Except, he's missing quite a few facts.

Like, that there's still unsolved engineering problems with finding naterials that survive the corrosive environment within of molten salt reactors.

Like, that he didn't inckude waste storage at all.

Like, that he calculated prices for renewables based on buying the most expensive panels, but for nuclear, based on the lowest estimates.

In reality, the ROI level of renewables has already won against nuclear.

And then there's the problem that also plagued plants in Germany, massive corruption and bribing of investigators because it's cheaper than actually building safe plants. It's no coincidence that the only success story for nuclear is a country where all reactors are government-owned and don't have to turn a profit necessarily.

And this brings us to the last problem:

Most new CO2 will soon be produced in developing countries, with unstable governments and unreliable public systems.

And you want to give those governments nuclear reactors? Iran, Pakistan, Nigeria, etc?

Your point about the materials challenge for MSRs is very true: for the designs I've seen the 'solution' is super corrosive. However, I don't think it's an insurmountable challenge, and its more a question of durability over time (and then subsequent maintenance costs). Our water plumbing systems will eventually succumb to the rigours of time; it will just be a very long time...

On waste storage, MSRs product much much less waste. A solid fuel, light water reactor achieves maybe 3% fuel efficiency, whereas it's somewhere north of 90% for the thorium MSR. In addition, the byproducts it does produce are mostly useless for nuclear weapons. Dr Carlo Rubbia (former director of CERN and 2016 Nobel Laureate) has stated one of the reasons funding for MSRs was cut by the US in the 70s was precisely because it's difficult to use them to produce nuclear weapons. There are even proposals for MSRs that are fuelled specifically by nuclear waste transuranics (e.g. http://www.transatomicpower.com/wp-content/uploads/2015/04/N...).

Honestly, I'm not so sure the ROI for renewables is anywhere close to 'winning', unless you ignore stuff like renewable intermittancy (meaning you need to build a crapload more highly expensive transmission infrastructure, or attach highly cost-effective batteries to the grid i.e. lead-acid batteries atm).

As for German nuclear, I suppose that won't be a problem once they've completely shut down all their nuclear reactors. Pity that the government official estimated cost of this is 55 billion Euros over the next decade. Although the (frankly more credible) 'unofficial' estimate puts the cost at 250 billion euros over the next decade. Not to mention the fact it's resulted in their co2 emissions increasing in both 2015 and 2016 (https://www.cleanenergywire.org/factsheets/germanys-greenhou...). Their own environment ministry now predicts they will probably not make their 2020 co2 reductions targets.

So other than the enormous cost and increase in co2 emissions, good policy.

Yeah the:

- 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.

Yeah it's possible he's overstating the case, although he seems to have gone in to quite a bit of detail estimating the costs. But you'll see from some of the comments above those might be distorted.

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...

I didn't know the Nixon stuff. Nick Touran has a good summary of the practical difficulties if you haven't seen it https://whatisnuclear.com/reactors/msr.html#problems

Thanks for the link, it's an interesting read.

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...

My experience with development cost estimates suggests the easy parts are easy and the interesting parts might as well be splatter-paint.

> It's no coincidence that the only success story for nuclear is a country where all reactors are government-owned and don't have to turn a profit necessarily.

I don't have the numbers, but I would also have to imagine that the government-owned reactors are the most expensive on paper because there will always be political powers that will want to make it appear more expensive than it is.

Well, France has no reason to do that – they (with > 80% nuclear) have every reason to make them seem as good and effective as possible.

> And you want to give those governments nuclear reactors? Iran, Pakistan, Nigeria, etc?

No, we want to sell them access to our grid and cheap power.

> parcels of land totalling the area of Indiana

This to me is one of the biggest issue with solar (and wind) is you need huge amounts of land if you want to really try to replace what is being generated by fossil fuels. Sunlight just doesn't provide all that much energy per square foot. When we stand in the sun on a clear day, we feel comfortable warmth not searing heat.

Do all the cost estimates showing Solar is competitive include the land acquisition costs?

This is a good question. I suspect the answer is no, given most of the exuberant headlines about renewable costs don't even factor in the renewable intermittancy problem (which massively increases the total cost of generation and transmission/storage).

It's the same problem when people propose building giant solar arrays in the desert. Unfortunately, our large population centres tend to be concentrated in coastal areas, and transmission infrastructure isn't free (it's actually super expensive).

EDIT: Just to be clear, I have nothing against renewable energy (I suspect I'm coming off as a little negative here). I just hope that people aren't misled when costs aren't properly factored in (or overstated, in other cases). I actually think there's something rather special about solar PV: it's one of the few forms of electricity generation that isn't just another method of making heat to spin a turbine. Hell, despite its sophistication, fission is just a fancy way of boiling water...

One reasonable approach is, for example, Tesla's solar powered roofs along with its home integrated power wall for storage.

They claim that their solar roof costs less than a traditional roof, and lasts longer. And produces sufficient power for the house.

This approach nicely works around a lot of the very real problems associated with the traditional, more centralized solar generation methodologies.

Yes, that is a huge amount of space if you concentrate it, but think about all the idle space out there -- the many many acres of rooftops of a city, for example, or the acreage of interstate highways. There are a number of places where solar can be placed where it can be ubiquitous, but at the same time, invisible. Nuclear needs a significant amount of dedicated space, whereas solar can coexist with a significant amount of uses.

