It’s probably not your ideological thing, or mine. It’s about object-level claims about how data centers impact the communities where they’re built, many of which are wrong.
Yes. While I'm sure there are ideologues in every group, you indeed get the impression that if the hyperscalers said, "Fine, we'll build an even bigger one in Ohio!", the protesters would say, "Fine, you do you, just keep it away from our town." They aren't protesting data centers in general, or "AI slop", or "hire an artist, you thieves". They are protesting this data center, here, or - if there are coordination benefits - at least in the sate.
The drumbeat of panics about AI/ML energy/water consumption that started well before ChatGPT does seem like the main culprit, but I'm still amazed at the deep traction this got. There's some perfect storm of emotional and purity triggers here that makes the issue very resonant.
(The Masley article that describes taking the data center income and using it to offset environmental pollution/damage in other ways / elsewhere / for greater benefits, even changed the mind of my two kids. That's a winning explanation. (a) The things are getting built somewhere, (b) likely someplace much more pollution friendly, (c) so take the revenue HERE and apply it to mitigation more efficiently. It's an obvious win.
What do you think explains “years of media coverage of object-level local environmental and economic harms,” including stories that are “strange isolated statistics or just flat out fake” or that “just share people’s broad concerns without seeing whether they stand up to scrutiny”? Why is there so little incentive to correct these stories, and why do they get amplified so much on social media?
To my mind the explanation for this has a lot to do with ideology.
I wonder if the decline in journalism as a profession and the general sense of precarity that many journalists must feel as a result is an abstract cause of this
I suspect it’s just general “if it bleeds, it leads”, or more generally “if it sounds like a story about big shadowy forces destroying the pure natural environment and farmers, it leads”. People have a natural cognitive bias to caring about stories that sound like these.
Their actual result is just that people are ticking either "much more supportive" or "somewhat more supportive" on a multiple choice question when given the argument about water recycling and asked "Does knowing this information change your view of data centers?"
What to make of that result? I'm not sure as I don't have intuitions for how to interpret these kinds of polling questions. I'm very suspicious of social desirability bias here. Other arguments put up similar "net shift" numbers in this format, such as data centers creating jobs, paying taxes, or (weirdly) "AI can fight inflation".
But consider Amazon’s HQ2 plan in 2019 in New York, which did draw a backlash that felt quite similar to today’s data center backlash.
To be fair, that backlash was on the left, whereas the data center backlash is somewhat more bipartisan.
On the other hand, I suspect “an Amazon distribution center” polls better in part because people think of their own personal benefit from getting Amazon packages quicker. AI data centers provide no similar benefit to locals.
Also, AI data centers are new and scary and the subject of active discourse in a way that Amazon warehouses are not. That makes them more likely to be an outlet for generalized distrust.
I think the the data center backlash is probably a bunch about AI and "big tech", in addition to environmental impacts / resource concerns — many factors that interact and reinforce one another. Polls seem limited as an instrument to understand the data center backlash, since they collapse
See Jasmine Sun on Nate Silver's substack, she traveled around the Midwest interviewing people about data centers: https://www.natesilver.net/p/why-does-everyone-hate-data-centers . For example she thinks the practice of having farmers sign non-disclosure agreements significantly worsened the problem.
One other point is that people weigh costs and benefits when deciding to support or oppose data centers, and I think the felt benefits are particularly weak for AI. I think a lot of Americans see AI as something that's being forced down their throats and which doesn't meaningfully improve their lives. (This may account for why we see datacenters-for-AI polling meaningfully worse than datacenters-for-other-stuff.)
I agree with this and I hate that I agree with it because it says something so depressing about the world we live in and how the average person in it operates
This was very good. The only thing I’d add is that the object-level concerns *themselves follow long-established patterns* as a function of the salience of various issues. The gap you point out between “Big Tech” (polls negative) and “specific ‘Big Tech’ companies” (polls positive) is illustrative in this regard, as it suggests a certain exposed-nervedness around the overall milieu of the centrality of the tech revolution even as individual expressions of that prove desirable. I’d argue that this is indeed a fairly coherent slate of opinions! https://zachill.substack.com/p/the-whole-big-data-center-fight-is
Complaints about immigration are mostly about immigration not disguises for economic anxiety. Complaints about new developments are complaints about those new developments and their characteristics - not expressions of environmentalism etc.
The question is why so many self-styled sophisticates believe people’s stated preferences are misrepresentative…
Andy, I do agree with you that the data center concerns come from concerns about "resources" which in turn have something to do with the media coverage but I do think there is something to analyze about "larger concerns" voters might have. I wrote a whole post about a study in our field that I love (because it really gets into the voters' minds) but where I think the analyst does draw conclusions that his own data doesn't warrant: https://computingandsociety.substack.com/p/brian-wynnes-study-of-cumbrian-sheep. In this case, the farmers' concerns come from their general distrust of the nuclear industry; but the particular government-owned nuclear power plant in their vicinity.
