Goodbye TikTok, Ni Hao RedNote? + A.I.'s Environmental Impact + Meta's Masculine Energy

“This is truly one of the funniest and most unexpected stories of the young year so far.”

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  Support for this podcast comes from Dell Technologies. It's time to put AI to work. Is your business ready? The challenge isn't getting started with AI. It's scaling beyond pilots while maintaining control of your data, governance, infrastructure, and costs. Dell AI Factory provides the integrated foundation you need to scale AI securely and efficiently with one trusted partner, Dell, AI on your terms. From idea to impact. Learn more at dell.com/yourwaytoAI. That's dell.com/yourwaytoAI.

• 0:34Kevin Roose?

  Hi, Casey.

• 0:35Casey Newton?

  Hello, Kevin.

• 0:36Kevin Roose?

  How's it going?

• 0:37Casey Newton?

  Well, you know, it's going all right. There's a lot of sad news in the world right now. Of course, you know, my thoughts are with everyone affected by these LA wildfires. And whenever something bad has happened this month, I just think, I can't believe they're doing this to us during Dry January. You know what I mean? I'm doing Dry January this year. And, you know.

• 0:58Kevin Roose?

  How's it going?

• 0:59Casey Newton?

  Well, the unfortunate news is that it's going fantastic. I had sort of assumed that, you know, each day I would wake up thinking like, oh, you know, it would be nice to crack open a beer with my friends tonight or something. But instead, I'm just like, I feel so incredibly well rested.

• 1:15Kevin Roose?

  Oh.

• 1:15Casey Newton?

  So that has been an interesting learning.

• 1:18Kevin Roose?

  Do you plan to continue it beyond January?

• 1:20Casey Newton?

  Well, yeah, I mean, I don't think I'm going to go 100% dry, but I have actually been thinking about like, what if I did like a Dry February, too? So I don't know. I don't know. Changes could be on the horizon.

• 1:28Kevin Roose?

  What would happen to your wine tap?

• 1:30Casey Newton?

  I might have to replace it with seltzer or something dramatic. It could be crazy.

• 1:36Kevin Roose?

  The horror.

• 1:42Casey Newton?

  I'm Kevin Bruce, a tech columnist at The New York Times. I'm Casey Noon from Platformer. And this is Hard Fork. This week, TikTok enters its final hours in the United States, and Americans are flocking to a new Chinese app. Then, hugging faces Sasha Lucioni joins us to help us understand the environmental impact of AI. And finally, my idea is to bring more masculine energy to Meta. Type a man up, Kevin.

• 2:05Kevin Roose?

  Let's go, bro.

• 2:17Kevin Roose?

  Well, Casey, the major, major news in tech this week is that we might actually get a TikTok ban.

• 2:24Casey Newton?

  It's true, and I feel like we've probably started at least seven segments of the show that way over the years, Kevin, but this really is looking like the final, final, final version of yes, TikTok could be banned.

• 2:36Kevin Roose?

  Yes, this is TikTok ban v. 12, final, final, use this one, docx.

• 2:42Casey Newton?

  That's right.

• 2:43Kevin Roose?

  We have talked so much about how we came to this point, the law that Congress passed to ban TikTok.

• 2:51Casey Newton?

  PAFaka.

• 2:51Kevin Roose?

  PAFaka, technically to force it to sell. That law is supposed to go into effect on January 19th, which is this Sunday, the day before Donald Trump's inauguration. And as of this taping, it appears that barring some last-minute intervention, TikTok's time as one of the most popular social media apps in the United States may be coming to an end.

• 3:15Casey Newton?

  And you want to talk about a roller coaster. I mean, the number of twists and turns this story has taken, from when Donald Trump tried to ban TikTok in his administration, to Joe Biden putting the brakes on that and exploring other alternatives, to that process going off the rails and Congress passing the first piece of tech-related legislation in the past decade to make this thing happen, then Donald Trump reverses himself, says, I'm going to save the app, then that doesn't work, and finally, last week, it all ends up at the Supreme Court. So, yes, if you wonder why we keep doing this segment over and over again, it's because very few things have changed as much in the past half a decade or so as the fate of TikTok.

• 3:58Kevin Roose?

  It's truly nuts. So, when we started recording this segment on Wednesday, the Supreme Court had not made a decision yet on this case, but as of Friday, the justices did issue an opinion upholding the law and denying ByteDance's challenge. So, Casey, we're going to talk about all this, but let's start by analyzing a little bit of what happened at the Supreme Court and then talk about where we go from here.

• 4:20Casey Newton?

  Okay.

• 4:21Kevin Roose?

  So, last Friday... The Supreme Court of the United States heard oral arguments in TikTok v. Merrick Garland. Merrick Garland, of course, is the Attorney General. And this was a lawsuit brought by ByteDance to try to get the Supreme Court to step in at the last minute and overturn this law and basically say ByteDance is allowed to own TikTok in the United States, or at least to delay the ban from going into effect. And did you actually listen to the oral arguments? I know you're a noted Supreme Court watcher.

• 4:49Casey Newton?

  I did not listen to the Supreme Court arguments live, Kevin, but I did catch up on them later. And I have to say, I was surprised by the tenor of the discussion.

• 4:59Kevin Roose?

  Say more.

• 5:00Casey Newton?

  Well, I sort of just thought that we would see a bit more deference to the First Amendment that we got. Justices seem to think that the speech issues involved in the case were not relevant because the way that the law is written says that as long as ByteDance divests this app, all of the speech on the app remains, right? So they sort of swept that away. And again, if this is somewhat surprising, we're talking about it is only because the court did not have to hear this case, right? The last time we talked about this potential ban, we said, hey, look, the court could just defer to the lower court, not hear this argument and just let the law go into effect. But instead, they did take it up, which made some people think, aha, maybe they have something to say about it.

• 5:43Casey Newton?

  But I did sort of predict at the time that really you just had a lot of justices that kind of wanted to like, give TikTok the finger, and that does seem like what happened last week.

• 5:51Kevin Roose?

  Yeah, so let's live in the world right now in which all of this goes as it looks like it's going to go. ByteDance is forced to comply with this law. What happens next?

• 6:04Casey Newton?

  So ByteDance has said that it will block access to the app for Americans beginning on Sunday. So you will open your TikTok app and it will not refresh. It will not be populated with new content. Now, under the law, ByteDance doesn't actually have to do that. The way the law is written, it is actually intended to force Apple and Google, the two big app store providers, to remove TikTok from the App Store. But ByteDance got ahead of that and said, we're just gonna Shut the thing off, which some people have speculated is essentially a move to get some leverage because you're going to make so many Americans so mad that maybe that might generate some, you know, political goodwill for the company.

• 6:44Kevin Roose?

  Right. I'm imagining something like what Uber used to do back in the day when it was fighting with governments, where you'd open your app and you'd see a little pop up that said like, we don't have Uber in your city. If this makes you mad, like, here's some numbers to call to contact your local legislators and yell at them about it.

• 6:57Casey Newton?

  Yeah. And, you know, ByteDance has tried stuff like that in the past in the United States, and it has backfired. But, you know, at this point, what does it have to lose?

• 7:04Kevin Roose?

  Right. So the other possibility for an outcome here, I suppose, is that ByteDance could agree to sell TikTok, that it would divest as this law was intended to, to force them to do, and that that would be how TikTok survives. Now, we should say, I think time is running out for any kind of deal like that. But maybe we should run down a few of the possible ways that this could end with a new owner of TikTok in the United States, rather than the app just going dark.

• 7:33Casey Newton?

  Let's do it.

• 7:33Kevin Roose?

  So, one group that has lined up to say that they are interested in possibly acquiring TikTok is a group of investors led by the billionaire Frank McCourt, who is the former owner of the LA Dodgers. He sent ByteDance a letter last week expressing his interest in acquiring TikTok. He said that he would acquire it even without the algorithm that determines what people see in their For You pages. And he said that he would cobble the money together for the sale from private equity funds and other ultra wealthy investors, including Kevin O'Leary, the investor who goes by Mr. Wonderful on Shark Tank.

• 8:08Casey Newton?

  Sharks, I'm coming before you today because I'd like to buy TikTok from the Chinese Communist Party. That would be a great episode. I'd watch it.

• 8:19Kevin Roose?

  Yeah, it sounds like this is a hastily arranged sale meant to avert a catastrophic legal outcome. And for that reason, I'm out.

• 8:29Kevin Roose?

  Another potential buyer is the YouTuber celebrity Mr. Beast, who said that he had had billionaires contacting him about buying TikTok after he posted about it on X. I don't know how serious this is, but I think we should just say Mr. Beast would be, I think, a pretty good owner of TikTok.

• 8:47Casey Newton?

  I mean, I don't know. After I heard what happened on his Beast Games show on Amazon Prime, where many of the contestants were lucky to survive, I'm not sure we want this man running a large tech platform, Kevin.

• 8:57Kevin Roose?

  Well, I don't think we have to think about it that hard, because I don't think it's gonna happen.

• 9:00Casey Newton?

  Me either.

• 9:00Kevin Roose?