Apparently the "source" (a Forbes blogpost) of the number of deaths stretched quite far to find people who died from wind power:

http://mediamatters.org/blog/2013/03/27/forbes-reaches-to-fi...

The fully-baked lifetime costs of nuclear power are quite vigorously debated; it's hard to get accurate numbers, since in general there are unpriced government insurance policies backing the plants (i.e. the government will step in to pay the $10B cleanup costs in the unlikely event that something goes wrong).

One way of attempting to answer this question is to look at France, where 75% of their power is nuclear; their price per KWh for consumers up to 2015 was lower than the average for Europe (12th-cheapest in Europe according to Wikipedia, and 2nd-lowest for commercial), but that almost certainly doesn't price the government insurance mentioned above [2].

Note that due to a lot of electric heating in residential stock, France is very susceptible to increased demand in cold winters; in such situations they have to import from the rest of Europe, sometimes at peak hours. And as of 2016, a bunch of French reactors were shut down over safety concerns, so the prices there are spiking dramatically [3].

Another approach; let's look at cost per installed MW of power for current projects. For solar, that's now beating wind (and half the price of coal) for the newest installs, at $1.65M/MW [4]. According to the World Nuclear Association (who I'm going to assume will give the best possible numbers that can be said with a straight face), nuclear installs are ranging from ~$2.5M/MW to $7M/MW of capacity, with most cases closer to the high end of that range [5].

Now, load factor appears to be around 20% for US desert installs [6], so you need 4.5x more installed capacity to meet nuclear's load factor of 90%. Load-factor-adjusted costs: solar $8.25M/MW, nuclear $2.8M/MW - $7.8M/MW. That makes it look a lot closer than your numbers suggest. A deeper analysis would include pricing energy storage (which increases the cost of solar more than nuclear, but does need to be considered for an all-nuclear mix) and subsidies (which predominantly increases the cost of nuclear).

I haven't done more than skim your link, which seems interesting reading, and I'll dig into that later. But the from-first-principles analysis doesn't seem to be using the right costs for solar; it has `Cost: $12.3 Billion` for a 500MW CSP installation, or $24.6M/MW. That's a factor of 15 above the observed cost of installing a modern solar plant. It might be accurate for a CSP plant, but that's not the cheapest solar available.

In summary, I'm somewhat agnostic on this issue, but the energy storage issues with solar/wind seem like they could push the cost up substantially as we start to get towards majority-renewable on our grid, so I tentatively support building some new nuclear as a short-to-medium-term hedge. But we should be sure to do so with as little subsidies as possible, so that the installation price reflects the true cost, rather than kicking the can down the road on billions of dollars of decommissioning costs. And entirely new reactor designs (e.g. fast breeders) could change the calculus entirely.

Caveat emptor; I'm not in the energy business, so there's probably some missing steps here. Please let me know if you spot one.

[1]: https://www.bloomberg.com/news/articles/2016-12-15/world-ene... [2]: https://en.wikipedia.org/wiki/Nuclear_power_in_France#Manage... [3]: http://www.zerohedge.com/news/2016-10-27/french-shocked-powe... [4]: https://www.bloomberg.com/news/articles/2016-12-15/world-ene... [5]: http://www.world-nuclear.org/information-library/economic-as... [6]: http://euanmearns.com/estimating-global-solar-pv-load-factor...

My thinking is pretty similar to yours. Especially the point on subsidies, although I would modify it slightly: if subsidies are employed (explicitly, or implicitly in the case of 'nuclear insurance'), governments should aim to keep them as technologically neutral as possible. I'm not arguing against subsidies (especially at this point): if it's a choice between climate catastrophe or handing out subsidies I'd definitely go with the subsidies. However, I think it's counter-productive (and more costly in the long-run) for governments to distort alternative energy markets by 'picking winners' (as they always seem to do). In other words, subsidise equally.

Carbon pricing always seemed like the elegant, 'neutral' solution to me, but it doesn't seem very popular. Also people seem to have trouble understanding that it makes perfect sense to cycle the revenue back to individuals (e.g. tax cuts), as the point is to make co2 emission intensive activity relatively more expensive (meaning people efficiently substitute away).

On the solar costs, it sounds like you've made a fair calculation (although I'm by no means an expert either, so not really qualified to make any kind of judgement). Although one other factor (which you might have already built in to your unit price) is the cost of transmission infrastructure: renewables (particularly wind) need to be geographically distributed to ameliorate their intermittancy issues (i.e. the wind only blows in some parts or the country at any given time, the sun is only visible from some perspectives etc.). Either that or, as you pointed out, you need quite a lot of energy storage capacity. Most likely there's some optimal mixture of both.

There's two other bits of fission tech that could be real game-changers: small modular reactors (https://en.wikipedia.org/wiki/Small_modular_reactor) and sub-critical reactors (https://en.wikipedia.org/wiki/Energy_amplifier). Although I get the sense the latter is a bit more speculative (even though we have all the bits of technology to make it work in theory).

And I should probably correct something that I think I (mistakenly) said earlier as well: nuke plants, although they can scale up and down to try and match demand, it apparently is not a fast process (meaning at least some energy storage or 'peaking'/'energy sinking' capacity is required, somewhat mitigated by good predictive models of demand).