WEBVTT - The Necessary War on Data Centers 0:00:02.120 --> 0:00:07.800 Zone media, welcome to it could happen here. 0:00:08.080 --> 0:00:10.280 I'm Robert Evans. Now. 0:00:10.560 --> 0:00:13.320 A few days ago I had the experience of listening 0:00:13.320 --> 0:00:15.239 to somebody that I've known for a while and known 0:00:15.280 --> 0:00:18.680 as a fairly mainstream liberal, someone who does not, in 0:00:18.760 --> 0:00:22.400 general support the personal ownership of firearms, certainly not weapons 0:00:22.400 --> 0:00:26.920 like ar fifteens, react to the recent race riots, bagrams, 0:00:27.320 --> 0:00:30.760 you know, whatever term I guess is most appropriate in Belfast, 0:00:31.240 --> 0:00:35.960 where racist mobs ran through the streets attacking people in 0:00:36.000 --> 0:00:37.879 their homes, pulling them out of their homes, lagging their 0:00:37.880 --> 0:00:41.320 homes on fire, targeting these people because they were not white. 0:00:41.479 --> 0:00:44.040 And this person, who I've known for a while, reacted 0:00:44.080 --> 0:00:47.520 to watching footage of this and reading headlines in their 0:00:47.560 --> 0:00:50.159 general shock and horror by saying something along the lines of, 0:00:50.800 --> 0:00:52.800 I guess, if this kind of stuff is on the table, 0:00:52.840 --> 0:00:57.200 I understand why people need ar fifteens. Now, nothing had 0:00:57.240 --> 0:01:00.000 actually changed about the way the world works to all 0:01:00.360 --> 0:01:04.360 this person's opinion of guns or of reality. They were 0:01:04.400 --> 0:01:07.240 simply forced to confront an aspect of reality that had 0:01:07.240 --> 0:01:10.600 previously been obscured to them, because in earlier days they 0:01:10.600 --> 0:01:13.840 had assumed that in any civilized nation and they considered 0:01:13.880 --> 0:01:16.720 Northern Ireland to be part of a civilized nation. Police 0:01:16.720 --> 0:01:20.080 would stop this kind of behavior. And if you feel 0:01:20.120 --> 0:01:22.639 like the rule of law is something you can trust 0:01:22.680 --> 0:01:25.360 on pretty immutably, then someone else saying no, I would 0:01:25.440 --> 0:01:27.360 rather have a gun of my own than trust a 0:01:27.440 --> 0:01:31.480 cop might sound like a mania. Today's episode is not 0:01:31.600 --> 0:01:34.399 about guns or gun rights. It is, in the most 0:01:34.400 --> 0:01:37.400 direct sense, about data centers and the current people's war 0:01:37.520 --> 0:01:41.000 on building more of them. I see the current vast 0:01:41.040 --> 0:01:43.720 groundswell of support for this kind of activism and this 0:01:43.760 --> 0:01:46.400 fight in general, as the most hopeful change in domestic 0:01:46.440 --> 0:01:49.480 politics in quite some time. And I think this movement 0:01:49.520 --> 0:01:51.760 could become a weapon that drives a stake into the 0:01:51.760 --> 0:01:54.480 hearts of the silicon oligarchy that currently rule quite a 0:01:54.520 --> 0:01:57.960 lot of our world. The problem is any mass movement 0:01:58.080 --> 0:02:01.520 like this, especially one built on what is essentially also 0:02:01.680 --> 0:02:05.680 like the biggest name in the news right now, AI, 0:02:06.200 --> 0:02:09.680 is going to involve a lot of misinformation and even disinformation. 0:02:09.919 --> 0:02:13.200 People don't just hate data centers for good reasons, They 0:02:13.320 --> 0:02:15.600 hate them for bad ones too, Or at least it 0:02:15.600 --> 0:02:17.560 would probably be more accurate for me to say a 0:02:17.560 --> 0:02:20.280 lot of people hate them for reasons that don't reflect 0:02:20.320 --> 0:02:24.520 something data centers actually do or make worse. The broader 0:02:24.560 --> 0:02:28.240 reason that I've seen some people have issues with this fight, 0:02:28.320 --> 0:02:30.840 with this mass movement against the construction of new data 0:02:30.840 --> 0:02:35.600 centers is that data centers are utterly necessary modern infrastructure 0:02:35.760 --> 0:02:39.040 and incredibly crucial to the maintenance of what we would 0:02:39.080 --> 0:02:42.800 consider basic daily life. This is something that's true even 0:02:42.840 --> 0:02:45.520 without AI, and this is something that was true prior 0:02:45.560 --> 0:02:48.200 to the introduction of what Silicon Valley likes to call 0:02:48.280 --> 0:02:51.639 AI into all of our lives. Data centers are what 0:02:51.720 --> 0:02:54.880 make the age of cloud computing possible, and starting in 0:02:54.919 --> 0:02:57.640 the twenty tens, we decided that the cloud was where 0:02:57.680 --> 0:03:01.480 everything compute wise was increasingly going to happen in the future. 