In other words, nuke cost has only ever increased, however much was built.

On a per-MW basis nuclear power dropped in cost during the 1950s. See the small blue dots round the late 1950s and early 60s? Compare that with the cluster of red dots.

Early, low-hanging fruit?

That, and larger plants. Things like concrete containment vessels have costs relative to the surface area of a hemisphere, while power output scales with the volume. But the big driver was economies of scale. Building multiple copies of the same or similar design means you can have longer production runs of steam generators, pressure vessels, turbines, etc.

I.e., cost has risen monotonically since the 1950s.

Incorrect, it fell from the 1950s through the 1960s.

I.e., cost has risen monotonically since the '60s. Costs were all over the map in the 50s.

In the early days, vendors didn't know how much NPPs would cost, so they priced them to sell, so they could gain that data. Later prices reflected that experience.

One can view the increase over time of nuclear costs as the replacement of wishful thinking with actual experience.

Thanks for the paper. Quick summary (of a quick read): Most research has studied US and France; this paper adds other countries. Costs have greatly increased in US and France, but not always in other countries. They've decreased recently in South Korea.

> (economies of scale)

Why do you blame economies of scale? The paper doesn't say that, afaict.

Also they say, "increased environmental and safety regulation ... may have led to cost increases", which does not sound conclusive.

Also, I think we really need to be talking about lifetime cost, including construction, operation, and decommissioning. In many things, spending more up front reduces later costs.