AI compute costs are forcing a fundamental rethink of how designers budget intelligence into products. The core argument: tokens are not free, and treating them as if they are produces bloated, economically unsustainable UX.
The arcade economics framing is the piece's sharpest tool. Just as quarter-per-play constraints shaped tight, deliberate game design, token scarcity should pressure designers into precision. Every prompt, every agentic loop, every context-stuffed API call has a price. The article makes the case that ignoring that price is a design failure, not just a cost problem.
The full piece is worth reading for how it maps token budgeting onto existing design decision frameworks, and where it identifies the specific expectations, around AI verbosity, real-time responsiveness, and always-on features, that rising compute costs will force the industry to abandon.
[READ ORIGINAL →]