My own very crude back of the envelope, for replacing all worldwide coal generation with AP-1000s (at $7bn a pop) puts the cost somewhere in the $6T range. This is assuming I got the unit conversions correct...
As for MSRs, there's some very interesting history there. There is, of course, the fact that MSRs weren't as good at producing the necessary byproducts for making nuclear bombs (remember this was during during the cold war). But the other accident of history was simply that Oak Ridge National Labs (who ran a prototype MSR for a few years) just happened to be in the wrong part of the country: the story goes that Nixon apparently wanted to shore up political support in (California?...) and so pushed money more in the direction of the LMFBR.
Nixon also fired the director of ORNL too, apparently because his support for MSRs and advocacy for increased nuclear safety didn't gel too well with that administration. At that point MSRs were kind of forgotten about...
It definitely seems like the big challenge for MSRs is of the materials/chemical engineering variety. I guess some of these could be designed around: for instance I've heard Kirk Sorensen (LFTR advocate) suggest situating the entire reactor and plumbing complex in a drain/basin of sorts. In the case of a pipe break the fuel(s) simply drains in to a passively cooled storage tank (deep underground).
Also some of the chemical handling would require great care. Fluorine, for instance, is not something I'd want to mess around with...
- We could do it for less than $3 Trillion ...
- We could do it for $1 Trillion with liquid-fueled Molten Salt Reactors
Seems very dubious. The first successful MSRs were built over 60 years ago. The reason none are in commercial use is probably because the costs would work out higher than conventional reactors.