Hacker News Reader

Comment by epistasis | Hacker News Reader

The problem isn't that nobody wants nuclear in their backyards, lots of people do (the people that already have nuclear in their backyard).

The problem with nuclear is the same one that it's always been: it's too expensive. Nobody wants to go bankrupt.

Meanwhile we have cheap batteries and super super cheap solar and wind to charge them.

View on HN

I think this is one of the major limitations of framing everything in terms of profit. Sometimes a society needs to build services and run them at-cost for it to make sense.

If nuclear was our only choice for zero-carbon energy, I would say that we should change the subsidies and prices to make it profitable, or at least at-cost.

The problem with nuclear isn't that it's at-cost, the problem is that construction projects balloon into financial disasters that are too risky. Check out what happened to France's Areva due to constructing the Olkiluoto. It is literally owned by the French government to bail it out from their financial disaster. Or what happened to Westinghouse and parent company Toshiba due to trying to construct AP1000s.

We have better alternatives that are cheaper and are not such massive single points of failure. Right now the US government is throwing money hand over fist at nuclear, and if the industry can make a go of it and manage to construct some more, great, I have no problem with that.

But nuclear is fundamentally inefficient, expensive, and bad source of power compared to the alternatives. It's always always overpromised and underdelivered. Seemingly nobody in the industry has every made an honest construction estimate. I mean, of course there were a few successful construction projects, but on average it's been a disaster. Throw another $100B at it, see if somebody finally figures it out after 70 years, but it seems pretty doubtful. Nobody left in the industry seems to be capable or smart enough to deliver.

Yeah, transit advocates say that too, but it's the same thing: once you advertise that you don't want to make a profit, you'll find that people can make sure you make a loss. At-cost is the same as cost-plus and has the same problem: the only way for the guy building something for you to extract value from the thing is to raise the cost high enough.

It's a weird thing. American transit advocates will cry themselves hoarse that public transit should be free or run at a loss but almost every system in the world that's good (TfL, HK Metro, JR East) makes money. But every American system loses money hand over fist and sucks.

This makes it quite clear to me. It's very easy to be shitty losing money. It's very hard to be shitty and make money.

Most of those systems tax property increase when infrastructure is built and operate in dense cities with extremely anti-car regulation that gives them a guaranteed consumer base. Do the same in the US and it will magically become profitable, because there will be no alternative.

Sure. There are policies you can change. But if you believe these things shouldn't make a profit you won't put those things into place. It looks like anyone who wants transit to not make a profit just wants it to suck.

Levying taxes directly to the transit agency and punishing car usage doesn't incentivize the transit agency to make the transit not suck, it's just a different way for accounting in public funding.

I'm sure you have a complex model for this, but transit agencies that aim to make losses all suck and the ones that aim to make profits are good. So if someone is telling me the agency should make a loss, they want it to suck. The correlation is strong and whatever your model is, it isn't capturing some variable.

I don't have a complex model, it's just your test that is capturing the wrong variable. Every network you listed is "profitable" because the subsidies they get are in the "income" column instead of the "subsidy" column. The distinction between profitability or making a loss is purely nominative. In the US for example, the HK transit authority would be "aiming to make a loss" because its largest source of income are the taxes it is allowed to levy. Since in the US it's the government that levies the taxes which are counted as subsidies, the HK transit authority in the US would be classified as loss-making. Additionally the MTR in HK has pseudo-governmental powers in that it gets to expropriate, rezone, and resell land at a higher price, which would also be counted as a subsidy in the US and therefore it would be considered to be loss making.

So the problem isn't that my model isn't capturing a variable, it's that your test isn't capturing a variable and is taking in erroneous data points due to diverging definitions of your metric.

Where are the cheap batteries? List prices are gradually coming down but availability remains limited for any large projects. And if you actually want to store a useful amount of energy, that takes a lot more than just a big box full of battery cells.

Where are the cheap batteries? Getting installed right now. Grid assets are usually measured in MW or GW, but multiply by 3-4 to get the GWh.

EIA said developers expect to add 30GW in 2024 here:

https://www.eia.gov/todayinenergy/detail.php?id=61202

Note the 7GW of batteries already in California, 3GW in Texas. Texas will move much more quickly than California now because it's a profit based market with few barriers to entry.

There were other estimates in 2024 were for 14.1 GW for this year, but IMHO 30 GW is much more likely:

https://www.eia.gov/todayinenergy/detail.php?id=61424

We are at the start of an exponential growth curve. As prices fall, more demand appears. There are plans for massive amounts of battery production capacity. BNEF has tracked 7.9TWh/year of planned factory capacity by the end of 2025 (probably only 50%-70% of that will get developed):

https://www.bloomberg.com/news/newsletters/2024-04-12/china-...

However, this is exactly on track with the estimates I heard, 3 years ago now, of 10x growth in production capacity every 5 years. By 2031 we should see 20-30TWH/year of production.

So...what's the story with grid-scale battery disposal? Are we heading for another series of super-fund cleanups? That's a lot of toxic material that's going to have to be disposed at end of life.

It will be recycled, there's tons of recycling facilities that are just waiting for recyclable material to show up. So far it's just defective new batteries that go into them.

Plus, they're really not that toxic. Not as bad as fly ash from burning coal, or fracking fluid from natural gas, for example.

https://electrek.co/2024/04/11/north-america-just-got-its-fi...

https://www.forbes.com/sites/alanohnsman/2023/11/02/redwood-...

https://www.redwoodmaterials.com/

(Redwood Materials was founded by Tesla's CTO co-founder JB Straubel)

I'm a nuclear proponent, but similar arguments can be made for why we cannot solve our woes with nuclear. I'd also suggest scaling battery production is going to be less challenging than scaling nuclear production.

It's too expensive because of the cost of endless lawsuits and review process. This is fixable.

I’m pro-nuclear, but the cost isn’t fixable. It is improvable, and we need to have power sources that ramp up and down on demand, and to fill seasonal gaps. Nuclear is the best long-term bet for that.

However, as it is, coal plants already cost more than solar or wind. Almost all of that cost is in the turbines.

Unfortunately, nuclear plants use similar turbines, and the coal industry has already cost-optimized the death out of them. It is unlikely there are significant improvements coming.

So, even if the rest of the nuclear plant was free, it would still be uncompetitive on a raw $/kwh basis.

> Almost all of that cost is in the turbines.

Cost of mining and transporting the coal?

Coal loses on cost before you even build the furnace, let alone put something in it to burn.

(“Turbine” also includes stuff like the boiler, generator, etc, etc. I should have written “cost of the plant, minus the coal furnace and the fuel”)

What about the 10 years being mired in the courts?

"The licensing process – something that can take more than a decade to complete – consists of a series of approvals by the NRC, including issuance of design certification, early site permits, limited work authorizations, construction permits, operating licenses and combined licenses for new nuclear power facilities."

Reply on news.ycombinator.com