  But the real wild card here, the one that I actually do take somewhat seriously, is that on Monday Bloomberg published a story suggesting that the actual preferred acquirer for TikTok might be Elon Musk. The Wall Street Journal shortly afterward put out their own reporting with largely the same information. According to the Bloomberg article, senior Chinese officials had already begun to debate contingency plans for TikTok as part of an expansive discussion on how to work with Donald Trump's administration, one of which involves Musk. This is according to anonymous sources who asked not to be identified, revealing confidential discussions.

• 9:38Casey Newton?

  Yeah, I bet they did.

• 9:40Kevin Roose?

  So Casey, what do you think of this theory that Elon Musk could acquire TikTok?

• 9:46Casey Newton?

  You know, every once in a while, something happens in the universe, and I think, was this done to upset me specifically? That's how I felt when I read the story that said that Elon Musk was gonna take over a second beloved social platform in the United States and, you know, presumably apply his signature brand of content moderation and other fun tricks to the app. So, you know, I have to say, Kevin, of everything that has happened in the TikTok story so far, this truly might be the craziest because of the different players here, who we think knows what, what has been said about it. I can absolutely see a world where this is plausible. I can just as easily see a world where this is a nothing burger,

• 10:31Casey Newton?

  and we're just gonna have to get a little bit more information. But what was your reaction?

• 10:34Kevin Roose?

  So I was somewhat skeptical when I first. Saw this in part because TikTok came out right away and said that this was pure fiction. But it was written in such a way that it felt like the the message of the the stories that I read was that actually TikTok may not be involved in these discussions, right? This may be happening at the level of the Chinese government, who is sort of deliberating about what to do. And that's very telling. Obviously, there's this theory that I subscribe to and that I think a lot of people subscribe to, that ByteDance is not really in control of its own destiny here, because there's this sort of, you know, there's this government control of all Chinese social media platforms, but especially this one, which seem strategically important to the Chinese government.

• 11:18Casey Newton?

  And under Chinese law, ByteDance would need the government's permission to divest TikTok. So that's a very real thing.

• 11:24Kevin Roose?

  Right. I can also see this making sense for Elon Musk. He's said before that he wants X to operate more like TikTok. It's obviously a social network that's very popular. They've obviously cracked the code on sort of algorithmic presentation of content. I also think that that he would be a a more palatable American acquirer for the Chinese government than any other potential acquirer.

• 11:49Casey Newton?

  So yeah, so sketch this case out, like, what would the Chinese government have if Elon owned TikTok?

• 11:55Kevin Roose?

  So I think one thing that we know about Elon Musk is that he does a lot of business in China, right? Tesla has a lot of operations and entanglements in China. Elon Musk has also been very deferential to the Chinese government when it comes to doing things that he needs to do to continue operating his companies in China. And I think if you're the the Chinese government looking at, well, OK, which American acquirer would allow us to continue to exert influence over the user base of TikTok in America? I think they they have points of leverage on Elon Musk that they do not have over, say, Frank McCourt.

• 12:33Casey Newton?

  Right. On one hand, Kevin, it seems a little crazy to me that, you know, the Chinese government thinks essentially we will turn Elon Musk into like kind of a soft Chinese agent to like do our bidding in the United States like that seems a little bit far fetched. On the other hand, if you look at Musk's behavior over the past few years, which I think has been really erratic in a lot of ways. He's always extremely careful about what he says about China. He truly almost never says anything remotely critical of that government. And, um, you know, if you're the Chinese government, maybe you look at that and you appreciate it and you think, yeah, sure, let that guy take it.

• 13:10Kevin Roose?

  Totally. So those are the sort of acquisition scenarios, but I think we should say like, I don't believe any of these are likely to happen by the time this deadline hits. I think that no matter what, acquirers might be interested in buying TikTok. A ByteDance does not seem interested in selling it. B, the Chinese government does not seem interested in letting ByteDance sell it. And C, I don't think anyone could put together a deal quickly enough to actually get this done by the 19th.

• 13:37Casey Newton?

  Yeah, that's right. Also, I just want to like acknowledge the American centricness of this conversation. TikTok is available in many other markets where it is not banned. And while the United States is a very lucrative market and a great place to run an e-commerce operation, the way that TikTok is doing. It is operating in many other countries in the world. I read this week that its largest market right now is actually in Indonesia. There are more users in Indonesia than in the United States. And so if you're ByteDance, just from a pure dollars and cents perspective, you may just look at this and think, we can actually make more money operating in the countries where we already exist and just sort of give up on America and we'll be fine. So that's another scenario here.

• 14:12Kevin Roose?

  Yeah. So now let's talk about the user response because this has been truly wild. So I'm sure, I know you are not like a TikTok addict, but presumably.

• 14:21Casey Newton?

  Despite my best efforts.

• 14:22Kevin Roose?

  Yes.

• 14:23Casey Newton?

  I tried to get addicted to it on this very show.

• 14:25Kevin Roose?

  You did.

• 14:25Casey Newton?

  Yeah.

• 14:26Kevin Roose?

  But the people on TikTok are starting to, I would say, panic over the impending possible loss of their favorite app, their favorite platform. And a lot of content creators on TikTok are starting to try to bring their audiences over to other platforms like YouTube or Instagram. Some are saying they're going to use VPNs to get around this ban. And the most fascinating development to come out of all this, in my opinion, has been that there is this now, this new trend of TikTok refugees downloading a Chinese. ...social media app called Zhao Hongxu or Red Note.

• 15:02Casey Newton?

  Yes, this is truly one of the funniest and most unexpected stories of the young year so far.

• 15:09Kevin Roose?

  Yes, so as of yesterday, at least when I checked, the Zhao Hongxu app was the number one free app on the iOS App Store. It has gotten a ton of downloads from people who are saying, basically, screw the US government, screw this TikTok ban, I'm going to protest this by going over to this explicitly Chinese app that does not even have an English name in the App Store.

• 15:36Casey Newton?

  That's right. Now, obviously, I have installed this app, which I'm going to call Red Note, even though I believe that that's just kind of an American nickname for it.

• 15:44Kevin Roose?

  Yes, the literal translation is Little Red Book, which is also not subtle. That is also the name given to the book of quotations from Chairman Mao that was distributed during the Cultural Revolution.

• 15:56Casey Newton?

  So tell people what Red Note is, Kevin.

• 15:58Kevin Roose?

  Okay, so Red Note is essentially a TikTok-like app. It is not owned by ByteDance, but if you open it, you see a feed that looks very much like the For You feed of trending videos. It basically has the same platform mechanics as TikTok. And until this week, most of the content there was people in China speaking Chinese and talking about China.

• 16:21Casey Newton?

  Absolutely. And, you know, I downloaded it, I installed it, signed up for an account, and immediately started watching a bunch of videos from, you know, refugees, as they're calling themselves, from TikTok over to Red Note. And they seem like they're having a great time over there. But, you know, in addition to this migration, Kevin, what was truly so funny was there were so many posts on X of people bragging about how they were racing to share all of their data with a new Chinese app. They were posting screenshots of themselves with the Apple App Tracking Transparency Screen. You know, Apple sends you this big scary warning, "Hey, are you sure that you want to share your data?" And people are like, "Hell, yeah, I want to share it, brother. Like, I'll give you everything." I saw TikTok from this one girl who was like, "I would fly to China and hand my social

• 17:06Casey Newton?

  security number to Xi Jinping before I would ever use Instagram reels." That's where the user base is at, Kevin.

• 17:14Kevin Roose?

  Right. So I also thought I should probably download and install this Zhahongshu Red Note app just to see what all the fuss is about. And can I tell you the first three videos that I saw on my feed? Number one was a clip from Modern Family.

• 17:32Casey Newton?

  Which that is the kind of like, that is the sourdough starter of the modern video-based social network. It's just clips of Modern Family.

• 17:40Kevin Roose?

  Number two was a Chinese language clip of a dog at the vet having its anal gland expressed.

• 17:46Casey Newton?

  Oh, perfect. Express yourself.

• 17:49Kevin Roose?

  And number three was someone making latte art of Luigi Mangio, the suspect in the murder of the United Healthcare CEO.

• 18:01Casey Newton?

  So I think it just goes to show you how quickly you can create an American social network. It really, in just three videos, you've captured a shocking amount of the zeitgeist, Kevin.

• 18:09Kevin Roose?

  But I would say after I scrolled more and more, I did start to see these so-called TikTok refugee videos. These Americans who are coming over to Red Note from TikTok and basically trying to make sense of this new thing and sort of participating in almost a cultural exchange. So why don't we play a couple videos that have been making the rounds on Red Note? Let's do it.

• 18:30Speaker 03

  Hi, guys. I got sent here from TikTok and I was hoping that you guys can welcome me. I really like this app and I love the makeup. I tried to do it today. So, thank you.

• 18:45Speaker 04

  I'm not gonna lie to y'all, bruh. I can't read shit on this app.

• 18:52Speaker 04

  Somebody help me out. I need some followers too. So I say, go ahead, go ahead and hit that follow button too. I need that. Say, who else a TikTok refugee? Y'all let me know in the comments or something.

• 19:05Casey Newton?

  Xiaohongshu.

• 19:06Kevin Roose?

  So it inserts the little Xiaohongshu at the end of every video. That's like the watermark.

• 19:09Casey Newton?

  That's great, because now I actually feel like I can pronounce it.

• 19:13Kevin Roose?

  So Casey, what do you make of the Xiaohongshu Red Note trend?

• 19:18Casey Newton?