0:03:02.120 --> 0:03:05.440 There were pragmatic reasons for this. Having words save your 0:03:05.520 --> 0:03:09.000 document to a cloud file that makes it instantly accessible 0:03:09.000 --> 0:03:12.440 from any machine with Internet access should you desire. That's 0:03:12.440 --> 0:03:16.239 a big leap forward in capability, at least in one direction. Likewise, 0:03:16.280 --> 0:03:18.880 having like a watch on your body that cannot just 0:03:18.919 --> 0:03:21.720 take down your biometric data during a workout, but can 0:03:21.760 --> 0:03:24.400 store it and analyze it over long periods of time. 0:03:24.840 --> 0:03:28.240 That offers people real utility there's a reason products like 0:03:28.280 --> 0:03:32.200 this are very popular. You can't, however, have the entire 0:03:32.240 --> 0:03:35.320 world mapped out and accessible for turn by turn navigation 0:03:35.800 --> 0:03:39.320 without needing a shitload of data centers somewhere and in fact, 0:03:39.440 --> 0:03:43.240 in quite a few somewheres to make that possible. Now, 0:03:43.240 --> 0:03:45.920 of course, there are also privacy trade offs for all 0:03:46.040 --> 0:03:49.160 of these kinds of products. That's been the entire logic 0:03:49.240 --> 0:03:52.040 of the Internet of Things era. You hand over your 0:03:52.160 --> 0:03:56.560 data and we deliver utility. Privacy advocates have had issues 0:03:56.600 --> 0:03:59.800 with this from the jump, but most people didn't because 0:03:59.840 --> 0:04:03.040 the benefits were obvious and most people don't like to 0:04:03.080 --> 0:04:05.360 read or think very hard if we can avoid it. 0:04:06.040 --> 0:04:09.520 For almost twenty years, consumers largely ignored the rollout of 0:04:09.600 --> 0:04:13.279 data centers around the world, even sometimes into our backyards, 0:04:13.480 --> 0:04:17.000 because big tech had reasonably good pr and most people 0:04:17.040 --> 0:04:20.080 were pretty happy with their gizmos. All in all, you 0:04:20.120 --> 0:04:22.640 get a hint of how uncontroversial this used to be. 0:04:22.800 --> 0:04:25.080 In a twenty twenty two article I found published on 0:04:25.160 --> 0:04:30.040 Microsoft's source EMEA website, written by Bill Briggs. The title 0:04:30.200 --> 0:04:33.839 was critical to our modern society, how data centers power 0:04:33.920 --> 0:04:37.119 everyday necessities. Now, there was a little bit about AI 0:04:37.200 --> 0:04:39.000 in there, but this is twenty twenty two and the 0:04:39.080 --> 0:04:42.040 hype cycle hadn't nearly hit its peak at that point. 0:04:42.279 --> 0:04:45.400 The article spent much more time emphasizing how integral data 0:04:45.440 --> 0:04:49.640 centers are in things that people like quote. Simply put, 0:04:49.839 --> 0:04:53.440 they are the physical infrastructure behind cloud computing, and across Europe, 0:04:53.640 --> 0:04:56.360 Microsoft data centers are operating around the clock to support 0:04:56.360 --> 0:04:59.320 a wide spectrum of critical services, from the life saving 0:04:59.360 --> 0:05:01.920 work of doctors and first responders to essential services like 0:05:02.000 --> 0:05:05.080 groceries and online banking. At the same time, data centers 0:05:05.080 --> 0:05:08.719 also empowered everyday necessities like food deliveries, remote work, and 0:05:08.839 --> 0:05:12.240 video calls to family. The article goes on to include 0:05:12.240 --> 0:05:15.520 a quote from Rahil Nasir, an associate research director at 0:05:15.520 --> 0:05:19.839 a market research firm, who called data centers the invisible infrastructure, 0:05:20.240 --> 0:05:23.160 which I find interesting because absolutely no one would call 0:05:23.200 --> 0:05:23.960 it back today. 0:05:24.279 --> 0:05:24.520 Right. 