If I might hypothesize somewhat, I think what seems to have happened here is that there was a "techlash" *among progressives* which was reflected in the media coverage around data centers and which in turn has now created a national-level data center backlash around resources. So I think "big tech" thing is a big reason but it's not because the general population is averse to big tech but because a large segment of progressives are and those progressives have driven the media coverage (they are also the main consumers of media).
Am I correct in understanding that the Harvard Cap Harris Poll you refer to when indicating the views of Americans' favourability/unfavourability towards big tech is from 2021 and uses a sampling of about 2,000 people?
Yeah it would have been helpful to have the year listed right next to that somewhere. I suspected it was out of date if Israel had net positive approval.
AI is not stupid; it’s just being used stupidly. There’s nothing new about that. Problems with technology are pretty much by definition the fault of the user.
Yeah it’s for sure the datacenters themselves. Try polling the favorability of any project with the square footage and resource usage of a factory that creates 10 jobs instead of hundreds. Honestly people can tend to have nimby views of factories and those at least produce tons of jobs
Yep — though I do think there’s probably a class element too: anger at the economic winners who seem to be taking advantage of the people and places that lie outside their power orbits in NYC and SF. The whole transaction feels extractive rather than reciprocal or productive.
So the anger we see in public comment isn’t really about frustration over the site per se, or about AI’s consequences, or big tech. It’s about people feeling like they’re being exploited and defiantly standing up against that.
That class-resentment explanation has more predictive power than the pure object-level concerns, but the object level concerns are what people reach for when they try to express the underlying feeling of powerlessness and exclusion from a future they recognize will increasingly alienate their way of life. It’s not like a factory that produces widgets or meat that the town can feel some ownership over. Instead, it represents the apotheosis of disembodied automation. And noisy to boot.
All of the online complaints about data centres appear to come from people getting angry on behalf of other people. It’s always “people’s bills are going up” and never “my bills are going up”. It’s always “how would you like to live next to a data centre?” and never “I live next to a data centre”. It’s always “people’s taps are running dry” and never “my taps are running dry”.
Sure, reporters have gone and found people saying those things but the number of the people with complaints is a tiny fraction of the number of people complaining.
I think if it were about data centres, you’d be hearing less from people getting angry on behalf of other people and more from people getting angry on behalf of themselves.
There are significantly more data centers planned than currently exist. Obviously the number if people worried about future impact is going to be higher than the number of people currently impacted
Are there? How many exist and how many are in the planning process?
You can swap out the tense in everything I said for things like “I *will* be living next to a data centre” and my point remains the same. The vast majority of people seem to be complaining about things that aren’t affecting them.
Thanks for writing this up. I wonder if the people who cite environmental reasons for opposing data centers are stating those reasons but have deeper-held beliefs that motivate them. Even if presented with data that things like water and energy usage are overstated/fake, people may choose to reject them because of lack of trust in these projects and the people building them, especially given the usage of NDAs with local officials during these projects. (This is my takeaway from Jasmine Sun's reporting on data centers: https://jasmi.news/p/no-data-centers-in-my-backyard). I agree that most people don't care enough to look into the resource usage info or even know about the NDA stuff, but I suspect stories of resistance like those in local communities echo to the wider world in the same way that the resource drain stories do. People hear about local communities fighting against domineering tech companies, and who is going to instinctively side with the company in that scenario? They might like Google, but the conception of Google on a survey is different from a conception of a big company planting a data center where the community doesn't want it.
First, what happens if we don't build data centers? What if there were a nationwide moratorium? How would that affect American society? It feels like the media isn't sharing the cost of not building data centers.
Second, how does race fit in? I don't mean to play the race card, but I am interested in how most (but not all) data centers are being built in suburban or rural areas, which tend to be mostly white. Would we have these issues if they were in predominantly black urban areas?
Yes. While I'm sure there are ideologues in every group, you indeed get the impression that if the hyperscalers said, "Fine, we'll build an even bigger one in Ohio!", the protesters would say, "Fine, you do you, just keep it away from our town." They aren't protesting data centers in general, or "AI slop", or "hire an artist, you thieves". They are protesting this data center, here, or - if there are coordination benefits - at least in the sate.
The drumbeat of panics about AI/ML energy/water consumption that started well before ChatGPT does seem like the main culprit, but I'm still amazed at the deep traction this got. There's some perfect storm of emotional and purity triggers here that makes the issue very resonant.
