Big Tech's AI Datacenters Demand Electricity. Are They Increasing Use of Fossil Fuels? (msn.com) 56
The artificial intelligence revolution will demand more electricity, warns the Washington Post. "Much more..."
They warn that the "voracious" electricity consumption of AI is driving an expansion of fossil fuel use in America — "including delaying the retirement of some coal-fired plants." As the tech giants compete in a global AI arms race, a frenzy of data center construction is sweeping the country. Some computing campuses require as much energy as a modest-sized city, turning tech firms that promised to lead the way into a clean energy future into some of the world's most insatiable guzzlers of power. Their projected energy needs are so huge, some worry whether there will be enough electricity to meet them from any source... A ChatGPT-powered search, according to the International Energy Agency, consumes almost 10 times the amount of electricity as a search on Google. One large data center complex in Iowa owned by Meta burns the annual equivalent amount of power as 7 million laptops running eight hours every day, based on data shared publicly by the company...
[Tech companies] argue advancing AI now could prove more beneficial to the environment than curbing electricity consumption. They say AI is already being harnessed to make the power grid smarter, speed up innovation of new nuclear technologies and track emissions.... "If we work together, we can unlock AI's game-changing abilities to help create the net zero, climate resilient and nature positive works that we so urgently need," Microsoft said in a statement.
The tech giants say they buy enough wind, solar or geothermal power every time a big data center comes online to cancel out its emissions. But critics see a shell game with these contracts: The companies are operating off the same power grid as everyone else, while claiming for themselves much of the finite amount of green energy. Utilities are then backfilling those purchases with fossil fuel expansions, regulatory filings show... heavily polluting fossil fuel plants that become necessary to stabilize the power grid overall because of these purchases, making sure everyone has enough electricity.
The article quotes a project director at the nonprofit Data & Society, which tracks the effect of AI and accuses the tech industry of using "fuzzy math" in its climate claims. "Coal plants are being reinvigorated because of the AI boom," they tell the Washington Post. "This should be alarming to anyone who cares about the environment."
The article also summarzies a recent Goldman Sachs analysis, which predicted data centers would use 8% of America's total electricity by 2030, with 60% of that usage coming "from a vast expansion in the burning of natural gas. The new emissions created would be comparable to that of putting 15.7 million additional gas-powered cars on the road." "We all want to be cleaner," Brian Bird, president of NorthWestern Energy, a utility serving Montana, South Dakota and Nebraska, told a recent gathering of data center executives in Washington, D.C. "But you guys aren't going to wait 10 years ... My only choice today, other than keeping coal plants open longer than all of us want, is natural gas. And so you're going see a lot of natural gas build out in this country."
Big Tech responded by "going all in on experimental clean-energy projects that have long odds of success anytime soon," the article concludes. "In addition to fusion, they are hoping to generate power through such futuristic schemes as small nuclear reactors hooked to individual computing centers and machinery that taps geothermal energy by boring 10,000 feet into the Earth's crust..." Some experts point to these developments in arguing the electricity needs of the tech companies will speed up the energy transition away from fossil fuels rather than undermine it. "Companies like this that make aggressive climate commitments have historically accelerated deployment of clean electricity," said Melissa Lott, a professor at the Climate School at Columbia University.
They warn that the "voracious" electricity consumption of AI is driving an expansion of fossil fuel use in America — "including delaying the retirement of some coal-fired plants." As the tech giants compete in a global AI arms race, a frenzy of data center construction is sweeping the country. Some computing campuses require as much energy as a modest-sized city, turning tech firms that promised to lead the way into a clean energy future into some of the world's most insatiable guzzlers of power. Their projected energy needs are so huge, some worry whether there will be enough electricity to meet them from any source... A ChatGPT-powered search, according to the International Energy Agency, consumes almost 10 times the amount of electricity as a search on Google. One large data center complex in Iowa owned by Meta burns the annual equivalent amount of power as 7 million laptops running eight hours every day, based on data shared publicly by the company...
[Tech companies] argue advancing AI now could prove more beneficial to the environment than curbing electricity consumption. They say AI is already being harnessed to make the power grid smarter, speed up innovation of new nuclear technologies and track emissions.... "If we work together, we can unlock AI's game-changing abilities to help create the net zero, climate resilient and nature positive works that we so urgently need," Microsoft said in a statement.