0:05:24.880 --> 0:05:28.040 People only saw them as invisible back then because there 0:05:28.080 --> 0:05:31.080 hadn't been much public discussion about these things, and there 0:05:31.120 --> 0:05:33.640 weren't nearly as many of them, to be fair. In 0:05:33.720 --> 0:05:37.320 twenty twenty two, the del Oro Group estimated global data 0:05:37.320 --> 0:05:40.480 center capex to have been about two hundred and forty 0:05:40.480 --> 0:05:45.159 one billion dollars. McKenzie is currently predicting global spending on 0:05:45.240 --> 0:05:49.200 data centers will reach seven trillion dollars by twenty thirty. 0:05:50.320 --> 0:05:52.880 The number of data center projects in development or under 0:05:52.920 --> 0:05:56.440 contract exploded pretty much right after Microsoft put out the 0:05:56.520 --> 0:05:59.200 article I quoted from earlier, and this put millions of 0:05:59.240 --> 0:06:01.839 people face to face with the reality of data center 0:06:01.880 --> 0:06:06.599 construction projects, many of them revolted. Q one of twenty 0:06:06.600 --> 0:06:09.760 twenty six saw a record seventy five data center projects 0:06:10.000 --> 0:06:13.760 blocked or delayed nationwide, which made it, in NBC's words, 0:06:13.960 --> 0:06:17.200 the most blocked and delayed quarter for data center projects 0:06:17.200 --> 0:06:20.840 on record. Researchers at Data world watchtold journalists they did 0:06:20.839 --> 0:06:24.919 not consider this a cyclical spike polygon, instead a structural 0:06:25.000 --> 0:06:29.880 shift caused as quote communities internalized an opposition playbook, legislative 0:06:29.920 --> 0:06:33.559 sessions introduced formal regulatory uncertainty, and the number of active 0:06:33.600 --> 0:06:36.200 opposition groups more than doubled to eight hundred and thirty 0:06:36.279 --> 0:06:37.960 three across forty nine states. 0:06:39.120 --> 0:06:39.400 Now. 0:06:39.600 --> 0:06:42.520 The rapid and obviously organic nature of this growth has 0:06:42.560 --> 0:06:46.560 alarmed certain democratic and Republican politicians, largely the ones who 0:06:46.600 --> 0:06:49.120 have taken money from the tech industry. Some of these 0:06:49.160 --> 0:06:52.839 folks have panicked and insisted that Chinese government propaganda is 0:06:52.880 --> 0:06:55.880 behind the growth and opposition to data centers. The New 0:06:55.960 --> 0:06:59.520 York Times published a piece by sociologist Tressy McMillan Cottam, 0:06:59.720 --> 0:07:02.320 who argue that Democrats need to make the issue of 0:07:02.400 --> 0:07:05.960 data centers their issue because she believes that these protests 0:07:06.160 --> 0:07:08.680 could have a significant impact on the midterms, and in 0:07:08.720 --> 0:07:11.520 twenty twenty eight, in her column, she states that she 0:07:11.600 --> 0:07:15.040 wasn't initially sold on the value of data center protests, 0:07:15.200 --> 0:07:17.679 but time around activists convinced her that she'd been wrong, 0:07:17.720 --> 0:07:20.280 in part because this was not an issue that inherently 0:07:20.360 --> 0:07:24.120 drew urban or suburban or rural voters together, but all 0:07:24.160 --> 0:07:26.880 of them like it didn't inherently pull out Republican or 0:07:26.880 --> 0:07:29.880 Democratic voters, and thus it was a way to bring 0:07:29.920 --> 0:07:33.600 people together and get them working together uncommon cause, which 0:07:33.640 --> 0:07:36.480 is a really beneficial thing to do if you're trying 0:07:36.480 --> 0:07:40.360 to pull people closer into like a broader political alignment 0:07:40.480 --> 0:07:43.280 with you. And I would add to that point, just 0:07:43.320 --> 0:07:45.680 based on my own research and what I've seen, I 0:07:45.840 --> 0:07:48.280 feel like the fight against data centers is something that 0:07:48.320 --> 0:07:50.400 has what I like to call a high likelihood of 0:07:50.440 --> 0:07:54.600 fundamentalism in it or HLF, and an HLF issue is 0:07:54.960 --> 0:07:57.480 a kind of political fight that if you get drawn 0:07:57.520 --> 0:08:00.960 into it, you're really likely to become some to fundamentalist 0:08:01.040 --> 0:08:03.520 on the matter. And abortion would be a good example 0:08:03.560 --> 0:08:05.320 of this sort of issue, right. That's why back in 0:08:05.360 --> 0:08:08.640 the nineteen seventies the Republican Party adopted abortion as a 0:08:08.680 --> 0:08:11.960 central concern, because if you get someone to become an 0:08:11.960 --> 0:08:16.320 anti abortion activist, they'll vote just based on that right. 