Analysis — 분석
The Business Economics of the Loop — Earning and Surviving on Tokens
Controlling cost alone does not sustain the loop. Read the revenue structure as token unit cost and margin, and choose how to survive among three paths — self-sustaining, investment, ecosystem. This piece lays out each path's realistic problems and best- and worst-case scenarios in the vocabulary of reopt architecture's evolution stages.
Control is only the start
The ways to handle token cost are already deep. Per-path model selection, per-agent budgets, prompt caching and batching — the Cost Control pattern provides the structure to press the cost curve down.
But pressing the curve is a start, not a finish. Even if you halve the cost, if you don't know what that cost earns, the loop stops the moment an external budget runs out. Control answers "how much do we spend," but sustainability must answer "what does that spend return."
From token unit cost to margin
The next question is in the unit. How many tokens does one turn of this agent cost, and what does that turn produce. The moment you measure unit economics — token cost per feature and per agent — cost shifts from something to be controlled into a management metric.
Place the value a turn produces (a resolved inquiry, a passed deploy) next to that turn's token cost, and per-feature margin appears. Run the positive paths more; push the negative ones down to a lighter model or hand them back to a human. This judgment happens in SHARPEN — the cost-control principle that cost signals feed loop tuning extends into a management decision: which path to keep alive.
Surviving is not one thing
After seeing margin, the next question is "so how do you survive." There is more than one path to keep the loop alive. Three strategies exist, and each has different realistic problems and best- and worst-case scenarios.
Self-sustaining — live on margin
Grow by repaying your own cost with the loop's positive margin and ancillary revenue (offering an API, productizing data). It is the most solid structure, running on unit economics without outside capital.
The realistic problem is speed. Early margin is thin and growth is slow, and leaving even one feature whose token cost exceeds its revenue drags the whole loop into the red. The best case is positive margin compounding so you scale without external dependence and buy your own next evolution. The worst case is margin hardening negative, the loop stalling as a cost center, and being passed by a competitor because you cannot raise growth capital.
Investment — buy time with capital
Raise outside capital to weather losses and secure scale first. Lower token unit cost through economies of scale, and build a moat with data and share.
The realistic problem is that scaling while unit economics stay negative only burns runway. Miss the moment to turn self-sustaining (positive margin) and you are captive to the next round. The best case is economies of scale and the learning loop pulling unit cost into the black, with market share giving you pricing power. The worst case is failing to reach self-sufficiency before the capital runs out — shrinking sharply or being acquired at a discount.
Ecosystem — become the standard, grow the pie
Grow the loop into a platform or standard others ride on, not a single product. Let the token economy flow across the whole ecosystem rather than locking it inside, creating larger value.
The realistic problem is enduring a long J-curve until network effects cross their threshold. Lose the standards race and you become captive to someone else's ecosystem. The best case is becoming the ecosystem standard — more value is created on top, and a self-reinforcing share flows back to you. The worst case is losing the standards race or the ecosystem never reaching critical mass, leaving only the cost of opening up and a dependency.
Revenue maturity arrives with the stages
Whichever of the three strategies you choose, to be able to choose it you must first be able to see revenue. A revenue structure matures with the evolution stages. At stage one you make cost simply visible; at stage two per-agent unit cost separates the first revenue paths from the loss-making ones; at stage three per-feature margin lets you pick a sustaining strategy; at stage four the loop reaches self-sustaining unit economics where it repays its own tokens.
Choosing a strategy is not deferred to stage four. Whichever you aim for, making the next measurable number at each stage is what makes that strategy executable. From control to unit cost, from unit cost to margin, from margin to choice — that is the order in which the business hardens.
See also
- This theme among the environment's three faces: Business
- Sister pieces: Token Strategy · The Data Flywheel
Tags