The lesson from the latest delay to Finland's planned new nuclear reactor, announced last week, is either build lots of them or don't build any at all.
It is a general economic principle that costs per plant decline the bigger a programme.
Historical data support that for nuclear, where it may apply more acutely given its highly specialised and often unique supply chain and engineering skills.
It makes no sense to take a suck-and-see approach, building incrementally in a modular fashion, especially for new technologies such as Finland's advanced pressurised water reactor.
But plans by developed countries including Britain, Finland, Poland and the Czech Republic are for just that.
It is not surprising that developed countries are wary of big nuclear rollouts, given the high capital cost and the present credit squeeze for both banks and impoverished electricity consumers.
The danger is that they are caught in an economic no man's land, afraid to cancel programmes altogether in the face of the falling costs of modular approaches such as wind power, biomass or gas, or to ramp them up.
Of countries with definite plans, Britain supports industry plans for 16 gigawatts (10 or so large power plants) by 2025. This target looks stretched given construction has not started on any, while developers of 6 GW have already abandoned their interest.
The Czech Republic has immediate plans to build two new reactors, Poland to build one by 2023, and Finland also one more, all potentially involving advanced technology.
Contrast that with China's plans to build 50 or so new reactors, and India's for 60 GW by 2030 (according to World Nuclear Association estimates).
Consultants Parsons Brinckerhoff estimated the capital costs of a "first of a kind" plant (built without an established supply chain) at nearly a fifth higher than an "nth of a kind", in a report to the British department for energy and climate last year.
The difference may be greater, however.
Precise estimates of the cost of nuclear power seem to go in circles depending on the views of the advocates, given so many variables including the discount rate applied to finance; the present and future cost of rival technologies; and an assumed carbon price.
It is useful to review the scale of actual problems faced by first of a kind plants being built using advanced pressurised water technology under construction in Europe.
Finnish utility Teollisuuden Voima on Monday said its Olkiluoto 3 reactor would no longer meet its 2014 completion date, which itself was five years behind schedule and is also massively over-budget at about €4,125 ($5,200) per KW.
The Finnish new build plans follow those of France's new Flamanville 3 reactor.
French utility EDF last year estimated the overnight cost (excluding cost of finance) of its new reactor, the first to be built in France in 15 years, at €3,600 ($4,600) per kilowatt, up from an initial estimated €2,000 ($2,500).
The Flamanville reactor will be over-due by four years.
Britain has the boldest plans for nuclear new build in the developed world.
The country will want to avoid these kinds of cost overruns and delays, from a lack of supply chain and skills, but that is precisely the situation the country presently faces.
A report published last November by a UK parliamentary panel found that Britain's nuclear R&D workforce had declined from more than 8,000 in 1980 to less than 1,000 now, following the closure of Government nuclear laboratories.
The R&D programme received government funding of £300-350 million a year in the 1980s, compared with £11 million annually now, it found.
It is hard to see how a rebuilding of such intellectual capital can be cost-effective unless applied to more than three or four plants, leaving it dithering between shelving its plans or committing to a large, prescribed target.
This article was originally published by Reuters.