Everyone seems to be latching onto nuclear power. Nuclear power now delivers just a meager 4% percent of global power use and they want to expand it at least 10 fold.

Now it is written (https://www.scientificamerican.com/article/how-long-will-global-uranium-deposits-last/):

"According to the NEA, identified uranium resources total 5.5 million metric tons, and an additional 10.5 million metric tons remain undiscovered—a roughly 230-year supply at today's consumption rate in total. Further exploration and improvements in extraction technology are likely to at least double this estimate over time."

230 years at today's consumption rate is 23 years at 10 times current consumption. Double that at great cost is roughly 50 years before every last grain has been dug up. That is, uranium will be spend before the last power station comes online. And a ten fold expansion will still not cover the worlds current energy use, let alone expected use in the future. That means, massive amounts of money are spend building power stations that will be out of fuel in 20 years or so.

But use uranium from sea water, it is said?

It takes many cubic miles of seawater to extract enough uranium to power a country. Anyone guess how to process cubic miles of seawater (cost and energy) efficiently?

Thorium reactors? That is 1950's "technology" that is still not in production. Why? Maybe because it is a radioactive nightmare?

It's possible to use nuclear reactors to make more fuel. Breeder reactors aren't commercially interesting because uranium fuel is currently cheap, and many people are unhappy with the idea of plutonium becoming widely available - it's easier to concentrate Pu to a purity you could use in a nuclear bomb than with U-235. Reprocessing 'spent' fuel is also possible and has the same issues. Either would be more practical than scraping up the least concentrated ores let alone trying to extract it from seawater.

Never seen any numbers on that. Until then, I have to consider this speculation.

I do know that breeder reactors and reprocessing plants get awfully radioactive over time. France does a lot of reprocessing and has to cycle personnel very quickly in their plants to keep them from getting radiation sickness.

Some of this is due to the design of the plants. Certainly Sellafield was built as a bomb grade plutonium production facility and little thought was given to the future. The plants in the USA were they made the plutonium had some absolutely shocking cut corners and suffered quite a few 'events'. The race to make bombs will forever haunt the civilian nuclear power industry.

Most of the spent fuel we have stored around the world will primarily contain Pu240 rather than Pu239 as it will have been in the reactor too long so the worries about bombs is not actually a real worry.

I would hazard a guess that those figures are based on a once through fuel cycle. Reprocessing increases that by a factor of 20-30x.

A move away from thermal neutron reactors might add another zero to the end of that multiplier.

Sellafield has several (many?) thousand tonnes of un-reprocessed mostly once-through magnox fuel sitting in its ponds. And the US is stacking up its 'spent' fuel in dry stores. If we can come up with a way to make use of all that then we have a ready supply already out of the ground.

Thorium reactors? That is 1950's "technology" that is still not in production. Why? Maybe because it is a radioactive nightmare?

Like most things nuclear related, it's only a 'radioactive nightmare' to people who've been mislead by decades of anti-nuclear FUD. Or more recently, the 'renewables' lobby who'd much rather we wasted money on windmills and solar instead of clean, ultra low carbon nuclear.

Problem with LFTRs is although test reactors were built in the '50s, they didn't help with the nuclear weapons programmes, so were abandoned. Now there's renewed interest because they're potentially safer, thorium's very abundant and they can recycle nuclear 'waste' as fuel.

As for uranium, Tim Worstall, lately of these parts explained that people mistake current proven reserves for potential uranium availability. If there was no demand for uranium, nobody went prospecting for it. Now demand is increasing, more activity is taking place.

Sounds good to me. The fact that we've been using relatively unsafe reactors does not mean we have to keep using them, or stop trying new designs.

Nuclear power now delivers just a meager 4% percent of global power use and they want to expand it at least 10 fold.

As long as there are coal power plants (40% of total), we should do everything we can to find other sources. It's much worse than nuclear on all dimensions, even radioactivity, but especially in the number of deaths it causes.

I would have preferred to see a statement by Google that they will improve functionality/power ratio of AI by a factor a two every two years at least until 2035. I mean, that's their core talent right? And nuclear energy isn't.

Their core talent is getting as much as possible for as little as possible.

If they can throw the media a few bones about how they are cleaning up their energy needs then it can be business as usual building out data barns.

"Its designs are scoped to produce 150 MWE – 150 million watts of energy"

A watt's a unit of power; a Watt of energy isn't really a meaningful expression.

Possibly meant to be megawatt electric (MWe) - one million watts of electric capacity?

Watt = Joule/second, energy/second. That is the correct way to describe the output of a power plant.

Possibly meant to be megawatt electric (MWe) - one million watts of electric capacity?

I was wondering that. I've also seen reactors specified as MWt, so their thermal output. MWE seems like a rather non-standard and redundant metric. Then again, a lot of energy related press releases confuse power and energy, or deliberately mislead with units like MWh or 'enough to power X homes'.

I also think that energy wasters like AlphaGoo should be buying reactors, not power purchase agreements. But I guess if they're doing their sums right, buying 50% of the output should be paying for 50% of the cost of the reactor. Risk is if the PPA ends up distorting the price to other consumers, but that's something for energy regulators to deal with. I guess the good thing is if AI bubblers are funding increased generating capacity, when that bubble bursts, there might be some excess generating capacity and then lower energy prices.

Thermal power is what comes out the reactor or whatever is used to raise steam, electrical power is what comes out the generator. Not all of the thermal energy converts to electrical hence cooling towers etc..

If AI can kick start SMR's it'll be the most useful thing it can do for humanity. I suspect though that once contractors and govts gold plate everything, a piddly little 150MW(e) installation will be totally uneconomical due to the thinking, at least in the UK, revolving around massive 2GW+ plants. You could group a bunch of them together and share logistics but that defeats the advantage of them being distributed and close to points of consumption.

Just leave that to the Third Parties that you sell the Plutonium you've manufactured to.