3 ms·

Pretty much everything that's being manufactured today can be manufactured for 10 to 100 times cheaper. But you can't drive the cost down if you can't iterate e

Pretty much everything that's being manufactured today can be manufactured for 10 to 100 times cheaper. But you can't drive the cost down if you can't iterate enough. A few hours ago I bought a smartphone for $35. It's an LG smartphone, and while it doesn't hold a candle to the latest iPhone 12 Pro, it will smoke any phone manufactured just 7 years ago. And it's only $35. Can you imagine that? Some hundreds of millions of transistors, two cameras, some wireless antennas, a GPS chip, an FM receiver, a screen with millions of colored pixels, etc, etc, for only $35. All these things would not have been available in such a small package 20 years ago, and they would have cost thousands of dollars in any size. Now they fit in something that's 100 grams heavy and costs $35. So, yes, a large reactor costs $20 BN today, a small reactor is probably of the order of $1 BN, but a small reactor allows you to iterate. Once you are able to iterate, there's nothing that stops that cost from going down by a factor of 10 or 100, just like the cost of solar and wind went down.

baybal2 6y ago
Past the reactor containment building, a nuclear power station look pretty much like any other rankine cycle plant, which were already delivered on scales those guys can only dream about. This only reinforces the rationale for much bigger nuclear plants, not reduces. > But you can't drive the cost down if you can't iterate enough. > A few hours ago I bought a smartphone for $35. A low-end price for a smartphone SoC tapeout on the latest node is $100-$150m. Iterate that. All of what you spoke about, and microelectronics in particular, could've never been done without sisyphean effort to nail hardest engineering problems in one go, because the nature of the problem does not afford "a half-solution." You absolutely must solve such kind of engineering problems in fullest in one go, or it's not going to work. A reactor vessel is not like a piece of software you can let crashing for a few beta versions. It either has no holes, and it work, or it does have holes, and it leaks.
- credit_guy 6y ago
  I don't know much about nuclear reactors and Rankine cycles, just like I don't know much about rockets. But somehow Musk found a way to drive down the cost of launching things into space quite a lot.
- imtringued 6y ago
  >All of what you spoke about, and microelectronics in particular, could've never been done without sisyphean effort to nail hardest engineering problems in one go, because the nature of the problem does not afford "a half-solution." Why is history littered with half solution nuclear powerplants then? If Fukushima or Chernobyl were small modular reactors you could have shut them down and replaced them long before they had failed. That's something we do in almost every industry, yet nuclear is pardoned despite the high impact of a nuclear disaster. >You absolutely must solve such kind of engineering problems in fullest in one go, or it's not going to work. That just means the plants will have critical construction flaws from the start or develop critical flaws over time. >A reactor vessel is not like a piece of software you can let crashing for a few beta versions. It either has no holes, and it work, or it does have holes, and it leaks. The nuclear industry is worse. Instead of removing the beta versions they just pretend that they are release ready and keep them around for decades until the inevitable high profile failure.

baybal2 6y ago

Past the reactor containment building, a nuclear power station look pretty much like any other rankine cycle plant, which were already delivered on scales those guys can only dream about. This only reinforces the rationale for much bigger nuclear plants, not reduces. > But you can't drive the cost down if you can't iterate enough. > A few hours ago I bought a smartphone for $35. A low-end price for a smartphone SoC tapeout on the latest node is $100-$150m. Iterate that. All of what you spoke about, and microelectronics in particular, could've never been done without sisyphean effort to nail hardest engineering problems in one go, because the nature of the problem does not afford "a half-solution." You absolutely must solve such kind of engineering problems in fullest in one go, or it's not going to work. A reactor vessel is not like a piece of software you can let crashing for a few beta versions. It either has no holes, and it work, or it does have holes, and it leaks.

credit_guy 6y ago
I don't know much about nuclear reactors and Rankine cycles, just like I don't know much about rockets. But somehow Musk found a way to drive down the cost of launching things into space quite a lot.
imtringued 6y ago
>All of what you spoke about, and microelectronics in particular, could've never been done without sisyphean effort to nail hardest engineering problems in one go, because the nature of the problem does not afford "a half-solution." Why is history littered with half solution nuclear powerplants then? If Fukushima or Chernobyl were small modular reactors you could have shut them down and replaced them long before they had failed. That's something we do in almost every industry, yet nuclear is pardoned despite the high impact of a nuclear disaster. >You absolutely must solve such kind of engineering problems in fullest in one go, or it's not going to work. That just means the plants will have critical construction flaws from the start or develop critical flaws over time. >A reactor vessel is not like a piece of software you can let crashing for a few beta versions. It either has no holes, and it work, or it does have holes, and it leaks. The nuclear industry is worse. Instead of removing the beta versions they just pretend that they are release ready and keep them around for decades until the inevitable high profile failure.