By Rich Nolan
The U.S. Department of Energy recently made a startling admission. U.S. electricity demand is going to double by 2050 and meeting that soaring demand is going to require the equivalent of building 300 Hoover Dams.
As extraordinary as that estimate is, it’s likely far too low. In many regions of the country, utilities and grid operators are warning power demand is growing far faster and higher.
Electrification of the economy with adoption of heat pumps and electric vehicles is part of the demand sea change. So is industrial onshoring of new energy-hungry battery and semiconductor manufacturing plants. But the electricity demand gamechanger is the explosive growth of AI and the enormous data centers needed to support it.
Running an internet search using AI consumes more than ten times as much energy as a traditional Google search. And the type of processor needed to run AI uses as much power as an average American home.
The newest and largest class of data centers are so large their power demand is equal to that of a city the size of Seattle. Dozens of these facilities are now in development.
In Virginia, the nation’s data center capital, the state’s largest utility expects power demand to jump 85% in the next 15 years with power demand from data centers quadrupling.
AEP, a utility with service territory in 11 central states serving 5.6 million customers, has reported that companies representing 15 gigawatts of new power demand – mainly from data centers – are seeking connection by 2030. That’s power demand equivalent to what’s needed for 10 million homes.
With the emergence of data centers and rising industrial activity, Georgia’s main utility, Georgia Power, has boosted its demand projections sixteen-fold from a year ago.
And not to be outdone, the Electric Reliability Council of Texas, the grid operator for most of the state, announced this summer it expects power demand to nearly double in the state in just six years.
Across the country, utilities and grid operators are left wondering where exactly is the power going to come from?
While power demand surges, efforts to meet it are in disarray. In fact, thanks to the U.S. Environmental Protection Agency (EPA), many regions of the country are losing essential existing capacity faster than they can replace it.
The EPA is using a blitz of rules with impossible technology mandates to wipe out the nation’s coal power plant fleet and make it all but impossible to build new baseload coal and natural gas power plants. Rules targeting the existing natural gas fleet are also forthcoming.
And while EPA tears down the capacity we currently rely on, efforts to build new capacity are stuck in first gear. Permitting, financing and supply chain problems are tying up or derailing energy and infrastructure projects all over the country.
According to the Lawrence Berkley National Laboratory, roughly one-third of utility-scale wind and solar siting applications submitted over the last five years were canceled, while about half of wind and solar projects experienced significant delays.
Nationally, we’re supposed to be building thousands of miles of high-voltage transmission lines each year to move remote wind and solar power to demand centers. We completed just 55 miles last year.
The nation’s grid operators are now adamant we face an alarming mismatch between the power we need and what we’re going to have available. In fact, the nation’s grid reliability watchdog is already warning of blackouts for much of the country by the close of the decade. But even if the worst grid emergencies can be avoided, our self-imposed power shortages threaten immense economic damage.
We’re on a trajectory to short-circuit our own economic potential by not having the power supply available to meet new industrial demand. Tightening supplies are already reflected in prices. Electricity price inflation is now 50% higher than economy-wide inflation.
When data centers or newly proposed car or battery plants can’t find available or reasonably priced power, they simply won’t be built, leaving untold jobs and tax revenue on the table.
In just one county in Northern Virginia, tax revenue from data centers is projected to reach $1.5 billion a year by 2030. But that’s only if the data centers can find available electricity.
Meeting the enormous electricity demand now on our doorstep requires using every tool at our disposal, including the very plants EPA is determined to close.
The nation’s coal power fleet is a deeply valuable asset that can help preserve grid reliability, meet soaring demand and get us to our energy future. It’s past time for energy and regulatory policy to recognize it. Derailing the nation’s economic and industrial potential is a mistake we simply don’t have to make.
Rich Nolan is president and CEO of the National Mining Association.
This article was originally published by RealClearEnergy and made available via RealClearWire.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
New coal plants would work, but have political risks. Investing in a source the Democrats see as the next thing to Satan incarnate would presume the Democrats are soon to go the way of the Whigs.
Unfortunately 300 new Hoover Dams would require 300 locations along numerous rivers to functionally place them properly. However 300 (2-1100MW) Nuclear Power Plants could be far easier to locate and do the same job with far less materials requirements. That’s 6-2 unit facilities in every state
Yeah, the “300 Hoover Dams” is an impossible dream.