The tech giants say they buy enough wind, solar or geothermal power every time a big data center comes online to cancel out its emissions. But critics see a shell game with these contracts: The companies are operating off the same power grid as everyone else, while claiming for themselves much of the finite amount of green energy. Utilities are then backfilling those purchases with fossil fuel expansions, regulatory filings show... heavily polluting fossil fuel plants that become necessary to stabilize the power grid overall because of these purchases, making sure everyone has enough electricity.
The article quotes a project director at the nonprofit Data & Society, which tracks the effect of AI and accuses the tech industry of using "fuzzy math" in its climate claims. "Coal plants are being reinvigorated because of the AI boom," they tell the Washington Post. "This should be alarming to anyone who cares about the environment."
The article also summarzies a recent Goldman Sachs analysis, which predicted data centers would use 8% of America's total electricity by 2030, with 60% of that usage coming "from a vast expansion in the burning of natural gas. The new emissions created would be comparable to that of putting 15.7 million additional gas-powered cars on the road." "We all want to be cleaner," Brian Bird, president of NorthWestern Energy, a utility serving Montana, South Dakota and Nebraska, told a recent gathering of data center executives in Washington, D.C. "But you guys aren't going to wait 10 years ... My only choice today, other than keeping coal plants open longer than all of us want, is natural gas. And so you're going see a lot of natural gas build out in this country."
Big Tech responded by "going all in on experimental clean-energy projects that have long odds of success anytime soon," the article concludes. "In addition to fusion, they are hoping to generate power through such futuristic schemes as small nuclear reactors hooked to individual computing centers and machinery that taps geothermal energy by boring 10,000 feet into the Earth's crust..." Some experts point to these developments in arguing the electricity needs of the tech companies will speed up the energy transition away from fossil fuels rather than undermine it. "Companies like this that make aggressive climate commitments have historically accelerated deployment of clean electricity," said Melissa Lott, a professor at the Climate School at Columbia University.
AI is ... (Score:5, Funny)
Re: (Score:2)
Hopefully someday AI will get rid of stupid clickbait.
AI in 2023 consumed an estimated 15 TWh. Bitcoin consumed 91TWh. Video games consume 230TWh. Can we stop pretending that it's having any meaningful impact on current power consumption?
These "huge power numbers for AI" are based on hypothetical futures where AI is in everything and is earning trillions of dollars of revenue and this is funding this massive AI arms race going on to provide the best model and that everyone always wants to use the best mo
Re: (Score:2)
"The roots of these developments lie in recent forecasts that show the power needs of Northern Virginia data centers growing by 7,500 megawatts by 2028 – a growth of about 35% over the 2021 load."
https://protectpwc.org/2023/08... [protectpwc.org]
35% growth equivalent to adding 50% more housing....in 7 years, just for the data centers.
We have *eleven* data centers going in within a 1/4 of our development. https://dmvcommunitymedia.b-cd... [b-cdn.net]
Probably (Score:4, Insightful)
Unless you can show that datacentres are getting their power from different sources than the general public (or power customers in other industries), all you have to do is look at expansion in power production in a given region and understand that datacentres are contributing to part of that expansion. If there's even one natgas or coal plant coming online, datacentres have something to do with it.
Re: (Score:2)
That should read:
Unless you can prove that the datacentres are getting their power from renewable sources that could never ever be used to produce power for the general public.
Re:Probably (Score:4, Interesting)
Higher demand does attract new investment, and that can be good for renewables. The main costs of solar power and wind are capital costs, and a huge new source of demand can attract investment to bring new renewable and grid storage projects online.
The problem with that rosy scenario is this: the industry is in a break-neck race to achieve dominance in a rapidly emerging and purportedly ultra-lucrative field. You could build your datacenter around a renewable energy project that would cut your electricity costs dramatically, but if supply chain issues with the energy end of your project delays the datacenter operation by a few months, all the money you spent might have been for nothing. If somebody else gets there first, you get nothing.
So moving in next to an old coal power plant that can be brought on line in a few weeks must sound pretty attractive, even if that means you're actually paying more for power. If you really believe in the near-term potential of AI, this could be the last opportunity ever to win big. If you believe AI will outperform humans, it will level the intellectual playing field, which means the people with the most resources will always prevail from now until the end of time. Note carefully, I'm describing fever-dream tech bro paranoia. If we're fortunate they'll all fail miserably, and we'll just be stuck with twenty or thirty gigatons of CO2 in the atmosphere for no good reason.
