Amazon buys stake in nuclear energy developer in push to power data centres
ft.com
ft.com
It's a good thing to have a diverse portfolio of energy solutions to help avoid single points of failure in the network. Especially as something like solar continues to become more and more efficient.
https://www.wired.com/story/nuclear-power-plants-struggling-...
"Amidst a slow-burning heat wave that has killed hundreds and sparked intense wildfires across Western Europe, and combined with already low water levels due to drought, the Rhône’s water has gotten too hot for the job. It’s no longer possible to cool reactors without expelling water downstream that’s so hot as to extinguish aquatic life. So a few weeks ago, Électricité de France (EDF) began powering down some reactors along the Rhône and a second major river in the south, the Garonne. That’s by now a familiar story: Similar shutdowns due to drought and heat occurred in 2018 and 2019. This summer’s cuts, combined with malfunctions and maintenance on other reactors, have helped reduce France’s nuclear power output by nearly 50 percent."
France is about the best existing case for nuclear, incidentally.
Furthermore, nuclear plants don't need to be cooled with potable water. They can be cooled with ocean water, or with waste water. In fact, seawater cooling is the most popular form of cooling. Only 15% of nuclear plants are cooled with river water.
https://world-nuclear.org/information-library/current-and-fu....
> Nuclear power plants in Europe have been forced to cut back electricity production because of warmer-than-usual seawater.
And YES It is a problem which needs to be fixed, increases risk and costs. Don't say its not a problem.
Renewable generation itself is cheap. But what's expensive (or straight up unfeasible) is everything required to mitigate the intermittent production. Storage at the scale of tens of terawatt hours can't even be feasibly built with current technologies. Moving electricity over thousands of miles, across mountain ranges, would require HVDC lines to be constructed in very rugged terrain.
solar on roofs doesn't take anything from local ecosystem. Solar above car parks neither.
Solar on a home is such a simple, affordable and save solution, why are you 'non-fan'? which indicates you hate it? How much are you against it? So much that you prefer burning coal over it?
1. http://large.stanford.edu/courses/2018/ph241/duboc2/
2. https://world-nuclear.org/information-library/current-and-fu...
And even in France this was only a problem because of their terrible delayed maintenance.
> France is about the best existing case for nuclear, incidentally.
No it isn't. France has done essentially nothing for 30+ years. Has done little maintained, their reactors aren't up to date.
The generation after the generation that built the reactor has always resented the system and wanted to rip it out. They literally decided to retire it by 2035 despite having no plan to replace it.
I mean, we aren’t disagreeing.
They’re the best case, and that’s not great. It a bit like true communism - real success is just over that next hill.
Lakes can dry up. So that's no better than a river I would think.
Either way, there are no fully safe and permanent sources of cooling water, is what I'm taking away from this.
Which ones? The North American Great Lakes, maybe? Anything else?
The Aral Sea was the third largest lake in the world. Within the human time scale of the last six decades, it's lost 90% of its area.
Hopefully this changes once every nuclear project isn't some complex bespoke thing that is likely to be late and over budget.
In fact, any country that has built them in mass figured out how to do it cost effective and quickly.
The reality is, its only not cost effective and quick when a country only builds a single reactor as a vanity project to keep the industry alive.
No country that seriously tried to quickly increase production with nuclear has failed.
"Yes sir, all went according to plan!!"
Even though it obviously did not all go according to plan.
The market doesn't want it, banks don't want to finance it, researchers aren't interested and startups can't afford to.
You can't fight against market dynamics when you're talking about capital expenditure this high.
Edit: also, sodium fast reactors have existed for 20 years. The R&D has mostly already been done for that tech. But the lack of projects make it stuck to TRL 8.
Thats just not the case.
Solar is used together with wind and storage.
And there are very little energy systems which rwould relia only on solar. More north you go, there is often water.
All moot points if they last indefinitely where you are, serious points if they don't.
And those dumb turbines and solar panels produce power at a fraction of the cost of nuclear. And every year they get cheaper and more efficient. Which is why the market has continually decided over and over again to go in this direction.
Sadly, also for a fraction of the time.
Until storage is solved in a satisfactory way, solar and wind will remain auxiliary. And the capacity should be in gigawatt-days.
But trying to hit some purity target of 100% is irrelevant. There are lower hanging fruit for decarbonisation than the last few percent of power generation.
https://www.sciencedirect.com/science/article/pii/S030626192...
We need a battery that is safe and mandate every home to have one. Or provide incentive for it to be installed. We shouldn't wait for super battery to be install in a central place, instead every home could afford one an provide high quality electrical power.
Surely instead of a giant Water boiler in a lot of homes we could have a giant battery instead.
And storage is being solved as we have more batteries being deployed and grid supply capable EVs becoming more popular.
The market is deciding where money should be allocated and it's simply not going towards nuclear.
Because you'd get paid for it, and you like free money.
And it should be pretty trivial to set an option to ensure you always have the necessary range for your daily commute by a little bit before you leave your home.
And why wouldn't it be? The whole point here is for power companies to avoid buying as many batteries themselves, and their own batteries depreciate too.
This isn't hard -- the power company adjusts the price it pays minute per minute, and you set the threshold at which it is profitable for you, taking battery cycles into account. And it becomes a classic supply and demand curve -- it's Econ 101.
I have a 100kWh battery and can drive with this 3 weeks around without charging at all.
Why would i not want to leverage this?
And results from storage systems show that you can charge and discarge car batteries a lot more often without real degeneration when you do this a lot more stable than when driving.
Also it reduces the overall straine to the power grid. If you fill your cars battery with local solar, you are transporting less energy across the whole grid. If you discharge it locally, again less overall energy which needs to be transfered across the whole grid.
How this would work at scale? easy: in my city for example there is one local power company and they offer a charging solution for my EV. They have a few powerplants locally here too. They have everything they need.
Also overall solar energy prediction for the next day is very good. You can easily save a lot of money by leveraging this up front.
https://assets.solar.com/wp-content/uploads/2023/01/Carbon-f...
I'd say 1-5% of coal is pretty good.
Also as we make the power grid cleaner and switch to electric vehicles, the CO2 used in manufacturing goes down.
Energy return in investment is <10x for solar, but more like 100x for nuclear. We have an energy budget to solver climate change. This is nothing to do with how much things cost.
Rooftop solar is worse, but 4x worse than wind/nuclear is still quite good (and non-rooftop solar is almost certainly better than that). Solar's energy return on investment is higher than 10x by my searching, and we should be mass producing all three technologies at this point.
