Internalizing the negative externalities with Carbon Fee and Dividend with a CBAM (ie. pushing a strong US carbon price around the world) is 50% of achieving our 1.5°C climate goal according global climate policy modeling of En-ROADS by MIT and Climate Interactive:
I was already opposed to overreliance on AI and crypto because I don't trust them. This only makes me feel more secure in not getting involved with them.
Both crypto and AI are at heart very wasteful ways to use computers. Crypto because everyone’s competing to find a hash for the next block, essentially being wasteful by design. With AI it’s more that the data sets that are being parsed are now bigger and bigger, requiring more and more computing resources to make sense of them. In earlier days you might just decide it’s not worthwhile but in an AI arms race it becomes a question of “you’ll lose out if you don’t participate”. Upside down incentives.
It’s very important here to understand that sovereign funds from the gulf states have played a large role in the startup ecosystem for the last decade. Most startup investors don’t invest in what they will be profitable, but what they think the “bigger fish” investors are interested in on the presumption that ultimately someone will turn the company into a profitable business even if lots of people get rich regardless of profitability.
Oil states can use these mechanisms to increase demand for oil in a bunch of ways. For instance ride share and the promise of autonomous vehicles have slowed the buildup of quality transit. Food delivery has created entirely new traffic and fuel demand. And now AI is a new way to increase energy demand.
The key here is that this can happen in the shadows and it doesn’t need to be made super explicit. If companies that pursue growth with weak climate plans are consistently rewarded and clean tech companies consistent struggle it sends a pretty clear message to the entire innovation sector about where to focus.
Well done as usual. More emphasis on the climate impacts of crypto-currency mining (which have no demonstrated social benefit and have been banned in China) would be helpful.
Really glad you wrote about this, since it has been coming up in discussions recently and because even though I don't like to say it, I disagree with the main argument in this article and that AI is "bad" for the climate.
I think trying to "ration" electricity in this way, or pick and choose "good or bad" uses of electricity, is a complete fool's errand considering what is necessary to fight climate change by electrifying everything and simple growing energy demand.
All that does imo is spread a narrative that electricity isn't something to just be abundantly built out, both to help the climate and power new things, but that every solar panel is some precious resource that will keep electricity prices high. That I think directly helps fossil fuels arguments against a clean energy transition because they can position themselves as the only abundant cheap source of power.
Like I'm trying to genuinely understand the other side of this debate, but when I read articles like this,
and also read articles like this about AI being bad for the climate because of electricity demand, I'm having difficulty reconciling the two.
If the problem is electricity demand growing faster than renewables can be built, and that is having the negative impact of using or prolonging fossil fuels, to me the problem to be addressed is how to build clean energy at a larger scale, faster. Not try to fight an uphill battle about specific uses of electricity when we need to electrify everything and massively increase power supply regardless.
And I think when a large part of my climate argument to people is to decouple electricity from fossil fuels, that isn't helped by also saying certain uses of electricity are bad for the climate because they use fossil fuels. Like the article mentions this isn't beef or direct use of fossil fuels for energy, this is electricity.
I don't believe you can make the argument that a clean energy transition is possible because we can have largely the same lifestyle, if not more energy usage, because of cheap clean power. But then also say things like data centers and their usage are something that needs to be near the forefront of concern in fighting climate change. I think people are reasonably going to ask, "ok well then what does that mean for the 286+ million vehicles on the road that we want to electrify, if we are worried about data centers?". I don't think you can put forward both arguments at the same time.
To be clear, I'm not against energy efficiency. In fact that is part of my argument because of Jevons Paradox. Like say ChatGPT 5 becomes more efficient as a result of increased transparency about energy usage. Who is to say that doesn't just lead to a bigger model because of those efficiency gains?
I think it is also worth considering how much of this comes from groups or people, like Friends of the Earth, that are against nuclear power. If one sees an increasing demand for electricity, but is against nuclear power and the GWs of clean power it brings, what else is there to do but be critical of data centers or whatever else? But the power demand is the power demand, which is why I think it is a fools errand to try and "ration" electricity in this way instead of just building more power.
