Comment by AnthonyMouse
12 hours ago
> They take too long to build and cost too much.
The global average to build one is ~7 years. People have been saying they take too long to build as an excuse for not building them for what, two decades or more? It seems to be taking longer to not build them than to build them.
> By the time a new plant is ready, alternate sources (likely solar + battery and long-distance HVDC) will have eaten its lunch.
Neither of those have the same purpose. Solar + battery lets you generate power with solar at noon and then use it after sunset. It doesn't let you generate power with solar in July and then use it in January. More than a third of US energy consumption is for heating which is a terrible match for solar because the demand is nearly the exact inverse of solar's generation profile both in terms of time of day and seasonally.
HVDC is pretty overrated in general. It does nothing for the seasonal problem and it's expensive for something that only provides a significant benefit a small minority of the time, i.e. the two days out of the year when the entire local grid has a shortage but a far away one has a surplus. It's also hard to secure because it inherently spans long distances so you can't have anything like a containment building around it and you end up with an infrastructure where multiple GW of grid capacity is susceptible to accidental or purposeful disruption by any idiot with a shovel or a mylar balloon.
The issue with them in addition to time is a huge capital expense that needs to be amortized. Nobody wants to hold 30-80 year debt on giant capital projects that could be rendered obsolete.
For commercialization, solar makes more sense as there is a much better return on capital.
If I were king, I’d do socialized power and have the government capitalize and own the nuclear plants, and bid out the operations to private entities. Government has better debt economics and doesn’t care about return in monetary means.
Even then, relatively small tweaks to tax law and some grid investment would create a solar boom at lower cost. Every Walmart parking lot and some road infrastructure should be covered with solar. Interstates could be utility and generating corridors - they aren’t because federal law makes any multimodal use very difficult.
> Nobody wants to hold 30-80 year debt on giant capital projects that could be rendered obsolete.
There isn't really an "obsolete" after it comes online because things get built when expected revenue exceeds construction costs + operating costs, but once built (or close enough to completion) they continue to operate as long as revenue exceeds only operating costs because by then the construction cost is in the past. When the construction cost is large, the amount the price of electricity would have to decline to fall below operating costs is equally large. And investing in something where you expected a positive ROI and you ended up with a slightly negative ROI clearly isn't what you'd have preferred, but it isn't nearly as bad as the -100% ROI you'd get from shutting down the plant instead of selling it for slightly less than what you put in. There's a reason the US is not only continuing to operate 20th century nuclear plants but even looking to reactivate some of the ones that have already been decommissioned.
Moreover, solar has the same problem. You invest in a solar farm because you're expecting to profitably sell power at current prices, but if e.g. the AI thing turns out to be a bubble then there will be oversupply and current prices won't stick. Solar also has the added "everybody is doing it" risk. If you and everybody else add solar then the price at times when solar output is highest is going to be lowest and vice versa, i.e. if too many people invest in the same type of generation then your output gets inversely correlated with the market price, which is bad for ROI.
I think you’re misunderstanding the economics at a fairly basic level. The cost to build is funded through debt that’s paid off over time. The construction costs aren’t in the past; they’re in the present, and in the future, in the form of debt payments.
Think of it this way: if you buy a house, the “operating cost” is fairly small: upkeep and painting, mostly. Does that mean you can buy a house, move out of your apartment, and quit your job, because your cost of living has just dropped a few thousand a month?
No, of course not. Upkeep isn’t the real cost of buying a house. The real cost is the monthly mortgage payment. Unless you were already independently wealthy, you have to keep your job. Sorry.
The cost of energy for a nuclear plant is the cost of paying back the loans. As other forms of power generation get cheaper, those loans stay the same, making it harder and harder for nuclear to compete. As they get squeezed out of the energy market, they have to raise their per-watt prices in order to continuing to service the loans.
Think of it like this. You rent your house to your cousin, who pays you enough to cover the mortgage. But then your cousin finds a sweet deal couch-surfing in the tropics in the summer. He stops paying you for June, July, and August. You can’t get anybody else in your house during that time, so you say, “Sorry dude, you have to pay more for the rest of the year. I’ve got bills to pay.” That works great until your cousin gets tired of your high prices and moves out, and now you’re left with a mortgage to pay and no one renting it.
1 reply →
> It doesn't let you generate power with solar in July and then use it in January.
That’s not necessary. Solar panels are so cheap that you can massively overprovision for winter and still come out ahead of nuclear.