0:08:16.440 --> 0:08:19.000 And if one party is against abortion and one parties 0:08:19.040 --> 0:08:21.720 for it, they'll never consider the party that is even 0:08:21.800 --> 0:08:25.480 open minded to it being legal in some limited sense, right, 0:08:25.560 --> 0:08:28.600 they become an absolutist on the matter. These are great 0:08:28.720 --> 0:08:31.200 issues to pull people into like a voting block with, 0:08:31.360 --> 0:08:33.560 because if you can get them to associate that issue 0:08:33.559 --> 0:08:36.280 with your party, you can kind of immunize that chunk 0:08:36.280 --> 0:08:39.000 of your voter base from economic concerns or other points 0:08:39.000 --> 0:08:42.960 of attack, at least in theory. Now, different issues tend 0:08:43.000 --> 0:08:46.319 to create fundamentalists for different reasons. People who are organizing 0:08:46.400 --> 0:08:49.679 against data centers aren't doing it usually because they think 0:08:49.720 --> 0:08:52.840 the idea of a data center is wrong, But many 0:08:52.880 --> 0:08:55.439 of them do believe that AI is evil and the 0:08:55.559 --> 0:08:58.120 moral and AI is the reason that so many new 0:08:58.200 --> 0:09:01.320 data centers are going up. Many more people oppose new 0:09:01.360 --> 0:09:03.679 data centers for the specific reason they see them as 0:09:03.679 --> 0:09:06.679 a threat to their own personal environment and to their 0:09:06.679 --> 0:09:09.240 own power bills, and to the cleanliness of their own 0:09:09.280 --> 0:09:12.440 water and air. It may seem as if this is 0:09:12.440 --> 0:09:15.520 the kind of support that would tend to breed shallow activists. 0:09:15.559 --> 0:09:18.199 If you just don't want a data center in your hometown, 0:09:18.280 --> 0:09:20.120 that doesn't mean you give a fuck about one going 0:09:20.240 --> 0:09:22.280 up in Mississippi or wherever you don't live. 0:09:22.559 --> 0:09:25.559 After all, none of these people presumably. 0:09:25.000 --> 0:09:27.080 Cared about the data centers that had been going up 0:09:27.120 --> 0:09:29.800 five years ago, ten years ago, before the issue became 0:09:29.880 --> 0:09:33.320 salient to them. But as treci Cottam wrote, the realities 0:09:33.320 --> 0:09:35.080 of doing this kind of organism on the ground have 0:09:35.120 --> 0:09:38.800 a tendency to transcend the reasons individuals start getting into 0:09:38.800 --> 0:09:42.079 the fight quote. I have been watching this new groundswell 0:09:42.160 --> 0:09:46.000 of descent firsthand in community meetings, organizing sessions, and civic trainings. 0:09:46.000 --> 0:09:49.480 Here in North Carolina. The resistance has lifelong joiners, alumni 0:09:49.559 --> 0:09:52.680 from environmental and housing movements, and young organizers. There are 0:09:52.720 --> 0:09:54.600 also a lot of people who have never dreamed of 0:09:54.640 --> 0:09:57.400 being disagreeable in public, much less considered joining a rock 0:09:57.440 --> 0:10:00.160 as social movement. The imminent risk of living next to 0:10:00.160 --> 0:10:02.200 a data center maybe why they show up for a meeting, 0:10:02.360 --> 0:10:04.840 but they're committing to the issue for bigger, deeper reasons. 0:10:05.040 --> 0:10:08.560 Political corruption and corporate mouthfeasance make them feel politically impotent. 0:10:08.880 --> 0:10:12.400 Voicing their objections, sharing their anxieties with others, recalling politicians 0:10:12.400 --> 0:10:14.959 who override them, and in some cases beating the opposition 0:10:15.280 --> 0:10:18.200 is giving them something. Few politicians are offering a taste 0:10:18.200 --> 0:10:40.880 of political power. And we'll continue, but first, here's some ads. 0:10:33.600 --> 0:10:34.400 And we're back. 0:10:35.040 --> 0:10:37.960 I think it's really important to emphasize something that Cottam 0:10:38.120 --> 0:10:39.800 said at the end of that last quote, which is 0:10:39.800 --> 0:10:42.720 that part of why the anti data center movement is 0:10:42.840 --> 0:10:44.960 potent and why it tends