I suppose that instead of eliminating the processes that produce large amounts of CO2, we should embrace them, learn to capture the CO2 where it's produced in high concentrations (like steel plants or cement furnaces), and either bind the carbon in non-volatile ways, like in plastics, or produce fuel from it again and close the loop.
Energy return in investment is <10x for solar, but more like 100x for nuclear. We have an energy budget to solver climate change. This is nothing to do with how much things cost.
And if we could build nuclear properly in a reasonable fast way around the globe, i would be for investing it heavily anyway i'm just not controling it in anyway and the only thing actually working is solar, wind and battery.
Low risk, broad spectrum investment (private, small companies, big, basically everyone can)
If you need baseload, and need it carbon-free, your only option currently is nuclear. It's terribly encumbered, but apparently it's still less of an impasse that large-scale electricity storage currently is.
What about geothermal, tidal, hydro, etc?
- Hydrothermal: great if you're in Iceland, or near Yellowstone, or other such place blessed by heated rock being close to the surface. Not as great if you need to drill 7 km to reach it.
- Hydro: great if you have a lot of mountains and rivers, like Switzerland or Norway. Harder if you don't, like in much of Texas, to say nothing of Florida.
- Tidal: great if you have a sea shore, preferably with narrow bays / fjords. But if you're in a place like Turkmenistan (one of the two doubly-landlocked countries), you resort to drilling for and burning methane %)
* most of the "good sites" have already been taken
* hydro will regularly silt up, requiring constant dredging
* hydro does not work as well or at all in drought conditions, which are increasingly common
* hydro dams require reservoirs, which in addition to the obvious disruption of the displacement of thousands or even a million plus people in the case of Three Gorges, can end up emitting large amounts of carbon as flooded vegetation decomposes.
* hydro is extremely disruptive to migratory fish. the success rate of interventions like fish ladders is in the low single digit percents. so you can get a bunch of carbon free power but also destroy an ecosystem in the process, not to mention any downstream fishing. and you see similar effects with other things that downstream users might want from the river like fresh silt.
That they don't work when the sun doesn't shine or the wind doesn't blow is such a tired and trite remake.
Yes, they don't provide consistent output on their own, but storage exists. If you need electricity store it in batteries or pumped hydro. If you need it for heating store it in thermal batteries.
I think this varies depending on where you live. In California, Cal-iso has a really cool dashboard that seems to show that they have enough battery storage to hold about 90 minutes of daytime solar generation.
Definitely not something that can handle a fully cloudy day yet (presumably it's cheaper to keep some natty gas plants ready to spin up or import than to store more excess solar?)
We need 30x, or maybe 100x more storage. After that, we could live on renewables only, and keep gas-fired plants only as a disaster-recovery tool, like diesels in datacenters.
Until then, we need stable generation which does not spew CO2, which is, well, nuclear. It's hard to tell if ubiquitous cheap utility-scale batteries emerge in 5 years or 50 years.
It’s just insufficient. Solar + storage is cost competitive with nuclear. The problem is we have a bajillion needs for batteries, a Balkanised global market and wholly insufficient production forecasts over the next decade without gas or nuclear support. Voters seem to like gas. Private participants are choosing nuclear where they can.
Good to be a gas exporter for the next half century or so.
We need batteries that cost, say, $10 / kWh, are not (as) flammable, not toxic (not Pb, not NiCd), and don't degrade too fast.
With that, they can be large, heavy, have low specific charge, require high or low temperatures (within reason), etc. For a large utility-grade installation all these qualities are not hugely important. There is plenty of space under solar panels and around wind turbines anyway.
Before that, selling methane remains a very good business indeed.
China is building a lot more, too. It will probably have 80% of the world’s energy storage capacity in a decade.
We just don’t talk about them in the West for some reason. I guess battery tech is more appealing somehow but it’s not cheaper nor more environmentally friendly, all in all.
I hope that now once there are some serious users of power, they might build some of these plants, and then governments in the West will wake up to how good the plants are.
Cool. Yes. Now scale that by two orders of magnitude.
Here’s the kicker: getting approval to put an SMR in Wyoming is easier than flooding another valley or buying massive contiguous plots. Nuclear, if these guys can execute, which is a big if, promises to be as agile as gas generators. Our current default when we have energy shortfalls.
My ancestors (along with natives) were run off from their valley home in PA when a dam was built. Out here in the Sonoran Desert, we've got a lake where the river should be, and the indigeneous (among others) are still pissed about the dam that was pretty much politically forced on all of us.
Conversely, there are groups who absolutely rejoice when a dam is decommissioned and demolished. It's party time when that land is reclaimed, and the water flows again naturally.
A nuclear plant in the USA takes oh about 18 years to build. That's 18 years of solar, wind, and storage growth and cheapening.
In my region wind generation dipped down to only 1pc of capacity for a day or two last winter which seems to be the floor as far as I can tell.
That tells me that we only need to increase capacity by 100x to get the coverage we want, and as a bonus we get all that surplus capacity most of the time to use on non-essential industry.
Try doing the figures, we can't scale batteries to cover these drops.
Especially on continent-sized electrical grid scales. We put the farms in windy spots, and we pick a variety of spots.
First of no one is saying its just wind and solar. You say wind, solar and energy storage (batteries).
And second, if more people would buy EVs, we would already have A LOT MORE storage available. My EV has 100kWh and could heat a whole modern build house for 2 full days in winter.
But we need, as you said, a baseload. Nobody is really discussing how to create this, they're just saying we should magically invent an incredibly dense battery (more energy-dense than hydrocarbons, ideally, or let's say within 50% as useful) using technologies nobody can even fathom. Meanwhile, we're seeing breakthroughs in fusion tech, miles ahead of the usefulness of fission, which is miles ahead of anything else we have right now.
I just don't understand the aversion to a clean, renewable power source.
Solar growing about 30% a year, nukes about 0%.
Storage is coming along too.
(some source stuff https://news.ycombinator.com/item?id=41601621)
This Amazon investment aims to only have 5GW of nuclear on the grid 15 years from now. That's not a solution to climate change, it's not even a rounding error. It rises to a rounding error on the suddenly discovered need for new energy for AI, but it's a decade late for that. In 15 years, the grid will be completely decarbonized by solar and batteries and wind, existing nuclear/hydro, and probably some new tech like enhanced geothermal. This is 2039 energy environment that nuclear will enter into, and it will have to compete with 15 more years of prices falling on batteries and renewables. SMRs can't compete on price with today's price for new large nuclear, todays new large nuclear can't compete on price with today's solar/wind/batteries, and today's solar/wind/batteries sure as hell won't compete with 15 years of prices falling. SMRs like this have no hope to be a competitive product.
