# The New Cost Curve

**Track:** Foundations of the AI-Native Venture — AI for Entrepreneurship — complete (29)
**Framework / surface:** venture strategy
**Level:** Beginner
**Prerequisites:** What AI-Native Means

**In one line:** The marginal cost of intelligence is collapsing; plan for the model six months out — without needing it to survive.

## Theory, aesthetics & inspiration

The startling economic fact of this era is not that machine intelligence exists but how fast its price falls: Stanford's AI Index measured a more-than-280-fold drop in the cost of GPT-3.5-level inference in roughly eighteen months, and Andreessen Horowitz's Guido Appenzeller dubbed the broader pattern "LLMflation" in 2024 — roughly tenfold cheaper per year at constant capability, faster than Moore's law by a wide margin. Two consequences follow. First, anything uneconomic today at the margin — reading every document, reviewing every line, personalizing every interaction — should be planned as if it will soon be nearly free. Second, William Stanley Jevons's paradox from "The Coal Question" (1865) applies: when using a resource gets more efficient — and thus cheaper — total consumption of it explodes rather than shrinks. The founder's discipline is double-entry: plan for the model six months out, but stress-test the venture against the curve stalling — the business needs today's viable case with capability and cost as upside, never an exponential as a requirement merely to exist.

**Founder question:** If intelligence becomes ten times cheaper next year, what becomes possible in your product?