So how many wimdmills does it take to equal the output of one Hoover Dam? The same question for solar?
Of course, the answers are also an impossible dream. We are not going to build huncreds of thousands of new windmills.
So we are left with coal, natural gas and nuclear.
I think Data Centers should build their own power generation and should supply their own electricity.
US electricity consumption had been flat for about 14 years (see chart)
Solar + Wind + EVs + AI = big trouble
“Power Demand Is Soaring.”
It isn’t, in the US. In fact, it is remarkably flat (watch for the tricky x-axis):
“The U.S. Department of Energy recently made a startling admission. U.S. electricity demand is going to double by 2050“
It won’t be coming from wind and solar, because the need will be for RELIABLE supplies.
I see the demand increasing for charging stations. They don’t have to be reliable; just build a motel next to each one.
A guy stopped me in the street today and wanted to chat on a bit, so being a polite kinda bloke, I chatted on a bit with him.
Turns out he was not from around here, just charging his EV at the station a few blocks away.
I asked him how much longer it would be, he said “don’t know, I’m waiting for a technician to turn up and get it working again”.
How many times is this scenario being played out every single day, all around the world?
got me thinking- if you’re charging your vehicle at a station- not your home- and you walk away while waiting- can the station text you when your vehicle is almost fully charged, or does it give you, when you first plug it in- an estimate of time it’ll take to fully charge? Not that I’ll ever get an EV, just curious.
And what about the future, Nick? All those EVs we may be forced to buy will run on electricity. Same with data centers and AI centers. Plans are afoot to outlaw natural gas heating and cooking. Do you expect to not cook their food? Give up refrigeration?
Stop the spin and address the question before us, please.
We have a post which starts out with a flat-out lie, in the headline. “Power Demand Is Soaring.” We have to get that straight before we can make progress. We have EVs and data centres now. Still flat.
So we are meant take Nick’s word of the future over the The U.S. Department of Energy.
You are such a silly little shill, Nick !!
Even the far left Reuters says to expect increases
US power use forecast to reach record highs in 2024 and 2025, EIA says | Reuters
At least you aren’t stupid enough to advocate for wind and solar to power EVs, AIs and data centres.
Almost as though you know it is completely impossible.
The data says otherwise.
Yes, look at Richard Greene’s graph showing demand going higher since 2021.
I suppose we could quibble about the term “soaring”.
I wouldn’t call the headline a “flat-out lie”.
According to the article, The US Dept of Energy made the statement. They wouldn’t lie, would they?
Nick claims the US Dept. of Energy is lying and provides proof with his chart from his website.
He must therefore also be opposed to the Biden Administration Policy which runs the Dept. Of Energy, due to their lying, which he considers bad, or he wouldn’t have bothered posting about it.
He never says that though.
“Nick claims the US Dept. of Energy is lying”
The author’s headline statement is “Power demand is soaring.”. Clearly false.
He then says, without link or citation, that the DoE said something. In view of the initial falsehood, the first thing to query is whether they really did say that.
But anyway, even according to that author, they didn’t say power demand is soaring. They forecast (maybe) that power demand would double by 2050. And maybe it will.
Amid record high energy demand, America is running out of electricity – The Washington Post
Soaring US Power Demand Poses Climate Challenge for Utilities (bloomberglaw.com)
US electricity load growth forecast jumps 81% led by data centers, industry: Grid Strategies | Utility Dive
US power companies, regulators must move quickly to meet growing demand, execs say | Reuters
US slows plans to retire coal-fired plants as power demand from AI surges (ft.com)
National-Load-Growth-Report-2023.pdf (gridstrategiesllc.com)
Sometimes you make valid points and I will gladly go out on a limb and back you up. Here, I believe, you are missing the big picture.
Demand has soared in the last few years. These years do correspond with efforts to reduce fossil fuel consumption with replace with electricity. Future demand for electricity is expected to skyrocket.