to draw people in much 0:10:45.000 --> 0:10:47.440 deeper than maybe the initial shallowness of why they got 0:10:47.520 --> 0:10:50.080 in would suggest, is that it offers them a sense 0:10:50.080 --> 0:10:52.400 of political power and of agency, and an age in 0:10:52.440 --> 0:10:54.920 which people are being trained more and more to feel 0:10:54.920 --> 0:10:57.400 as if they have no agency. You can't stop this, 0:10:57.760 --> 0:11:00.760 you can't fight this, or alternatively, this will do the 0:11:00.800 --> 0:11:03.800 fighting for you. Right, people are getting it from all 0:11:03.840 --> 0:11:06.560 sides wherever they actually land in the political aisle. And 0:11:06.600 --> 0:11:10.240 that's quite frankly, not very much fun. It's really fun 0:11:10.520 --> 0:11:12.280 to feel like you're a part of a movement like 0:11:12.320 --> 0:11:14.280 this that's kicking ass and taking names. 0:11:14.320 --> 0:11:14.560 Now. 0:11:15.040 --> 0:11:17.439 I don't share the long term goals that led Cottam 0:11:17.440 --> 0:11:19.360 to write that article, at least not all of them. 0:11:19.400 --> 0:11:21.080 I get the sense that she wants to give the 0:11:21.160 --> 0:11:24.040 Democratic Party a powerful new long term voting block. 0:11:24.480 --> 0:11:26.360 I don't care so much about that now. 0:11:26.400 --> 0:11:28.800 I do, of course, want to see the Republican Party 0:11:28.840 --> 0:11:32.400 beaten electorally, and more Democratic votes is the only current 0:11:32.440 --> 0:11:35.040 way to do that. But the Democratic Party is not 0:11:35.200 --> 0:11:38.559 morally or logistically capable of doing anything but betraying these 0:11:38.600 --> 0:11:41.160 people in the long run. I really do believe that 0:11:41.640 --> 0:11:44.000 the reason why is the reason why the party has 0:11:44.040 --> 0:11:47.520 been incredibly slow to embrace these activists in any cohesive fashion. 0:11:47.960 --> 0:11:52.240 This movement is death for Silicon Valley, and Silicon Valley 0:11:52.280 --> 0:11:54.839 has a lot of fucking money. And when I say death, 0:11:55.000 --> 0:11:58.199 don't think I'm exaggerating. Data Center Watch is calculated that 0:11:58.280 --> 0:12:00.880 the total value of data centers block in early twenty 0:12:00.920 --> 0:12:02.880 twenty six. This is just the first three months of 0:12:02.920 --> 0:12:05.520 this year was around one hundred and thirty billion dollars, 0:12:05.720 --> 0:12:07.559 compared to a total of one hundred and fifty six 0:12:07.600 --> 0:12:11.679 billion dollars in projects blocked in all of twenty twenty five. Now, 0:12:11.880 --> 0:12:14.679 a huge month last year for data construction expenses in 0:12:14.720 --> 0:12:17.480 the US was last July, which was actually a historic month, 0:12:17.600 --> 0:12:21.080 and that represented about fourteen billion dollars in expenses. So 0:12:21.120 --> 0:12:24.880 the value of the projects that have been stopped are significant, 0:12:25.040 --> 0:12:28.199 even in tech industry terms. More to the point, nearly 0:12:28.240 --> 0:12:31.520 all the money that underlies and underwrites these corporations and 0:12:31.559 --> 0:12:35.320 their owners is fake. Elon Musk could not produce a 0:12:35.360 --> 0:12:38.560 trillion dollars of real liquid US dollars if he wanted to, 0:12:38.840 --> 0:12:41.360 because that simply is not how assets work. Most of 0:12:41.360 --> 0:12:43.720 his wealth is in stock valuation, and those stocks are 0:12:43.720 --> 0:12:46.000 worth what they're worth because of consumer sentiment as much 0:12:46.000 --> 0:12:49.960 as anything objectively real. The entire AI bubble is being 0:12:50.040 --> 0:12:52.720 underwritten by the belief that this will be worth it. 0:12:53.040 --> 0:12:55.480 There are trillions in value here and we just need 0:12:55.480 --> 0:12:58.080 to spend a few trillion more to unlock them. None 0:12:58.120 --> 0:13:00.679 of that value can be unlocked without data, and if 0:13:00.679 --> 0:13:02.720 you throw a wrench in their plans to build more, 0:13:02.800 --> 0:13:05.920 you can hobble the whole effort. Any delay or serious 0:13:05.960 --> 0:13:09.800 setback creates the risk of a panic, and panic is 0:13:09.840 --> 0:13:14.280 their greatest enemy. All it takes is one sufficiently disastrous 0:13:14.280 --> 0:13:18.440 crash to kill many of these big overvalued companies. SpaceX 0:13:18.440 --> 0:13:21.440 and Anthropic and OpenAI and Microsoft and Meta can't all 0:13:21.480 --> 0:13:25.040 survive a market collapse, just as they won't sadly all die, 0:13:25.320 --> 0:13:28.960 nor can their billionaire bosses all remain masters of the universe. 