The single most under-discussed fault lines in left of center spaces is between taxes+transfers vs regulation as the primary government tool for addressing problems.
I like the idea of taxing data centers if they are producing a negative externality. But those on the other side of the debate just want them banned or want their buildout slowed down via red tape. Both approaches largely do the same thing, but only one raises revenue.
As a first approximation, markets increase freedom. As a second approximation, reducing poverty increases freedom. As a third approximation, greater equality increases freedom by reducing the coercive powers of elites. Statecraft is optimizing the coefficients of these three vectors, and deciding how many leaky redistributive buckets to build.
For me, it's one thing I hate (Pigouvian taxes), and another I'm ambivalent about (data centers). Just throw the kitchen sink at building as much cleaner energy as possible.
But I think the ultimate problem that data center perception faces is that many, perhaps most, people don't agree with Matt that "on its face, artificial intelligence seems to clearly be an economically valuable technology". Their reasons seem to be all over the place beyond using electricity--water consumption, IP infringement, job loss, particularly to the creative class, output deemed to be bad, investors deemed to be bad, existential risk to the human race. Any of these may not be well founded, but the fact that they are present suggests that not enough people do, in fact, find that this technology is clearly economically valuable. As for me, I find the value to be plausible, but not clearly demonstrated, hence my ambivalence. Perhaps that means that the companies building the data centers should work harder on their PR to demonstrate to the public that they are, indeed, creating economic value.
I strongly believe that the tax code should not be used to pick on people or things people don't like, as it breeds a chain reaction of such picking upon that feeds into polarizing division and zero sum mentality. Instead, I strongly believe that if we're going to have things collectively provisioned, we should just accept that we're in all in this together as a society, and raise our tax rate to pay for those collective provisions commensurate with our proportion of production.
This seems like a reasonable argument against taxing things that some people like but lots of people think are morally bad, like alcohol or gambling. The kinds of taxes discussed here on things like pollution or traffic or carbon emissions are in somewhat different situations where everyone agrees that those negative impacts are bad.
I don't agree that everyone agrees that the impacts of GHG emissions are bad. Or at the very least, that the "social costs" calculations that economists make are much higher than everyday people would assign. Most people are thinking of the here and now and how useful burning fossil fuels and other activity that emits GHGs are. That can have bad consequences, too, but it exists.
My point is that the carbon emissions are bad, which can be balanced against the valuable things that they accomplish. Just like traffic jams are bad, even though that's balanced against all the individual people getting where they want to go. There are indeed some people who deny that carbon emissions are bad at all, but they're just wrong.
Do alcohol taxes lead to polarization? I don't see it. Or tobacco? Again, I don't see it. It's like anything if you think there's a negative externality on an activity you make an empirically that there are bad side effects and people agree or not.
But if you look at alcohol prohibition or weed legalization it's the pigouvian tax that made compromise easier, since you can always compromise over dollar amounts.
Maybe it's too easy for people to propose new pigouvian taxes on straws or sugar, and whatever but most of those don't go anywhere.
Also. taxes in proportion to your consumption is better than production, imo.
I've spent my entire career (1994-now) working for Big Computer, but share your skepticism about the long term value of AI. I have no doubt AI is a valuable technology, but it's very unclear how valuable it is, and I'm 99% confident there are serious bubbles in the market. The rise of the internet also created big bubbles too and some early investors lost quite a bit of money, but in the long-run the value is clearly positive. Time will tell the net value of AI.
Moreover, I don't understand Matt's read of Bernie's proposal being motivated by environmental concerns. It seems just as plausible to arrive at Bernie's policy proposal through a straightforward "your LLMs were created by consuming public knowledge and now you're trying to privatize the value of the product. No sale." The Open Source community/movement always had a "standing on the shoulders of giants" mentality. That is completely missing from AI leaders now.
Dario Amodei has done a great job of making me think he is a wise natural aristocrat who can be trusted with outsized leverage over the fate of humanity. His writings acknowledge tradeoffs and avoid boosterism. He seems to be basically content with the extraordinary wealth he already enjoys and focused on the value of his technology to our species and sentient life generally. I’m really impressed and the captains of industry rarely inspire me.
However, I am far enough from the median voter than turning me into a fan boy is not necessarily evidence of PR savvy.