So sorry if this comes off as too critical but I get frustrated when I see a real opportunity for true clean power abundance and all the benefits that brings, and the opposite view says that basically isn't the goal. Imagine if we had GWs of clean power available and what that would mean for power hungry processes like desalination or direct carbon capture. I think working towards that future is better not just for the climate but for a broader society that would revolve around essentially free, abundant power. And this view I think works against that, which is why I disagree with it so strongly.
Like I said I'm glad this article was written because I think this debate is important, so thank you Arielle!
This is a very thought-provoking argument! Does not come off as too critical. It's definitely true that if we had true clean power abundance this wouldn't be much of a problem for the climate. That's why the first suggestion for what to do is "rapid transition to renewable energy." However, I do think there is still a resource issue re: the concept of "unlimited renewable and nuclear energy." Even in a 100 percent net zero energy economy, I think there will be resource issues if we aren't smart about what the energy is used *for*. The idea that we can continue to consume energy limitlessly so long as it's net zero... I am still pretty skeptical of that. But willing and curious to have my mind changed.
My comment was long and ranty but what I'm trying to say is that I think we as a climate movement should be able to say is "yes we will be getting rid of fossil fuels, but in return you will have abundant clean electricity". And I think arguments that counteract that won't help the movement and might work against it. I agree resources should always be considered in the context of a net zero economy, cryptocurrency is an excellent example just a dumb and wasteful thing I forgot to mention.
I think in setting the main argument for a clean energy transition to people from the outset, it should be centered around the idea telling them that they really don't have to worry about resource management of electricity, because we want people to embrace electrifying everything. And I think that by saying AI is "bad" for the environment, data center electricity usage growth, with using hours watched of tv as an example, could just lead people to turn against the energy transition "because it all uses fossil fuels anyways so what's the point".
Maybe I'm just looking at it in sort of a way this comes across to the public. I enjoy internal debates about this, my worry is how this reads in the mainstream and how that affects the public's perception of climate action.
And I know it isn't yours or Arielle's intention but in some way it does feel like telling me I shouldn't even try. When I read articles like the one I linked about India's AC usage growth, I look at that as a challenge to myself, thinking "I believe I can meet that incredible electricity demand with clean electricity and still have extra". And if that happens, then data centers in the US would be a rounding error right? I could be wrong but I want to at least try first. And I think critical focus on uses of electricity right now, work against that.
Sorry for another long and ranty comment. I enjoy HEATED for articles like this and the discussion it brings!
And to reiterate crypto is just dumb and stupid and should be banned if not heavily regulated to the point of practical non existence for lots of reasons. So any energy savings are welcome I agree.
Wow...I did not know a lot of what is in this article other than the stuff AI and crypto currency. This was a great piece. Thanks for this report and keep up the great work! The planet needs you!
But yeah, stay away from bitcoin always and "generative AI" as much as you can!
With summer coming, in much of the country, one great thing individuals can do to clean the grid and encourage a transition to a cleaner grid is to use as little electricity as possible during the peak evening hours. If you have AC, crank it in the sunny PV-saturated afternoon, and let your abode warm up during dinner. Don't charge the car then; turn off air cleaners and other optional items and get a timer for any tank electric DHW you might have. Don't freak about using the e-range, but if you have the incredibly efficient Instapot, set it to do its thing mid-day.
Externalities! This is the first essay I have seen about the true costs of the internet.
Crypto and AI should be heavily taxed for their energy use until it can be shown that all of it is supplied by solar & wind.
Internalizing the negative externalities with Carbon Fee and Dividend with a CBAM (ie. pushing a strong US carbon price around the world) is 50% of achieving our 1.5°C climate goal according global climate policy modeling of En-ROADS by MIT and Climate Interactive:
https://docs.google.com/presentation/d/1QBWniFFCgtFMRabyRhPD-YqYV4IeU0FaCA2U578aJTE/edit?usp=drivesdk
I was already opposed to overreliance on AI and crypto because I don't trust them. This only makes me feel more secure in not getting involved with them.
