US Regulator Rejects Bid To Boost Nuclear Power To Amazon Data Center (thehill.com) 29
The Federal Energy Regulatory Commission (FERC) blocked Amazon's bid to access more power from the Susquehanna nuclear plant for its Pennsylvania data center, citing grid reliability and consumer cost concerns. The Hill reports: In a 2-1 decision, the FERC found the regional grid operator, PJM Interconnection, failed to prove that the changes to the transmission agreement with Susquehanna power plant were necessary. The regulator's two Republican commissioners, Mark Christie and Lindsay See, outvoted Democratic chair Willie Phillips. The chair's two fellow Democratic commissioners, David Rosner and Judy Chang, sat out the vote. "Co-location arrangements of the type presented here present an array of complicated, nuanced and multifaceted issues, which collectively could have huge ramifications for both grid reliability and consumer costs," Christie wrote in a concurring statement.
In a dissenting statement, Phillips argued the deal with Amazon "represents a 'first of its kind' co-located load configuration" and that Friday's decision is a "step backward for both electric reliability and national security." "We are on the cusp of a new phase in the energy transition, one that is characterized as much by soaring energy demand, due in large part to AI, as it is by rapid changes in the resource mix," Phillips wrote.
Amazon purchased a 960-megawatt data center next to the Susquehanna power plant for $650 million earlier this year. Following the announcement, PJM sought to increase the amount of power running directly to the co-located data center. However, the move faced pushback from regional utilities, including Exelon and American Electric Power (AEP).
In a dissenting statement, Phillips argued the deal with Amazon "represents a 'first of its kind' co-located load configuration" and that Friday's decision is a "step backward for both electric reliability and national security." "We are on the cusp of a new phase in the energy transition, one that is characterized as much by soaring energy demand, due in large part to AI, as it is by rapid changes in the resource mix," Phillips wrote.
Amazon purchased a 960-megawatt data center next to the Susquehanna power plant for $650 million earlier this year. Following the announcement, PJM sought to increase the amount of power running directly to the co-located data center. However, the move faced pushback from regional utilities, including Exelon and American Electric Power (AEP).
Good (Score:5, Interesting)
Seems like a good decision. Nuclear doesn't integrate well into grids, because it doesn't have the ability to rapidly alter output. Everything else has to work around it, and consumers are forced to pay whatever rip-off prices the nuclear subsidies force them to.
Re: (Score:3)
If your rationale is correct, then it would seem like a good idea to send more of the output of a nuclear plant to a constant, directly connected local load, and less to the public grid, whereas this decision does the opposite of that.
Re:Good (Score:5, Insightful)
1) A datacenter is not a constant load. The load varies daily with use.
2) Grids are based on socializing the cost of infrastructure. Short circuiting this will not work.
You cannot get power from heavily subsidized nuclear generation, and then not pay the common infrastructure, while anybody else has to pay grid overhead, whatever the distance to the power plant of each customer is.
Especially Amazon does not need to be indirectly subsidized, and can afford to pay fair price, and should also reduce it's overly hungry power consumption.
Re: (Score:3)
1) A datacenter is not a constant load. The load varies daily with use. 2) Grids are based on socializing the cost of infrastructure. Short circuiting this will not work.
You cannot get power from heavily subsidized nuclear generation, and then not pay the common infrastructure, while anybody else has to pay grid overhead, whatever the distance to the power plant of each customer is.
Especially Amazon does not need to be indirectly subsidized, and can afford to pay fair price, and should also reduce it's overly hungry power consumption.
Another thing is what happens if Amazon, or Microsoft, or whoever, decides "Hmm, maybe this AI stuff isn't all that." and simply stops using the rebuilt plant, and says "Here ya go, Pennsylvania, we're outa here!" after PA restores the plant, "You have a nice shiny new reactor, we're not going to pay for it now, since we don't need it any more."
Re: (Score:2)
Second, It is a viable point to make that companies leave states holding the bag all the time, see stadiums, see sports teams, see any manufacturing. Hell, Tesla is a great example of this
States do not add provisions ALL THE TIME to the things they do for businesses, because Government works in good faith, and businesses do not.
We have had 200 years of business pulling this stu
Re:Good (Score:5, Informative)
Seems like a good decision. Nuclear doesn't integrate well into grids, because it doesn't have the ability to rapidly alter output. Everything else has to work around it, and consumers are forced to pay whatever rip-off prices the nuclear subsidies force them to.
True on both points. A lot of people do not understand the tapdance that turbine based electricity generation has to perform in general, and nuc plants don't like sudden increases or decreases in output. Your second point is especially galling. Nuclear often turns out to be a great way to increase electricity bills for nothing. The nuc plant in Long Island that never opened, the Virgil Summer fiasco in South Carolina that gave South Carolina citizens a 9 billion dollar bill while never being finished.
The New York plant was foolishly sited in a place that would have required everyone east of it to be evacuated by sea is it had a severe problem, and the Summer plant was a litany of incompetence. My "favorite" part is that Westinghouse - the main contractor - claimed to be immune from South Carolina law for it's design process, and had unlicensed and unqualified people doing the designs without an engineer signing off of them. And turns out that didn't work well. Turns out that you should have qualified people designing nuclear plants, because much of their work was faulty.