I don’t understand the focus on the government “owning stock”. Aren’t policy makers aware that Congress can legally expropriate whatever percentage of corporate profits they choose? It’s called an “income tax”.
It’s like these discussions are taking place pre-16th amendment!
Ownership rather than taxing has a bunch of benefits. For example, then you don't have to worry about the structure of companies and their revenue relative to taxes, or whether they want to do dividends versus stock buybacks or other things like that. There's a reason that many countries have moved to having sovereign wealth funds.
Income taxes don’t capture the value of appreciation which is currently more than 100% of the value these companies are capturing since they’re actually operating at a loss.
Because you feel cool and awesome like you're part of an angel investor community. It's why the government wanted to act like VCs in the 00's and we got Solyndra. (And dumping piles of money on it with the same result as an amateur farmer dumping piles of fertilizer on their crops.) You get to pretend you're one of the creators making it happen instead of the party coming along later to tax the creators.
Data centers are remarkably efficient. Even with the energy use increase they account for ~0.5-1% of GHGs (IEA #s). Agriculture and livestock account for 12% and are almost universally excluded from carbon taxes and emissions trading schemes like cap&trade sorts of programs. Cement and concrete production account for 7%, and the US similarly does not have carbon taxes on the industry. Steel & iron production have the same protections at 8 or 9% of CO2 emissions. Airlines are 2-3% of global emissions writ large and are allowed "carbon offset" programs rather than forced into a direct carbon tax.
I could go on and on. The exemptions are always for industries that are crucial for international industrial competitiveness. AI is even more lucrative and produces even less emissions. The answer therefore is extremely obvious. No.
There are quite literally thousands of data centers. A single Microsoft data center produces about as much emissions as a neighborhood restaurant. They are, again, quite literally some of the most efficient businesses on the planet.
Additionally the problem that the public is reacting to principally - the electricity price increases is not something that is really purely data centers
Electrification of industrial processes is driving significant demand
The ACTUAL PROBLEM is the GRID - and the ability to add more Gigawatts faster
Upgrade the grid and liberalise for installs so both of RnEn plus storage can be added fast and the problem goes away.
Yes. The book The Grid by Bakke is a great deep dive into this. It's why tech companies building data centers are also investing hundreds of billions into grid reforms in & around their construction sites. It's actually an extraordinarily pro social energy development. If the Dept of Energy were doing it versus Google and Amazon etc the same people decrying it would love it.
and as my literal real job is financing this shit, I have been climbing the wall for years as the idiot Green Left and Biden Admin neglect / ignore.
Upgrade all of Transmission Grids and Distribution Grids and you solve 80% of your problems.
It's legit to look at surcharges to big users / off-takers to upgrade and accelerate grid modernisation - completely idiotic to talk "carbon taxes" -it's degrowth logic that I found shocking coming from MY
Well, there's the perception problems that data centers are facing right now, even if unfounded. It's obvious to people that beef is tasty and airplanes get them to far away places fast, but not obvious to them what data centers do to help them, so it'd be helpful if they worked on closing this perception gap.
If the question is “should we punish data centers because people are uninformed and mad” the answer is no. If the question is “but does it make sense based on their emissions contribution” the answer is also no.
If you want to tax them, tax them. This idea that punishing data centers with random other taxes that make no sense will lead people to be satisfied is ridiculous. People that feel this way won't be satisfied until they're entirely banned outright. Rather than give ground to these degrowth ideas we should just be earnest and honest about the value of economic growth and the pro social regular profit taxes that growth can help stimulate.
What pitches would you give to convince people that data centers are unequivocally creating economic growth to get them to resist the siren song of degrowth?
Nothing. That's impossible. I would just sell them on the possibility of revenue and reassure them that the data centers will be in the middle of nowhere, which most of them are.
We should AI companies profits and use that money for various pro social causes. But carbon taxes? Makes no sense. There's normal ways to get at what you're saying
It doesn't make good sense to me either, and I think most people would be bewildered at Matt's take here. But my point is if companies deemed to be in the AI industry don't want to be picked on in whatever manner by the government, they need to do better with their PR in selling the benefits of their product to the general public.
Matt buries the lede. He spends most of this piece on carbon — which he correctly identifies as real but manageable. The issue that will actually determine AI's political fate is unemployment. The carbon debate is an intra-elite skirmish between tech optimists and climate institutionalists. The unemployment debate will involve everyone.