The "cost" of not building them is remote and abstract to people. This is a cheap easy way to oppose something and claim that you're doing good.
People finding communion in irrationally hating things is both so real and so bleak
These people obviously haven’t been constantly running into their five-hour limit on Claude!
(The Masley article that describes taking the data center income and using it to offset environmental pollution/damage in other ways / elsewhere / for greater benefits, even changed the mind of my two kids. That's a winning explanation. (a) The things are getting built somewhere, (b) likely someplace much more pollution friendly, (c) so take the revenue HERE and apply it to mitigation more efficiently. It's an obvious win.
What do you think explains “years of media coverage of object-level local environmental and economic harms,” including stories that are “strange isolated statistics or just flat out fake” or that “just share people’s broad concerns without seeing whether they stand up to scrutiny”? Why is there so little incentive to correct these stories, and why do they get amplified so much on social media?
To my mind the explanation for this has a lot to do with ideology.
I pontificate a bit here, no confident takes https://blog.andymasley.com/i/198417196/the-decline-of-science-journalism
Find that a lot of journalists are more sympathetic than average to very localist/degrowth stories
I wonder if the decline in journalism as a profession and the general sense of precarity that many journalists must feel as a result is an abstract cause of this
I suspect it’s just general “if it bleeds, it leads”, or more generally “if it sounds like a story about big shadowy forces destroying the pure natural environment and farmers, it leads”. People have a natural cognitive bias to caring about stories that sound like these.
Unfortunately the 31pp result you cite from the Rainey Center is extremely misleadingly framed by them.
They say "support swings by a net +31 points" but what they mean by this, if you look at page 5 (https://cdn.prod.website-files.com/663a60b91deeae06e8ae1d1c/6a317d2d221255ef40039b0a_RaineyCenter_DataCenters_May2026_Memo_8.5x11_r1_digital.pdf?utm_source=substack&utm_medium=email), is that upon hearing the water recycling argument, 37% of respondents become *more supportive* and 6% become *less supportive*, leading to what they call "net shift" of 37 - 6 = 31 points.
Their actual result is just that people are ticking either "much more supportive" or "somewhat more supportive" on a multiple choice question when given the argument about water recycling and asked "Does knowing this information change your view of data centers?"
What to make of that result? I'm not sure as I don't have intuitions for how to interpret these kinds of polling questions. I'm very suspicious of social desirability bias here. Other arguments put up similar "net shift" numbers in this format, such as data centers creating jobs, paying taxes, or (weirdly) "AI can fight inflation".
Oh great thank you! Will circle back and edit
But consider Amazon’s HQ2 plan in 2019 in New York, which did draw a backlash that felt quite similar to today’s data center backlash.
To be fair, that backlash was on the left, whereas the data center backlash is somewhat more bipartisan.
On the other hand, I suspect “an Amazon distribution center” polls better in part because people think of their own personal benefit from getting Amazon packages quicker. AI data centers provide no similar benefit to locals.
Also, AI data centers are new and scary and the subject of active discourse in a way that Amazon warehouses are not. That makes them more likely to be an outlet for generalized distrust.
I think the the data center backlash is probably a bunch about AI and "big tech", in addition to environmental impacts / resource concerns — many factors that interact and reinforce one another. Polls seem limited as an instrument to understand the data center backlash, since they collapse
See Jasmine Sun on Nate Silver's substack, she traveled around the Midwest interviewing people about data centers: https://www.natesilver.net/p/why-does-everyone-hate-data-centers . For example she thinks the practice of having farmers sign non-disclosure agreements significantly worsened the problem.
One other point is that people weigh costs and benefits when deciding to support or oppose data centers, and I think the felt benefits are particularly weak for AI. I think a lot of Americans see AI as something that's being forced down their throats and which doesn't meaningfully improve their lives. (This may account for why we see datacenters-for-AI polling meaningfully worse than datacenters-for-other-stuff.)
I agree with this and I hate that I agree with it because it says something so depressing about the world we live in and how the average person in it operates
Helpful! Thank you!
This was very good. The only thing I’d add is that the object-level concerns *themselves follow long-established patterns* as a function of the salience of various issues. The gap you point out between “Big Tech” (polls negative) and “specific ‘Big Tech’ companies” (polls positive) is illustrative in this regard, as it suggests a certain exposed-nervedness around the overall milieu of the centrality of the tech revolution even as individual expressions of that prove desirable. I’d argue that this is indeed a fairly coherent slate of opinions! https://zachill.substack.com/p/the-whole-big-data-center-fight-is
This generalises doesn’t it.
