Tokenomics: The house always wins, unless you count.

Taking the gamble out of AI spend

Most organisations can't answer three simple questions about AI. What did we spend last month? Which teams spent it? What did we get back?

In this 24-minute talk, Aidan Grace, Head of Applied AI at Cloudscaler, explains why token-based AI costs behave so differently from the cloud bills we already understand. Usage swings far more sharply, and the token itself means little to the people paying for it. Using a casino analogy, he sets out a practical system for control: train staff properly before handing them powerful tools, set hierarchical spending limits, match the model to the task, and accept that many pilots won't reach production.

His biggest point? With AI, counting cards is easy. Invocation logs show exactly what people are doing and whether it worked. From there, Aidan introduces two measurable alternatives to chasing a mythical ROI: net AI leverage and effective AI leverage.

Watch to learn how to turn AI from a gamble into an investment.