenses on a daily or weekly basis to allow for timely intervention. Second, employees must be trained to understand what drives these costs so they recognise when they are entering risky territory. Third, organisations should review their existing contracts to ensure they have the best possible agreements, noting that it is possible to negotiate individual terms with major AI players rather than simply accepting standard business subscriptions.

These precautions matter because managing software and cloud expenses is already a complex challenge, and AI introduces new economic hurdles known as tokenomics. Without proper oversight, training, and contract negotiation, public and private entities face the prospect of sudden, dramatic price hikes and exhausted IT budgets as the artificial intelligence market matures.