Nuclear has overpromised and underdelivered for 60 years. We should have kept existing plants running. If our anscestors had lit money on fire to produce a bunch of reactors 40 years ago, we'd be in a much better position.
But we'd also be in a much better position had they not pooh-poohed solar and made bigger investments sooner, driving forward solar revolution by 15-30 years from where it is now.
We'd have been in a much better position if there had been a massive investment in battery technology 15-30 years earlier, making EVs feasible sooner, and revolutionizing the grid with storage sooner.
Solar and batteries will be the foundation of the future, because they are technology that gets cheaper the more we invest in them. Nuclear might be around in 50 years, maybe not, but it won't be any cheaper or more affordable. It's a technology that barely moves the more we spend on it, and can sometimes be economically efficient in the best of cases. But the average case for building nuclear is big price overruns, and it's not uncommon for utilities to be brought to the point of bankruptcy.
We have better technology available, today, to be deployed. Let's do it. Nuclear is a pipe dream from people that haven't run the numbers, or have run the numbers they have gotten them drastically wrong. The entire history of nuclear energy in this country is of people not running the numbers or running the numbers and getting them drastically wrong.
Take a look at what Germany did to itself. Compare France.
The anti-nuclear wind/solar proponents are dangerously wrong.
https://www.cleanenergywire.org/factsheets/germanys-energy-c...
Isn't that the goal? If they can do that without nuclear power - great.
Despite France's >50% nuclear power they have a much higher ratio of "dirty" power compared to Germany.
https://en.wikipedia.org/wiki/Energy_in_Germany#/media/File:...
Germany's energy is 80% fossil fuels. France's is 50%. I'm not sure how you reached the conclusion that Germany's energy mix is less "dirty" than Frances.
Where did you get that number from?
https://www.iea.org/countries/germany
> Renewables 39.4% share of power generation, 2022
vs. France's 26.1%.
Edit: I guess you might be talking about consumption whereas I'm talking about production.
Climate change fact #2: energy consumption is not a cause of climate change, except for when it causes CO2 emission.
Climate change fact #3: reduced CO2 emission will not stop climate change.
Reducing energy usage is a losing climate strategy in three ways 1) it doesn't solve the problem, cutting emissions 50% does jack shit, we need to get to 0% emissions and then negative emissions. 2) It's politically ineffective because only some people and countries will actually reduce emissions. 3) It's political suicide because it validates false anti-climate change propaganda. 4) It's highly socially regressive because these dictates demand that the burden be placed mostly on developing countries.
Reason #1 is enough to disqualify the "reduce energy usage" as a climate strategy, but really there's zero way to look at it and think that there's any validity. Please find better sources and stop spreading this counterproductive idea.
The reality is that this is all nuclear can rest its hopes on. Industry specific applications, because for general electricity generation it is hopelessly non-competitive.
Thermal/Fast Breeders are needed to use those fuel sources. Only Russia commercialiced them, because western nations stopped funding r&d (EBR2 in the US and Superphênix in France)
There's a good article on the technical problems here https://thebulletin.org/2022/06/molten-salt-reactors-were-tr... One quote from that:
>During its operational lifetime, the Molten Salt Reactor Experiment was shut down 225 times. Of these 225 interruptions, only 58 were planned. The remaining interruptions were due to various technical problems, including: “chronic plugging” of the pipes...
>...These problems remain relevant. Even today, no material can perform satisfactorily in the high-radiation, high-temperature, and corrosive environment inside a molten salt reactor.
I'm not sure anyone has made one that powered anything or produced electricity. The MSRE just intermittently heated the air when it was working.
Apparently although it only ran from 1965 to 1969 and produced a peak of 8MW, it's still costing about $10m per year to manage the waste produced. In the 1990s they found uranium in the waste had migrated leading to the possibility of an accidental criticality.
Still the Chinese supposed to be running an experimental one similar to the MSRE in the Gobi desert next year. I'm not holding my breath for commercial operation though.
I applaud these efforts. The amount of energy that's going to be burned by these things is crazy. Whether or not that energy use is wasted is an open question.
I think it's a positive thing if the companies making these bets don't offload the costs of those bets onto public utilities, and also that they're looking at energy sources that have less of an adverse environmental impact than things like coal.
Kudos, guys.
Wonder why that is?
> I think it's a positive thing if the companies making these bets don't offload the costs of those bets onto public utilities
This is written as if with concern for public utilities which makes me wonder: Do you think they’ll stop using public utilities? Do public utilities want that? How long until these companies try to outright replace public utilities?
Still want that public money.
https://www.washingtonpost.com/business/2024/10/03/nuclear-m...
> Whether or not that energy use is wasted is an open question.
Slop is a waste for us, but a boon in: data harvesting, surveillance, labor and union costs; for them.
> also that they're looking at energy sources that have less of an adverse environmental impact than things like coal.
You think these companies made this decision with altruistic concern for the environment?!
M$FT makes recall a required dependency in explorer.exe -> “I'm so glad I can finally compliment these guys for caring about the environment.”
Make it make sense.
That said, it's as important to acknowledge when they do something positive, even a little bit, as it is to call them out when they do bad things. More important, even, since if all a person can do is condemn then that person becomes less effective in terms of encouraging change. You want both carrot and stick.
> You think these companies made this decision with altruistic concern for the environment?!
Of course not. Nothing any of these companies do is altruistic. It's all about the Benjamins, baby.
They likely will stop using electric utilities. It's just simpler this way, more predictable.
> Do public utilities want that?
Maybe they do; a huge consumer usually negotiates special rates, and may require more costly infrastructure work because the capacity is not sufficient for them.
> made this decision with altruistic concern for the environment?!
No, more likely with an egotistic concern for the environment. Billionaires don't have a spare copy of Earth (Mars is even worse), and if you've got billions and can actually move the needle of the climate change to keep Earth in a better shape for you and your children, won't you?
* A major advance in spaceflight [0]
* a ton of private investment in nuclear power [1]
* AI models performing at PhD-level on some tasks [2]
I know it can feel low-status to admire these accomplishments -- it feels like we're aligning with/submitting to the people behind them -- but I perceive technological growth to be accelerating across a bunch of fields that matter to me.
[0] https://news.ycombinator.com/item?id=41827362
the way I see it, virtual goods (games, digital goods, etc) will continue to be ever more free and lower cost, while nothing much changes in the real world.
All this stuff used to be high-end research with unstable and hardly usable results like 10 years ago.
I've noticed the exact opposite over the last year or so. They've become riddled with errors of the sort that actively mislead people about what was actually said.
> Does it make your very real life easier?