Same.
Both crypto and AI are at heart very wasteful ways to use computers. Crypto because everyone’s competing to find a hash for the next block, essentially being wasteful by design. With AI it’s more that the data sets that are being parsed are now bigger and bigger, requiring more and more computing resources to make sense of them. In earlier days you might just decide it’s not worthwhile but in an AI arms race it becomes a question of “you’ll lose out if you don’t participate”. Upside down incentives.
It’s very important here to understand that sovereign funds from the gulf states have played a large role in the startup ecosystem for the last decade. Most startup investors don’t invest in what they will be profitable, but what they think the “bigger fish” investors are interested in on the presumption that ultimately someone will turn the company into a profitable business even if lots of people get rich regardless of profitability.
Oil states can use these mechanisms to increase demand for oil in a bunch of ways. For instance ride share and the promise of autonomous vehicles have slowed the buildup of quality transit. Food delivery has created entirely new traffic and fuel demand. And now AI is a new way to increase energy demand.
The key here is that this can happen in the shadows and it doesn’t need to be made super explicit. If companies that pursue growth with weak climate plans are consistently rewarded and clean tech companies consistent struggle it sends a pretty clear message to the entire innovation sector about where to focus.
Super smart take
Well done as usual. More emphasis on the climate impacts of crypto-currency mining (which have no demonstrated social benefit and have been banned in China) would be helpful.
Wow, really insightful. Thank you for sharing this! I didn't know how bad it was.
Really glad you wrote about this, since it has been coming up in discussions recently and because even though I don't like to say it, I disagree with the main argument in this article and that AI is "bad" for the climate.
I think trying to "ration" electricity in this way, or pick and choose "good or bad" uses of electricity, is a complete fool's errand considering what is necessary to fight climate change by electrifying everything and simple growing energy demand.
All that does imo is spread a narrative that electricity isn't something to just be abundantly built out, both to help the climate and power new things, but that every solar panel is some precious resource that will keep electricity prices high. That I think directly helps fossil fuels arguments against a clean energy transition because they can position themselves as the only abundant cheap source of power.
Like I'm trying to genuinely understand the other side of this debate, but when I read articles like this,
https://economictimes.indiatimes.com/industry/cons-products/electronics/indias-demand-for-electricity-for-acs-to-exceed-total-power-consumption-of-africa-iea/articleshow/104669858.cms?from=mdr,
and also read articles like this about AI being bad for the climate because of electricity demand, I'm having difficulty reconciling the two.
If the problem is electricity demand growing faster than renewables can be built, and that is having the negative impact of using or prolonging fossil fuels, to me the problem to be addressed is how to build clean energy at a larger scale, faster. Not try to fight an uphill battle about specific uses of electricity when we need to electrify everything and massively increase power supply regardless.
And I think when a large part of my climate argument to people is to decouple electricity from fossil fuels, that isn't helped by also saying certain uses of electricity are bad for the climate because they use fossil fuels. Like the article mentions this isn't beef or direct use of fossil fuels for energy, this is electricity.
I don't believe you can make the argument that a clean energy transition is possible because we can have largely the same lifestyle, if not more energy usage, because of cheap clean power. But then also say things like data centers and their usage are something that needs to be near the forefront of concern in fighting climate change. I think people are reasonably going to ask, "ok well then what does that mean for the 286+ million vehicles on the road that we want to electrify, if we are worried about data centers?". I don't think you can put forward both arguments at the same time.
To be clear, I'm not against energy efficiency. In fact that is part of my argument because of Jevons Paradox. Like say ChatGPT 5 becomes more efficient as a result of increased transparency about energy usage. Who is to say that doesn't just lead to a bigger model because of those efficiency gains?
https://en.wikipedia.org/wiki/Jevons_paradox
I think it is also worth considering how much of this comes from groups or people, like Friends of the Earth, that are against nuclear power. If one sees an increasing demand for electricity, but is against nuclear power and the GWs of clean power it brings, what else is there to do but be critical of data centers or whatever else? But the power demand is the power demand, which is why I think it is a fools errand to try and "ration" electricity in this way instead of just building more power.