AI will only cause unemployment to the extent it raises productivity and generates wealth , so we are tautologically dividing a bigger pie. The bad news is that dividing a bigger pie requires political mechanisms that don't currently exist. AI skepticism can unite the existing precariat with professional class families watching office jobs — paralegal, junior analyst, mid-level coder — evaporate. That coalition, once it organizes, will make anti-data-center NIMBYs look like a rounding error.
The solution is the pre-commit to taxing AI based upon excess unemployment (roughly the spread between actual unemployment and 4%) and industry revenues. AI should pay higher taxes if unemployment is 12% and it is bringing in $4 trillion a year than if unemployment is 4% and it is bringing in $1 trillion. Pre-commitment could take the wind out of AI luddism.
While the policy presented here is sound, it’s unfortunate that it says nothing about Gaza or Israel, and therefore will not gain any attention or advocacy from environmental or climate organizations.
I genuinely struggle with what to do about the polar bears and reefs and the huge human impositions on animals that really seem like a much bigger deal to me than to the average person.
The cost inflicted on animals for humans seem really awful.
If we're discussing a carbon tax then that should also encompass the supercomputer mining centers that supply the energy for crypto currencies, which I would argue have hit a wall in terms of financial value, utility, and public opinion. A.I. data centers may prove to be a net social good but you can't say the same for crypto. Some data mining centers have incorporated the use of renewables into their operations which is good but on the whole I would lump these facilities in with A.I. data centers.
I think this is close very close to a good idea, but not quite right. The government should select locations for data centers in a competition. The winners should be the center developers that will pay for right to build the center. The government should use the money to pay for free interconnection of all renewable power projects to the grid. It’s an auction where the money is recycled back to grid upgrade that lowers prices to consumers.
I always thought Bernie Sanders had a lot of interesting ideas but wouldn't make a good president. I'm not sure the way he suggests going about it is the best way but putting AI into a quasi-public/private research arena might be a good idea, similar to the Bell (et al) research entities post WWII. So much data center usage goes to dumbing down the public on social media, maybe we could temper that proclivity and put more emphasis on important uses like medical innovations.
Surprised Matt didn't mention noise, the newer processing data centers are incredibly noisy apparently. The ones that actually get stopped it's generally because they are too close to residential. And 2nd the tax exemptions stop them as people question the amount given for something that generates a bunch of noise, few permanent jobs and possibly increases your electricity bill.
My worry with data centers is that they’ll encourage AI to develop in a bad way where everything will be centralized in the cloud with enough memory to store and analyze everything you’ve done forever and also be less available to poor nations which can’t afford to build out these data centers. A better future is one where AI is developed for efficiency so that it can be run locally, which significantly curtails the risk of surveillance or centralized control and also allows the technology to diffuse to poorer countries that cannot afford massive amounts of hardware. Constraints on available data centers force AI developers to develop in this direction (like they are in China) and that is good.
One of my somewhat crank habits is that I am very averse to putting fieles on the cloud without a very good reason, and I never do it automatically. A recent example is that I tried to use Microsoft's built in video clip editor, but said he'll no quickly when they recently changed it to force you to save your project files on their cloud service that they keep relentlessly pushing on you. I promptly found an open source program after that.
Countries control who can use them though, like we saw with the Anthropic export controls.
Fixing the fundamentals would require a Time Machine to 1491. Literally the only former poor country that has caught up to the US on a per capita basis without a resource bonanza is tiny Singapore and it did it by being so specialized that it’d probably be in pretty big trouble if globalization seriously broke down. It’s hard to blame individual countries for not catching up when effectively none of them have.
They are not the same thing in any way. My data center project in a certain non-USA-non-EU LMC geography is very happy to have ReEn gigawatts and Chinese or another model renting platforming. US cutting off its own nose to spite its face only matters if they rent compute time from me.
And I have zero interest in your bad-retread of 1970s Tiers Mondisme, I have real investment to do.
"The problem is that if you start with a temperature target of 1.5 or 2 degrees Celsius, and then derive from that a global emissions target, and then divide that into national and local emissions targets ..."
