Constellation Energy plans to invest $1.6B to revive the Three Mile Island nuclear power plant in Pennsylvania and sell all the output to Microsoft for AI energy demands.
Constellation Energy plans to invest $1.6B to revive the Three Mile Island nuclear power plant in Pennsylvania and sell all the output to Microsoft for AI energy demands.
Constellation described the agreement with Microsoft as the largest power purchase agreement that the nuclear plant operator has ever signed.
Three Mile Island was the worst nuclear accident in US history. Was mainly caused by poor design of human feedback systems which caused operational confusion and lead to a catastrophic failure.
You're viewing a single thread.
Seriously, we need the less carbon-emitting plants to replace the dirty coal ones, not come online just to power the AI hype :smh:
If/when the ai hype train crashes, it would already be online and therefore a good argument can be made to redirect the power to the grid instead of the then-defunct project
That's true, but this reactor had been operating for decades, until 2019. Methane was too cheap (and didn't have to pay for its waste) and made it unprofitable, so they stopped. Presumably the price MS is willing to lock in at makes it profitable for them again. I assume they predict prices going up or they're getting it cheap.
Keep in mind, MS has only signed an agreement to buy 100% of the output, not the reactor itself.
Maybe true. But if we have increased energy demand it might as well be nuclear.
Halting ai development might be nice according to many people, but we cant make that happen. Fraud alone is magnitudes more rampant. Its here to stay and we have to deal with it. I think this is a big win.
You're assuming "AI" isn't fraud?
This is a pretty fucking stupid comment lmao
What the hell are you talking about?
Probably they mean the way "AI" is presented to the public / investors. Most things that claim to be "AI" aren't, they're just regular boring ol' software, and even those things that do use AI are usually just using them for one tiny facet that doesn't actually require them.
There's potential for some incredible advances in AI right now, but outside of some novelty examples we really haven't figured out what it's actually useful for quite yet. It's one of the more interesting questions in the field.
Agreed, but the point of the original reply is that AI, like fraud, is magnitudes more rampant and here to stay. For now.
And that fraud is specifically rampant because of ai... so if companies arent using ai to combat it then they would never be able to keep up.
If the general bulk of appreciable-quantity tech investors suddenly found out in granular detail what is actually being done with their money as opposed to what they were told, it would probably cause the biggest crash since the Great Depression.