So far, generative AI has not made my very real life any easier. It has made it more uncertain, though, and has made it harder to trust anything I am not seeing/hearing in person.
It seems to be an incredible feat of engineering, no doubts there, but how significant are the practical implications?
I found this from Musk, which is his usual over the top optimistic self:
In an interview to YouTube channel Everyday Astronaut, Musk said that his vision is that Mechazilla will one day be able to turn around and set a rocket back on the launchpad, perhaps as little as 30 minutes after touchdown.
Heavy Booster is 232 feet tall, and 29.5 feet across, and weighs something like 275 tons
These things are no easy build, they take time, high skill levels, resources, etc.
30 minutes turn around on reuse is BS for now, they'll take weeks of fine toothed retesting and examination to clear them for reuse - but that'll still be a considerable saving of time, money, and resources compared to a complete new build from scratch.
What's your Engineering background? Civil, Mechanical? Ever worked on any of those 380+ metre tall North Sea Oil and Gas platforms?
I work in Software but I'm a massive tinkerer and have done a lot of work wit my hands. I'm not a structural or mechanical engineer. Most of my knowledge is just things I've picked up doing renovations, fixing my car etc.
The Falcon 9s have been landed, checked out, refueled and flown again for a while now. The record is currently 23 flights from a single booster. I don't see how this is fundamentally different - reusing the booster that is. The upper stage is a different matter.
I'm not saying that AI is inherently evil and has no potential, but just ignoring the cost is misleading. It's like saying oil is wonderful and pretending global warming isn't a thing.
The tech may be "hard" but that doesn't mean it's a net positive for society, which to me is what really matters. Of course this is all subjective and nobody can predict the future, but I just find this blind optimism to be self-destructive.
But anyway; let's take it at the best case curiosity. I do like spaceflight. Is it a problem to you what people like? Does everything you like specifically address human needs? I find this confusing.
Regardless, all life on Earth is going to die. I see it as a "major human need" to avoid death of our civilization and all other life we know of in the whole universe, if it becomes possible. And it does seem more possible now - because of the advances SpaceX has made with hard rocket tech in the last few years, specifically.
I would say that space flight is cool as a potential experience -- I hope that it's safe and cheap enough one day that I can go -- and I also think that life is good [0], so if we can spread life to more planets, that's good.
[0] https://www.overcomingbias.com/p/this-is-the-dream-timehtml
Any metal you can think of exists in phenomenal abundance there, and should be able to be exploited at minimal cost in the long run. Having huge supplies of cheap platinum, gold, nickel, cobalt etc would be extremely good for humanity. It also means we don't need to have dirty and ugly mines on earth.
There is also the military angle. If the west lets China control orbit, we're in big trouble.
If you're building a megastructure in space, that's a different story. But please understand that mining gold or platinum from asteroids, at (literally) astronomical cost, will not do anything to advance the state of life on Earth except, at best, reduce the price of gold and platinum, which are not societal bottlenecks.
This article tells a very different story: https://www.cnet.com/science/rare-asteroids-near-earth-may-b...
Platinum has a lot of industrial uses today at $1000/oz: https://market-news-insights-jpx.com/ose/commodities/article...
At $10/oz it could be used for vastly more purposes.
We should be beyond ready by the time that comes. It's not an immediate or daily human need, but it IS a human need.
On that timescale we could use a gravitational tractor to fix it, if we insist keeping the planet around.
250 million years is an estimate for when the formation of a new supercontinent results in sufficient volcanic activity to drastically increase CO2 in the atmosphere and probably kill off all large mammals. 500 million is when C3 photosynthesis stops being possible and virtually all plants are gone, which would collapse all terrestrial ecosystems and leave behind very little animal life, probably none. There seems to be a very high likelihood of extinction level asteroid strikes happening well before either of these.
But we're talking here about a span of time that is a thousand times longer than anatomically modern humans have existed up to this point. Given how far we've come since then, I don't know how you can possibly speculate what kinds of capabilities we might have by then to synthesize breathable air and food from raw disintegrated atoms of anything. If you look billions of years into the future, then it's going to get hot enough to sterilize the planet of any life whatsoever, which we probably can't overcome. If we can terraform other planets, we can terraform Earth itself, which would seemingly overcome any other challenge short of triple the heat that is eventually coming.
It seems maybe a bit premature to think this is something currently living humans should worry about figuring out how to escape from.
We might also note that, given the compartively short time it took humans to come about after the K-Pg event, it's probably reasonable to expect there is more than enough time before these "possibly all life killer" type far future things happen for some other kind of intelligent life that develops civilization and technology to replace humans if we go extinct by some means other than the planet being totally destroyed.
A lot of time, but it is irrefutably a requirement, and as we research it, we'll learn lots of things to use here on Earth, too.
And, personally, I want a space telescope a thousand times bigger than James Webb. That's my biggest need after food, shelter, health and human connection.
An asteroid-ravaged Earth is still more habitable than any planet in the solar system.
b) There is not a ton of private investment in nuclear power. It is a handful of LOI that history has shown usually doesn't translate to much.
c) AI models do not perform at PhD level. They can solve some PhD level tasks but as Apple showed in their research the minute you swap out variables or add irrelevant information they fall apart. So clearly not evidence of intelligence.
Not to be negative on anything because I do think we are in an incredible era but these aren't what I would consider the best examples.
I'm not sure they proved it but they put out a paper arguing that.
Google commits to buying power generated by nuclear-energy startup Kairos Power
https://news.ycombinator.com/item?id=41840769
Three Mile Island nuclear plant restart in Microsoft AI power deal
This announcement has downside. If it fails, Amazon's investment goes to zero. Which means they likely did significant due diligence. This is a real commitment.
https://news.ycombinator.com/item?id=41505514
That makes four big tech companies on board with nuclear in 30 days.
In other words, a lot of engineers could work quite happily in both fields.
Inference is somewhat elastic - people want relatively low latency - they might be able to tolerate an extra round-trip around the world, but probably not waiting until a time when there is more total capacity.
However, the big impediment to using cheap and green power is the capital cost of the training hardware; that can't be moved around in a hurry, so its capacity goes unused when the sun isn't shining and the wind isn't blowing. Much of the cost of the high-end data centre oriented GPU hardware is likely not incremental cost for nvidia - it is recovering fixed costs, or profit. In addition, people buying the hardware fear it will depreciate, so they have a limited time frame to use it.
So it is fair to say that it is nvidia's pricing strategy that is a significant driver of Google, Microsoft and Amazon investing in nuclear.
Therefore it makes sense to use even very expensive electricity to run GPUs. Thus Musk insanely running generators to keep his new DC running. A little money can solve really really bad planning. And these SMRs, if ever built, will be very expensive electricity, a long long time from today.
