You have economies of scale (size) squaring off against economies of scale (production).

For one-offs, the economies of size win, because then the economies of production don't actually get a chance to do their thing. So kick-starting is a problem. However, the problems with the EPR and the AP-1000 hint (do not prove) that we may be hitting size limitations.

The strategy that Rolls-Royce is pursuing seems promising: factory-built microreactors (so economies of production) that you can plop into existing coal-powered plants to upgrade them to nuclear. That way you reduce the issues with the non-factory-built parts.