Amid explosive demand, America is running out of power

https://www.washingtonpost.com/business/2024/03/07/ai-data-centers-power/

by thm • 2 years ago

36 76 2 years ago

washingtonpost.com

h2odragon2• 2 years ago

https://archive.ph/xbA0V

h2odragon2• 2 years ago

We don't all even have our electric cars and at home quick chargers yet.

The electric utilities have been singing "We need more generation capacity" for a couple decades.

thelastgallon• 2 years ago

EVs provide flexible demand, they are parked 22 hours+ a day, can charge whenever power is negative (which happens 100s of millions of times/year) or power is zero. Right now, there is excess production from renewables, which is wasted. EVs can absorb all that.

We don't need quick chargers at home, unless you mean 240V chargers? Which is in some of the countries with 110V/120V.

We don't need more generation capacity. Utilities are keen on increasing costs, their profits are a percentage of cost, there is a nexus between generation, transmission and distribution companies to keep adding capacity. This was pointed out by many commenters, a recent out: https://news.ycombinator.com/item?id=39487714

Energy demand will down with EVs and renewables: https://www.sustainabilitybynumbers.com/p/iea-energy-scenari...

Whats needed is demand response. But utilities won't do it, because it won't increase their profits.

quantified• 2 years ago

How does the amount of energy necessary to recharge the vehicles compare to the amount of energy required to run the additional industry?

ZeroGravitas• 2 years ago

Note the energy demand going down is due to the inefficiency of fossil based transport and heating that gets replaced with electrical versions.

But it'll still mean an increase electricity production/consumption.

It's not a big drama, and it's already planned for, but it'll still be an increase (even after accounting for efficiency gains).

bryanlarsen9• 2 years ago

Electric cars will take ~30 years to transition (~10 years before almost all new cars are electric, and then another 20 years before the old cars are off the road). They will increase electrical demand by 20% over 30 years, or under 1% a year increase to demand.

The US increases its generating capacity by about 4% per year. (50GW added per year, 1.3TW total).

EV's are a very small part of demand increase.

foobarian2• 2 years ago

The other day I was at a Target parking lot plugging in the EV at an island of about 6 superchargers. I looked over the rest of the parking lot at hundreds of ICE cars and it just struck me that getting the whole fleet electrified has got to be a substantial power demand increase, but maybe not. Which raises the question, what the heck is all the power being used for? :-)

luch• 2 years ago

I don't think so, the adoption rate has done nothing accelerating since WWII. Once EV will be mainstream enough the switch will probably be quite brutal.

I don't think it is economically viable to maintain two sets of power distribution (electricity and petrol) at the same time so countries will probably "push out" traditionnal petrol stations once they think EV distribution is okay enough

Georgelemental• 2 years ago

50 years ago yesterday, French Prime Minister Pierre Messmer announced an ambitious plan to build 13 nuclear power plants to fulfill most of the country's energy needs: https://en.m.wikipedia.org/wiki/Nuclear_power_in_France#Mess...

Time for America to do the same?

bryanlarsen9• 2 years ago

Good idea 20 years ago. Announcing a program that'll solve the problem in 20 years won't solve today's problems.

vianneychevalie• 2 years ago

At least in France we’ve been saying that for 8 years.

The second best time to plant a tree is now.

falserum3• 2 years ago

https://www.sustainabilitybynumbers.com/p/nuclear-constructi...

It takes around 6 to 8 years to build a nuclear reactor.

(Note: There is no data on planning part though; not so easy to methodically measure probably)

Point still valid, but 20 years seems excessively pessimistic.

TrueGeek• 2 years ago

The best time to build a nuclear power plant is 20 years ago, the second best time is now.

andruby• 2 years ago

true, but that's the short-term thinking that gets us here

zpeti• 2 years ago

Highly doubt this is a good political move atm, so it’s very unlikely to happen

foobarian2• 2 years ago

Why not though? I thought the anti-nuke stuff kinda petered out by now.

blueferret• 2 years ago

While I agree that it would make a lot of sense, as another user commented, it's too much of a political football. More's the pity. Nuclear is an incredibly useful power generation method. With current tech, it would take much less time to stand up than it used to.