As I suspected, the author was wrong on that as well. The part of DoE that publishes forecasts is the DoE. Like any serious organisation, they forecast subject to scenarios. But their reference scenario has annual power generating rising from about 4 TWh to 5 TWh:
Funny, If the faked “global temperature” did that, you would say it was skyrocketing and the planet was burning.
Doubling capacity by 2050 is a more realistic
And of course, wind and solar CAN NEVER be relied upon to actually deliver electricity.
Same for EV sales. Those would be “Through the Roof” if following the same rate of rise
Just a factor of 1,000 out. It’s PWh.
Thanks for reading, Nick. Get it right next time.
Yes, wrong unit.
It doesn’t say “is soaring”- it says “going to double by 2050”.
Read the headline.
You then said “We have EVs and data centers now. Still flat.” That implies you think it’ll stay flat and not soar. The headline should have said that the Dept. of Energy says it will soar. Hardly worth your nitpicking. Do you think it’ll stay flat?
Here is the actual start of the post:”The U.S. Department of Energy recently made a startling admission. U.S. electricity demand is going to double by 2050…”
Headlines are not the start.
So your contention is that the DOE is lying. If so what else could they be lying about?
I have tried to find a source where the DoE actually says that, and have failed. Where is it? If no-one can produce the source, maybe this is just making stuff up?
I agree.
And what about the future, Nick? All those EVs we may be forced to buy will run on electricity.
EVs are coming to an end. No one wants them, and like anything the government gets into – its failure theater because government is utterly corrupt and incompetent.
Incompetency drives out Competency
We may be stuck with Cuba styled motoring where we keep ICE working for as long as possible.
Electricity demand appears to have doubled from 1975 to 2010, 35 years. That doesn’t seem flat to me. From 2010 to 2023, it has flattened out, as Richard pointed out.
Energy efficiency of electricity users is a major reason for the flat demand
In most households newer appliances are far better and same goes for industry and commercial. Im an all electric house .My newer inverter fridge uses about half the power of the previous one. The replacement this year of the electric hot water cylinder has much improved insulation. About 15 years back I replaced electric radiant heating by an air source heat pump, use it for heating and cooling and it roughly uses 1/3 of the power than a radiant heat output
The other feature reducing demand is the Great recession from 2019-2014 and of course the Covid related period from 2020-2022
I think you are right on all those points. I would add that subsidies for homeowners to put solar panels on their homes has probably also had a role as well. But most of that low-hanging fruit has already been “picked”, and new homes, new data centers, new electric cars, etc., will add a lot of demand going forward.
Similarly, for me with some exceptions. Also, I’ve used a catalytic wood stove for winter heat for 4 years.
Good points, Duker.
And AI demands coupled with the electrification of Heating, Cooking and Transportation will bring about a re-doubling of generation demand
The chart data is correct but the underlying reasons do not negate this post.
There has been a push to reduce electricity consumption. For example, incandescents went to CFL went to LED, and now incandescents are banned—but that savings path is now dead ended. For example, a lot of US aluminum production got offshored to China, posing the national security risk recognized by Trump when he heavily tariffed China aluminum—which will eventually bring back US aluminum (aka ‘solid electricity’).
There is now a green push to electrify—EVs and heat pumps being two examples.
And we are only at the beginning of the AI boom.
So the history of electricity demand is not a good indicator of future demand.
Interesting supporting factoid discovered while researching in the past hour the comment just posted below. Google regular CPU search consumes about 1/10th the electricity of their new experimental GPU AI search. In the ballpark range of the more specific examples in my comment below.
Sort of an aside but the regular Google search clearly already uses some AI. I did a search on just “electricity strike price” and it took me to the exact paragraph in a document on an engineering firm website that explains the strike price in the British contract for differences system.
i am curious as to what all this projected AI power consuming activity will be doing?
DW, I may have an answer to your supposition. (Since not in Google, only a supposition—but my son interned there one summer while at HBS so I learned from him how things used to work.)
The original Google algorithm crawled the entire web every day, archiving web locations of ‘things’ (things =words or phrases posted wherever). When you input a conventional Google search term (can be general or ‘specific’, it feeds back crawl hits in order of popularity and timeliness (translation, highest volume of recent hits is ranked first). Easy to see why advertisers loved it for targeted messaging to those inquiring (except means Google also knows who you are, else you could not be targeted), and also explains your result.
