The Simulator
Manufacturing follows Wright's Law: every time cumulative production doubles, unit cost falls by a fixed percentage — the learning rate. Drag the sliders to see how learning rate and production growth reproduce the observed price collapse, and where prices could head next.
Model: cost = start × (1 − learning rate)^log2(cumulative units ÷ initial units). Illustrative — actual prices reflect mix, subsidies, and strategy as well as cost.
Where the 70% Actually Came From
Actuator commoditization
Harmonic drives, planetary rollers, and high-torque motors moved from bespoke robotics suppliers to mass-market EV and drone supply chains, cutting the biggest single cost block.
Sensor price collapse
Depth cameras, IMUs, and lidar rode the smartphone and autonomous-vehicle volume curves. A sensor stack that cost thousands now costs hundreds.
Design-for-manufacture
Fewer machined parts, more castings and injection-molded structures, simplified wiring harnesses — the same playbook that drove down EV costs.
Volume and amortization
Fixed R&D and tooling spread across thousands of units instead of dozens. This is why margins rose even as sticker prices fell 70%.
How to Read the Curve
With a 25% learning rate and production nearly tripling each year, the model lands close to the observed $25,000 average within two years — no exotic assumptions needed. That is the quiet takeaway: humanoid robots are not defying economics, they are obeying it, on the same curve that solar panels, lithium batteries, and flat-panel displays rode before them.
Try lowering production growth to 50% and watch the curve flatten: without volume, learning stalls. Then push the learning rate up and see how quickly a sub-$10,000 humanoid — cheaper than a used car — stops looking like science fiction.