OMG! This is an incredibly stupid way to approch reducing the harm of CO2 emisions! One uses the temperature target to derive the global carbon tax that woud produce it and then either have a carbon tax or mimic it with subsidies paid for with the lowest DWL tax possible and or using that tax value in the cost benefit analysis of public investment. Local, even national emission targets (why not zip code targets?) make no sense at all. That is COP thinking and McKibbenism that has got us to where we are!
"Consolidated Edison is the largest property taxpayer in both New York City and New York State. The utility recovers this expense directly from its users by embedding the tax costs into the "Delivery Charge" on monthly energy bills, a pass-through mechanism authorized by state regulators. Government taxes and fees account for roughly 30% of a standard Con Edison bill. The property and special franchise taxes are the primary drivers of this, comprising roughly 17% to 20% of the total amount a customer pays each month."
My belief in rational taxation on energy and climate issues is at a low ebb. Why this charade of taxing the utility in order to tax the consumer?
If a utility is using 1000 units of a public good, and a private actor using 1000 units of a public good would get taxed $X, then the utility should also pay $X. It is more transparent. People looking at setting up some other kind of utility that is perhaps more or less land-efficient can compare dollars directly.
I wanted to say MY is wrong on Polar Bears but I am not sure, it is one of the issues with a huge partisan divide. Conservative sources are saying Polar Bear populations are exploding, leftwing ones say it is complicated some are getting fatter and healthier, some subpopulations are doing well but other might not be.
I think Polar Bears might be fine with global warming but birds they eat as an alternative to seals in warm weather might be really upset by data centres.
The single most under-discussed fault lines in left of center spaces is between taxes+transfers vs regulation as the primary government tool for addressing problems.
I like the idea of taxing data centers if they are producing a negative externality. But those on the other side of the debate just want them banned or want their buildout slowed down via red tape. Both approaches largely do the same thing, but only one raises revenue.
As a first approximation, markets increase freedom. As a second approximation, reducing poverty increases freedom. As a third approximation, greater equality increases freedom by reducing the coercive powers of elites. Statecraft is optimizing the coefficients of these three vectors, and deciding how many leaky redistributive buckets to build.
For me, it's one thing I hate (Pigouvian taxes), and another I'm ambivalent about (data centers). Just throw the kitchen sink at building as much cleaner energy as possible.
But I think the ultimate problem that data center perception faces is that many, perhaps most, people don't agree with Matt that "on its face, artificial intelligence seems to clearly be an economically valuable technology". Their reasons seem to be all over the place beyond using electricity--water consumption, IP infringement, job loss, particularly to the creative class, output deemed to be bad, investors deemed to be bad, existential risk to the human race. Any of these may not be well founded, but the fact that they are present suggests that not enough people do, in fact, find that this technology is clearly economically valuable. As for me, I find the value to be plausible, but not clearly demonstrated, hence my ambivalence. Perhaps that means that the companies building the data centers should work harder on their PR to demonstrate to the public that they are, indeed, creating economic value.
What's with the Pigouvian Tax hate? Is it personal, like you're a big smoker, drinker and gambler, or more philosophical?
I strongly believe that the tax code should not be used to pick on people or things people don't like, as it breeds a chain reaction of such picking upon that feeds into polarizing division and zero sum mentality. Instead, I strongly believe that if we're going to have things collectively provisioned, we should just accept that we're in all in this together as a society, and raise our tax rate to pay for those collective provisions commensurate with our proportion of production.
This seems like a reasonable argument against taxing things that some people like but lots of people think are morally bad, like alcohol or gambling. The kinds of taxes discussed here on things like pollution or traffic or carbon emissions are in somewhat different situations where everyone agrees that those negative impacts are bad.
I don't agree that everyone agrees that the impacts of GHG emissions are bad. Or at the very least, that the "social costs" calculations that economists make are much higher than everyday people would assign. Most people are thinking of the here and now and how useful burning fossil fuels and other activity that emits GHGs are. That can have bad consequences, too, but it exists.
My point is that the carbon emissions are bad, which can be balanced against the valuable things that they accomplish. Just like traffic jams are bad, even though that's balanced against all the individual people getting where they want to go. There are indeed some people who deny that carbon emissions are bad at all, but they're just wrong.
And my point is that plenty of people think that balance is net positive, even if we disagree.
