Inference got cheap and the bill still went up. Tokens keep falling in price while the workflows built on them consume far more of them. The gap between unit cost and total cost is now the whole story in AI economics.
- Tokens are cheap. ~$60 to under $12 per million output tokens, median.
- Workflows are expensive. Agents burn 15x-1,000x the tokens of a chat, and every step compounds.
- The model is ~10% of the cost. Process, governance, and integration make up the rest.
The AI Infrastructure report covers where the money actually sits, including the shift from token price to the cost of a whole workflow.
Get the next one before it is old news
Independent analysis of cloud-native infrastructure, Kubernetes and data centre economics. No vendor spin.