Re: (Score:2)
That's 20 to 30 years out for current solar technology and perhaps 20 for the wind turbine; the time adjusted value of that money is nil, and by that time the AI training gold rush will be long over.
Re: (Score:2)
Well, I don't know if it's typical human thinking. It's engineering. Nothing lasts forever, you factor the cost of parts replacements into the cost of your design.
Re: (Score:3)
Unless you can show that datacentres are getting their power from different sources than the general public (or power customers in other industries), all you have to do is look at expansion in power production in a given region and understand that datacentres are contributing to part of that expansion. If there's even one natgas or coal plant coming online, datacentres have something to do with it.
I know some folks in the utility business and tehy say teh data center's approach is "we don't care where the power come from, just give us what we need." Despite all the green claims, in the ned they'll ask to fire up an old coal plant if they need the power. They also say they can't build out fast enough to meet the project demand.
Of course yes, for anything that increases demand. (Score:2)
Energy is energy (Score:3, Insightful)
Re: Energy is energy (Score:4, Insightful)
I believe the question is important because AI applications are quickly becoming the number one consumer of power in the industry
Re: (Score:2)
We should care of an industry that is attracting massive amounts of investment and massive profits (well, at least for Nvidia) is creating a lot of demand that is being met by fossil fuels. In such a case it's not unreasonable to ask them to contribute a little more to make sure that renewables are developed to meet the demand.
It can't be nuclear because it will take too long to build.
every improvement in civilisation does this (Score:3)
Why would compute resources for AI be any different?
Re: (Score:3)
If we truly want an electric-driven society that doesn't also need fossil fuels, we need to plan to includ
Re: (Score:3)
The "and store" is the issue people keep hand waving away. If you are content to turn the data centers off at sundown much of the problem goes away.
It's when you insist on running them through is 16 hour winter night there is a problem. Then the next day is overcast and the PV panels are putting out a whole 7% of nameplate, so you need 14 times the number of panels you thought you did to run the data center plus enough extra to recharge the batteries for the next 16 hour night. The good news is cooling the
Nuclear is the solution (Score:4, Insightful)
Re: (Score:1, Insightful)
as someone pro nuclear you're not wrong but part of being an effective advocate is being humble when things are turning your way and build upon those victories as a way to gain more advocacy
the recent legislation is a big win, things are shifting but now is the time to convince more folks to support nuclear and also not blow what is probably the industries last shot at any survival long term and being so confrontational isn't it, this makes you no better than the anti nuclear advocates you talk about
just be
Re: (Score:3, Interesting)
Nuclear power in this age is simply too expensive. Look at what the latest plants are costing France. https://www.reuters.com/busine... [reuters.com]
Re: (Score:2)
Re: (Score:2, Insightful)
FTFY
The problem as with anything isn't the actual cost but all the hands in the cookie jar.
Getting to the bottom of it all (Score:2)
The tech giants say they buy enough wind, solar or geothermal power every time a big data center comes online to cancel out its emissions. But critics see a shell game with these contracts: The companies are operating off the same power grid as everyone else, while claiming for themselves much of the finite amount of green energy. Utilities are then backfilling those purchases with fossil fuel expansions, regulatory filings show... heavily polluting fossil fuel plants
This is why carbon taxes don’t work and never will.
Re: (Score:2)
Yes of course they are (Score:2)
And we already know Bitcoin and crypto in general which used the same basic processing power devoured electricity to the point where it was using more power than all of Argentina. Of course AI is going to guzzle down power.
The frustrating thing is as the AIs are taking our jobs and rendering us homeless it's also going to be taki
Remember Andrew Yang? (Score:2)
Re: Remember Andrew Yang? (Score:2)
Post-scarcity isn't happening as long as people believe in private property. Our culture on this planet isn't anywhere near the point where they would embrace the idea.
Re: (Score:2)
Oh shut up (Score:5, Insightful)
First off, yes they do. They require energy, and anyone who drills down enough will find oil / natural gas / misc pollution in the process.
The problem is that this complaint lacks perspective. The fossil fuels aren't being used in leaded gasoline fueling 6MPG cars. They're being used to power datacenters that offset *other* areas of fossil fuels.
