Let's do some math. You can build a 3 MW wind turbine for around £2 million. Which means you could build around 11,500 wind turbines with the amount of money spent on HPC. Which in turn means a maximum production of 34,500 MW. So even under the assumption that these will run at below 10% of their capacity you're still producing more electricity than HPC. Now in reality this number averages around 30% actually. So the price per MWh is orders of magnitudes lower compared to nuclear.

This is still not equivalent because we haven't solved storage (peaky energy is not as valuable as constant energy) [and because you've compared build costs to total costs].

Bit in [] is false; apologies.

Yes, there needs to be more storage or backing by gas plants. But nuclear power needs the same gas plants as you cannot change the output levels quickly enough to drive the grid. In a sense, constant output is almost as bad for the grid as peak energy, which can easily be throttled. In both cases you need additional flexible plants or storage. In the moment, when there is true surplus renewable energy production, storage buildup will increase quickly.

Side note: electric cars will be a large part of the grid balancing in the future.

> Side note: electric cars will be a large part of the grid balancing in the future.

Do you have any good reading for this?

In my naive mind, it’s doomed by the fact that we drive to work in the day, then all come home, together, in the evenings, when the sun is on its way down. Or is the plan to have two way power chargers at businesses too? Otherwise, all these batteries are disconnected when you need them.

Considering how much storage capacity becomes available when connecting cars to the grid - even if the part used for grid storage is only 10kWh/car (typical batteries have 70 or more kWh) - I would assume there will be a huge push by utility companies to get as much cars connected as possible.

Sure, but assuming all of the office buildings of the world don't retrofit their parking structure with grid connections, these cars will only be connected to the grid in the evenings. I suppose this would be great for taming the dinner time spike, but it seems to put the availability, for them to store excess, where it's needed least: https://www.researchgate.net/figure/California-daily-utility...

Why would you assume that no office building would retrofit their parking structure when there are huge incentives to do so? For sure not all of them might, but if enough do, it will be a major benefit. But yes, if a car isn't connected to the grid it cannot participate.

> Why would you assume that no office building

I'm definitely not assuming that none will, I'm just assuming it will be a some smallish fraction for the foreseeable future. Nearly 50% of California's working population works for small business. It's not just the cost of the equipment, but also cost up upgrading the panels and, almost certainly, service lines.

Where I am (a city in California), you have to pay for replacing the service lines/conduits, including digging it up/repairing the street and sidewalk, even if they run 1 city block.

Do the math on this too. How much storage would half of $21 billion buy?

https://insideevs.com/news/458106/battery-cost-110-kwh-price...

Using a pessimistic $150/kWh (it's already under that) I get 70,000,000 kWh or 70 gWh, enough to run the entire state of California for a little over an hour. So you could build over 10GW of wind power capacity and 70 gWh storage for the cost of one nuclear plant.

Nuclear isn't dying because of hippies. It's dying because it's expensive. You're right that the safety concerns are over-inflated, but failure to deliver on its cost promises are the major failure of nuclear (fission) power.

If we make fusion work and can scale it, that's a totally different ballgame... but it won't happen fast enough to impact climate change.

I really don't think this isn't about maths with money. People are constantly working on this as-yet unsolved problem. Those batteries aren't suitable. Perhaps some will be one day, and that will be the fix.

But in the meantime, the best option is nuclear.

this does the sums with storage (for a Georgia nuclear plant)

https://pv-magazine-usa.com/2021/08/05/youve-got-30-billion-...

massively over provisioned PV+Storage is still half the price of Nuclear. It even throws in extra money for grid upgrades.

I've compared the build costs of HPC to the build costs of wind turbines. Not total costs.

You're right. Will edit.

But! But. What's the source for the turbine build cost? Now I need to try and attack that :)

> 11,500 wind turbines

That's a lot of space used.

And don't forget those wind turbines will have to be replaced after 20 or 30 years while your nuclear plant is good for 50 to 60.

The largest onshore windfarm has 240 wind turbines (https://www.wikiwand.com/en/F%C3%A2nt%C3%A2nele-Cogealac_Win...).

You can't just easily build 11500 wind turbines without huge impact on environment.

That’s not quite fair. If we were building 11,500 nuclear power plants their unit price would be much lower too.

Part of the problem is that nobody has built the things in the past 30-40 ish years.

Though falling prices would be intuitive, with nuclear that's been found to be false, both in the US and in France. Building subsequent versions of the same design increase in cost, rather than staying the same price or going down.

France: https://www.sciencedirect.com/science/article/abs/pii/S03014...

US: https://news.mit.edu/2020/reasons-nuclear-overruns-1118

This is truly remarkable phenomenon that occurs with almost no other technology. And this happening in France rules out the possibility that it's merely increasing regulatory burden that causes increasing costs.

Those who want us to build new nuclear need to grapple with the issue of their cost if they want it to happen, yet I have not found a single nuclear advocate on HN that will acknowledge that these issues even exist, much less bother to propose some solution.

At least for the "safety" issues, we can have a discussion about how designs can mitigate that. But there also needs to be discussion on how design, workforce training, logistical management, or something can improve the construction issues.

If you build 11,500 nuclear plants they will have to be breeder reactors (or, we'd have to deploy 2 million square miles of uranium collector fields in the ocean if that could be made cheap enough to supply U for burner reactors.) We don't use breeders because they've proven to be more expensive than LWRs, not less expensive.