0:13:29.280 --> 0:13:31.120 If they fall off their perch in a ray that 0:13:31.200 --> 0:13:33.880 shatters the global economy, we won't even have to bring 0:13:33.920 --> 0:13:36.040 all of these guys to court. Some of them will 0:13:36.080 --> 0:13:39.160 get done and by their former friends overseas. There's a 0:13:39.200 --> 0:13:41.880 reason why some VC types have already started trying to 0:13:41.880 --> 0:13:45.480 describe anti data center activism as terroristic. They are in 0:13:45.520 --> 0:13:49.400 fact terrified of it. This segment from an ArsTechnica article 0:13:49.440 --> 0:13:51.679 by Ashley Bellinger should give you an idea of what. 0:13:51.640 --> 0:13:53.000 I'm discussing quote. 0:13:53.280 --> 0:13:56.160 The researchers suggested that the back half of twenty twenty 0:13:56.160 --> 0:13:59.040 five marked a turning point, as data center opposition emerged 0:13:59.040 --> 0:14:01.720 as a national level narrative that showed the AI industry 0:14:01.720 --> 0:14:04.839 can no longer see the fights as individual zoning disputes. 0:14:04.960 --> 0:14:08.920 It is now reshaping elections, regulation, and site viability nationwide. 0:14:08.960 --> 0:14:12.440 Data Center Watch reported last year, where before officials were 0:14:12.480 --> 0:14:16.880 criticized for quietly signing deals without discussing construction with nearby residents, 0:14:17.080 --> 0:14:20.040 now they're encountering backlash before any deal is in the books. 0:14:20.120 --> 0:14:24.080 Data Center Watch found in some cases, researchers reported opposition 0:14:24.120 --> 0:14:27.800 mobilized before any project was officially filed. The mere rumor 0:14:27.840 --> 0:14:32.160 of a data center was enough to trigger organize resistance. Now, folks, 0:14:32.440 --> 0:14:35.880 when you hit this level of cultural inflection with a movement, 0:14:36.240 --> 0:14:39.880 you are in an extremely powerful but also very dangerous position. 0:14:40.400 --> 0:14:43.160 There's a reason that Steve Bannon's also made waves recently 0:14:43.200 --> 0:14:45.440 by encouraging his side to start reaching out to the 0:14:45.440 --> 0:14:48.520 anti data center people. He knows this energy can be 0:14:48.600 --> 0:14:51.160 co opted, or at least he believes it can. For 0:14:51.240 --> 0:14:53.400 my part, I want to see this energy harnessed. I 0:14:53.400 --> 0:14:56.080 want to see it burn brightly like the working flame 0:14:56.120 --> 0:14:58.760 of an oxyocettylene torch, because I want to see it 0:14:58.880 --> 0:15:01.520 used to cut directly into the belly of the industry 0:15:01.520 --> 0:15:03.960 that I hate. The best way for us to do 0:15:04.000 --> 0:15:06.840 this right now is to create as many moments as possible. 0:15:06.920 --> 0:15:10.120 We're projects to construct new data centers clashed with protesters. 0:15:10.600 --> 0:15:12.840 Each of these conflicts has the potential to spark a 0:15:12.880 --> 0:15:15.440 panic in the market, and every piece of bad press 0:15:15.480 --> 0:15:19.000 contributes to the overall weight of public opinion and expectation 0:15:19.360 --> 0:15:22.880 against AI and against big tech. We never know which 0:15:23.000 --> 0:15:25.800 piece will be the one that breaks this current incarnation 0:15:25.880 --> 0:15:28.840 of the industry. The name of the game for us, then, 0:15:29.000 --> 0:15:30.760 is to create as many moments as we can that 0:15:30.880 --> 0:15:33.760 endanger the future of this market and the industry behind it. 0:15:33.800 --> 0:15:36.720 And we can do that entirely legally by forcing them 0:15:36.720 --> 0:15:40.040 to confront reality. None of these data center projects are 0:15:40.080 --> 0:15:43.240 sold to towns and cities based on real, accurate information 0:15:43.440 --> 0:15:46.600 about how they will impact the environment and the local community. 