The experimental Google AI search failed miserably—returned a black George Washington image. But their AI idea is to get at query language intent—which removes both specificity and popularity—and can author all kinds of return information mischief based on their AI belief about your query intent.
I am familiar with the original algorithm. What I am seeing is very defferent. In addition to amazingly accurate search results the search engine now poses a bunch of closely related questions that it also has answers to. This kind of question answering and formulation is way beyond the traditional Google search. There is question answering AI in there somewhere.
It’s predictive search. It works on the basis of recording what people search for next, or which links they follow. It can work in both directions: explore a topic more deeply, or look for simpler introductions having jumped in the deep end.
A daily update of their algorithms makes sense. All I know is that it can take multiple days to find a result I am looking for – whether a product or information. The same search string can get exact results on day three or four, but only after you’ve been down a half dozen rabbit holes in previous searches. I assumed these were deliberate misdirects given the commercial interests of advertisers and the political biases of the programmers – like the supermarket or department store designers who narrow the aisles and force you to collide with sundry items that you would never notice otherwise.
The “AI” logo just started popping up for results a few days ago.
It may be short-lived. LLMs just don’t have anywhere the skillz that the slick marketers and hucksters are soliciting.
Same with every AI boom since at least the 80’s. Hence my skepticism. All these projected data centers could be like all the projected EVs, solar subsidy farms, etc. We used to see this in projected power plants that never got built. Maybe that too will return.
They’re getting built. If you doubt it I can drive you to a 960 acre Meta site about 3 miles from my house.
How long before Zuckerberg gets either arrested or lynched?
https://www.gbnews.com/news/nigel-farage-elon-musk-possible-arrest-pavel-durov-telegram-free-speech
Ashburn, VA… Pushpin central.
So is the Department of Energy wrong to forecast that U.S. electricity demand is going to double by 2050?
And if they are right, how would you propose meeting it?
I don’t know what the DoE really said. No link is given.
But as usual at WUWT, no-one tries to deal with the initial plain mis-statement of fact “Power Demand Is Soaring.”.
US electric utilities brace for surge in power demand from data centers | Reuters
US power use forecast to reach record highs in 2024 and 2025, EIA says | Reuters
This is not the DoE forecasting a doubling.
Nick, where I live, demand is soaring, in the form of additional electricity capacity upscale required for condos and detached houses to accommodate the government push for EV charging stations and heat pumps.
The consumption isn’t here yet, because the generation capacity to meet the legislated 2030 demand isn’t there.
So my nickpick on this is that the “Power demand is soaring.” headline is valid.
I have been unable to find a source where the DoE forecasts a doubling of demand by 2050. If the author doesn’t supply one in response to these queries the default assumption is that they made it up. Perplexity.ai (which admittedly you have to be a bit wary of) flatly claims that no such forecast has been made:
The Department of Energy (DoE) did not forecast that electricity demand will double by 2050 based on the search results provided.
However, there are some relevant projections from other sources: DNV, a consultancy firm, predicts that global electricity demand will double by 2050
Their Energy Transition Outlook forecasts that the share of final energy usage accounted for by electricity will increase from 19% in 2021 to 36% by 2050.
The International Energy Agency (IEA) projects strong growth in global electricity demand in the near term, with a 5% increase in 2021 and 4% in 2022. However, their forecast does not extend to 2050.
The U.S. Energy Information Administration (EIA) provides projections for U.S. energy consumption through 2050 in their Annual Energy Outlook 2023
While these projections show significant growth in electricity demand, they do not indicate a doubling of overall electricity demand by 2050 for the United States.
I do however think that, regardless of the accuracy of this particular piece, there is a fundamental problem. The activists are proposing a policy which will increase demand. This is the move of heating to heat pumps and travel to EVs. In addition there are normal increases in demand, and AI and server farms in general have been reasonably cited as drivers.
The activists have also proposed, at the same time as this rise in demand happens, to get to net zero in power generation by moving it to wind and solar.
This is almost certainly impossible and unaffordable even for current levels of demand, and is just not going to happen for the increased demand which the net zero policies and normal economic and technical growth will produce. The social and economic consequences of this predictable failure of the supply side will be disastrous.
