2010: https://www.sciencedirect.com/science/article/abs/pii/S03014...
2020: https://arstechnica.com/science/2020/11/why-are-nuclear-plan...
So in a few decades France has forgotten how to build reactors on time and in an affordable way?
Everything has been debundled, privatized, consolidated, streamlined, modernized. This is not "bad" in itself, but doing big public projects this way is not efficient, because this way of organization needs big volume to become efficient.
Basically we need to order 100+ reactors to get them cheap and fast. (And we need to have them use truly standardized interchangeable (modular) components. Naturally, modularity has an overhead, but it pays off if there's a long term for it to pay off. As long as we only build a few plants treating each of them as different allows for some local optimizations, but that is against large volume manufacturing, maintenance, etc.)
This has some useful charts.
It shows that nuclear still has economy of scale to improve, if we give it the chance.
The big difficulty for nuclear is public pressure/fear of accidents. People are much more afraid of "big scary nuclear accident" vs. probabilistic deaths caused by air pollution, even if the deaths from the latter is in the tens of thousands and the former is in the dozens.
That fear of public/policy backlash makes raising capital in order to actually benefit from those economics of scale very difficult.