  I mean, I have to say, like, this is why I love Americans. Like, the absolute irreverence that they are bringing to this conversation, I find like such a refreshing change of pace, you know, so much of the discussion, and certainly I participate in this, is in terms of like, the rights involved, the equities, like, speech versus security, what are the national security implications? And Americans are truly just rolling their eyes out of the back of their heads with this discussion. And they're saying, oh, you want to ban one Chinese app? We are going to flock to another one, and I just think it's amazing.

• 19:52Kevin Roose?

  Totally. I mean, it's also very interesting to me that the response from American TikTok users is not the response that TikTok had hoped for, which was that Americans get outraged that their favorite social media app is disappearing and band together to like storm the streets and try to save TikTok and overturn this law. It's like, no, we're just gonna like pack up and move to another platform. I think it really speaks to the fragility of social media right now and the fact that like, these platforms are seen as somewhat interchangeable and commoditized. And so like, if one of them gets banned by the government, you just pack up and like move over to another one.

• 20:26Casey Newton?

  Yeah, although of course, there are several American apps that folks could have chosen to move over to. And I think it is extra funny that instead of doing that, Americans were like, no, find us something that looks exactly like TikTok, but is even more Chinese.

• 20:38Kevin Roose?

  Even more Chinese and even less, makes even fewer promises about data privacy.

• 20:44Casey Newton?

  Yeah, exactly. Now, you know, some people might ask, doesn't Puffaka ban Red Note as well? And my understanding is that no, Puffaka primarily applies to ByteDance and TikTok. But when it comes to other apps that are owned by a company in a country that the United States considers a foreign adversary, it is up to the president to decide that it represents a national security threat. And I imagine it's gonna be some time before Red Note is coming. To be seen that way.

• 21:15Kevin Roose?

  Right. So in addition to packing up and moving over to Shaohongshu, other TikTok users are participating in the Chinese spy meme. Have you seen this one?

• 21:25Casey Newton?

  I love this meme so much.

• 21:27Kevin Roose?

  OK, explain what's going on.

• 21:28Casey Newton?

  So people are saying goodbye to their Chinese spy. This is another sort of irreverent American joke. Basically, they're making fun of the fact that members of Congress are constantly saying that the Chinese government is using TikTok to spy on Americans. And the Americans are now just making videos saying, hey, I'm so glad we got to spend this time together, my Chinese spy, I will always remember you. And then you also have Chinese people, you know, and these could be, you know, Chinese people from all over the world, but they're making memes pretending to be the Chinese spies, saying, hey, I, you know, really loved spying on you for all these years. And, you know, maybe call your mother a little bit and, you know, this message goes out to Laura from New York and that sort of thing.

• 22:08Casey Newton?

  So, um, yeah, saying goodbye to my Chinese spy, I do think is one of the best TikTok memes of all time.

• 22:14Kevin Roose?

  Yeah.

• 22:14Casey Newton?

  Coming in hot right at the end.

• 22:15Kevin Roose?

  It's so good.

• 22:16Casey Newton?

  Yeah.

• 22:16Kevin Roose?

  So we've talked a lot in the past about the free speech implications about all this, about whether this is sort of the first of many conflicts between the US and China over emerging platforms. But Casey, like, where are you right now, days away from the likely end of TikTok as a major presence in US social media? What are you thinking about?

• 22:37Casey Newton?

  I mean, my feeling has been, and I've kind of gone back and forth on this, but where I netted out was, I do think there are good reasons for the United States to restrict foreign ownership of these kinds of apps, particularly from its adversaries like China. But I really hate the way that they went about it. And I worry about the implications for other speech platforms in the future. You know, people are saying, well, this one is really easy because of the Chinese control angle. But I don't know, you know, if, um, if this, this incoming administration decides it doesn't like a lot of the content on an American-owned app and says,

• 23:21Casey Newton?

  you know what, we're actually just going to make you change ownership. Same way that we did with ByteDance and TikTok. And now the Supreme Court has essentially rubber stamped that argument and said, yeah, there are no speech concerns because as long as you sell the app, all that speech can remain. You can imagine a lot of really dark outcomes for that kind of thinking. So, you know, I personally, as an older American who tried and failed to get addicted to TikTok, am not going to miss it that much day to day. But TikTok was an engine of culture that we are going to miss out on in this country. Those folks are going to have to find a new home. It really sucks for all of those creators. And so, yeah, I think there's going to be a lot of really sad fallout from this. What do you think, Kevin?

• 24:01Kevin Roose?

  So I still think there's a chance that Donald Trump decides to intervene and try to save TikTok in the United States. This week, we got some news that Cho Chu, the chief executive of TikTok, is going to be at Trump's inauguration, sitting with a bunch of other VIPs. And some people have interpreted that as Trump saying he supports TikTok and might try to save it. He obviously made promises about saving TikTok during his campaign. Obviously, a lot's changed since then. But I do think that he understands that a lot of young Americans care deeply about the fate of TikTok. And then maybe he can build some goodwill with those young Americans by stepping in at the last minute to sort of heroically save TikTok.

• 24:44Kevin Roose?

  Now, there are some different ways he could do that. He could instruct the Justice Department in his administration not to enforce the ban on TikTok. He could also try to arrange some kind of deal, potentially selling TikTok to Elon Musk or someone else that he, who he trusts and sort of say, that is enough of a divestment for me, that satisfies the requirements of Puffaka.

• 25:07Casey Newton?

  And he is, after all, Kevin, the author of The Art of the Deal.

• 25:11Kevin Roose?

  Exactly. So I do think there's a chance that Donald Trump sort of keeps TikTok around in some form after all. But I'm not sure about that, and I think it's equally plausible that TikTok actually does sort of go away and that it becomes this kind of free-for-all in the social media world as different companies race to hoover up the users who had previously spent hours a day on TikTok.

• 25:38Casey Newton?

  Yeah, and when TikTok was banned in India, we saw what happened there, which was that YouTube Shorts and Instagram Reels, which were Google and Meta's answers to TikTok, exploded in popularity. So, one way that you should be thinking about this is if this goes into effect, this is truly one of the greatest gifts for Google and Meta that you can imagine. And that is just really interesting, given the strong bipartisan feeling in Congress that Google and Meta specifically ought to be reined in and actually even broken up. I mean, when you think about who is on TikTok, it is the younger generation of Americans. So if Meta and Google can now go out and further entrench themselves into the lives

• 26:22Casey Newton?

  of Generation Z, they're going to have essentially monopolies over those folks, at least in terms of short-form video consumption, for the foreseeable future.

• 26:30Kevin Roose?

  Yeah, I think that's totally possible. I think there are probably a lot of executives at Meta who are licking their chops about this, who are very excited about the potential because their platforms, Facebook and Instagram, are aging. Facebook is for boomers and Gen X, Instagram is for millennials. And until now, Gen Z has been the TikTok generation. And if Meta can sort of suck up those users, it can sort of extend its dominance for another generation. But I think that the past week has made me less sure that that's going to be the outcome here. Because what we have seen on TikTok as this ban has approached is not people saying, "Oh, everyone move over to Instagram Reels." It's saying, "Let's

• 27:12Kevin Roose?

  move over to this obscure Chinese-language app that no one's ever heard of. That's how badly we don't want to be on Instagram." I think part of the Gen Z identity is about not just embracing TikTok as a platform, but rejecting the platforms that people older than you. And so I think it's equally plausible that those younger users do not go to Instagram reels or YouTube shorts, that they instead go to some new app that may have most of the features of TikTok, but is different in some way. Maybe we're finally going to get a new American social media app.

• 27:47Casey Newton?

  You know, I would love to believe everything that you're saying, and I think that it absolutely could come to pass. But I also think it's true that most members of Gen Z who have TikTok on their phone probably have Instagram as well. And it's just going to be really hard for them to avoid taking a look at that as they look elsewhere to get their fix. But at the same time, we're also seeing sort of separately from all of this, a boom in the Fediverse and people building on protocols. And that is rooted in the exact same frustration with these apps that are controlled by billionaires and giant faceless corporations. So I agree with you, there is a lot of frustration among all sorts of Americans on that point. And so who knows, maybe we do get an American-owned alternative to TikTok that is not YouTube or Meta.

• 28:35Casey Newton?

  When we come back, we answer one of the most common questions we get from listeners. What is the environmental impact of AI?

• 28:51Speaker 00

  Support for this podcast comes from Dell Technologies. It's time to put AI to work. Is your business ready? The challenge isn't getting started with AI. It's scaling beyond pilots while maintaining control of your data, governance, infrastructure, and costs. Dell AI Factory provides the integrated foundation you need to scale AI securely and efficiently with one trusted partner. Dell, AI on your terms. From idea to impact. Learn more at dell.com/yourwaytoAI. That's dell.com/yourwaytoAI.

• 29:25Speaker 05

  Building and scaling AI today can feel pretty chaotic, mostly because teams are juggling too many fragmented tools at once. That's where Outsystems comes in. Outsystems gives you one unified agentic systems platform to manage your entire portfolio of agents, apps, and workflows. You get to innovate at the speed of AI, but with the security, governance, and control your enterprise actually needs. And the results are already there. A major manufacturer used Outsystems to build a multi-agent system that analyzes blueprints. And a global bank cut their development time by 60%. Own your agentic future today at outsystems.com/nyt.