"Zomg iPhones pollute, and so do Spotify's datacenters!" Yes, but they replaced ipods, which themselves replaced CD players. You can't convince me that pallets of AA batteries, powering a Discman that used twice as much plastic as an iPhone, to play CDs made of plastic, which were shipped by trucks to retail stores..is *better* for carbon emissions than a datacenter streaming music to a device with a rechargeable battery. You can't convince me that Netflix and its datacenters pollute more than DVD manufacture and transport...but we don't issue carbon credits for those.
Even if we bring it a bit closer to AI, lots of modeling is used to optimize things that themselves would require fossil fuels in one form or another. Not all of it, certainly, but there's no meaningful way to quantify the amount of pollution *saved* by the output of datacenters.
Surely, there are areas where AI (and energy, by extension) is used wastefully...but this is the sort of math that only works if it's selectively chosen to reflect the desired outcome. To be fair, this works in reverse; one can certainly show negative carbon emissions by calculating solar energy usage and ignoring the amount of energy used to build and transport the panels. ...but ultimately, this reads like an unnecessarily simple calculus, in turn sounding a lot more like an intent to shame datacenters into buying more carbon credits, rather than an actual concern to reduce pollution.
Re: Oh shut up (Score:2)
Consumerism causes unnecessary pollution.
Units (Score:2)
7 million laptops
Is the million-laptop unit going to substitute the ten-thousand-household/small city unit? I mean which kind of laptops, are we talking CISC or RISC, web browsing or video editing? So many questions.. How many million laptops will fit into a football stadium, I wonder? I for one, would prefer, as a more compatible unit, the energy needed to burn enough cd-roms to hold the whole Library of Congress. That would be an easy to grasp concept.
Not an issue here in TX (Score:3)
With the deregulated energy market consumers can require their retailer purchase renewable to cover their energy just like these datacenters. The more consumers who make that choice the more it shrinks the pool of usage that can be backfilled with fossil energy. Sure you can show all the studies of them escaping to other parts of the room today but eventually the renewable centipede is going to box the whole the screen off or it would if they weren't constantly packing on more renewable energy.
Most of the new datacenter builds I'm aware of also produce a significant amount of power to offset their consumption which isn't just green but saves them money. For instance if they can build in highly efficient cooling that uses little to no electrical power from the grid, sure you say they consume X million watts of cooling energy but since it is provided by them on site it isn't taking anything from anyone else.
Re: Not an issue here in TX (Score:3)
Re: (Score:2)
You've been listening to propaganda. The Texas grid is near the top for stability nationally.
Re: (Score:1)
Yes, Texas is near the top in reliability, if you start counting in the mid 20's. link [usnews.com]
Re: (Score:2)
If you correct for the scale of infrastructure California and Texas are at the top and get to fight over how you tally the details. That list uses a flawed methodology which fails to do so which is why small states and low population states top the list. If you think about it, it would hardly be reasonable to compare RI apples to apples with either when you can rubber stamp RI dozens of times in either. Personally I give the edge to TX in the CA vs TX ranking because Texas is the only state which reports ev
Re: (Score:2)
unlike other states which can avoid reporting a failure because their neighbors bailed them out on the national grid.
Hey! Don't forget about Canadian hydro (well mostly hydro)!
https://www.cer-rec.gc.ca/en/d... [cer-rec.gc.ca]
The real question... (Score:1)
Re: The real question... (Score:2)
Yes, because every company is looking for tools of enshitification and AI promises replacement of customer service with smarter FAQs.
If this means consumer rights are trampled on and more extractive industry is required, that's someone else's problem.
They are not increasing the use of fossil fuel... (Score:3)
The more relevant question to ask would be: "Are AI Datacenters a good use of energy - compared with other purposes?". To which I would respond: "No, they produce way too much bullshit with that huge amount of energy". And independently of that, one may ask: "Is there any chance any country may decide not to unearth their available fossil fuels? Or use them in some non-CO2-emitting way?".
This is fine. (Score:2)
Sounds great. Carry on as usual.
Re: (Score:2)
We *could*... (Score:2)
Water (Score:4, Insightful)
Yeah thatâ(TM)s the problem. (Score:2)
AI is the solution (Score:2)
AI is the solution. The problem of reducing fossil fuel emissions can be put to the wise AI. After due consideration, it was determined that it was best solved by AI suicide.
Require new industry to run off new renewables (Score:2)