Investing in nuclear has nothing to do with satisfying current AI power needs. This announcement aims to have 5GW running in 15 years. A small amount of energy too far away.
When has any tech company planned for energy 15 years away? Or any other specific R&D need 15 years away? Never.
These announcements are about vibes and politics. I hope it means that the nuclear industry, for once in its history, actually delivers on its promises and doesn't fail. But the history of nuclear is to fail fail fail at every opportunity, so I'm not optimistic.
Also, note the investment size, $500M. Basically nothing. That would buy a DC a bunch of power right now, in the form of solar and the batteries to convert the solar to constant output. Building the 5GW of SMR nuclear is going to cost a minimum of $50B, probably $100B or $150B. It's not going to be competitive with today's prices on solar and batteries. The nuclear purchase is all virtue signalling, not based in actual needs or solutions.
There are some great things about Helium though. One is that Helium-4 (which represents 99.9998%) is the only isotope in the universe that has exactly zero neutron absorption cross-section. All other coolants absorb some of the neutrons in a reactor, but helium doesn't. It is also an inert gas, so it does not pose the various corrosion problems posed by other coolants and moderators (like water for NuScale, FLiBe for Kairos, sodium for Natrium, lead for Westinghouse).
China did already put in production two reactors of the same type as Xe-100, the HTR-PMs. They appear to be working just fine.
Good luck to X-Energy.
Same reason it makes sense to build as much wind and solar with batteries plus nukes as possible: scale. (And false dichotomy: they’re also investing massively in wind, solar and batteries.)
We didn’t produce enough batteries (or solar panels or wind turbines, for that matter) fast enough before . For solar and wind, moreover, approval difficult scales with land area. You’d think nukes would be harder, but it’s apparently easier to get a state to let you YOLO on land rights when it’s on a small footprint.
Solar and wind (with batteries) is still cheaper. Gas is politically problematic for Big Tech, so they go with nuclear. The voting public in America and Europe, on the other hand, chose gas.
Do you have sources for your time scale?
The acceleration is massive and from a massively tiny baseline. AI is predicted to add 0.8% energy growth a year to American power use [1]; lots of that is going into gas.
> unless you have huge overcapacity of all of them
There is the political reality that what’s been built has voting employees and tangibility in a way what is to be built does not. That’s the danger in our deployment of gas turbines. If we hit an energy surplus, the first to get cut will be things not yet built, even if those are renewables. (Think: phase out of subsidies, maybe even grid charges.) Some enlightened jurisdictions will continue shutting down gas turbines to replace them with solar, maybe even nukes, but most won’t.
[1] https://www.goldmansachs.com/insights/articles/AI-poised-to-...
And given past performance, the nuclear projects being proposed here are projected to add 0% energy capacity to the grid within the next 20 years.
So the question then is: why do the nuclear deals make Hacker News and the solar deals don't?
Price and availability of electricity and power is more or less global, however datacenter customers are in the situation where they need to power a city with electricity in a location where there is neither an existing city nor its generation capactiy.
Law of diminishing marginal returns: you don’t. You either need to pay a premium for expedited delivery or eat the time value of delays.
I was under the impression that whatever drop in consumer usage exists overnight is made up for with ML training, video transcoding, etc. That there's never any shortage of tasks to run.
Do you have a source for that, how wide is the variation for different reactor types and operators, does it apply to individual reactors or whole sites when there are several reactors on one site?
https://pris.iaea.org/PRIS/WorldStatistics/ThreeYrsEnergyAva...
Thanks.
I know what I would bet on.
Or you can buy 480 Tesla MegaPacks (about 6 acres) and 1600 acres (it's roughly 1 MW per 5 acres) of solar panels to run the DC and charge the batteries during the day. Sure some can go on the building, but you'd still need way more acreage than that can provide.
I know what I would bet on too.
Yes. Uranium is (currently?) cheap and plentiful, to the point that nuclear 'waste' fuel reprocessing isn't generally economically worth it: it's cheaper to store in and buy new fuel.
If the price of uranium does ever spike or becomes harder to get to, then reprocessing can become an option.
https://news.ycombinator.com/item?id=41661768
which includes a risky method of profiting from the adoption of nuclear power.
More detail here:
Quite easily chemically separated.
The meltdown at Fukushima amounted to a very small proportion of the total damage incurred by the tsunami. No deaths due to radiation besides a couple plant workers are expected. The exclusion zone was cleaned up and reopened after about a decade. The learnings from this event should be that nuclear isn't actually all that risky, relative to how much carbon-free energy it generates.
We already have a perfectly good fusion reactor. It's about 1 AU away, and it's beaming more power to us than any Kardashev Type I civilization could ever use (and small-scale fusion will be useless to any Type II civilization anyway).
So a fusion reaction will gradually irradiate the generator that is creating the reaction.
A molten salt reactor which consumes virtually all of its nuclear fuel as the closest approximation to your desire to have nuclear reactors without nuclear waste.
Edit: I recall that helium 3 fusion may actually be pretty neutron free. But you got to get your hands on a lot of helium 3
No privatizing the profits but then socializing the cost of the waste disposal (and leaks).
If the public has to pay for it as a EPA superfund site for cleanup, well then Amazon should then be sued out of existence for this.
From Admiral Rickover's 'Paper Reactor' memo (1953)[1]: "An academic reactor or reactor plant almost always has the following basic characteristics:
1. It is simple. 2. It is small. 3. It is cheap. 4. It is light. 5. It can be built very quickly. 6. It is very flexible in purpose (“omnibus reactor”) 7. Very little development is required. It will use mostly “off-the-shelf” components. 8. The reactor is in the study phase. It is not being built now." (Emphasis mine, again.)
I've seen so many of these come (and go), I think they actually use Rickover's paper as a blueprint for their marketing pitch. Xe's site on the Xe-100[2], "A simple design & modularized components" where, "Our elegant and simple design maximizes the use of off-the-shelf components manufactured and shipped to site using existing road & rail." <shakes head>
I think safe and affordable nuclear power could do a great deal of good. But I've seen "small modular reactor real soon" (including pebble-beds) for thirty years, so I'm reflexively skeptical. I checked to see if I could find any evidence this one is beyond the paper reactor stage, but my search-fu was insufficient to find it, presuming it exists.
[1] https://whatisnuclear.com/rickover.html (The whole paper is well worth reading, IMHO) [2] https://x-energy.com/reactors/xe-100
Let's begin with a quote from Yann LeCun (Vice-President, Chief AI Scientist at Meta):
AI datacenters will be built next to energy production sites that can produce
gigawatt-scale, low-cost, low-emission electricity continuously.