Do alcohol taxes lead to polarization? I don't see it. Or tobacco? Again, I don't see it. It's like anything if you think there's a negative externality on an activity you make an empirically that there are bad side effects and people agree or not.
But if you look at alcohol prohibition or weed legalization it's the pigouvian tax that made compromise easier, since you can always compromise over dollar amounts.
Maybe it's too easy for people to propose new pigouvian taxes on straws or sugar, and whatever but most of those don't go anywhere.
Also. taxes in proportion to your consumption is better than production, imo.
I've spent my entire career (1994-now) working for Big Computer, but share your skepticism about the long term value of AI. I have no doubt AI is a valuable technology, but it's very unclear how valuable it is, and I'm 99% confident there are serious bubbles in the market. The rise of the internet also created big bubbles too and some early investors lost quite a bit of money, but in the long-run the value is clearly positive. Time will tell the net value of AI.
Moreover, I don't understand Matt's read of Bernie's proposal being motivated by environmental concerns. It seems just as plausible to arrive at Bernie's policy proposal through a straightforward "your LLMs were created by consuming public knowledge and now you're trying to privatize the value of the product. No sale." The Open Source community/movement always had a "standing on the shoulders of giants" mentality. That is completely missing from AI leaders now.
Dario Amodei has done a great job of making me think he is a wise natural aristocrat who can be trusted with outsized leverage over the fate of humanity. His writings acknowledge tradeoffs and avoid boosterism. He seems to be basically content with the extraordinary wealth he already enjoys and focused on the value of his technology to our species and sentient life generally. I’m really impressed and the captains of industry rarely inspire me.
However, I am far enough from the median voter than turning me into a fan boy is not necessarily evidence of PR savvy.
I don’t understand the focus on the government “owning stock”. Aren’t policy makers aware that Congress can legally expropriate whatever percentage of corporate profits they choose? It’s called an “income tax”.
It’s like these discussions are taking place pre-16th amendment!
Ownership rather than taxing has a bunch of benefits. For example, then you don't have to worry about the structure of companies and their revenue relative to taxes, or whether they want to do dividends versus stock buybacks or other things like that. There's a reason that many countries have moved to having sovereign wealth funds.
Income taxes don’t capture the value of appreciation which is currently more than 100% of the value these companies are capturing since they’re actually operating at a loss.
Because you feel cool and awesome like you're part of an angel investor community. It's why the government wanted to act like VCs in the 00's and we got Solyndra. (And dumping piles of money on it with the same result as an amateur farmer dumping piles of fertilizer on their crops.) You get to pretend you're one of the creators making it happen instead of the party coming along later to tax the creators.
Data centers are remarkably efficient. Even with the energy use increase they account for ~0.5-1% of GHGs (IEA #s). Agriculture and livestock account for 12% and are almost universally excluded from carbon taxes and emissions trading schemes like cap&trade sorts of programs. Cement and concrete production account for 7%, and the US similarly does not have carbon taxes on the industry. Steel & iron production have the same protections at 8 or 9% of CO2 emissions. Airlines are 2-3% of global emissions writ large and are allowed "carbon offset" programs rather than forced into a direct carbon tax.
I could go on and on. The exemptions are always for industries that are crucial for international industrial competitiveness. AI is even more lucrative and produces even less emissions. The answer therefore is extremely obvious. No.
"Industry X emits 1% of GHG, Industry Y emits 2% of GHG, therefore Industry X is more efficient" is skipping about 3 really important steps.
There are quite literally thousands of data centers. A single Microsoft data center produces about as much emissions as a neighborhood restaurant. They are, again, quite literally some of the most efficient businesses on the planet.
Additionally the problem that the public is reacting to principally - the electricity price increases is not something that is really purely data centers
Electrification of industrial processes is driving significant demand
The ACTUAL PROBLEM is the GRID - and the ability to add more Gigawatts faster
Upgrade the grid and liberalise for installs so both of RnEn plus storage can be added fast and the problem goes away.
Yes. The book The Grid by Bakke is a great deep dive into this. It's why tech companies building data centers are also investing hundreds of billions into grid reforms in & around their construction sites. It's actually an extraordinarily pro social energy development. If the Dept of Energy were doing it versus Google and Amazon etc the same people decrying it would love it.
100%.
and as my literal real job is financing this shit, I have been climbing the wall for years as the idiot Green Left and Biden Admin neglect / ignore.