0:15:47.120 --> 0:15:49.960 One good example of this recently from the United States 0:15:50.080 --> 0:15:52.800 is a story earlier this year about a data center 0:15:52.840 --> 0:15:56.880 project in Box Elder County, Utah. The proposed buildings together 0:15:57.160 --> 0:16:00.480 would have been three times the size of Manhattan, stretched 0:16:00.480 --> 0:16:04.240 across multiple sites. The project was championed by celebrity billionaire 0:16:04.320 --> 0:16:07.000 Kevin O'Leary, who was also the VC behind it. For 0:16:07.120 --> 0:16:10.680 a separate article in Ours, Technica Quote. Resident's top concern 0:16:10.760 --> 0:16:13.440 was the straight out data center project draining local waters, 0:16:13.600 --> 0:16:15.480 and they were willing to pay to protect them most, 0:16:15.560 --> 0:16:18.560 especially the vulnerable Great Salt Lake. Many locals paid a 0:16:18.600 --> 0:16:21.280 fifteen dollars fee to register comments to block the transfer 0:16:21.280 --> 0:16:23.720 of nineteen hundred acre feet of water from a ranch 0:16:23.880 --> 0:16:27.440 to the hyperscale data center. Other concerns include electricity bills 0:16:27.480 --> 0:16:30.640 rising and potential risks to air quality, local wildlife and land. 0:16:31.000 --> 0:16:34.160 Venture capitalist Kevin O'Leary, chair of O'Leary Digital and Shark 0:16:34.200 --> 0:16:37.080 Tank Investor, is behind the construction of the project. He 0:16:37.120 --> 0:16:39.560 told a local ABC affiliate that he regrets not working 0:16:39.560 --> 0:16:41.880 with state officials to be more transparent about the project 0:16:41.960 --> 0:16:44.880 from the beginning. We really screwed it up, O'Leary set, 0:16:44.920 --> 0:16:47.640 while confirming he was not expecting this intense kind of 0:16:47.640 --> 0:16:50.280 blowback from the public. He claimed that he and state 0:16:50.320 --> 0:16:53.240 officials anticipated that people would be excited about the major 0:16:53.360 --> 0:16:56.120 local investment and made huge mistakes by not involving the 0:16:56.120 --> 0:16:59.520 public more in discussions. Based on that assumption. We pissed 0:16:59.520 --> 0:17:01.480 off a lot of people, and that's not the way 0:17:01.480 --> 0:17:05.679 I do business. O'Leary said, that's not Now, given that 0:17:05.720 --> 0:17:09.520 he undeniably did business this way, I disagree, but that's 0:17:09.520 --> 0:17:13.640 beside the point. The resistance was successful. Now, the project 0:17:13.680 --> 0:17:16.880 wasn't killed entirely, but it has been massively scaled back 0:17:16.920 --> 0:17:19.560 and may yet die on the vine. The fact that 0:17:19.640 --> 0:17:22.400 resistance to this huge project and many like it, has 0:17:22.440 --> 0:17:25.320 caused so much damage to the industry has prompted a 0:17:25.440 --> 0:17:28.720 scale response. In late May and early June of twenty 0:17:28.760 --> 0:17:30.879 twenty six, I noticed a drip and then a flood 0:17:30.920 --> 0:17:33.640 of new articles and viral content critical of the anti 0:17:33.720 --> 0:17:36.840 data center protests. One good example of this would be 0:17:36.880 --> 0:17:39.520 a June tenth article on the website bar Charted by 0:17:39.640 --> 0:17:42.880 Nash Riggins. It was titled data centers get a bad 0:17:43.000 --> 0:17:45.639 rap for using too much water and energy. It turns 0:17:45.640 --> 0:17:49.199 out al meends suck up even more now. In this 0:17:49.400 --> 0:17:53.000 incredibly snotty little shit of an article, Nash is reporting 0:17:53.040 --> 0:17:56.400 on the fact that a bunch of pro AI influencers 0:17:56.400 --> 0:18:01.320 had with shocking coordination, started posting online responding to complaints 0:18:01.320 --> 0:18:04.120 about data center water usage by pointing out how wasteful 0:18:04.119 --> 0:18:06.040 it really is to grow almonds. You know, the almond 0:18:06.040 --> 0:18:08.120 industry uses even more water than AI does. 0:18:09.000 --> 0:18:11.159 And you know, if you want a good response to that. 0:18:11.359 --> 0:18:13.600 Just say, sure, put them out of business. 