• 30:01Speaker 06

  AI spend is growing fast, but most businesses still can't see exactly where that money is going. Ramp gives businesses visibility and control across AI providers with real-time guardrails by team and model, helping ensure spend goes to the highest ROI areas. Stop guessing where your AI spend is going. Get started with Ramp. And if you're new to Ramp, you can get $250 off at ramp.com/hardfork. Go to ramp.com/hardfork. Ramp.com/hardfork.

• 30:34Kevin Roose?

  Well, Casey, for basically the entire time we have been making this podcast, we have gotten emails from listeners who want us to talk about the environmental impact of AI.

• 30:44Casey Newton?

  Yeah, this might be the question that we have gotten the most, that we have not yet devoted a segment to.

• 30:48Kevin Roose?

  Yeah, and I would say my own reluctance to talk about this topic on the show so far has been some insecurity on my part about not being an expert in climate science or the relevant information here. But also, it is very hard to get good and authoritative data about this subject in particular.

• 31:08Casey Newton?

  Yeah.

• 31:08Kevin Roose?

  It is just not something that there is a large body of reliable literature about.

• 31:13Casey Newton?

  Yeah, the companies that have the best data, by and large, are not disclosing any of that data. And so that means that a lot of what we talk about when we talk about the environmental impact of AI is based on estimates that may or may not be close to the mark.

• 31:26Kevin Roose?

  Yeah, but I'm sure you have observed, as I have, that the issues around the environment and AI have only gotten more important to people. This really came to a head last week when the wildfires started burning in Los Angeles. I saw so many people posting on social media about what they viewed as a link between AI use and the wildfires. And I'll just read you one meme I saw in my feed that was liked and shared millions of times. This was posted by a guy named Matt Bernstein, and I'll just read it to you. It said, "One search on ChatGPT uses 10 times the amount of energy as a Google search. Training one AI model produces the same amount of carbon dioxide as 300

• 32:10Kevin Roose?

  roundship flights between New York and San Francisco and five times the lifetime emissions of a car. We don't need AI art. We don't need AI grocery lists. We don't need AI self-driving cars. We don't need ChatGPT or Gemini or Grok or Dolly or whatever revolutionary technology already exists inside our own human brains. We need the Earth." And then below this meme was a picture of a blazing fire. So clearly, this idea has taken root in culture that there is some kind of link between the disasters that we are seeing in places like Los Angeles and the use of AI for basic everyday tasks.

• 32:48Casey Newton?

  Yeah. And I think today, we want to see what we can find out about how true some of the ideas in that post are.

• 32:56Kevin Roose?

  Yeah. So to shed some light on this very hot topic of AI and energy use, I realized that I just used light and heat. That was not intentional.

• 33:05Casey Newton?

  But we're going to hope we'd shed more light than heat in this discussion, Kevin.

• 33:09Kevin Roose?

  Yes. Today, we are talking with Dr. Sasha Lucioni. She is an AI researcher and the climate lead at Hugging Face, which is an AI company that offers tools to developers for building AI models. She has been researching and talking about AI's environmental impact for many years, and also developing tools to help developers understand the impacts of their own systems on the environment.

• 33:32Casey Newton?

  Yes, and Kevin, this might be a good time to dust off my shiny new disclosure, because when we talk about AI issues, I will sometimes remind people that my boyfriend is a software engineer at an AI company called Anthropic. My full ethics disclosure is that platform or not, news slash ethics.

• 33:47Kevin Roose?

  And my fast disclosure is that I work at the New York Times Company, which is suing OpenAI and Microsoft over issues of copyright violations.

• 33:52Casey Newton?

  Perfect. All right. Let's bring in Sasha Lucioni.

• 34:01Kevin Roose?

  Sasha Lucioni, welcome to Hard Fork.

• 34:03Speaker 07

  Thanks for having me.

• 34:04Kevin Roose?

  So I'm very excited to have this conversation. This is one we've been looking forward to for a while and are frankly overdue in having. And I want to start by reading you an email that we recently got from a listener. This comes from a listener named T. Morris, and it says the following. "As a tech content marketer, I feel increasingly conflicted about using AI. On the one hand, it's been an amazing writing partner for big tasks like brainstorming and editing tech articles and smaller copywriting tasks like drafting social media posts. On the other hand, I see climate disasters like the North Carolina floods and LA fires linked with the amount of water and natural resources it takes to sustain AI infrastructure

• 34:47Kevin Roose?

  and feel myself rationing my AI use, questioning whether the time saved is worth the environmental trade-offs. How do I navigate this new world where AI is everywhere while staying true to my environmental values?" So Sasha, we'll dive into some of the specifics in just a minute, but I want to just start with this question from our listener. What advice would you give T. Morris?

• 35:10Speaker 07

  I'm generally very skeptical of like individual culpability when it comes to the climate crisis. Like, yes, of course, we all contribute, but I think that we're all also part of systems and we have professions that require usage of technology. Some people drive for a living and we can't spend our time feeling bad. I'm much more of a fan of requiring accountability from companies and requiring transparency because I think that especially around climate change, but also a lot of aspects of society, we just don't have the numbers to make informed decisions. And that doesn't mean you need to care, but you should have the information necessary for caring. So I'm more about like, ask for accountability, ask for transparency when using these technologies instead of like kind of psyching yourself out about them.

• 35:55Kevin Roose?

  Got it. So I thought a one way to sort of frame this discussion would be to split it into essentially two parts, the the micro and the macro, micro being this question of like, what do we know about the environmental impact of AI at the level of the individual user, the individual question that you might ask to ChatGPT or Gemini or Claude and getting a response to that. And then macro being this larger question of like, what is the AI sector's energy footprint more broadly? What do we know about where all the energy is coming from to run these very powerful models? And what can we do as sort of a society

• 36:35Kevin Roose?

  and as big corporations to position ourselves better for the future? And with your permission, Sasha, I wanted to start with the micro. So one of the statistics that people will often throw out when talking about the energy demands of AI is this figure that a ChatGPT query or something like it costs somewhere in the neighborhood of 10 times more energy than a traditional web query on something like Google. Now, I asked Google about this figure and they wouldn't say exactly how much energy it takes to query Gemini versus to run a traditional web search, but they did say that those numbers are much larger than what they've seen internally. But Sasha, where did that figure come from? And what do we know about how accurate it is?

• 37:18Speaker 07

  I think the initial Google search query is actually pretty old. And it was part of a of a study to like greening the web kind of type situation. And they made an estimate. And once again, they didn't really have the numbers, but they they tried to extrapolate. And then for ChatGPT, it was a similar kind of assuming that somebody is querying a model that is running on this type of hardware and assuming that the latency is X and blah blah. And they kind of extrapolated that. Like, there are other models that do similar things. So maybe even if you don't know exactly ChatGPT, you know, inherently, you have other models that will do similar tasks. And so you can get a range. And I think that that range is more interesting than trying to like to to chase down the exact number and compare the two. And also, it's probably not a single number anyway. And so that's why it's so hard to like pin down this number. And that's why it's going to be always possible

• 38:03Speaker 07

  for them to say, oh, no, that's not the number. That's not the exact number.

• 38:06Casey Newton?

  Right, and I think that your point earlier that what one of the things that we need on this subject is just a lot more transparency is really well taken. I know that Google has folks who work on climate issues, but I'm curious, as like you look across the industry, maybe it's some of the newer, smaller AI labs, or just, you know, I don't know, companies other than Google, do you get the sense that people are paying attention to this, that they are taking these sort of measurements, that they even have a sense of like, the per query energy usage of one of their products?

• 38:39Speaker 07

  Definitely, because unlike Google, or Microsoft, or any of the big tech companies, usually smaller companies are a lot more compute restrained. So they're doing more with less because they have to. They don't always come at it from a sustainability perspective. They're not like, oh yeah, we want to protect the planet. But there is a part of that. It's like frugality. It's like, we want to be more efficient because we only have 100 GPUs to work with.

• 39:02Casey Newton?

  Right. I mean, and this seems like maybe sort of one positive thing that I'm hearing so far, is that it sounds like the incentives for all of these companies are to get the amount of compute and energy that they are using over time down as quickly as they can.

• 39:17Kevin Roose?

  Another claim that you often hear from people who are worried about the environmental impacts of using AI on a micro or personal level, is about water use. There's this statistic, I'm sure you've seen it around, that using an LLM is like pouring out a bottle of water, or half a liter of water, I've seen going around. Where did that figure come from? And why do these AI models need water? And is that statistic true?

• 39:43Speaker 07

  So that that paper is kind of, once again, an extrapolation. It takes some of the work that I did about an open source model, where we measured how many kilowatt hours of energy were being used to query it. And essentially, what happens in data centers is that they have an amount, like a liter of water per kilowatt hour of energy. I mean, it's like a water efficiency, they call it water use efficiency. And essentially, depending on

• 40:06Casey Newton?

  Because, sorry, why do they need water in the data centers?

• 40:09Speaker 07

  Mostly for the cooling. This hardware heats up. I don't know if you've ever visited a data center. If you can, I highly recommend it. It is like an overwhelming experience. The noise, the heat, and just like the general like buzz of electricity is pretty overwhelming. Anyway, so you need a lot of cooling. And essentially how that's usually done is with water cooling, like you pump in cool water. And there's a bunch of pipes and it goes through all of the hardware and then it either a part of it evaporates completely and a part of it has to be cooled down before either reused or put back into into nature or whatnot. And so that whole process is hugely water consumptive. And of course, it's not like not all the water evaporates, but a fair amount of it does just because the hardware heats up so much.

