We have a drawer in the kitchen that has become a graveyard for many things, including lighters. Every time I fire up either the fireplace or the barbecue I cannot find one, so I buy 3 more, and then somehow all the old ones turn up at once. Anyway, bear with me. Last week I picked one up and actually looked at it, because I had spent the evening reading about solar panels and my brain was in a strange place.
A disposable lighter is basically a pressurised fuel tank. It has a igniter, a spark gap, a metal flame guard thingie, an adjustable valve, and the better ones have a refill port. It is a small chemical engine you can carry in your pocket. It survives being dropped, sat on, and left in a hot car for a month (mine live in a drawer with the batteries and the elastic bands, so they survive worse). In the supermarket it costs about 25 cents, and that price already contains the factory, the shipping, the importer, the wholesaler, the shop’s own margin and the BTW, that’s insane!
The skill I want to point out is not the lighter
So yes I did some old school googling (well duckduckgo ftw) and Wenzhou made 80% of the world’s lighters at its peak. It’s hundreds of manufacturers sitting on top of each other in one city. The interesting thing is not that they learned to make a lighter, because anyone can make a lighter. The interesting thing is that they learned to take a hundredth of a cent out of a valve seat, and then go and do it again on the next part, and the next one. fractional cent improvements because at those volumes, that is still money.
That is a completely different skill from making things. It is the skill of grinding down a cost curve across the value chain, and once you succeed with that, you can point it at anything.
And they did, they pointed it at solar. Chinese solar module capacity now runs to almost double what the world actually wants. Those factories sat at 44% utilisation last year and did not stop. Panels went out below cost for years, and as a result European module makers went away. Now the consolidation everybody predicted is here, and there are some nuggets in the whole story. The 6 biggest producers put together a plan worth 50 billion CHY to buy and shut down a third of their industry’s capacity, and in january their own antitrust regulator blocked it because of monopoly grounds. Even the people who built the overcapacity cannot switch it off!
The same thing then happened to cars. That is where you get the drone footage of the parking lots… nobody needs a caption for those. And then to battery cells, which is where it stops being somebody else’s problem. Northvolt is gone. 11 of 16 planned European-led cell plants have been delayed or cancelled. CATL’s CEO was asked about it and said, roughly, that the Europeans had the wrong design, then the wrong processes, then the wrong equipment, so how exactly were they going to scale. That is a brutal thing to say and I have not seen anyone credibly argue he was wrong, though I would love to be shown otherwise.
Drie keer niks
Look at what Europe did in all 3 cases, because we answered with subsidy every time. Panels got subsidised, factories got subsidised, cells got subsidised. And it did not do much all 3 times, and I now think it is for the same reason each.
My opinion, and I have gone back and forth on this, is that we keep mis-understanding or mis-interpreting what is being aimed at us. That is what I meant to express with the title. I am by no means a China expert, far from it. And not a complete tinfoil madhatter, but the overcapacity is not an accident, or a planning failure, or a bubble that will pop and teach everyone a lesson, it is a form of weapon. Building far more than the market wants, needs, and sitting inside that kind of pain for a decade, is a strategy that works. The fact that it hurts the people doing it is not the counter-argument we keep hoping it is.
Where this ends up
Back in 2014 I wrote a short piece asking what the continued IaaS price cuts were actually signalling, and I concluded they were a loss leader. Get people into the store on cheap compute, sell them the layer above. That was 12 years ago and it was, if I am honest, a slightly obvious observation at the time. But hey I was still learning to write and have somewhat of an opinion of my own.
I am now starting to think it was the same mechanism, just wearing a cloud suit.
So the thing I would take to a leadership team is this. When a supplier’s price is falling faster than your cost model can explain, stop congratulating yourself on the negotiation, because you are not a good buyer. You are watching a market get restructured around you, and the restructuring is the product. Ask what layer they are actually protecting, because it is never the layer they are discounting.
Which brings me to the bit I did not expect. I started writing this convinced I could point the whole thing straight at AI models in one go. Then I went and looked at what Geely actually did with Volvo. The dates are the wrong way round from how I had them in my head, and that changed the argument enough that it needs its own piece.
So that is next week. Bring a lighter.
Leave a Reply