Thread by @ramez

1. A short thread on a fundamental trend in economics of physical stuff that's highly relevant to the conversation about nuclear & renewables:
A. Manufacturing things gets cheaper with scale.
B. Constructing things in the field gets more expensive with time.

2. The learning rate / Wright's Law is now a widely understood phenomenon in clean energy. Every doubling of scale of simple (low part count), factory-manufactured goods leads to a fairly constant rate of cost decline. E.g., in solar: rameznaam.com/2020/05/14/sol…

3. Similar phenomena to solar appear in semiconductors, automobiles, aircraft, batteries, wind power, consumer electronics, and pretty much any mass-manufactured good. (With differing slopes.) The seminal paper on this was written in 2008. santafe.edu/research/resul…

4. *Construction* costs, however, don't show such a trend. Brian Potter at Construction Physics has a post today asking, "Does Construction Ever Get Cheaper?" The answer is pretty much no. constructionphysics.substack.com/p/does-constru…

5. What's worse is that mega-project construction shows not just an increase in cost, but also a much higher likelihood and size of cost overrun when compared to highly modular projects made of repeatable units built in a factory.

6. The implication of this is that, if we want low costs, we should build everything we can in factories, and do as little assembly in the field as is humanly possible. Of course, it's sometimes not possible to get field assembly down to zero. But minimize it.

7. This topic comes up frequently for me when talking to next-gen nuclear startups. One of my first questions is always: How much do you build in the factory vs assemble in the field? A shocking number aren't able to answer this question. Yet it's key for costs.

8. This doesn't just apply to nuclear. Wind power has a slower learning rate than solar in part because there's more in-field assembly. Ocean energy techs also tend to involve more in-field assembly. For all, my advice is: Build all you can in a factory. Minimize assembly.

A couple addenda to this thread, specific to nuclear:

9. I don't consider anything to be a "Small Modular Reactor" unless it's almost entirely built in a factory. The term SMR is now being stretched and abused, IMHO.

10. I do still have hope for small modular nuclear reactors (SMRs). But it's chicken and egg. At some critical number of units manufactured, costs will get low and predictable. But who will order the first few tens (to maybe hundred) of expensive units? That's the question.

11. Jumbo size isn't the only reason that megaproject nuclear (gigawatt scale) has gotten more expensive. Regulation & public fear play a role. And the decimation of the supply chain plays a role. If you build a very small number of reactors, of any size, they'll be expensive.

12. If you want nuclear to thrive and get cheap, you have to have a plan for how specific reactor designs will get built again and again and again, possibly hundreds of times, to get that learning rate.

13. IMHO, some current efforts at the DOE on GenIV nuclear ignore this dynamic, by aiming to vastly expand the number of reactor designs that get approved. I understand that impetus. But unless you get high volume production on a specific design, don't expect it to get cheap.