• 40:50Speaker 07

  But once again, it's so I go back and forth on this a lot, like whether putting out statistics like this on that are based on estimates or not is, I guess, useful for the conversation. Because on one hand, it's really easy for the companies to say, no, you guys are you guys are tripping. That's not at all the the true number, which, you know, then it kind of cuts the conversation. And on the other hand, they do become like urban legend. And so now I hear this 500 milliliter per conversation number a lot. And it's like, well, actually, it'll depend on so many different things. So it's definitely not systematically 500 milliliters, but it is a non negligible amount of water. And depending on where the data center is located, that can become an issue. So we've seen places where the data centers have put a strain on like

• 41:33Speaker 07

  the towns around them that have water shortages because the water is being pumped into a new data center that has been, you know, powered up.

• 41:41Casey Newton?

  So when you and you know, you said earlier, understandably, that you're not a huge fan of thinking about these issues at the individual level. I'm still curious when you are considering your own personal use of AI, where water usage fits into things like, is that for you a reason to send a few queries to chat GPT or an equivalent?

• 42:01Speaker 07

  I'm in general such a like I don't use AI that often, I mean generative AI that often. The one use case that I found was really kind of something that actually is useful in my life is when I read an article or a research paper, like putting in the abstract and getting a fun title. Like, I'm so bad at generating fun titles, but ChatGPT is really good and, you know, it can come up with like puns and stuff like that. But what really kills me is is like people who switch to generative AI for things that don't really need it. Like, like my my pet peeve example is is calculators. People use ChatGPT as a calculator now, and that's really like, it's really terrible. Like, you really don't need it. Not only is it bad at arithmetic, like it's literally not made to do math,

• 42:43Speaker 07

  but it's also like orders of magnitude more energy and a crazy amount of water for something that, you know, doesn't need water.

• 42:49Casey Newton?

  Well, I have to say, I'm gonna, you know, I'm going to admit something, which is I've talked before on the show about how I use this app called Raycast, which is plugged into to OpenAI's model. And I can just summon it on my keyboard with command space, and I do probably ask it four or five questions a day. And it is, I am definitely using it for things I can Google, like, you know, one of my toxic traits is I'm curious how old people are. And so sometimes I'll just be like, how old is Billy Crystal or whatever?

• 43:15Speaker 07

  Do you check the answers and they're all accurate?

• 43:17Casey Newton?

  It's it's not that I check the answers, it's that I don't really care that much. So when when the LLM says, you know, that Billy Crystal is, you know, I don't know, 70 or whatever he is, I'm like, yeah, that's that's the right ballpark, you know? And then I move on. I'm I'm realizing that I sound very silly as I'm saying this, but I'm saying it because I suspect other people may be doing the same thing. And I think there is a case that, you know what, maybe I should actually like Google that in part for environmental reasons.

• 43:41Kevin Roose?

  I want to just sort of, you know, embody the other side of this, because what I'm what I'm hearing from you is like, generative AI is not useful enough in many cases to justify the energy costs of engaging with an LLM. And I'm a person who uses AI every day. I generally find it quite useful in my life. I use it to accomplish a lot of tasks that I could not use equivalent tools for. I don't just run like how old is Billy Crystal searches over and over again.

• 44:12Casey Newton?

  To be clear, I only ran it once.

• 44:15Kevin Roose?

  And I would say that my own, you know, usage of this is to do new things that I couldn't do before mostly. And I think if people don't find generative AI useful, they shouldn't use it. But if people do find it useful but are worried about the environmental costs, I'm just not entirely convinced that we're thinking about the costs of AI in the in the sense of energy at the right scale. I recently was reading a Substack post by a guy named Andy Masly, where he basically broke down the best data and estimates we have about the environmental costs of using AI. And he compared it to some other activities like sending emails, or streaming a video on Netflix, or driving a car a very short

• 44:59Kevin Roose?

  distance. And basically, what he found is that compared to all these other activities, the energy required to to to generate an answer on ChatGPT or a similar system is just infinitesimally small. That if we are worried about our own personal environmental footprint, we could do much more to help the environment by cutting out meat from our diets, or by taking fewer trips in cars or on airplanes. And basically, the argument that he made that I am tempted by is that all of this sort of talk about personal responsibility is just neglecting to look at AI use in the context of all the other things that we do in our lives that require energy. And I'm wondering, Sasha, what you make of that argument?

• 45:40Speaker 07

  I mean, it kind of builds upon what I said at the beginning. But like, in general, when you talk to people around the issue of climate change and mitigation, it's like we're bound by the structures in which we operate and live and, you know, the constraints that we have. So of course, I'm not going to be like, oh, yeah, don't take that plane to take a well-deserved vacation and, you know, spend your time worrying about climate change, because that's not a productive, you know, state of mind. But on the other hand, we can make decisions with the environment in in our minds. So for example, like nowadays, a lot of people have ChatGPT open as the de facto source of information on the internet. And I do think that, yes, of course, little by little, that the individual like energy consumption of each query is not that much. But if we start using it as

• 46:24Speaker 07

  Literally like our rubber duck and our bouncing board and our companion. And then people also will use chat GPT to build tools, right? Nowadays, people are building like therapists and and whatnot companions using. And and then like that incrementally becomes a deal. Personally, I try to focus on like a like a specific task you want to do. For example, searching the internet or or answering a question. And then comparing like what you would use like option A and option B. And then what's the difference? And then it's up to you to decide whether that difference is worth it based on the advantage that the technology gives you. But I don't think it makes sense to compare like meat and email or Netflix and taking your car. Because I feel like they're like incomparable actions.

• 47:06Casey Newton?

  Right, like people aren't choosing between like, well, should I drive to work today? Or should I ask chat GPT?

• 47:10Speaker 07

  Yeah, exactly. So so so I feel that that like I understand where he's coming from in in his argument. But I don't I don't feel that that helps us make choices any better. It kind of makes us feel bad all around.

• 47:20Kevin Roose?

  So Sasha, can I try to sort of summarize what I'm hearing from you on the point of individual use, the sort of micro question about the environmental impact of AI? What I'm hearing you say, I believe, is that the individual costs of using L L Ms may not move the needle on climate one way or the other, but that people should be conscious of what they are using AI for and maybe use the the smallest model that will allow them to get the task done that they are looking to do. And that maybe we shouldn't be tearing our hair out over people using chat GPT if they're using it to do stuff that is genuinely useful to them.

• 48:04Kevin Roose?

  Is that an accurate reflection of your sentiment?

• 48:07Speaker 07

  It's a great reflection, and I think that often we forget our our power as consumers and users of technology. And I think that putting pressure on companies and being like, hey, we care. We want this number. Stop like bullshitting us like you have the number somewhere of of the average energy. You know, even if it's not a single number, if it's a range, give us the range. And then we'll make our informed decisions because like people are more and more aware of like relative comparisons, like, you know, a mile driven in a car or like a steak or whatever. Like, we need to add, even if it's a range, AI to those options that people have so that they can actually make informed decisions. We can, we should stop just like feeding them shit and keeping them in the dark.

• 48:44Kevin Roose?

  Right. Okay. So that is the sort of micro picture of the AI energy story. Now let's talk about the macro. There was recently a report just last month from the Lawrence Berkeley National Laboratory about the power that is currently needed to run the data centers in this country and the power that will soon be needed as the AI boom sparks demand for more and more of these data centers. This report said that between 2018 and 2023, the power to run data centers around the US went from 1.9% of total annual electricity consumption to 4.4%, more than double. And this report estimated that the energy demands of data centers,

• 49:29Kevin Roose?

  of which AI is a major part, will continue to increase over the next few years and could by 2028 make up between 6.7% and 12% of total US electricity consumption. So Sasha, just let's zoom out a little bit and talk about the energy needs of the AI industry as a whole. Where are we? Do these companies know where they are going to get all this energy to build these incredibly powerful AI models?

• 49:55Speaker 07

  Yes and no. I mean, we currently have a certain infrastructure, but the problem is, is the growth of the infrastructure is kicking into high gear. And so what's interesting is that the big tech companies are the largest purchasers of renewable energy credits, which are kind of like offsets for energy. And also they make a lot of power purchase agreements, which are essentially ways of kind of promising to buy energy, especially renewable energy into the future. So they've been kind of, I'll give them that, that they've been actually on top of things. But then this year, I mean, this past year, both Google and Microsoft actually put out reports saying that they're not meeting their own sustainability targets. Like they dropped the ball on their own like energy and carbon goals because of AI,

• 50:38Speaker 07

  because they were not ready themselves for the amount of energy that they would need and where that energy is coming from, like the renewable energy offsetting things weren't covering. And so I think that the latest and greatest in the trends in terms of energy generation has been nuclear. All the big tech companies have signed nuclear agreements, like power purchase agreements in the last couple of months. And the general messaging is that that's going to solve the issue in terms of energy like demand growth.

• 51:06Kevin Roose?

