Glossary term

Marginal cost of intelligence

The incremental cost of generating an additional useful AI output, decision, or completed task.

The right lens for a scaling decision: not "what does the program cost in total" but "what does the next unit of useful output cost," which can rise or fall independently of total spend depending on model choice, caching, and routing. Tokenomics treats this as a variable-cost curve to manage deliberately, not a fixed number to accept.