Upgrade all of Transmission Grids and Distribution Grids and you solve 80% of your problems.
It's legit to look at surcharges to big users / off-takers to upgrade and accelerate grid modernisation - completely idiotic to talk "carbon taxes" -it's degrowth logic that I found shocking coming from MY
More supply - and more supply needs the grid!!
Well, there's the perception problems that data centers are facing right now, even if unfounded. It's obvious to people that beef is tasty and airplanes get them to far away places fast, but not obvious to them what data centers do to help them, so it'd be helpful if they worked on closing this perception gap.
If the question is “should we punish data centers because people are uninformed and mad” the answer is no. If the question is “but does it make sense based on their emissions contribution” the answer is also no.
If the question is "should we punish data centers because people are uninformed and mad," the answer is no.
Sorry dude, that's politics. The alternative is likely that they just banned everywhere. I don't like it either but it is what it is.
If you want to tax them, tax them. This idea that punishing data centers with random other taxes that make no sense will lead people to be satisfied is ridiculous. People that feel this way won't be satisfied until they're entirely banned outright. Rather than give ground to these degrowth ideas we should just be earnest and honest about the value of economic growth and the pro social regular profit taxes that growth can help stimulate.
Surcharge for grid upgrading - this at least responds to the populist slop but at same time to data center operators can be quite actually rational
What pitches would you give to convince people that data centers are unequivocally creating economic growth to get them to resist the siren song of degrowth?
Nothing. That's impossible. I would just sell them on the possibility of revenue and reassure them that the data centers will be in the middle of nowhere, which most of them are.
But uninformed and mad people can make up the voters, and you have to deal with the voters you have, not the ones that always think clearly.
We should AI companies profits and use that money for various pro social causes. But carbon taxes? Makes no sense. There's normal ways to get at what you're saying
It doesn't make good sense to me either, and I think most people would be bewildered at Matt's take here. But my point is if companies deemed to be in the AI industry don't want to be picked on in whatever manner by the government, they need to do better with their PR in selling the benefits of their product to the general public.
So much more efficient than human brains once you account for heating, cooling, food and transport that humans need.
Matt buries the lede. He spends most of this piece on carbon — which he correctly identifies as real but manageable. The issue that will actually determine AI's political fate is unemployment. The carbon debate is an intra-elite skirmish between tech optimists and climate institutionalists. The unemployment debate will involve everyone.
AI will only cause unemployment to the extent it raises productivity and generates wealth , so we are tautologically dividing a bigger pie. The bad news is that dividing a bigger pie requires political mechanisms that don't currently exist. AI skepticism can unite the existing precariat with professional class families watching office jobs — paralegal, junior analyst, mid-level coder — evaporate. That coalition, once it organizes, will make anti-data-center NIMBYs look like a rounding error.
The solution is the pre-commit to taxing AI based upon excess unemployment (roughly the spread between actual unemployment and 4%) and industry revenues. AI should pay higher taxes if unemployment is 12% and it is bringing in $4 trillion a year than if unemployment is 4% and it is bringing in $1 trillion. Pre-commitment could take the wind out of AI luddism.
While the policy presented here is sound, it’s unfortunate that it says nothing about Gaza or Israel, and therefore will not gain any attention or advocacy from environmental or climate organizations.
I genuinely struggle with what to do about the polar bears and reefs and the huge human impositions on animals that really seem like a much bigger deal to me than to the average person.
The cost inflicted on animals for humans seem really awful.
If we're discussing a carbon tax then that should also encompass the supercomputer mining centers that supply the energy for crypto currencies, which I would argue have hit a wall in terms of financial value, utility, and public opinion. A.I. data centers may prove to be a net social good but you can't say the same for crypto. Some data mining centers have incorporated the use of renewables into their operations which is good but on the whole I would lump these facilities in with A.I. data centers.
I think this is close very close to a good idea, but not quite right. The government should select locations for data centers in a competition. The winners should be the center developers that will pay for right to build the center. The government should use the money to pay for free interconnection of all renewable power projects to the grid. It’s an auction where the money is recycled back to grid upgrade that lowers prices to consumers.