  And when Microsoft and Google said like, hey, you know, we're not going to make our targets, was it nuclear that they were pointing to? Like, did they say like, don't worry, we're going to fix this? Like, we have a new strategy? Or did they say like, we might just never hit these targets because our, you know, values have changed?

• 51:24Speaker 07

  So they were kind of like, well, these were very ambitious targets that we said that we said back in 2018 or 2019. And, you know, like, oh, technology moves at such a fast pace. But no, nuclear actually entered the chat relatively recently. I think that the reports came out in around May of last year of 2024. And then like a couple of months later, it was Microsoft announced that they are recommissioning Three Mile Island, Google signed a partnership with I don't remember what nuclear like generator. And they're also

• 51:53Kevin Roose?

  Kairos.

• 51:53Speaker 07

  Yeah, exactly. And so they're saying that, well, this is the new direction we're going. Because the thing is, I mean, sadly, like, building out renewable energy infrastructure does take time. And also, the problem with data centers and renewable energy is, is that data centers need energy 24/7. And the cycles aren't necessarily like predictable as like heating and cooling, for example, you know, when the, you know, when the temperature drops, people will turn on their their heating systems. Like, you have these models that that have worked pretty well historically. But for data centers, they don't work. And renewable energy tends to vary, you know, there's wind, if there's sun. And so there are a lot of challenges. You can't just let a bunch of solar panels and expect them to respond to the demand of your data centers. It's much more complicated. And so that's

• 52:37Speaker 07

  why nuclear has emerged as a potential solution.

• 52:40Kevin Roose?

  So my understanding, Sasha, is that a lot of the big AI companies are now just sort of racing to get as much energy capacity as they can. And that one of the worries is that they are sort of tapping out the infrastructure for clean or renewable energy. And so they are starting to go into these dirtier forms of energy that we know have these harmful environmental costs, because there just isn't enough renewable energy and adding more takes time, as you said.

• 53:09Speaker 07

  Yeah. And also, the thing is with data centers is that like, there are very concentrated, very intense energy sinks. So making that connection, like, I was talking to some energy grid operators in Paris, and they were saying, like, even if we did have the capacity, like the actual megawatt hours, distributing it in a way that all of that extra capacity goes towards the data center in whatever like rural area they build it in, is a challenge in itself.

• 53:33Kevin Roose?

  Hmm. One argument, I was talking with someone the other day who's works at an AI company. And one of the arguments that they made for why we shouldn't worry so much about the energy costs associated with AI is that basically our electrical grid in America is has been in desperate need of modernization, that we have this sort of creaky old electrical grid that we are that has not been growing nearly as quickly as it needs to. And that basically because AI now exists and demands all of this energy, we are starting to do things that we probably should have done a long time ago as far as investing in new

• 54:14Kevin Roose?

  sources of energy, in these mini nuclear reactors, in trying to scale up things like solar and wind power. And so, yes, these models are demanding a lot of energy, but they are sort of forcing us to modernize our infrastructure and our energy grid in ways that will benefit us as a country down the line. What do you make of that argument?

• 54:34Speaker 07

  Well, so what's interesting about the United States particularly is that it's not a single energy grid. It's there's a lot of energy providers in the States. There's a really nifty website called Electricity Map, and they map out electricity. And what's interesting, when you zoom in on the US, it's like a patchwork. There's some states that have like 12 different grids, and then there's some actually like multiple states have a single grid. What's interesting is that, I mean, for example, Canada is one per province. In Europe, it might be one per country, like France has a single one. And so it's yet they're probably right to an extent, but modernizing the US energy like system, network of grids is actually really difficult because it's so heterogeneous. And because, you know, even if you update one part of the grid, that doesn't mean like,

• 55:19Speaker 07

  for example, smaller energy grids don't have that much capacity, and the bigger ones will take time to update. So I think it's like, yes, in theory, it would be good to overhaul the US energy grid. But in practice, it's a lot of small problems that are harder to solve.

• 55:34Kevin Roose?

  One other thing I've heard from people who work in the AI industry or are not as worried about the environmental impact of AI is that, yes, this stuff costs energy. Yes, we need to find new sources of energy, but ultimately, AI is going to be more of a help in addressing the climate crisis than it will hurt. What do you make of that argument? Is that just self-serving?

• 55:58Speaker 07

  I don't think it's self-serving, but I think it's kind of a false dichotomy because the AI systems that are the most energy-intensive, like large language models, are the ones that have yet to prove their utility in fighting climate change. Like, I think that the issue here is that we're using these big models for tasks that are not helping the fight against climate change. And compared to that, the models that are helping climate change aren't the ones that are the issue. And so it's like, the problem with AI being an umbrella term kind of makes it very, very hard to have this discussion. But it's like, essentially, large language models are not solving the climate crisis anytime soon. And the models that are helping are not the ones that are contributing, like, most of the energy and carbon issues that we're seeing.

• 56:40Kevin Roose?

  Hmm. All right. One more argument that I want to have you address, which is about the efficiency of AI over time. We've heard from companies that they are making their models much more efficient because they're creating these algorithmic breakthroughs, doing things like model distillation. The chips themselves are also becoming much more energy-efficient. And so there's this argument that you'll hear from folks in the industry that actually, we're running out on outdated information when we say that AI is a risk to the climate because the energy needs are scaling down over time per use, and that actually, we're just worried because our information isn't up to date. So, do you

• 57:25Kevin Roose?

  think about efficiency in those terms, or how should we think about that?

• 57:28Speaker 07

  So, efficiency is interesting because I think that, like, a lot of what people talk about when they talk about, like, technological progress is some form of efficiency. It's like, oh, we're using less time, we're using less, I don't know, fuel, we're using less energy, for example. And I think in AI, we are seeing this. But what's interesting, I've been really going down the rabbit hole in terms of, like, macroeconomic literature on this. Like, there's this really interesting paradox, it's called Jevons' paradox. What Jevons observed in, like, the 19th century was that we're actually still using more and more coal despite using it more efficiently. And so, they, like, this kind of phenomenon has been observed a lot with different kinds of efficiency gains, whether it be time...

• 58:07Speaker 07

  Whether it be, you know, for example, cars like now that we can drive farther on the same amount of fuel will actually go to more places. And so I think what we're seeing a lot is in AI is this kind of rebound effect that, yeah, we can do more AI for the same, you know, amount of computer money. But that means we're going to do even more, we're going to put AI into even more things. And so those efficiency gains are kind of lost. Because now we're using L L Ms for things that we didn't use L L Ms for before.

• 58:36Kevin Roose?

  Casey, do you want to try repeating back what we've heard about the macro picture when it comes to AI and energy?

• 58:42Casey Newton?

  Well, the macro picture of AI and energy is that the construction of data centers does actually put a strain on the grid. We're seeing many more of them and that even as individual usage of AI gets more efficient. It seems likely that we'll just use a lot more of it. And so this is one that it seems like we do have to watch and take the environmental claim seriously. That's what I feel like I heard. Does that sound right?

• 59:14Speaker 07

  Yes, it does. I think you summed it up really well.

• 59:17Kevin Roose?

  Got it. And I think what we can agree on, whether or not we think that the, the individual or the, the macro use of AI across the economy is dangerous for the environment, is that I think AI companies should be required to disclose a lot more data about the energy use of their models. It just seems like the data we have, a lot of it is based on, you know, estimates from the outside, a lot of it is outdated. A lot of it has sort of gone through this game of telephone where all of a sudden, every time people use chat G P T, they think they're like burning down a forest. And it seems like this could all be solved by just having much better and more transparent data from the AI companies themselves about how much energy they're using.

• 59:58Speaker 07

  Agreed, and giving users more agency when it comes to generative AI. And, you know, even having a toggle when it comes to our AI generated summaries in Google, just like giving people a little bit more control over how they use. Like, we don't want to stop using Google or, you know, most people don't. So like, let us use Google in a way that is coherent with our values or the things that we, we want to optimize for.

• 1:00:21Kevin Roose?

  Well, Sasha, thank you so much for enlightening us on this subject is one I imagine we will return to. Because I don't think this debate is going away anytime soon, but I really appreciate your expertise and your time.

• 1:00:33Speaker 07

  Thank you for the great questions.

• 1:00:38Kevin Roose?

  When we come back, put on your gold chains, insert your zins, and let's do some jujitsu. We're talking about masculinity in the tech industry.

• 1:01:09Speaker 00

  Support for this podcast comes from Dell Technologies. It's time to put AI to work. Is your business ready? The challenge isn't getting started with AI. It's scaling beyond pilots while maintaining control of your data, governance, infrastructure, and costs. Dell AI Factory provides the integrated foundation you need to scale AI securely and efficiently with one trusted partner. Dell, AI on your terms. From idea to impact. Learn more at dell.com/yourwaytoAI. That's dell.com/yourwaytoAI.

• 1:01:43Speaker 05

  Companies are pouring money into AI, but few leaders can explain how those AI investments work with the systems you already have. You know, the ones that actually run your business? Without Systems, you don't have to start over with AI. You can choose the projects where AI will have the most impact. And OutSystems gives you the platform to make it agentic on your terms with complete control every step of the way. So you can trust AI with the mission-critical systems that run your business. Own your agentic future. Learn more at outsystems.com/NYT. That's outsystems.com/NYT.

