Cathie Wood's wright's Law cost curve
Apply Wright's Law to current cost, cumulative production, and an editable learning rate to see how production doublings change unit cost.
Cathie Wood Reviewed Aug 6, 2026 Rule located in ARK Invest, 2026 Portrait: Caroline Wood, CC BY-SA 4.0, via Wikimedia Commons. Self-hosted by JMM.
Put your numbers through the rule
The source establishes the rule. The values below belong to you, and the output is JMM's deterministic calculation.
Every field recalculates immediately. Changed values can be copied into a shareable URL.
- Production doublings
- 3
- Dollar cost decline
- $44.86
- Percent cost decline
- 44.9%
The curve advances with cumulative production, not calendar time.
0 doublings
- Unit cost
- $100
- Decline from today
- 0.0%
1 doubling
- Unit cost
- $82.00
- Decline from today
- 18.0%
2 doublings
- Unit cost
- $67.24
- Decline from today
- 32.8%
3 doublings
- Unit cost
- $55.14
- Decline from today
- 44.9%
4 doublings
- Unit cost
- $45.21
- Decline from today
- 54.8%
What the source says, and what the calculator adds
What changes the answer
Production, not the calendar, is the independent variable
The same cost decline can happen quickly or slowly depending on how fast cumulative units grow.
Learning rate sensitivity compounds
A few percentage points per doubling can create a large difference after several doublings, so the input should never be treated as a hidden constant.
A real empirical regularity, routinely used to smuggle in a demand forecast.
The underlying observation is well documented across manufacturing: cost per unit falls a roughly constant percentage for every doubling of cumulative output. Where it goes wrong is in the hands of an investment case. The law needs cumulative production to double, and production only doubles if someone buys the units, which is the thing being forecast. Feed a learning rate and a future volume into this page and you get a cost, not a company. JMM would use Wright’s Law to sanity-check whether a claimed price point is physically reachable, and never as evidence that it will be reached.
Not the calculator’s limits. The rule’s.
The learning rate is fitted, not given
Every published learning rate comes from a historical series for one product in one industry. Carrying a solar-module rate into batteries or a battery rate into robotaxis is a choice, and the sensitivity on this page shows how much that choice decides.
Curves flatten when the remaining cost is physics
Learning removes process inefficiency. It does not remove the cost of lithium, silicon or electricity, so a curve extrapolated past the raw-material floor produces a price nobody can build at.
What this model does not know
- The learning rate is a user input unless separately established from a historical series.
- The model says nothing by itself about demand, margins, competition, valuation, or stock returns.
- Future production can be lower than entered or fail to reach the modeled level.
Before you use the result
Is Wright’s Law the same as Moore’s Law?
No. This page relates cost to cumulative production, while Moore’s Law is commonly framed around computing density over time.
Does lower cost guarantee a good investment?
No. Cost curves do not determine demand, pricing power, capital structure, or valuation.
The next question this page cannot answer
- Cathie Wood, sourced rule Disruptive innovation exposure map
What a portfolio built on these cost curves actually holds, and how concentrated the exposure gets.
Open → - Calculator CAGR calculator
Converts a cost decline per doubling into the annual rate it implies at your production growth.
Open → - Guide Six ways a backtest lies, and the habit that catches them
The learning curve fits history well by construction. This is the reason that is not evidence about the future.
Open →