No battery farm can protect a solar/wind grid from an arbitrarily extended period of bad weather. If you have N days of battery storage and the sun doesn't shine for N+1 days, you're in trouble.

Nuclear fission is the answer.

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 (maybe 80 years, maybe more) and the uranium fuel is very cheap, perhaps 10% 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 you care about greenhouse emissions.

Sam Altman owns a stake in Oklo, a small modular reactor company. Bill Gates has a huge 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 (https://news.ycombinator.com/item?id=41505514).

In China, 5 reactors are being built every year. 11 more were announced a few weeks ago. 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. But you don't hear about it in the mainstream news yet. And many people (Germany, Spain, I'm looking at you) 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.

> Constellation, which plans to spend about $1.6 billion to restart the plant, is awaiting permits and expects the facility to come online by 2028 [and] would provide Microsoft with 835 megawatts of energy.

I don't quite understand how this makes sense financially for Microsoft. A 1 GW offshore wind farm costs about 1 billion [1]. The Gemini Solar + Battery Storage Project in Nevada is about 1.1 billion (690 MW + 380 MW battery). [2]

With solar, wind and battery prices continuously trending downwards, how does it make sense to invest in nuclear which nearly always has cost overruns, bad PR and unknown potential future costs?

[1] https://businessnorway.com/articles/cost-of-wind-turbines [2] https://commercialsolarguy.com/americas-first-gigawatt-solar...

More nuclear is a good thing. FAANG nuclear is even better, because they will be be ready and equipped to push back the insane NIMBY crowd(s) and also don't need to be worried about getting reelected.

Something that isn't discussed enough is that nuclear power is extremely safe. According to Our World In Data (https://ourworldindata.org/safest-sources-of-energy), it produces the fewest emissions per gigawatt -- solar and wind both have substantial overhead -- and the second-fewest deaths, after solar, of coal, oil, natural gas, biomass, hydropower, wind, solar, and coal. The misguided perception nuclear is more dangerous is due to its unfortunate habit of clumping deaths. People remember Chernobyl, but not Bill, who died mining coal, even if there are a thousand Bills all over, their total deaths dwarfing Chernobyl's.

The climate activists should really be advocating for nuclear. It's cheap, has the lowest emissions, and is really safe. But, of course, people will object. It's hard for PR to say that, even if you can show the data. People rarely change their mind because of data.

While this is better than loading up a coal power plant, the way I see it, it's still a tragic situation. Even if AI used purely fossil-free electricity (which it doesn't), it's still putting pressure on our grids and energy production. The green energy AI consumes if the green energy some other less lucrative industry will not consume. Big tech is currently scraping their net neutrality plans left and right...

Was anyone else horrified because they thought the powerplant was going to be run by a Microsoft AI? (TFA is about Microsoft purchasing nuclear power to run their AI)

I feel like we are living in an episode of the Simpsons at this point

Release from Constellation Energy: https://www.constellationenergy.com/newsroom/2024/Constellat...

Nuclear plant, microsoft and AI on the same phrase. What could possibly go wrong?

My sister is a geologist who works at nuke plants (regulatory stuff, determing if they're in legal areas to build, ground contamination, etc). She always complains that it's an old-mans (this was in her 20s, 20 years ago) group of staff everywhere with guys who had incredibly light educations on what they're actually working on and had been in the business, and sometimes at the same plants, for decades. Some 40-50+ years.

We might have a severe lack of people who can work in the field at plants and tech getting involved is probably a good thing on making that field more prominent for new students than it currently is.

"Nuclear Engineer for Microsoft" is a cool title.

This article says it will be the first restarted nuke. But there is also a restart in process in Michigan. Will this beat the Michigan one into full operation?

Presumably all the things that have gone wrong have been fixed? I can honestly say the micro reactors and pebble beds look far more fail safe than the typical PWR/BWR that must have power and a pump running as I understand things to prevent catastrophe.

Can someone explain why it's going to take so long to fully decommission TMI-2, the old reactor?

Reading the commission report, they have about $733m in the decommissioning fund, while decommissioning is projected to cost $1B. Active demolition projected to start 2046, completed in 2052. They've started some work on debris demoval, but it looks they're going to take a break 2030-2046 to let the fund accrue (at expected 2% real RoR) enough to get the fund to $1B.

Isn't that insane? So from 2030-2046, it will just sit there "mothballed" with a minimum spent on security and so on, just for the fund to gain value?

Wait and see if this will actually be up and running by 2028 at the costs indicated.

https://www.energytransitioncrisis.org/

If this story shocked you, do yourself a favor and check out Erik Townsend's docuseries on the need for nuclear to even come close to filling humanity's need for energy in the coming decades.

I'm generally in favor of nuclear, but the symbolism of reopening Three Mile Island for AI power is strikingly anti-human.

Maybe we should take some the money being put into AI and direct it towards fusion research. The rest of the funds can be spent on AI to help with fusion research. Once we have fusion we can forget about the awful fission era and also get on with healing the planet.

AI might be the thing that ends badly this time. Better than using fossil fuels for it though.

So how would this work? Does this mean Microsoft will build a directly connected data center nearby?

Or is Microsoft just buying power from the plant's owner on the energy market?

Classic SNL skit from when it was originally shut down.

https://vimeo.com/357925913

Feels like right out of a Westworld plot.

Wait, it literally was.

So someone is paying massive insurance premiums on this right?

Not "privatize the profits, socialize the costs" right?

Hmmm. 50 years old and out of action for half a decade. That’s going to require a lot of work to persuade the NRC it’s safe to operate.

Given how difficult it’s proven to build new nuclear around the world, what are the odds of this recommissioning being on time and budget?

Wow. AI has already accomplished more positive in a few years of big commercial stuff (admittedly as a side effect) than cryptocurrency has since, say 2009. :(

Should be renamed Three E's Island.

While I love to see nuclear get a comeback from new tech ventures and commercial impetus, it's also interesting to point out how one now mostly hears crickets from the people who screamed about the energy consumption of crypto. AI has so far mainly managed to generate enormous amounts of spam sludge, pollute the already toxic landscape of social media with auto-generated garbage, and lead to a level of data prying and mass data ripping by the big tech companies that makes their previous efforts seem almost quaint.

This shit is largely what this voracious drive for more power is helping build. Even crypto wasn't this shitty, since at least its fundamental application wasn't nearly so broadly parasitic. Yet now I hear very little about uselessly burning vast loads of energy.

I guess if the companies responsible for the above trash heap are largely the same ones that hire or indirectly subsidize the jobs of so many people in the tech world as represented by a site like HN, then the whole dumpster fire is okay.

I'm also fairly sure that OpenAI, Microsoft and others would happily burn baby seals in coal furnaces if it meant powering their precious AI data needs. It's only for PR that they make useful noises about using clean power.

> restart one of the units at the noted Three Mile Island nuclear plant in Pennsylvania

Ah yes, the ‘noted’ Three Mile Island. I wonder why this particular plant enjoys a degree of fame? The name certainly is familiar…

Seems like a rebranding might have made sense - like the UK switched Sellafield to Winscale.

There's something glorious about the optics of using one toxic disaster to power another.

It's unfortunate that Three Mile Island is a meme. Objectively this is a good thing, but it will create headlines and stoke more pearl clutching around AI power consumption.

Fascinating. How will we live ok back in, say five years, at this very funny period from, 2018ish to 2022ish when our echo chamber made us believe ESG was going to save the world.

How far we’ve come. Is it all due to the prospect of finally seeing a feasible road to AGI?!