NoMoreNicksLeft2• 2 years ago

There are legitimate technical/logistical issues. Nuke plants require alot of water for cooling, have to be sited at a river or lake or something, and the ambient temperature of that water determines whether or not it is sufficient for cooling needs. Even when this is the case, the plant can often raise temperatures enough with its discharge to murder native species living in it downstream of discharge. Like with hydro, there's only so many places nukes can go. Even without nimbyism and politics, nukes may not scale. Depends on how power hungry the next few generations become.

MichaelRo• 2 years ago

> ambitious plan to build 13 nuclear power plants to fulfill most of the (France) country's energy needs

Scaling the surface this means 200 nuclear power plants in the US. There already are 50 so another 150 to go. More chances for Fukushima / Chernobyl situations.

the_third_wave• 2 years ago

Don't build them on fault zones and tsunami-frequented coastlines, problem solved. There's sure to be 150 locations in the US where neither earthquakes nor tsunamis are frequent occurrences. Also, don't build graphite-core dual-purpose (power- and weapons-grade fissile material-producing) reactors without documenting the dual purpose and without telling the operating personnel about reactor behaviour under stress situations.

1oag2• 2 years ago

THE USA hates to think outside the box especially for long term planning..

grw_• 2 years ago

Great news- if you're a fan of Nassim Taleb: "I've seen gluts not followed by shortages, but I've never seen a shortage not followed by a glut."

2devnull• 2 years ago

Already priced in. That’s the problem with Taleb. His insight is not actionable.

whiterknight• 2 years ago

If it’s giving you true insight into how something works, then how is that not valuable?

Are you suggesting everyone already knows it?

diab0lic• 2 years ago

I have no take on whether or not this specific situation is priced in or not. However I’d challenge you on the second comment. I’ve found Taleb’s advice very actionable but it did take a lot of thinking about it to move from abstract thoughts into practice.

peteradio• 2 years ago

You've read infinite jest now try infinite jackass the story of a man with a stupid name and a single note to play.

stuff4ben• 2 years ago

This is an easily solvable problem. Make crypto-bros subsidize the electric generation they consume. Levy a tax on bitcoin mining and use that to build more power stations.

infecto2• 2 years ago

I don't think cryptos serve any use at the current moment but this comment is out of touch with reality.

pavlov3• 2 years ago

Why not levy a tax on Bitcoin transactions in general, including the new ETFs?

All Bitcoin was generated through energy waste. Taxing it over its lifetime will reduce incentives to continue producing it.

spacewrangler• 2 years ago

Who determines what is "waste" ?

wyager3• 2 years ago

Why not levy a tax on Bitcoin transactions in general

Do you have an idea of what this would entail at a legal, accounting, or technical level? This is sort of a bizarre suggestion from any of those perspectives

staringback• 2 years ago

Bitcoin already reduces incentives to continue producing it over its lifetime by design.

wyager3• 2 years ago

Bitcoin mining does subsidize electricity generation. It creates an extremely reliable floor price market for power, stabilizing demand and making power generation a more predictable business. There's a reason large renewable facilities are happy to contract with nearby bitcoin mines.

anankaie• 2 years ago

You're right. We should make it so they have to pay for the energy they consume. Maybe price it by kilowatt-hour or something.

mistrial9• 2 years ago

fat people demand donuts

agentultra2• 2 years ago

A source of the increasing demand are data-centres scaling up for AI (machine learning) workloads... and crypto mining is still happening I guess. They're also sucking up a lot of fresh water for these tasks. I'm curious how significant this portion of the demand is.

If it's big enough, maybe start making those companies pay for these externalities they're forcing on the rest of society.

