Solar crossed 3 terawatts - and got so cheap its makers are losing money

A strange kind of success

Solar panels have never been cheaper, and the companies that make them have never been in more trouble. Chinese manufacturers have lost money for about three years running. This year alone, polysilicon, the purified silicon at a panel’s base, dropped 42% below where it started. The world just installed its third terawatt of solar. And the makers are begging their own government to stop the fall.

That is a puzzle only if you think price follows demand. It doesn’t, not here. The price is following something else.

The engine under the price

There is a rule that shows up wherever people make the same thing over and over. Each time the total number ever made doubles, the cost to make one falls by a roughly fixed percentage. Not each year. Not each factory. Each doubling of the running total.

An aircraft engineer named Theodore Wright first measured it in the 1930s: every time cumulative plane production doubled, the labour per plane dropped about 20%. The same slope turned up in computer chips, in batteries, and in solar, where it fell so reliably it got its own name, Swanson’s law. Solar’s cost has dropped roughly a fifth for every doubling of how much has ever been built.

This is not the same as a technology catching on. That is a different curve, the slow-then-fast-then-flat spread of who adopts it. This one is quieter and harder. It is the reason the cost keeps falling at all. Not because more people want it, but because every doubling forces the same fresh rounds of learning, tinkering, scaling, and cutting.

Why doubling, and why it doesn’t stop

Each doubling is a full turn of the same crank. To make twice as many, firms redesign the line, automate a step, thin the silicon, squeeze the supplier, and undercut each other. Those gains repeat every time the pile of everything-ever-made doubles again. Early on, doublings come easily, because doubling a small number is cheap. So the price falls fast when the technology is young.

Here is the part that traps the maker. The crank measures cost, not profit. It keeps cutting the cost per panel whether or not that price still leaves anyone a margin. China built factories for twice the world’s demand, so the doublings kept coming, and the price sailed straight past the line where the makers break even. The curve did exactly what it always does. It just doesn’t care who is standing on it.

Who rides the fall

The pain sits in a few Chinese factory towns. The gain is spread across the world, most of it far from anyone who ever heard of Swanson’s law. Britain fitted more solar in six months than in any half-year since 2011, chasing lower bills. China now draws less than half its power from coal for the first time. Pakistan and Nigeria, hit by energy shocks, buy cheap imported panels by the container.

Look at how the early doublings were paid for. Rich countries subsidised the first terawatt, the expensive one, when panels were a curiosity. Those subsidies bought the early doublings that drove the price down. Poorer countries now buy near the bottom of the slope, riding a curve someone else’s taxes started rolling. A household putting panels on a roof this year is standing on decades of doublings it never paid for.

The whole of it

No single seat sees this straight. The factory owner sees ruin and calls it a price war. The regulator sees a race to the bottom and tries to set a floor. The household sees a bargain and takes it. All three are looking at the same slope, and each is right about their own end of it. The cheapness that empties a balance sheet in Jiangsu is the cheapness that fills a roof in Lagos, and neither one can see the other from where they stand. The curve connects them without ever telling them they’re connected.