Wright's Law: unit costs fall a consistent percentage for every doubling of cumulative production. Formulated in 1936 from aircraft data, and it has held up across an unreasonable number of industries.
Photovoltaics may be its cleanest modern demonstration — roughly 90% decline since 2010.
Lithium-ion cells are following a curve of remarkably similar shape, driven by the same mechanism. Which makes storage a useful natural experiment: we know what happened to solar's market after its cost collapse, and we can watch whether storage repeats it.
What happened to solar's market
The cost decline itself is well documented. The second-order effect is less discussed.
Wright's Law applies to manufactured goods with scaling production. It does not apply to a technician driving to a site.
Solar hardware fell 90%. Installation labour, permitting, interconnection, and service calls didn't — several rose with wage inflation.
So project cost composition shifted. Hardware, once dominant, became the minority component in distributed installations.
The inversion this produced
When hardware was expensive, choosing cheaper components saved a large absolute amount. Downside risk existed but was small relative to that saving.
Once hardware got cheap, the same percentage discount saved much less — while downside risk stayed constant. Replacing a failed component in year twelve costs whatever labour costs in year twelve.
Saving collapsed. Exposure didn't.
As hardware prices fall, the cost of buying badly rises relative to purchase price.
This is the inverse of how we evaluate most purchases. Usually falling prices lower the stakes of a decision. Here they raise them.
Why storage will repeat it
Same structure. Cells are getting cheap fast. Installation, integration, and eventual replacement labour aren't.
And storage has a variable solar doesn't: cycle life is operationally determined. A battery cycled to 80% depth of discharge degrades faster than one cycled to 50%. Two systems with identical nameplate capacity and identical warranties can have materially different useful lives depending purely on system design.
Which means the quality-selection problem is worse for storage than for solar. There's more to get wrong, and less of it is visible in a spec sheet.
Three modelling errors specific to storage
Round-trip efficiency omitted. Energy in doesn't equal energy out — 10–15% loss is typical for lithium systems. Models treating stored kWh as equivalent to generated kWh overstate returns by that margin, compounded across thousands of cycles.
Sizing against day-one capacity. Batteries lose capacity annually. A system covering evening load in year one may not cover it in year eight. Sizing against nameplate rather than end-of-warranty capacity builds in a shortfall discovered gradually.
Wrong comparison baseline. This is the big one. Storage evaluated against grid tariffs assumes the grid is always available. Where it isn't, the real comparison is against generator hardware, fuel, servicing, and downtime cost — a substantially different number.
The generalisable point
Any hardware category riding a steep learning curve will hit this inversion. Manufacturing cost falls; service cost doesn't; quality selection becomes more important precisely when the price signal says it matters less.
Solar hit it years ago and the buying advice never caught up. Storage is arriving there now, and the same advice is being recycled.
Worth building the correction in before the market does.
I work on solar and storage system design at SolarBazaar. Points above are provider-agnostic.
Top comments (1)
Solar Bazaar provides excellent vendor discovery options.