Cost falls with cumulative experience, not calendar time. Each doubling of total units made multiplies the unit cost by a fixed learning rateThe fraction that unit cost becomes each time cumulative output doubles — e.g. 85% means every doubling cuts cost to 85% of what it was. Also called the progress ratio. Named after Theodore Wright, who observed it in aircraft production in 1936.. On log-log axes, that power law is a straight line — and it's why chips, solar panels, and batteries got relentlessly cheaper as the world made more of them.
Down from €1 000 — about 5.05× cheaper. Average €258 per unit, €258k all in, across 1k units.
On log-log axes, Wright's law is a perfect straight line — the signature of a power law. The slope is the learning exponent (-0.23 here). Every tick on the x-axis is a 10× jump in volume; every tick on the y-axis is a 10× drop in cost. The line doesn't care how many years it takes — only cumulative output drives cost down.
| Cumulative units | Unit cost · % of first |
|---|---|
| 1 | €1 000100% |
| 10 | €58358.3% |
| 100 | €34034% |
| 1k≈ now | €19819.8% |
| 10k | €11511.5% |
Each row is a 10× jump in output. Cost drops geometrically — the bar is log-scaled so each step reads as the same visual distance. Click a row to load that volume above. The first unit is 100% by definition; every doubling brings it to 85%.