Basically, next to nuclear power plants.
The advantage is that there is no need for expensive and wasteful
long-distance distribution infrastructure.
Note: Yes, solar and wind are nice and all, but they require lots of land
and massive-scale energy storage systems for when there is too little sun
and/or wind. Neither simple nor cheap.
No battery farm can protect a solar/wind grid from an arbitrarily extended period of bad weather. If you have battery backup sufficient for time T and the weather doesn't cooperate for time T+1, you're in trouble.
Even a day or two of battery backup eliminates the cost advantage of solar/wind. Battery backup postpones the "range anxiety deadline" but cannot remove it. Fundamentally, solar and wind are not baseload power solutions. They are intermittent and unreliable.
Nuclear fission is the only clean baseload power source that can be widely adopted (cf. hydro). After 70 years of working with fission reactors, we know how to build and operate them at 95%+ efficiency (https://www.energy.gov/ne/articles/what-generation-capacity). Vogtle 3 and 4 have been operating at 100%.
Today there are 440 nuclear reactors operating in 32 countries.
Nuclear fission power plants are expensive to build but once built the plant can last 50 years (probably 80 years, maybe more). The unenriched uranium fuel is very cheap (https://www.cameco.com/invest/markets/uranium-price), perhaps 5% of the cost of running the plant.
This is in stark contrast to natural gas, where the plant is less expensive to build, but then fuel costs rapidly accumulate. The fossil fuel is the dominant cost of running the plant. And natural gas is a poor choice if greenhouse emissions matter.
Google is funding construction of 7 nuclear reactors. Microsoft is paying $100/MWh for 20 years to restart an 819 MW reactor at Three Mile Island. Sam Altman owns a stake in Oklo, a small modular reactor company. Bill Gates owns a stake in his TerraPower nuclear reactor company. Amazon recently purchased a "nuclear adjacent" data center from Talen Energy. Oracle announced that it is designing data centers with small modular nuclear reactors.
In China, 5 reactors are being built every year. 11 more were recently announced. The United Arab Emirates (land of oil and sun) now gets 25% of its grid power from the Barakah nuclear power plant (four 1.4 GW reactors, a total of 5.6 GW).
Nuclear fission will play an important role in the future of grid energy, along with solar and wind. Many people (e.g., Germany) still fear it. Often these people are afraid of nuclear waste, despite it being extremely tiny and safely contained (https://en.wikipedia.org/wiki/Dry_cask_storage). Education will fix this.
Nuclear fission is safe, clean, secure, and reliable.
This is exactly why highly energy intensive consumers are still connected to the grid, and NOT to individual power plants.
In a grid with increasing proportion of renewable energy (wind/solar), it becomes less and less appealing to build nuclear plants because the amount of time that those plants are not competitive increases (=> whenever wind/sun is available).
Even in China, basically the only country where nuclear power is being added at a non-negligible rate right now, nuclear output is being eclipsed by wind/solar already, and those are growing much faster, too: More wind power was added in China since 2019 than the total nuclear power right now (~400TWh/year), and absolutely no trend reversal is in sight.
Anyways you could double the power of the country with solar on just roofs and parking lots, so consider that before you go in on your land argument.
Solar and batteries will win
In Europe old paid off nuclear plants are regularly being forced off the markets due to supplying too expensive energy.
This will only worsen the nuclear business case as renewable expansion continues, today being a bonanza fueled by finally finding an energy source cheaper than fossil fuels.
Nuclear power is essentially pissing against the wind hoping the 1960s returns. It needs to come down by 85% in cost to be equal to a reliable renewable system.
> The study finds that investments in flexibility in the electricity supply are needed in both systems due to the constant production pattern of nuclear and the variability of renewable energy sources. However, the scenario with high nuclear implementation is 1.2 billion EUR more expensive annually compared to a scenario only based on renewables, with all systems completely balancing supply and demand across all energy sectors in every hour. For nuclear power to be cost competitive with renewables an investment cost of 1.55 MEUR/MW must be achieved, which is substantially below any cost projection for nuclear power.
https://www.sciencedirect.com/science/article/pii/S030626192...
China finished 1 reactor in 2023 and are in track for a massive 3 finished reactors in 2024.
On the other hand they are building enough renewables to cover their entire electricity growth.
Even China has figured out that nuclear power is not economically viable.
https://reneweconomy.com.au/chinas-quiet-energy-revolution-t...
Every dollar invested in nuclear today prolongs our reliance on fossil fuels. We get enormously more value of the money simply by building renewables.
The Real World™ disagrees with your model. Time to update your model.
> In Europe old paid off nuclear plants are regularly being forced off the markets due to supplying too expensive energy.
No, nuclear power is being forced off markets by insane subsidy schemes that lead to grids being flooded with electricity at negative (or just zero) prices, due to those producers being isolated from price signals by both (a) subsidised/guaranteed producer prices and (b) priority.
We are paying "renewable" produces to produce electricity that nobody wants, then have to pay consumers to take it off our hands AND wreak havoc with our reliable producers. And then we congratulate ourselves on a job well done.
> Nuclear power is essentially pissing against the wind hoping the 1960s returns
Nuclear power plants are licensee to print money. Unless you forbid them to operate or flood the market with subsidizes competitors and give those competitors priority.
> Even China has figured out that nuclear power is not economically viable.
LOL. That's why they are accelerating their nuclear program.
https://www.bloomberg.com/news/newsletters/2024-03-15/china-...
China’s Nuclear Energy Expansion Is Getting Even Faster
Beijing’s rapid deployment of atomic power rivals its growth in solar and wind, and the round-the-clock electricity is more beneficial for the grid.
As someone who is pro-nuclear, there are also a lot of pro-nuclear folks who are needlessly pessimistic on wind, solar and batteries.
The anti-nuclear wind/solar proponents are dangerously wrong.
We need to clearly explain why wind and solar are not enough, and why nuclear is complementary. We need to explain it so everyone can understand it.
We must not allow what happened in Germany to happen elsewhere.
The evidence for this is overwhelming and undeniable, yet some people here still seem to believe that nuclear is ‘cheap and safe’.
I truly don’t understand how it’s possible for rational humans to believe that.
The problem is that this is just a feeling. The bigger problem is that this feeling is still extremely widespread, completely contrary to the facts. For example...
> Nuclear energy has never been profitable anywhere in the world to date
This is completely false. Nuclear energy is profitable pretty much everywhere it is used. Almost obscenely profitable, if you are allowed to run the plants.
I ran the numbers for Hinkley Point C. At the rate they negotiated (14,8 Cents/kWh...or was it pence?) the profits are almost obscene.