0:18:13.880 --> 0:18:14.360 Fuck them. 0:18:15.400 --> 0:18:18.040 Another popular article in this genre was a blog post 0:18:18.080 --> 0:18:20.840 by Andy Masley. He could taque the argument that AI 0:18:21.000 --> 0:18:24.000 was particularly wasteful by making arguments like this, per an 0:18:24.000 --> 0:18:27.480 interview with him on asfamily dot Com, almost all of 0:18:27.480 --> 0:18:30.080 that one point seven trillion gallons of water per year, 0:18:30.119 --> 0:18:32.840 specifically water that is withdrawn in the return to the source, 0:18:32.920 --> 0:18:35.720 Masley told me, referring to a widely cited estimate from 0:18:35.800 --> 0:18:38.960 Karen House Empire of AI. The book claimed surging AI 0:18:39.040 --> 0:18:41.560 demand could use up to one point seven trillion gallons 0:18:41.560 --> 0:18:44.400 of fresh water globally by twenty twenty seven, or half 0:18:44.440 --> 0:18:47.560 the water annually withdrawn in the UK. Masley says the 0:18:47.560 --> 0:18:50.359 word use in that sentence is misleading. Data centers do 0:18:50.440 --> 0:18:53.000 evaporate most of the water they use on site for cooling, 0:18:53.160 --> 0:18:55.480 but on site use represents only about twenty percent of 0:18:55.520 --> 0:18:58.080 their total water footprint. The other eighty percent comes from 0:18:58.119 --> 0:19:00.760 power plants that heat water to spin turbine. That water 0:19:00.840 --> 0:19:04.440 is then returned, mostly unchanged, to its source. Masley's viral 0:19:04.480 --> 0:19:06.280 post got him interviewed by The New York Times and 0:19:06.320 --> 0:19:08.199 earned a correction from the author of the book, but 0:19:08.240 --> 0:19:11.080 his critiques about specific claims made there shouldn't be seen 0:19:11.080 --> 0:19:14.440 as having any overall legitimacy to the broader argument about 0:19:14.520 --> 0:19:16.879 data center water use. For one thing, he acts as 0:19:16.960 --> 0:19:20.600 if all data centers function the same way basically, both 0:19:20.600 --> 0:19:22.240 in terms of how they are powered and in terms 0:19:22.240 --> 0:19:24.240 of how they do cooling, and thus you can make 0:19:24.280 --> 0:19:26.560 the kind of assumptions about water use that is making 0:19:26.800 --> 0:19:30.240 and you just fucking can't now. The Environmental and Energy 0:19:30.280 --> 0:19:33.320 Study Institute has published an expanse of analysis of data 0:19:33.359 --> 0:19:36.080 centers and water consumption. They point out that this is 0:19:36.119 --> 0:19:38.840 all much more complex than Masley's quote might lead you 0:19:38.880 --> 0:19:42.960 to believe, and I'm going to read from that article quote. 0:19:43.040 --> 0:19:45.480 In the context of data centers, water consumption refers to 0:19:45.480 --> 0:19:48.200 the amount of water withdrawn from blue or gray sources 0:19:48.200 --> 0:19:51.280 minus the water discharged by the centers, primarily warm water 0:19:51.400 --> 0:19:54.240 left over from cooling the it racks. The consumed water 0:19:54.320 --> 0:19:56.800 is generally the water that evaporates or is otherwise taken 0:19:56.800 --> 0:19:59.680 out of immediate human usage. Withdrawal of fresh water from 0:19:59.760 --> 0:20:03.000 low streams or underground aquifers may lead to aquifer exhaustion. 0:20:03.240 --> 0:20:07.160 Particularly in water stressed areas. Approximately eighty percent of the water, 0:20:07.240 --> 0:20:10.359 typically fresh water with drawn by data centers, evaporates, with 0:20:10.400 --> 0:20:14.080 the remaining water discharge into municipal wastewater facilities. The large 0:20:14.119 --> 0:20:17.080 volume of wastewater from data centers may overwhelm existing local 0:20:17.160 --> 0:20:20.200 facilities which were not designed to handle such a high volume. 0:20:21.080 --> 0:20:23.000 And again, one of the things that Masley tends to 0:20:23.040 --> 0:20:26.280 ignore is that even when you're putting this water back 0:20:26.320 --> 0:20:28.400 into the system quote unquote, that doesn't mean it's going 0:20:28.440 --> 0:20:30.440 back to the aquifers that it was pulled out of. 0:20:30.520 --> 0:20:32.520 It doesn't mean that it's going back into the same way, 0:20:32.560 --> 0:20:36.680 which can still cause massive issues. Now, how water efficient 0:20:36.760 --> 0:20:39.199 a data center is is going to vary depending on 0:20:39.200 --> 0:20:41.840