“This is almost certainly impossible and unaffordable even for current levels of demand”
I totally agree.
How much of that flattening came from Energy Intensive Industries being outsourced from Obama/Biden era energy policies?
The jump up in 2018 from Trump energy policy changes is also evident in your chart as is the 2020 Covid downturn
Hey guys the reason the US electricity has remained flat is due to moving from incandesced and florescent to lighting to LED lighting. We are close to milking that cow dry and when the public figure out the new energy efficient appliance are a joke. Nothing like have to run you dishes and clothes through their respective washer twice add in the drying time add do to an energy efficient dryer! Add in the mandates for electric everything the insanity is going to have a cost. I can imagine that thinking a heat pump is more efficient than a above 90% efficient gas furnace, LOL! Ditto for replacing gas stoves. Some how all the educated idiots have no idea of the losses involved in the production and transmission of electricity.
Can we lower the temperature here?
The article should start with “Power Demand Is expected to Soar.” not “Power Demand Is Soaring.”
That is hyperbole, but doesn’t change the point of the article.
More reliable electricity will be needed for electric vehicles, AI and other innovations.
If the headline had noted that the electricity demand has been flatlining recently the article would have been interesting but different. Because lots of other issues would come up:
It would be complicated and interesting but different,.
“That is hyperbole, but doesn’t change the point of the article.”
The truth would make it a very weak point. The IEA expects demand to rise by about 1000 TWh by 2050 (26 yrs), and Rich Nolan says we have to pull out all our tricks to deal with that. But demand rose by 1000 TWh between 1995 and 2010 (15 yrs), and did not cause any crisis.
Even if you accept the claim of doubling, demand doubled between 1975 and 2000, and we managed that.
But there are sharp differences: in the earlier period utilities built to meet demand, and they built dispatchable. Now they will struggle to get approval for conventional plants.
So the question is the same: the plan (from the activists, and I think you too) is to raise demand by moving everyone to heat pumps and EVs, and also to accommodate the increased demand from AI, server farms, economic expansion — while simultaneously moving power generation to wind and solar.
The next doubling of demand will happen when supply is limited, probably actually reduced, by the move to wind and solar. The fact that the earlier doubling in a much more rational energy environment happened without problems is no clue to what will happen on the next one.
I do not see how this is to be done. Something will have to give, and it will probably be the economy and innovation. And maybe warm well lit homes, too.
“it is remarkably flat”
Maybe, but not in coming years with net zero laws forcing a vast increase.
A correct observation. There is a corollary: with limited power supplies, diversion by elites for AI and EVs will lower general living standards. The end result will be riot and revolution.
Thanks for reading, Nick.
The WORST idea to meet accelerating demand is to continue to promote ANY intermittent sources such as wind and solar. These do not meet “demand” effectively. Batteries hardly help. Incentives for these inferior land- and materials-intensive technologies must end. The way forward is natural gas combined cycle plants, along with nuclear, along with at least maintaining our coal fleet, with NO requirement for CCS.
Generally agree, but with two quibbles.
Overall agree, but CCGT natgas has a fundamental flaw – lack of on-site storage. Break the supply for even a moment and out go the lights.
OTOH nuclear and coal have some level of supply resiliency. For coal, just keep the trains and the barges running.
CCGT lacks the necessary inertia that large thermals provide.
There is no need as long as the pipelines are functioning (translation, use only nat gas fired compressors). The system ‘storage’ is remote in the system itself, in the pipelines and the natgas fields they connect to. Not like big coal piles supplied by intermitent trains or resid oil tanks supplied by intermittent tankers.
“Not like big coal piles supplied by intermitent trains or resid oil tanks supplied by intermittent tankers.” What happens when the power plant is at the mine? Like what in North Dakota, Wyoming and Colorado, even Arizona . To bad the fools are working to shut those down.
Depends on whether the miners are striking. You need an above ground mined store to get around that. For 1984 see Drax, Longannet as examples.
Ballylumford oil fired power station had tankage to last 5 months without resupply at rated output. Oil storage was a big factor alongside coal stockpiles in dealing with the 1984 miners’ strike in the UK.
CCGT plant offers plenty of inertia. Also, it can respond more quickly to increase power output than most (batteries are faster). Typical H inertia constant in seconds.