I always thought Bernie Sanders had a lot of interesting ideas but wouldn't make a good president. I'm not sure the way he suggests going about it is the best way but putting AI into a quasi-public/private research arena might be a good idea, similar to the Bell (et al) research entities post WWII. So much data center usage goes to dumbing down the public on social media, maybe we could temper that proclivity and put more emphasis on important uses like medical innovations.
Surprised Matt didn't mention noise, the newer processing data centers are incredibly noisy apparently. The ones that actually get stopped it's generally because they are too close to residential. And 2nd the tax exemptions stop them as people question the amount given for something that generates a bunch of noise, few permanent jobs and possibly increases your electricity bill.
My worry with data centers is that they’ll encourage AI to develop in a bad way where everything will be centralized in the cloud with enough memory to store and analyze everything you’ve done forever and also be less available to poor nations which can’t afford to build out these data centers. A better future is one where AI is developed for efficiency so that it can be run locally, which significantly curtails the risk of surveillance or centralized control and also allows the technology to diffuse to poorer countries that cannot afford massive amounts of hardware. Constraints on available data centers force AI developers to develop in this direction (like they are in China) and that is good.
One of my somewhat crank habits is that I am very averse to putting fieles on the cloud without a very good reason, and I never do it automatically. A recent example is that I tried to use Microsoft's built in video clip editor, but said he'll no quickly when they recently changed it to force you to save your project files on their cloud service that they keep relentlessly pushing on you. I promptly found an open source program after that.
Countries don't build fucking data centers.
People like me build the damn things.
And your "poor nations" need to solve their fundamentals.
I am perfectly happy and actually doing gigawatt project reviews in "poor'" places with a decent environment to do so.
Countries control who can use them though, like we saw with the Anthropic export controls.
Fixing the fundamentals would require a Time Machine to 1491. Literally the only former poor country that has caught up to the US on a per capita basis without a resource bonanza is tiny Singapore and it did it by being so specialized that it’d probably be in pretty big trouble if globalization seriously broke down. It’s hard to blame individual countries for not catching up when effectively none of them have.
Anthropic is a Model
Data center is an asset.
They are not the same thing in any way. My data center project in a certain non-USA-non-EU LMC geography is very happy to have ReEn gigawatts and Chinese or another model renting platforming. US cutting off its own nose to spite its face only matters if they rent compute time from me.
And I have zero interest in your bad-retread of 1970s Tiers Mondisme, I have real investment to do.
"The problem is that if you start with a temperature target of 1.5 or 2 degrees Celsius, and then derive from that a global emissions target, and then divide that into national and local emissions targets ..."
OMG! This is an incredibly stupid way to approch reducing the harm of CO2 emisions! One uses the temperature target to derive the global carbon tax that woud produce it and then either have a carbon tax or mimic it with subsidies paid for with the lowest DWL tax possible and or using that tax value in the cost benefit analysis of public investment. Local, even national emission targets (why not zip code targets?) make no sense at all. That is COP thinking and McKibbenism that has got us to where we are!
And Matt agrees that this is bad! In this article, he specifically says that "the entire exercise of reasoning backwards from targets is misguided".
"Consolidated Edison is the largest property taxpayer in both New York City and New York State. The utility recovers this expense directly from its users by embedding the tax costs into the "Delivery Charge" on monthly energy bills, a pass-through mechanism authorized by state regulators. Government taxes and fees account for roughly 30% of a standard Con Edison bill. The property and special franchise taxes are the primary drivers of this, comprising roughly 17% to 20% of the total amount a customer pays each month."
My belief in rational taxation on energy and climate issues is at a low ebb. Why this charade of taxing the utility in order to tax the consumer?
If a utility is using 1000 units of a public good, and a private actor using 1000 units of a public good would get taxed $X, then the utility should also pay $X. It is more transparent. People looking at setting up some other kind of utility that is perhaps more or less land-efficient can compare dollars directly.
You really need VAT
I wanted to say MY is wrong on Polar Bears but I am not sure, it is one of the issues with a huge partisan divide. Conservative sources are saying Polar Bear populations are exploding, leftwing ones say it is complicated some are getting fatter and healthier, some subpopulations are doing well but other might not be.
I think Polar Bears might be fine with global warming but birds they eat as an alternative to seals in warm weather might be really upset by data centres.
I don’t see how you can advocate for carbon taxes in the US and then not pair it with a carbon border adjustment.