Though for other industrial uses... does the U.S. have a plan for SMR research and policies for deploying them when the tech is ready? I'm a bit ignorant on that one. I know Canada has been making moves in that direction.

quirk2• 2 years ago

Externality of energy consumption? I'm pretty sure they are already paying for that. Or are you saying they should pay more per KWh than some other user for some reason? As for SMRs my guess is that local regulators will be the first hurdles to overcome. Cities and states will either embrace or reject the risk/reward there. I know the DOE approved one last year at the national level.

prirun• 2 years ago

My small city of 50K is soon to be home to an $800M 700K sq ft Meta data center that will employ 100 people. I don't think the planners involved in these deals care much about risks, like the power company might have to petition the state regulator to increase rates for everyone so they can build a new power plant to satisfy demand. IMO they are only focused on the reward part.

Or are you saying they should pay more per KWh than some other user for some reason?

I just checked, and residential electricity users pay more than commercial or industrial users:

"Residential Sector: The average price is approximately 17.08 cents per kWh. Commercial Sector: The average price is around 13.10 cents per kWh. Industrial Sector: The average price is about 9.47 cents per kWh."

https://shrinkthatfootprint.com/commercial-vs-residential-el...

The good thing about a data center is at least with fewer employees, there isn't a huge pressure to build new housing and schools.

jameshart• 2 years ago

I’m not even sure why we would need to ‘make’ them pay for this. As it says in the article:

A major factor behind the skyrocketing demand is the rapid innovation in artificial intelligence, which is driving the construction of large warehouses of computing infrastructure that require exponentially more power than traditional data centers. ... Tech firms like Amazon, Apple, Google, Meta and Microsoft are scouring the nation for sites for new data centers..

Sounds like these companies have capital investment available, and their demand is inducing increased supply of data centers.. which should also induce demand for electricity supply too.

If they really want to grow fast they are welcome to invest in generating capacity and grid infrastructure too.

agentultra2• 2 years ago

If they really want to grow fast they are welcome to invest in generating capacity and grid infrastructure too.

That's what I was getting at: if the grid is burdened by their demand, and the demand is outpacing other uses (preventing other productive use of that electricity like say... powering air conditioners in a heat wave) then maybe they should be contributing a portion of their capital to building more capacity.

Why does every one else, from residential users to small businesses, need to suffer rolling blackouts because a bunch of VPs need to, "win at AI?"

At this point it seems like the big tech companies are simply using up as much supply as they can get in order to grow and that there's only so much capacity to go around since funding for expansion comes from investments in projects that aren't being funded by the largest growing set of users creating that demand.

jeffbee• 2 years ago

Another article that misrepresents the situation with hyperscale datacenters. They are concentrating loads into extremely efficient cloud facilities, which shifts a lot of energy from one column to another in EIA's reports. It used to be if you had a 100kW corporate datacenter with a PUE of 3.0 and a cost utilization of 1%, that was counted under "commercial buildings" or "office". With those workloads now running in PUE 1.1 datacenters where the cost utilization is 80%+, the true situation is much better, but the energy use is in another column.

1oag2• 2 years ago

The same can be said about WATER and how little or clean drinking water is available.. If Nuke firms needs tons of water where is the water going to come from which already is scarce.. NOW! While I'm at it.. get rid of all mobile devices which literally suck energy.. kill off streaming and go back to landlines/cable TV. :-O See, the problems are too vast to say one solution is better than the other but yet here we are.. Go figure but I'm not.. For starters, I know how to cook without electricity and I dont depend on a mobile device.. Small things we'll need to know and learn when we don't have electricity anymore..

infecto2• 2 years ago

What is with the explosion of these manic sounding doomers? I appreciate conservation and am thoughtful of my use of resources but holy heck, these folks are pretty set on a lifestyle that we only read about in sci-fi books.

While nuclear generation plants do create waste, I was under the impression that most modern designs do a pretty decent job of recycling water that is contaminated or not having contaminated water to begin with so cooling water just gets ejected back out?

Amid explosive demand, America is running out of power

Related Stories