Writing · AI Economics

Why more intelligence doesn’t automatically mean more money

A short note on token maxing, output, and economic value · Published Apr 28, 2026

Originally published as a note on my Substack. Subscribe there for the latest essays, notes, and frontier work.

Infographic titled Why Token Maxing Doesn’t Move the Needle, explaining how AI token usage must convert into economic value
More AI tokens can create more output without creating more wealth. The conversion layer is the work.

The note

Why more intelligence doesn’t automatically mean more money. It’s great that you’re token maxing, but those total number of PRs, LOC, and refactoring are virtue signaling that “you know ai” if it doesn’t translate to economic value.

The signal

More AI usage is easy to measure. More prompts, more pull requests, more generated code, and more refactoring all look busy. None of them are the scarce thing by themselves.

The scarce thing is conversion: turning abundant model output into decisions, distribution, customer value, workflow change, revenue, cost savings, or reusable assets that compound.

  • Tokens are inputs. They only matter when they move a bottleneck with obvious dollar value.
  • Output is intermediate. A pile of generated work still has to survive integration, review, adoption, and timing.
  • Economic value is the proof. The frontier matters most when it changes what someone can sell, save, ship, or decide.

The bottom line is simple: intelligence is no longer the scarce resource. Conversion, execution, and distribution are.