A modern EPR will generate 1040 TWh of electricity over its lifetime, assuming 80 years operation and 90% capacity factor. Assuming the EDF-negotiated price is 14,8 cents, that's a cool € 150 billion worth of electricity. But that's just one reactor, whereas HPC is two. So € 300 billion. Puts even the completely ridiculous cost overruns for those two reactors into perspective, doesn't it?
China built their two EPRs for a total of $7.5 billion, and with the EPR2, EDF is pretty certain to get construction times and costs.
Here's an explainer of the economics.
https://www.youtube.com/watch?v=cbeJIwF1pVY
The annual reports for one of the Swiss plants are online. They produce electricity for 3-5 cents. Profitably. In Switzerland. And they have expenses like...oh...a new administrative building in one year. Switzerland is not cheap.
> has entirely disproportionate downsides
What are those "entirely disproportionate downsides", in your opinion? If you subtract the effects of radiophobia?
https://en.wikipedia.org/wiki/Radiophobia
> The evidence for this is overwhelming and undeniable,
Yes, the evidence for nuclear is overwhelming und undeniable, yet some people go right ahead and deny it.
> I truly don’t understand how it’s possible for rational humans to believe that.
Agreed. I truly don't understand how it’s possible for rational humans to believe that nuclear is unsafe and uneconomic.
Not if you consider the full costs. Every official source you find only shows the costs that the operators have to pay, but the operators never pay the full cost. No nuclear power plant that I have found documentation for anywhere is operated without immense national subsidies.
You mention Hinkley point and say that they will generate 150 billion euros in income over 80 years. Let's assume that's about right and look further.
It is currently estimated to cost over 40 billion to build. Now we're at 110 billion. The storage facility will likely be Sellafield, which has a yearly budget of around 2.5 billion per year, times 80 years is 200 billion. Now we're at negative 90 billion.
Negative 90 billion already while the plants are actually operational. But the waste has to be managed for longer than that. For centuries after the plants have been closed, in fact. So we're at negative what, 300, 400 billion over the lifetime of costs.
This is also without considering any cost of operations. We assume here that 1000 people operating the plant work for free, never need any equipment, the plant never need maintenance, and the calculation still has a debt of several hundred billion.
Now you might say wait, that makes no sense, then EDF (the operator) would be unprofitable and go bankrupt!
Yes, and they did, a few years ago.
France covered the bill of over 50 billion euros in debt and also the upcoming 20 or so more billions in repair and maintenance. EDF was unprofitable even when they neglected required repairs and maintenance.
Again, nuclear is not profitable. Now, since the costs are mostly covered by governments and future generations, short-sighted investors and quarterly earnings-focused companies will of course go along with the pretense that it is.
Amazon is buying the subsidized unrealistic purchase price because that's what's going to be relevant to them. When the operator goes bankrupt because the costs turn out to be way more than they calculated with, it's not Amazon's problem. It's the US taxpayer's problem.
A handful of people operating Chernobyl have caused costs that passed 600 billion euros already in 2016, and will keep costing more money for the foreseeable future.
Nations have international agreements to cover liabilities from nuclear power stations, since no company and very few individual nations can cover the costs of severe accidents.
That's disproportionate.
Yeah, nuclear power can be profitable when it is enormously subsidized.
The CFD for Hinkley Point C proves that nuclear power does not have a business case in 2024.
Then add on top that the French government is asking for more subsidies on Hinkley Point C since it might not be enough!
https://www.ft.com/content/c1e3bd19-763b-4ea1-b188-d2872cc36...
The reality of this is evidenced by Sizewell C. Even with complete political unity for nearly 20 years they haven't managed to come up with a working business case outside of essentially a cost plus contract.
Building nuclear power as a cost plus contract, there you have true insanity.
https://www.ft.com/content/2a5d9462-b921-4577-82c1-4eb508775...
But that is the reality of nuclear power today. It is vastly undercut by renewables and expensive to the point that even with near political unity the cost can't be defended when public scrutiny is applied down the line.
It's the grid which protects it and if the grid is broken nuclear is even worse since you're suddenly faced with far too much power on a small portion of the whole grid. Meanwhile with renewables, even if the grid is broken on several points, you are much better off due to the distribution of power generation.
Not only are the times of unavailablilty very rare and usually plannable, they are also uncorrelated. Seasons, day/night cycles and even large scale weather patterns are highly correlated.
Nuclear is among the least dependable generation sources from a long term capacity standpoint, with multiple incidents taking generation off for long periods. 3 mile island wasn’t a big deal from a health and safety standpoint, but at 4 AM TMI-2 went offline with zero warning and never came back. Similarly you can’t trust timelines for when exactly new generation will come online.
Less severe incidents don’t necessarily make the news, but losing 1.3 GW at some random period for weeks or months isn’t particularly uncommon. Sure major incidents are “rare,” but there’s not actually that many nuclear power plants ever constructed.
Nuclear has the highest reliability of all generation sources, with for example the largest fleet of reactors, the US fleet, consistently coming in at around 95% capacity factor.
Most of the remaining 5% is scheduled outages for maintenance and refueling. And both the rare scheduled as well as the even rarer unscheduled outages tend to be uncorrelated. Well, unless radiophobia grips a country and they decide to shut everything down.
Trust renewbros to always use a single example to bolster their claims.
Short term capacity factor is not the same as reliability, the grid needs power every day not some average over a year. Add some turbines and a given dams capacity factor drops but it becomes more valuable due to the added flexibility and redundancies.
Lifetime capacity factors of US reactors are a long way from 95% often in the low 80% or below that’s not because the grid doesn’t want their power but because nuclear is unreliable over the long term.
96.04% in 2017, lifetime 78.07% (unreliable). https://en.wikipedia.org/wiki/Browns_Ferry_Nuclear_Plant
95.7% in 2017 vs 80.25% lifetime (unreliable). https://en.wikipedia.org/wiki/Beaver_Valley_Nuclear_Power_St...
And before you think 2017 just had unusually high demand, nope 79.36% capacity factor in 2017. https://en.wikipedia.org/wiki/Arkansas_Nuclear_One
That’s just another unpleasant reality when trying to actually use nuclear as a major energy source.
Er, capacity factor and reliability are not the same thing, that is true. But not in the way you seem to think. Capacity factor is more stringent than simple reliability, as it also takes into account planned outages and even power reductions due to external circumstances, for example when nuclear output in Finnland was throttled to compensate for the over-production problems they had with their hydro last year.
Please do read up:
https://en.wikipedia.org/wiki/Capacity_factor
And you do need to make up your mind whether you think long term or short term is important, because you are contradicting yourself on that.