So after the good people of South Carolina were looking at 25 billion finished cost, it was terminated.
Guess who foots the bill?
Re: (Score:3)
It’s almost impressive how many logical fallacies can fit in one comment. From hasty generalizations and straw man arguments to appeals to emotion, this reads like a case study in how to misrepresent nuclear power.
A lot of people do not understand the tapdance that turbine based electricity generation has to perform in general, and nuc plants don't like sudden increases or decreases in output.
Logical Fallacies: Hasty Generalization and Straw Man Argument. Sure, not all r
Re: (Score:2)
It’s almost impressive how many logical fallacies can fit in one comment. From hasty generalizations and straw man arguments to appeals to emotion, this reads like a case study in how to misrepresent nuclear power.
What is your point? You use presumed fallacies in what I wrote ending up turning it into an ad hominem attack. For one so versed in fallacies that they ignore one of the worst, you kinda expose you got nothing.
What you have done is take huge insult, And rather than reply in a rational manner, you try the age old refutation by made up fallacies, while very specifically engaging in one. If your made up bullshit refutes everything I wrote, then your textbook ad hominem refutes everything you wrote. Homey do
Re: (Score:3)
Re: Good (Score:1)
Re: Good (Score:2)
The argument here fundamentally didn't have anything to do with nuclear power, it had to do with co-located generation. The goal was to build a datacenter tied to the station directly, and in so doing avoid a host of costs and delays associated with putting a consumer that large on the PJM interconnect (I think it was like 40 million dollars avoided? Something like that.). They were hoping to sidestep PJM entirely basically. The other utilities are pissed because the datacenter would effectively be taking t
Re: (Score:3)
Seems like a good decision. Nuclear doesn't integrate well into grids
I think it is the other way around.
I believe the article is saying that the nuclear plant *is* integrated into the grid. The commission rejected a plan to divert the power *off* the grid, directly to the data center. You are correct that nuclear doesn't have the ability to rapidly alter output, which is why you do want it integrated into the grid. The grid can flex, but the data center cannot.
Re: (Score:2)
The decision wasn't based on anything about 'Nuclear', it is based on the fact that the current situation is that the region is already very close to baseload capacity - so much in fact that consumer prices are up 30 percent in the region.
Amazon should be paying a consideration to help subsidize that price increase
Re: (Score:1)
Re: (Score:2)
Why would the load be static? Predictable perhaps, but certainly not static.
Not good. (was Re:Good) (Score:2)
Seems like a good decision. Nuclear doesn't integrate well into grids, because it doesn't have the ability to rapidly alter output. Everything else has to work around it, and consumers are forced to pay whatever rip-off prices the nuclear subsidies force them to.
Hmmm. Have to disagree.
Grid Flexibility and Stability: Since Susquehanna’s reactors can adjust their output (ramping up or down at 1% to 5% per minute), they offer flexibility that contradicts the notion that nuclear power can't integrate smoothly into grids. This load-following ability could be especially valuable in supporting the local grid around the data center, particularly with the rising demand driven by AI and data centers. Rejecting the proposal misses an opportunity to leverage a steady, l
Re: (Score:2)
Reduce power. (Score:4, Insightful)
Reduce your power needs.
There is no valid reason a datacenter should use 1GW.
Re: Reduce power. (Score:2)
Re: (Score:2)
Where human brain requires number of watts the AI needs megawatts.
That's a silly comparison. AIs learn many orders of magnitude faster. Starting from knowing only the rules of chess, Alpha Zero exceeded the ability of the best human grandmasters in nine hours.
Both humans and AIs are good at recognizing images, but an AI can process 10,000 images in the time it takes a human to process one.
Re: (Score:3)
not sure how a datacenter should output lasers, there are no sharks inside.
Re: (Score:2)
The Amazon drivers are already going 88mph most of the time. It's perfect.
AI God (Score:2, Interesting)
YOUR electric rates will skyrocket due to demand from the Transhumanists to bring about their AI God.
Project Independence is what really got Nixon booted.
Re: (Score:2)
YOUR electric rates will skyrocket due to demand from the Transhumanists to bring about their AI God.
Well, the only thing that is not pathetic about that "AI God" is its energy consumption...
Re: (Score:2)
They want to build an AI servant who's so pervasive in society and meticulous in it's enforcement of it's master's wishes that everyone will have no choice but to treat the AI's master as a God.
Re: (Score:2)
YOUR electric rates will skyrocket due to demand from the Transhumanists to bring about their AI God.
https://www.flickr.com/photos/... [flickr.com]
February (Score:2, Troll)
This will all get reversed around February, like it or not.
Not how large grids work (Score:2)
In a large electrical grid - and the NERC Eastern Interconnection is very large - there is no direct connection between even the largest generator and the largest load (typically aluminum smelters and electric arc steel mills, but some datacenters are now quite large). It is like a copper alloy sheet of varying thickness and inductance: energy added at one point will flow into the sheet and energy demanded at another point will flow out of the sheet but there are many paths by which the consumed energy is s