> Solar panels are so cheap that you can massively overprovision for winter and still come out ahead of nuclear.
Only you don't. In latitudes that get winter, solar output is only about a quarter as much in the winter as in the summer. You often hear things like "twice as much in the warmer half of the year" to try and stuff October and March into the "colder half" and disguise how screwed you are in December and January. Worse, if you electrify heating then it's not just that solar supplies less in the winter, you also have more demand in the winter.
By this point you're not just overbuilding by a bit, you'd need five times as much or more in January as in July. "Five times as much" is already over what it costs to use nuclear. Then it gets worse, because you now have a price of zero during the summer and even the spring and fall because of the massive oversupply and lower demand, so you now have to recover the entire cost of the overbuild during the three months when you're generating the least amount of power.
Then it gets worse yet, because heating demand is higher at night and we haven't yet added the cost of storage.
Okay, let’s say that we use solar + battery to cover everything but Nov, Dec, Jan, when the days are too short. Solar is cheaper than nuclear the rest of the time, so (due to the way energy markets work) we pay solar producers the cost of nuclear generation, creating strong incentive to build out more solar + battery.
So we end up using nuclear 1/4 of the time. But unfortunately, nuclear’s cost is in the capital expense, not the operating expense. We pay about the same amount for it regardless of whether we’re using it or not. So if we’re only using 1/4 the energy, the cost per watt of nuclear energy is effectively 4x larger.
This incentivizes further build-out of solar, catching those sweet winter profits (now 4x larger!), further squeezing nuclear’s usage, driving up its prices, and incentivizing even more solar.
Eventually nuclear gets squeezed out and solar’s profit margins go from “astronomical” (naturally, it’s power from the sun, nyuck nyuck) to “low margin.” But they’re still making money. Whoever built the nuclear plant is left with a very expensive stranded asset.
At least, that’s my understanding. I’m not a power company accountant. What I observe, though, is that power companies who do employ accountants aren’t building nuclear. They’re building shit-tons of solar. And I’m pretty sure it’s not because they’re hippies who hate nuclear.
1 reply →
Lets put down some Swedish numbers.
During the coldest winter month, solar energy produce (as per statistics from the solar industry in Sweden) somewhere around 3-7% of the amount produced during the warmest month. Households also consume around 2-4 times the amount of energy during the coldest month compared to the warmest month. Sweden is a country where only a small minority have air conditioning installed at home.
Those are the worst month vs the best month. Overall the winter is not that bad, but it is still pretty bad for solar. Talking with people who has had solar installed here, the general story is very similar. During periods where it do produce the market price is already exceptional low, so it isn't returning a major saving. When the market price is high, the output is low, forcing them to be connected to the grid and pay whatever the electrical company demand during the highest market peaks, as well as taxes and grid fees which themselves has increased to match the cost of high variability.
All this looks very different in countries with much warmer climates and where the major energy consumption from households are air conditioning.
Yeah but:
1. Sweden is just about the worst case, there's very few countries/people that far north.
2. There's this genius invention called "wires". HVDC has transmission losses on the order of 3.5% per 1,000km. You don't have to colocate the solar.
8 replies →
The nice thing is Sweden has lots of hydro, which works as natural long-term energy storage. Every bit of solar you generate means water is kept in the dam for use later in the year.
You also can't ignore wind power which should be part of any plan to "overbuild".
I selected random date in July 2025. During that time Finland produced about 10GWh of solar. I selected random one from February 2025. During that Finland produced about 0.5GWh. February also actually doesn't have shortest daylight hours, mid-December situation is even worse. Christmas Eve 2024 produced about 0.05GWh.
You sure overprovision factor of 200x is still cheaper? This is when looking at the peak generation. From what I understand solar has about 30-40% capacity factor in summer. Just to panels (I'm not sure about total cost of grid-scale solar) seem to be about $300k per rated 1MW or $750k per 1MW during peak. $150M per 1MW during December. OL3 cost about 11B € for 1.44GW (assuming 90% capacity factor) or 7M € per MW.
Unless there has been some huge overnight exchange rate change 7M € seems much cheaper than $150M. Latter of course would actually be much higher when you factor in rest of the equipment, labor etc. Some numbers I found say that it's probably 5x higher.
1. Overprovision as much as you want, solar still won't work at night.
2. Do you realize the consequences of casually overprovisioning solar capacity when it uses orders of magnitude more land than nuclear per kWh produced ? Source : https://ourworldindata.org/land-use-per-energy-source