• 1:02:18Speaker 08

  Ripple has been powering modern finance for a decade. It's not our first trip around the blockchain, and it won't be our last. We're leading the charge to bridge the gap between existing financial systems and the future of finance. You can have it all: legacy, stability, and digital innovation. Systems that work with your other systems. Because with the right partner, the question isn't crypto or traditional finance. The question is simply, why choose? A new standard of modern finance. It's happening with Ripple. For informational purposes only. Availability of products can vary and may be provided by different Ripple entities.

• 1:02:53Casey Newton?

  Well, Kevin, did you see the Mark Zuckerberg interview with Joe Rogan?

• 1:02:56Kevin Roose?

  I did, and I assume that that's why we're sitting here in our oversized baggy t-shirts and our gold chains.

• 1:03:02Casey Newton?

  That's right, yes. Thank you for agreeing to this costume change. Listeners should know that we are wearing very boxy, black t-shirts right now and gold chains to try to get us into the mindset of what I'm hoping we can do today.

• 1:03:14Kevin Roose?

  Yeah. Should we pop a zin too?

• 1:03:16Casey Newton?

  If you have one, any on you, go for it.

• 1:03:18Kevin Roose?

  I've got four. I've got, I've got upper deckies here, but if you want one, I'm happy to provide.

• 1:03:22Casey Newton?

  Upper deckies?

• 1:03:23Kevin Roose?

  You never heard of upper deckies?

• 1:03:24Casey Newton?

  No.

• 1:03:25Kevin Roose?

  Dog.

• 1:03:26Casey Newton?

  What's an upper decky?

• 1:03:27Kevin Roose?

  Let me teach you something about straight culture.

• 1:03:29Casey Newton?

  Please do.

• 1:03:29Kevin Roose?

  You know, you're always enlightening me.

• 1:03:31Casey Newton?

  Yeah.

• 1:03:31Kevin Roose?

  About gay culture.

• 1:03:32Casey Newton?

  Yeah.

• 1:03:33Kevin Roose?

  Uh, upper deckies are when you put a zin nicotine pouch in your upper lip.

• 1:03:38Casey Newton?

  That is perfect, Kevin. That is exactly the right spirit that I want to take into this segment. So you watched the Mark Zuckerberg interview on Joe Rogan?

• 1:03:46Kevin Roose?

  I did.

• 1:03:46Casey Newton?

  And what did you think?

• 1:03:47Kevin Roose?

  Uh, I thought it was very long.

• 1:03:49Casey Newton?

  Mm-hmm.

• 1:03:50Kevin Roose?

  That was, that was my main thing, was, um, this meeting could have been an email.

• 1:03:53Casey Newton?

  Well, I think that's a fair point, Kevin, but to me, I was so pleased to hear it because finally, someone in Silicon Valley was willing to say what we've all been thinking for years now, which is that this town does not have enough masculine energy.

• 1:04:07Kevin Roose?

  (laughing)

• 1:04:07Casey Newton?

  You know what I mean?

• 1:04:08Kevin Roose?

  (laughing)

• 1:04:09Casey Newton?

  Kevin?

• 1:04:09Kevin Roose?

  Yes.

• 1:04:10Casey Newton?

  Sometimes I will visit a company in Silicon Valley and see as many as one female executive. And finally, people like Mark Zuckerberg are starting to ask, when did things get this out of control? And I know you've thought the same thing. You've said that to me off mic.

• 1:04:24Kevin Roose?

  (laughing) I don't think I have, but go on.

• 1:04:28Casey Newton?

  Now, some people get confused because the most recent time that Meta shared numbers, it had about two men at the company for every one woman. But this just highlights how powerful feminine energy is, Kevin. What Zuckerberg is saying is that to counteract the presence of even one woman at Meta, at least three men are needed to restore balance. Now, just to give listeners a bit more of a sense of what we're talking about, I think we should play Mark Zuckerberg talking about masculine energy on the Joe Rogan experience.

• 1:04:58Kevin Roose?

  Let's do it.

• 1:04:59Speaker 09

  I just think we kind of swung culturally to that part of the, the kind of the spectrum, where it's all like, okay, masculinity is toxic. We have to like, get rid of it completely. It's like, no, like, it's, both of these things are good, right? It's like, you want like feminine energy, you want masculine energy? Like, I, I think that that's, like, you're gonna have parts of society that have more of one or the other. I think that that's all good, but, but I do think the corporate culture sort of had swung towards being this somewhat more neutered thing. And I didn't really feel that until I got involved in martial arts, which I think is still a more, much more masculine culture.

• 1:05:45Casey Newton?

  There is something about being punched in the face that makes you think, my culture has been neutered. You know what I mean?

• 1:05:51Kevin Roose?

  So, yes, I did hear this part of the interview. This went viral. Everyone on my feeds has been talking about this, this, these comments that Mark Zuckerberg made about masculine energy being missing from many of our greatest corporations.

• 1:06:06Casey Newton?

  Yes.

• 1:06:07Kevin Roose?

  And this is sort of in the context of all the moves that he's been making to try to make Meta more palatable to people on the right, including the incoming Trump administration. And this was sort of him saying to Joe Rogan, in a way that people mercilessly mocked, "The real problem in corporate America is that we've been letting this feminine energy take over, and we need to kind of assert masculine energy, and that's our path back to greatness."

• 1:06:31Casey Newton?

  Exactly, Kevin. And so, as we so often try to do on this show, I've spent all week thinking, how can we be part of the solution here? And so, I have come up with a list of ideas that we can bring to the Meta Corporation to help them restore masculine energy to Meta.

• 1:06:47Kevin Roose?

  Oh, boy.

• 1:06:47Casey Newton?

  We're going to give Meta a masculine makeover, and I would love to share some of the ideas that I have with you right now.

• 1:06:53Kevin Roose?

  Please.

• 1:06:54Casey Newton?

  Number one, modify the Facebook like button to display a bulging vein reflecting long hours spent in the gym. What do you think?

• 1:07:03Kevin Roose?

  I like it.

• 1:07:03Casey Newton?

  Something else? Whenever you tap it, your phone grunts.

• 1:07:09Casey Newton?

  Number two, let's just say the poke is gonna work a little differently now, but I can't say how on this podcast. Number three, transform every conference room at Meta into an octagon. Kevin, remind workers at every meeting that work is a combat zone and Mark Zuckerberg can strike at any time.

• 1:07:27Kevin Roose?

  I like this.

• 1:07:28Casey Newton?

  We're also changing the name of the finance department to MMA, mixed martial accounting.

• 1:07:34Casey Newton?

  Number four, Meta acquires 4chan. It's the largest repository of disturbed 17-year-olds in the world, Kevin, and they could be part of the solution, too. Now, the obvious thing to do would be to let them run the human resources department, but I'm proposing that Meta goes further and puts them in charge of content moderation. That'd be some masculine energy.

• 1:07:52Kevin Roose?

  I sure would.

• 1:07:53Casey Newton?

  Number five, no more of these beta team building activities like making pottery and volunteering, Kevin. Instead, we're going on a wild boar hunt.

• 1:08:02Kevin Roose?

  Yes.

• 1:08:02Casey Newton?

  As Mark shared on the Joe Rogan Experience, one of the greatest challenges in his life is that his ranch in Kauai is absolutely beset by an invasive species of wild boars. And for years now, Zuckerberg has been spending his downtime hunting them with bow and arrows. In fact, do we have a clip of that?

• 1:08:18Kevin Roose?

  Although my favorite is bow, bow and arrow. And that's, I think, like, the most... That feels like the most kind of sporting version of it.

• 1:08:27Speaker 10

  Yeah, if you want to put it that way, yeah. I mean, if you're just trying to get meat, it's not the most effective. The most effective is certainly a rifle.

• 1:08:35Casey Newton?

  If you work at Meta, I think this should be your problem, too. Whether you want to use a bow and arrow or a rifle, report to the Zuckerberg Ranch for further instructions.

• 1:08:43Kevin Roose?

  Now, do we know what happens if you are a Meta employee and you actually bring a bow, a hunting bow into the office?

• 1:08:48Casey Newton?

  That's actually one of the main ways to get promoted now. Number six, replace the water at Meta's data centers with Mountain Dew Code Red.

• 1:08:55Kevin Roose?

  Oh, I like this one.

• 1:08:56Casey Newton?

  Me, too. Number seven, in the 2019 film Joker, Kevin, Joaquin Phoenix's character does a famous dance down a set of stairs to signify that he is fully transformed into the Joker. My proposal, we bring those steps to the Meta campus in Menlo Park. You have a meeting with Mark Zuckerberg? Guess what, Kevin? You have to walk up the Joker steps.

• 1:09:14Kevin Roose?

  I like that.

• 1:09:14Casey Newton?

  Because Mark is the Joker now. Number eight. In what many people perceive as a cruel and pointless attack on trans people, Meta instructed managers to remove tampons from the male restrooms at their campuses. But this is a half measure, Kevin, because let's face it, real men don't use toilet paper.

• 1:09:29Kevin Roose?

  True.

• 1:09:29Casey Newton?

  Get rid of it.

• 1:09:30Kevin Roose?

  Yeah. Are we doing bidets or are we just going raw dog?

• 1:09:32Casey Newton?

  Bidets? Are you kidding me? There will not be one French thing in those restrooms. As long as I'm suggesting ideas.

• 1:09:39Kevin Roose?