Capacity factors are usually given for a year, and in recent years, the capacity factors for nuclear in the US have been in the 95% range for a while, though recently dropped to "just" 93%. What the capacity factor was 30 years ago isn't really relevant.
The idea that 93%-96% capacity factors somehow make nuclear power "unreliable" is ... "interesting".
If you find a more reliable electricity source, do present it here.
https://en.wikipedia.org/wiki/Capacity_factor#/media/File:US...
Individual participants making different bets is how the market decides what’s effective and what’s wasteful. But you can’t assume any subgroup is speaking for the total market, because they could be about to lose big. Nuclear could make a huge comeback in the coming decades, but until that happens we can only talk about the market in terms of what the market is actually doing.
It's because nuclear power plants are so durable, they last almost forever, but at least 80-100 years for modern plants.
If you build out too quickly, you end up in the same situation the French found themselves in after the Messmer-plan predictions turned out to vastly overestimate demand: they were done after 15 years.
With effectively no nuclear power plants to build for 45-85 years, their industry withered and they have had to re-learn building them. Not helped by the fact that there was a legal cap on the total amount of nuclear production.
China is already trying to use coal as a peaking generation something it’s terrible at. That’s something nuclear is even worse at and what’s ultimately limiting their nuclear ambitions. They slowed down the pass of nuclear construction despite nuclear only making up ~5% of their electricity supply it’s simply an issue with nuclear not regulations.
If you are aiming for a fleet of 200 reactors, you should be completing 2 per year.
However, don't discount renewables. Paired with energy storage -- which isn't only chemical batteries! --, intermittent generation can be extremely useful across vasts parts of North America.
Hydro is described as peaking power because it can be ramped up and down 100%-0%-100% multiple times in a single day. Very useful for tracking changing demand and intermittent energy from solar/wind.
Coal/Nuclear has issues doing the same due to thermal stress and heat loss when not in use. The energy used to get things back to working temperatures requires fuel which isn’t generating power, you also have ware issues from thermal cycling. Alternatively, you could keep things at operating temperatures but that’s again spending fuel without generating electricity.
On top of this Nuclear runs into if you try and ramp down very low very quickly, wait a few hours a then try and ramp back up. https://en.wikipedia.org/wiki/Neutron_poison To be clear 100%-30%-100% can be fine 100% to 0% to 100% isn’t.
Since intermittent renewables cannot do baseload, and get asymptotically more expensive trying, fans of intermittent renewables are trying to paint baseload as an obsolete concept by making intermittency primary.
That is silly. Baseload is real and a major component of demand. Intermittent renewables can be a good addition, as their supply curves can actually match the variable part of the demand fairly closely (thought obviously not perfectly).
It is not true that Hydro can be ramped arbitrarily. For example, just last year Finland had a huge problem with their plentiful hydro plants overloading the grid. Fortunately they also have significant nuclear that they were able to ramp down to compensate.
With ~60% baseload, there is no need to ramp nuclear plants down to 0 rapidly. That's only necessary if you screw up your grid by having too much intermittent renewables and giving them priority. Don't do that.
Alas, overloading their grid with renewables is exactly what some countries are doing, Germany for example. This is a really bad idea for a number of reasons. One is that due to even the average capacity factors being so low, you have to dramatically over provision in order to even achieve fully supply on average. However, that massive overprovisioning means that when weather is favorable, those massively over provisioned massively cannibalize each other (and other producers, if those still exist).
The more you overprovision, the worse this gets.
Reasonably ramping nuclear up or down is no problem. France for example has been using nuclear reactors almost exclusively for some time. Lower capacity factors for the nuclear plants than you'd really want, but otherwise not a problem. Their investment in renewables (aim seems to be around 30-40%) will likely increase the capacity factors of their nuclear plants. Good for them!
You can find that definition used regularly in older books dating well before renewables where a thing.
01 April 1967: “This paper states some of the basic principles concerning daily and annual load factor on power systems, and the use of load duration curves in coordinating the type and operating pattern of generating plant. Operating procedure is explained, and the types of plant in use or under construction in New Zealand are described. Overseas plant is divided into the categories of base load, medium load factor, and peaking, with an outline of the desirable characteristics in each case.”
Also, time of day pricing shifts users to periods of low demand so it doesn’t represent the minimum need, just the observed demand under specific pricing schemes. If daytime rates fall well below nighttime rates continually the actual usage at 3AM would fall dramatically.
'nuff said.
There’s an even older saying: “Better to remain silent and be thought a fool than to speak out and prove it.”
The article even goes on to use both, while making clear which definition the author is trying to promote: “Power plants that do not change their power output quickly, such as some large coal or nuclear plants, are generally called baseload power plants.[3][5][6]”
In the wider context grid operators don’t actually care about the specific absolute lowest demand number. They need to operate 24/7 for years so whatever the minimum number happens to be for a single second just doesn’t matter much compared to blackouts and other extreme situations. Bottom 5th percentile matters from an economic standpoint, but very temporary extremes only matter in terms of resiliency.
“Firm power” is more about the type of relationship between energy generation and consumers than the actual technology being used. https://en.wikipedia.org/wiki/Firm_service
Baseload is load, so demand side.
Firm power is generation, so supply side.
Add energy (speed) when available, use it to power things when not.
How cheap will wind, solar and batteries be then?
There is no universe where starting in on a new nuke today makes financial sense. They just take too long, cost to much to build and run.
China are currently building 27 reactors [1], and the average time is still 7 years for completion.
[1] https://itif.org/publications/2024/06/17/how-innovative-is-c...
Important point: even the most disastrous nuclear projects are better than "succesful" intermittent renewables.
Nuclear power plants are big and take a while to build. They also reliably deliver truly stupendous amounts of electricity over an amazing amount of time.
Totally worth it.
Nuclear: an expensive way to generate cheap electricity.
Intermittent renewables: a cheap way to generate expensive electricity.
Economists have the term market failure for configurations in which groups of individuals making decisions predictably produces bad outcomes [1].
The last decades seem to imply that there is a broader social failure when high-frequency, echo chambered, outrage-based discussion results in suboptimal debate. This is true from our political media to social media to the quality of private policy debate in America.
Nuclear power seems to exemplify that problem. If a public utility discusses nuclear power, every neighbour in the zip code will come out complaining about radiation. Even if their town is built on radioactive coal ash. Even if they live next to a missile silo, or near a port where our nuclear-powered fleet makes call. Private parties, on the other hand, can cut through the bullshit, which we usually see as bad, but plays a spoiler effect that maybe keeps the system from getting gridlocked.