[paulsutter]

It’s called the manufacturing learning curve, it was discovered in the 1940s relating to WWII aircraft production, and it’s not a special property of semiconductors. Just somehow “Moore’s Law” got popularized in the press.

[dgacmu]

No other technology stayed on as steep an exponential improvement curve for 60 years in the way semiconductors did. They were, in fact, quite remarkable.

[dalbasal]

I think transistors were unique in that we had/have a serious ability to consume more of them. Compuers are half price?!! I'll take three.

When the price of computing halved, we consumed more than twice as much, so that the industry still grew. Most other products/industries didn't have that kind of demand curve. Eventually, we had enough textile and we were happy to pay less for more (rather than pay more for way more). The textile industry stopped growing and learning/price reduction slowed too.

When the price of computers dropped, we paid the same for more powerful computers and more people decided to buy computers because the computers were now more useful. When compuers got powerful enough that we started paying less per computer, we still bought more computers in the form of ipods, VR goggles and smart toilets.

More demand, more production, more learning, lower price...question: If the product is cheaper, will you spend more or less in total? If the answer is "yes," rinse and repeat until the anser is no.

[VintageCool]

Coal was similar.

https://en.wikipedia.org/wiki/Jevons_paradox

When people learned how to use coal more efficiently in 19th century England, that didn't result in less coal being used. It meant that it became economical to do more things with coal, and overall coal consumption increased.

[dalbasal]

Yep. What stopped that learning curve was the availability of coal, not demand.

[_ph_]

If you consider the potential long term demand, the battery market is facing an increadible growth too. The current electric car production is below 1 Million/year, that is about 1% of the total car production, so the market for batteries for electric cars alone can grow about 100x, not counting in that the capacity per car might also grow (perhaps not much over the 100kWh for the top Teslas, but in average probably over 60kWh/car).

On top of that come all the new usages for batteries, especially energy storage, be it for individual houses or on the grid scale.

So potential market capacity is for long time no limitation for batteries. The unique thing of the semiconductor market development was though, that the chips grew in capabilities, but very little in size.

[emilfihlman]

This can be explained by the incredible applicability (and thus demand from business and opportunity creation). We literally had nothing alike before and then suddenly we could do so much more.