Complaints about immigration are mostly about immigration not disguises for economic anxiety. Complaints about new developments are complaints about those new developments and their characteristics - not expressions of environmentalism etc.
The question is why so many self-styled sophisticates believe people’s stated preferences are misrepresentative…
Andy, I do agree with you that the data center concerns come from concerns about "resources" which in turn have something to do with the media coverage but I do think there is something to analyze about "larger concerns" voters might have. I wrote a whole post about a study in our field that I love (because it really gets into the voters' minds) but where I think the analyst does draw conclusions that his own data doesn't warrant: https://computingandsociety.substack.com/p/brian-wynnes-study-of-cumbrian-sheep. In this case, the farmers' concerns come from their general distrust of the nuclear industry; but the particular government-owned nuclear power plant in their vicinity.
If I might hypothesize somewhat, I think what seems to have happened here is that there was a "techlash" *among progressives* which was reflected in the media coverage around data centers and which in turn has now created a national-level data center backlash around resources. So I think "big tech" thing is a big reason but it's not because the general population is averse to big tech but because a large segment of progressives are and those progressives have driven the media coverage (they are also the main consumers of media).
Am I correct in understanding that the Harvard Cap Harris Poll you refer to when indicating the views of Americans' favourability/unfavourability towards big tech is from 2021 and uses a sampling of about 2,000 people?
Yeah maybe a lot’s changed since but the Argument found pretty similar numbers more recently
Yeah it would have been helpful to have the year listed right next to that somewhere. I suspected it was out of date if Israel had net positive approval.
AI is not stupid; it’s just being used stupidly. There’s nothing new about that. Problems with technology are pretty much by definition the fault of the user.
Yeah it’s for sure the datacenters themselves. Try polling the favorability of any project with the square footage and resource usage of a factory that creates 10 jobs instead of hundreds. Honestly people can tend to have nimby views of factories and those at least produce tons of jobs
Yep — though I do think there’s probably a class element too: anger at the economic winners who seem to be taking advantage of the people and places that lie outside their power orbits in NYC and SF. The whole transaction feels extractive rather than reciprocal or productive.
So the anger we see in public comment isn’t really about frustration over the site per se, or about AI’s consequences, or big tech. It’s about people feeling like they’re being exploited and defiantly standing up against that.
That class-resentment explanation has more predictive power than the pure object-level concerns, but the object level concerns are what people reach for when they try to express the underlying feeling of powerlessness and exclusion from a future they recognize will increasingly alienate their way of life. It’s not like a factory that produces widgets or meat that the town can feel some ownership over. Instead, it represents the apotheosis of disembodied automation. And noisy to boot.
All of the online complaints about data centres appear to come from people getting angry on behalf of other people. It’s always “people’s bills are going up” and never “my bills are going up”. It’s always “how would you like to live next to a data centre?” and never “I live next to a data centre”. It’s always “people’s taps are running dry” and never “my taps are running dry”.
Sure, reporters have gone and found people saying those things but the number of the people with complaints is a tiny fraction of the number of people complaining.
I think if it were about data centres, you’d be hearing less from people getting angry on behalf of other people and more from people getting angry on behalf of themselves.
There are significantly more data centers planned than currently exist. Obviously the number if people worried about future impact is going to be higher than the number of people currently impacted
Are there? How many exist and how many are in the planning process?
You can swap out the tense in everything I said for things like “I *will* be living next to a data centre” and my point remains the same. The vast majority of people seem to be complaining about things that aren’t affecting them.
Thanks for writing this up. I wonder if the people who cite environmental reasons for opposing data centers are stating those reasons but have deeper-held beliefs that motivate them. Even if presented with data that things like water and energy usage are overstated/fake, people may choose to reject them because of lack of trust in these projects and the people building them, especially given the usage of NDAs with local officials during these projects. (This is my takeaway from Jasmine Sun's reporting on data centers: https://jasmi.news/p/no-data-centers-in-my-backyard). I agree that most people don't care enough to look into the resource usage info or even know about the NDA stuff, but I suspect stories of resistance like those in local communities echo to the wider world in the same way that the resource drain stories do. People hear about local communities fighting against domineering tech companies, and who is going to instinctively side with the company in that scenario? They might like Google, but the conception of Google on a survey is different from a conception of a big company planting a data center where the community doesn't want it.
I have two questions:
First, what happens if we don't build data centers? What if there were a nationwide moratorium? How would that affect American society? It feels like the media isn't sharing the cost of not building data centers.
Second, how does race fit in? I don't mean to play the race card, but I am interested in how most (but not all) data centers are being built in suburban or rural areas, which tend to be mostly white. Would we have these issues if they were in predominantly black urban areas?