Ok, that's too much coffee for you.
It is an interesting question. Eventually the supply chain of everything Tesla uses needs to double just as quickly as Tesla doubles. So pretty much anything can become a bottleneck at high enough scale.
One way to think about this is the learning curve. For example as the cumulative installation of solar doubles, the cost of solar per Watt comes down about 22%. This is very fast learning, and it is why solar is able to scale so quickly for so long, decades. Battery packs also are on a learning curve, and it may even be steeper than solar according to some researchers. So suppose that as cumulative battery pack production doubles, the producer cost comes down say 30%. On one hand this is very helpful because it allows packs to scale very fast as prices falls, but it may also impose a speed limit too. BNEF observes that pack prices are falling 20% per year. At 20%/y price decline it takes 1.60 years (=ln(.7)/ln(.8)) to achieve a 30% price reduction. So if the learning curve is at 30%, then the cumulative supply of battery packs can double every 1.60 years, which is an annual growth rate of 54% = 2^(1/1.60) - 1. So what happens if you try to grow supply much faster than that, say 100%/y? It could be that you simply accelerate annual decline in battery prices. But you could get ahead of the technology gains and obtain a lower learning curve rate. For example, costs may decline just 20% with every doubling of cumulative production which would remain in line with the 20% annual decline observed, but with production doubling every year. If you get seriously ahead of the technology gains and push say mineral supplies to double, you may see mineral prices spike up so fast that it undermine any gains with your technology. So in the extreme you could see the cost of batteries actually increase y/y instead of falling 20%/y. So it is hard to say how the learning curve changes if you push production to double faster.
I think mineral prices are a key constraint. If tight mineral supplies cause the annual pack cost to decline at a slower rate than 20%/y, then we may be trying to grow the pack supply too fast. Technology gains are critical because we need better technology to get better battery economy and performance from a given set of mineral prices. For example, if cobalt were ridiculously expensive, then we would need technology gains that show how to get more battery from less cobalt. We would also need higher investment in developing new cobalt resources, but that comes with higher prices. But without that specific sort of technology gain to minimize exposure to cobalt prices, then pressing the pack supply to keep doubling drives cobalt prices even higher. At some point a pack becomes too expensive for demand to keep doubling annually. For example, solar module prices have actually gone up about 25% because the supply of polysilicon has become tight. So this can slow up the growth of solar this year. More investment in polysilicon resource development and more research into how to minimize polysilicon per Watt are both needed to get solar prices back on track with long-term price declines, otherwise annual volume growth rates may slow. Mineral resources can be grown at certain rate for a certain price. To get them to grow faster, you have to pay a higher price. Technology gains are thus needed to avoid paying a mineral price higher that what pack demand can bear.
️ on Twitter