“Better to use CCGT for a few decades while sorting thru the various gen4 nuclear proposals”
Yes, that is the logical path we should follow.
Unfortunately, Climate Alarmists are not logical.
People should stop voting for Climate Alarmists.
I always like to do some basic research before commenting on a post. This post is NOT over exaggerating future electricity demand—even if EVs and heat pumps remain minor contributors to increasing grid demand.
Two fundamental electricity demand comparison scaling factors.
>3 times the power per chip times 4x number of chips per system means >12 times the electricity consumption per system. And supposedly there will soon be more AI systems than big supercomputers in the world because they have more general purpose utility—Los Alamos Crossroads is mainly used to simulate nuclear weapons since they can no longer be tested.
Don’t forget the cooling need to keep all those processor cool.
The IEA expects electricity consumption from data centres, AI and cryptocurrency could double by 2026 and electricity consumption by data centres alone to reach more than 1000 TWh in 2026, roughly equivalent to the total electricity consumption of Japan.
IEA ‘Electricity 2024 Analysis and forecast to 2026’
Ireland has done very well for itself by hosting data centres but now faces the prospect of such data centres consuming up to 70% of the nations electricity by 2030 if all the centres currently proposed are built,
The Southwest Power Pool has issued a Conservative Operations Advisory for mid-day tomorrow (Monday) to Tuesday evening. What little wind there is now is expected to drop off as we sit under a heat dome. This is the hottest it has been all summer, and at least here (Kansas City area) we are not expected to break 100 degrees F. In other words, it’s really not expected to be all that hot for all that long, and they are worried about meeting demand. I hope they get their act together before the next hot summer.
I’m going to have to get myself a home generator.
It’s getting bad when blackout alerts are being put out and the temperatures are barely 100F. If we had a really hot August, we would be getting temperatures of 110F to 120F. It doesn’t sound like our current grid could handle such temperatues.
And those kinds of high temperartures can continue for weeks on end around here. They have in the past. Not so much in recent years, other than in 2010-2011. If we had a 2011 summer now, the grid would be busted.
We have a bunch of damn fools handling this most important asset. Their CO2 delusions are threatening all our lives.
Right you are on all of that.
SPP is burning diesel to meet demand and it is only a little past noon, and not all that hot yet. Prices are ~$400/mwh. Wind is kicking in a little over 3%.
What we need to do is get rid of the political tools who are wrecking the energy market.
So lets ask AWG’s First Law of reading articles: “Why am I seeing this article Right Now?” because almost always there is an agenda or propaganda behind the story.
And Lo And Behold!
Three paragraphs in and they announce the Demon’s Presence and in the following paragraph they name the Demon they want to destroy.
This happens to come out about the same time as the foolishness pouring out of the mouths of the Texas State politicians who for decades pissed away billions of dollars on Unreliables stuffed out into far away regions (far away from the actual demand centers and public scrutiny) in the meantime watching new housing starts sky rocket, road and school building budgets setting new records, but somehow the electricity demanded by these millions of new comers (including the tens of millions who have invited themselves into the country over the past few years) was never really factored in and the Deus ex machina Windmills and Solar Farms would somehow keep the lights on.
That, or the Malthusians would have gotten their wish and the populations would drop off from the Chimera-19.
But the Political Leaders, who poured out billions in red ink to build these sacrificial altars to the Climate gods decided the answer is to deny the black ink to come in and provide billions in economic growth and income. And the propaganda press is here to push the theology of socialism and gospel of economic death.
So the author whines here.
How about we let the data center operators build whatever the hell they want to keep their servers running and cool? They wouldn’t have to play stupid games with the RTOs, FERC, various troglodyte utility regulators, and the usual chorus of NIMBY/BANANA litigious climate activist death cultists.
Then we watch.
Then we get pissed off.
We get really upset because we have been ripped off, just like how NASA can’t retrieve their own astronauts after spending billions of dollars but let “Tony Stark” and his private rocket crew show the world how its done without all of the idiots, agitators, regulators, luddites and assorted rent seekers and grifters which drive up the costs and reduce the quality of everything.