So sorry if this comes off as too critical but I get frustrated when I see a real opportunity for true clean power abundance and all the benefits that brings, and the opposite view says that basically isn't the goal. Imagine if we had GWs of clean power available and what that would mean for power hungry processes like desalination or direct carbon capture. I think working towards that future is better not just for the climate but for a broader society that would revolve around essentially free, abundant power. And this view I think works against that, which is why I disagree with it so strongly.
Like I said I'm glad this article was written because I think this debate is important, so thank you Arielle!
This is a very thought-provoking argument! Does not come off as too critical. It's definitely true that if we had true clean power abundance this wouldn't be much of a problem for the climate. That's why the first suggestion for what to do is "rapid transition to renewable energy." However, I do think there is still a resource issue re: the concept of "unlimited renewable and nuclear energy." Even in a 100 percent net zero energy economy, I think there will be resource issues if we aren't smart about what the energy is used *for*. The idea that we can continue to consume energy limitlessly so long as it's net zero... I am still pretty skeptical of that. But willing and curious to have my mind changed.
My comment was long and ranty but what I'm trying to say is that I think we as a climate movement should be able to say is "yes we will be getting rid of fossil fuels, but in return you will have abundant clean electricity". And I think arguments that counteract that won't help the movement and might work against it. I agree resources should always be considered in the context of a net zero economy, cryptocurrency is an excellent example just a dumb and wasteful thing I forgot to mention.
I think in setting the main argument for a clean energy transition to people from the outset, it should be centered around the idea telling them that they really don't have to worry about resource management of electricity, because we want people to embrace electrifying everything. And I think that by saying AI is "bad" for the environment, data center electricity usage growth, with using hours watched of tv as an example, could just lead people to turn against the energy transition "because it all uses fossil fuels anyways so what's the point".
Maybe I'm just looking at it in sort of a way this comes across to the public. I enjoy internal debates about this, my worry is how this reads in the mainstream and how that affects the public's perception of climate action.
And I know it isn't yours or Arielle's intention but in some way it does feel like telling me I shouldn't even try. When I read articles like the one I linked about India's AC usage growth, I look at that as a challenge to myself, thinking "I believe I can meet that incredible electricity demand with clean electricity and still have extra". And if that happens, then data centers in the US would be a rounding error right? I could be wrong but I want to at least try first. And I think critical focus on uses of electricity right now, work against that.
Sorry for another long and ranty comment. I enjoy HEATED for articles like this and the discussion it brings!
And to reiterate crypto is just dumb and stupid and should be banned if not heavily regulated to the point of practical non existence for lots of reasons. So any energy savings are welcome I agree.
Wow...I did not know a lot of what is in this article other than the stuff AI and crypto currency. This was a great piece. Thanks for this report and keep up the great work! The planet needs you!
TIL teens are already calling AI art "Boomer Art" and referring to it as cringe. So much for that "trend" 😅
I looked at a few of the references and some of the predictions of energy use seemed outside (above) the norm. I'm a fan of Jonathan Koomey, who has been debunking the 'internet/phone energy will eat the grid/planet" hype for 20 years. A recent article featuring him: https://www.latitudemedia.com/news/data-center-experts-on-energy-use-for-ai-calm-the-heck-down
But yeah, stay away from bitcoin always and "generative AI" as much as you can!
With summer coming, in much of the country, one great thing individuals can do to clean the grid and encourage a transition to a cleaner grid is to use as little electricity as possible during the peak evening hours. If you have AC, crank it in the sunny PV-saturated afternoon, and let your abode warm up during dinner. Don't charge the car then; turn off air cleaners and other optional items and get a timer for any tank electric DHW you might have. Don't freak about using the e-range, but if you have the incredibly efficient Instapot, set it to do its thing mid-day.