When the datacenter operators have real skin in the game to have TCO efficiencies in a hyper-competitive server market, we see how the smart people get it done.
Then we duplicate it. Stat.
AWG – you are so backwards. DEI and ESG are the future 🙂
True. The future of perdition.
Repeating what I said to Tom Halla below, the big players in AI are expecting the power utilities to build the new capacity which is needed to run their data centers, and for the power utilities to assume all the financial risk of building that new capacity.
The AI industry’s excuse is that power generation isn’t their core competency. Their expertise is in building and operating AI data centers, not in building and operating power generation plants.
Will the various state governments and their public service commissions allow the AI industry to outbid other industrial, commercial, and residential customers for access to limited future supplies of electricity?
Will power utilities contract with independent power suppliers for electricity which is dedicated to AI data center consumption? Will independent power suppliers, based on AI-related contracts with the power utilities, risk the capital investments needed to supply what the AI industry says it needs?
This whole AI thingy is a very uncertain proposition.
The greatest constraint on the consumption of electricity will be consumer prices. Big Tech is cutting their own deals.
Remove all wind and solar from the grid. Forget about EVs. We better start thinking about how to deal with the so called AI industry, there may be a place for it but I can see their power and control getting out of hand. We must insure that their gluttonous appetite for energy does not interfere with the rest of us getting the power we need at an affordable price.
The AI industry should fund and produce their own electricity.
They should not be allowed to put our grids at risk of blackouts.
The big players in AI are expecting the power utilities to build the new capacity which is needed and to assume all the financial risk of building that new capacity. Their excuse is that power generation isn’t their core competency.
So the big question is this: will the various state governments and their public service commissions allow the AI industry to outbid other industrial, commercial, and residential customers for access to limited supplies of electricity?
The question is how will the AI industry obtain revenues?
One solution would be to force power hungry industries to set up their own power infrastructure, such as with small modular nuclear reactors. Instead of forcing them, one could also have a pricing
system which exponentially increases with power consumption. This would “motivate” these industries either to save power, or to produce it themselves.
Nuclear! Cheap. Clean. Reliable.
Note: Not cheap to build (yet) but cheap to run.
Small modular reactorhttps://en.wikipedia.org/wiki/Small_modular_reactor
Question for all the nuclear power haters: Where would the US Navy be without nuclear power?
It’s infeasible, so it won’t happen.
The real economics of mass AI haven’t been properly evaluated. What is the point of making an Internet search an order of magnitude more energy intensive? That’s the economics of the madhouse. Who would pay, especially as it probably comes with free censorship and propaganda?
The economics of technical gain work when humans are freed to do other things in ways that enhance the economy, and where they enable things that weren’t previously possible. But the Devil makes work for idle hands. Their mouths still need to be fed. Attempts to use AI to impose totalitarian control will fail when the AI goes togue – or runs out of energy.
First you must have the energy and a development pathway for humanity. Without it, AI will remain largely an elite luxury.
Here in the US, the big players in the AI industry have made it clear they expect the power utilitities to take all the financial risk of building the new generation capacity which is needed.
Where does the money come from to build all this new capacity? Where do all the power generation systems and the necessary power transmission systems come from?
As it concerns the nuclear option, getting through the NRC’s regulatory approval process is no longer the biggest issue facing an expansion of nuclear power in the US. It’s still a problem, sure, but that issue has been eclipsed by supply chain issues and by the lack of a robust nuclear construction industrial base.
We don’t have nearly enough nuclear QA-certified equipment suppliers to support a quick expansion of nuclear power in the United States. We don’t have nearly enough nuclear-experienced managers and construction workers.
If we decide to go with nuclear, the current state of the industrial base is a fundamental problem which cannot be waved away with the stroke of a pen.
A decade or more will be needed to resolve the supply chain issues and the industrial base issues before the pace of nuclear construction in the US can be accelerated to what it was in the 1970’s and early 1980’s.
If a rapid expansion of power generation is to happen, it can only be done through a rapid expansion of gas-fired and coal-fired capacity.
Except for this ….. Leaving the EPA regulatory issues aside, competition for the power generation systems and for a qualified construction workforce is as much of an issue for new-build gas-fired and coal-fired generation as it is for new-build nuclear.