  Okay.

• 1:09:40Casey Newton?

  Number nine, employees will now get one extra day off a year to do one of the following three activities: mow the lawn, watch the game, or hang with the boys. Which one of those would you pick, Kevin?

• 1:09:51Kevin Roose?

  Hang with the boys, for sure.

• 1:09:52Casey Newton?

  Do you even have any boys?

• 1:09:53Kevin Roose?

  Come on, I've got you.

• 1:09:55Casey Newton?

  That's true. We can hang together on our day off. Now, I have one last suggestion to bring up the masculine energy at Meta, Kevin, and it goes like this: we're gonna have a hackathon for women. Doesn't that sound nice?

• 1:10:05Kevin Roose?

  Yeah.

• 1:10:06Casey Newton?

  Yeah. And at the end, we're gonna take all the best ideas from their hackathon and give them to Meta's male executives, because what kind of energy is more masculine than taking credit for a woman's idea? Anyway, just my thoughts, Kevin. Do you have any ideas as well?

• 1:10:23Kevin Roose?

  No, I think that basically covers it. I think with these changes, the Meta Corporation will be fully, uh, what's the opposite of emasculated?

• 1:10:31Casey Newton?

  It will be enmanulated.

• 1:10:32Kevin Roose?

  Enmanulated. And, um, and, and we will have a glorious future, uh, run by men. You know, it, it, there used to be a time at Meta when people like Sheryl Sandberg, uh, had a seat at the table and, and famously told women there to lean in.

• 1:10:45Casey Newton?

  Yeah.

• 1:10:45Kevin Roose?

  Uh, what, what's happening with that now?

• 1:10:47Casey Newton?

  I'm being given word that, that they're being asked to lean out, actually. Maybe all the way out. In fact, Mark Zuckerberg announced this week that he was gonna, uh, cut 5% of what he called the low performers at the company. And that is sort of the ultimate lean out is a layoff.

• 1:11:02Kevin Roose?

  Yeah. I, I did see some, uh, Meta employees posting that, that they were, the way they were gonna avoid getting laid off is by getting extremely jacked. Uh, so that's an idea there.

• 1:11:12Casey Newton?

  I mean, that is now something that we can respect in culture, as we can say, if you have visible muscles, maybe you belong around here.

• 1:11:19Kevin Roose?

  Yeah. Casey, uh, how do you, I have to ask, since we are in the, the Zuckerberg uniform now, minus the $900,000 watch, this is just my Apple watch, um, how do you feel? Do you feel more masculine sitting in the studio today?

• 1:11:34Casey Newton?

  I am having an almost uncontrollable desire to just wrestle you to the ground and force you to submit. How are you feeling?

• 1:11:42Kevin Roose?

  I'm feeling, uh, I'm feeling like I'm, I'm a little insecure, honestly.

• 1:11:48Casey Newton?

  Really?

• 1:11:48Kevin Roose?

  Yeah.

• 1:11:48Casey Newton?

  Why?

• 1:11:49Kevin Roose?

  Cause I don't think I can pull this off.

• 1:11:51Casey Newton?

  You can absolutely pull it off. Everyone looks good in a black t-shirt and a gold chain.

• 1:11:55Kevin Roose?

  Yeah, including me.

• 1:11:56Casey Newton?

  Yes.

• 1:11:57Kevin Roose?

  I'm not a big man jewelry guy.

• 1:11:59Casey Newton?

  You know what? I haven't been either, but then for our anniversary, my boyfriend and I got little chains.

• 1:12:03Kevin Roose?

  Yeah.

• 1:12:04Casey Newton?

  Isn't that so cute?

• 1:12:04Kevin Roose?

  That is cute.

• 1:12:05Casey Newton?

  Yeah. And manly in kind of a different way.

• 1:12:07Kevin Roose?

  Yeah.

• 1:12:07Casey Newton?

  In kind of like a gay manly way. This is what I love. You start up, you're talking about something super manly, but then you get into it in any degree of detail and you realize, no, it's, it's masculine and feminine energy together in the same place. Isn't that beautiful?

• 1:12:20Kevin Roose?

  Now Casey, the one serious thing that I do want to say about this is that I, it clicked for me when I heard Mark Zuckerberg on Joe Rogan talking about masculinity and masculine energy, that this is what Founder Mode was.

• 1:12:33Casey Newton?

  Yes.

• 1:12:33Kevin Roose?

  This, you can look back at our shows that we did about Founder Mode last year and to my recollection, not one of the people in Silicon Valley calling for the return of Founder Mode was a woman. And I believe that that is because Founder Mode was an elaborate way of saying, uh, we're, we're big boys and we would like to run our companies like big boys.

• 1:12:54Casey Newton?

  Yeah. And I mean, look, I don't want to completely dismiss the idea that people should get in touch with masculine energy. That is a fine thing to do, I think, no matter who you are. I get really concerned when somebody who employs tens of thousands of people starts talking about this in the context of corporate culture and amid a series of initiatives that includes killing off the DEI program and firing your quote unquote low performers. Like a clear message is being sent and the message is not, uh, women are welcome at Meta.

• 1:13:27Kevin Roose?

  One thing that also struck me as I was listening to Mark Zuckerberg is that it also reminded me of a conversation that Jeff Bezos had at the Dealbook conference just a few weeks ago that I heard where I was actually surprised, you know, Jeff Bezos was sort of the original sort of tech founder who kind of got super masculine, right? He, he turned from this like scrawny nerd into- This like jacked dude who lifts weights and you know, has these sort of bulging muscles and you know, just sort of embraced a masculine aesthetic, I think, earlier than a lot of other tech executives. But I was also struck by his comments at Dealbook where he basically

• 1:14:07Kevin Roose?

  talked about his feelings a lot and how he had started becoming more emotionally open at work about feeling scared or feeling vulnerable. And it just really struck me that like that is a person who is actually comfortable with masculinity. When you can talk about emotions in the context of a business meeting and you could talk about them on stage at a business conference. This sort of like LARPing that Mark Zuckerberg is doing where he is pretending to be super masculine all of a sudden and like enjoy bow hunting and hanging out with the bros. Like, it just feels very insecure to me and very like, very much like this is a person who has not yet actually become at peace with his own self.

• 1:14:48Casey Newton?

  Yeah, I think that there is something to that. I can't even make a joke about that because it's actually kind of terrifying. To be 40 and sort of still be trying to work out, hmm, what are my values? And could I just replace them wholesale almost overnight with a different set? That's some kind of a scary proposition for somebody who runs a set of platforms used by billions of people.

• 1:15:07Kevin Roose?

  Yes, and I hope that whatever Mark Zuckerberg is looking for, he finds it. And I hope that it does not come at the expense of a lot of boars who might needlessly die.

• 1:15:42Speaker 00

  Support for this podcast comes from Dell Technologies. It's time to put AI to work. Is your business ready? The challenge isn't getting started with AI. It's scaling beyond pilots while maintaining control of your data, governance, infrastructure, and costs. Dell AI Factory provides the integrated foundation you need to scale AI securely and efficiently with one trusted partner, Dell, AI on your terms. From idea to impact. Learn more at dell.com/yourwaytoAI. That's dell.com/yourwaytoAI.

• 1:16:16Speaker 05

  It's time to innovate at the speed of AI without compromising control. That's OutSystems. Take a look at what happened with one global retailer during a period of massive expansion. They used OutSystems to build AI agents to automate invoicing and translation processes, shifting 95% of their daily store operations to a new digital app, saving thousands of hours of manual work. OutSystems gives you the flexibility to deploy how you want and the governance to do it safely. Own your agentic future. Learn more at outsystems.com/NYT. That's outsystems.com/NYT.

• 1:16:51Speaker 08

  Ripple has been powering modern finance for a decade. It's not our first trip around the blockchain, and it won't be our last. We're leading the charge to bridge the gap between existing financial systems and the future of finance. You can have it all: legacy, stability, and digital innovation. Systems that work with your other systems. Because with the right partner, the question isn't crypto or traditional finance. The question is simply, why choose? A new standard of modern finance. It's happening with Ripple. For informational purposes only. Availability of products can vary. May be provided by different Ripple entities.

• 1:17:25Kevin Roose?

  One last thing before we go. Our colleagues over at the Matter of Opinion podcast just published an extensive interview with the tech investor Mark Andreessen about his support for Donald Trump and what he sees as the emergence of a new conservative tech right. If you're interested in checking out that show, you can search for the Matter of Opinion podcast or click the link in our show notes. Hard Fork is produced by Whitney Jones and Rachel Cohn. We're edited this week by Rachel Dry. We're fact-checked by Caitlin Love. Today's show was engineered by Brad Fisher. Original music by Rowan Nemisto and Dan Powell. Our executive producer is Jen Poyant. Our audience editor is Nell Galogly. Video production by Ryan Manning and Chris Schott.

• 1:18:08Kevin Roose?

  You can watch this full episode on YouTube at youtube.com/hardfork. Special thanks to Paula Schuman, Queuing Tam, Dahlia Hadad, and Jeffrey Miranda. As always, you can email us at [email protected]. Send us your ideas for how to make Hard Fork's masculine energy more palpable.

• 1:18:27Casey Newton?

  What if we had a third male co-host?

• 1:18:29Kevin Roose?

  Oh, no.

• 1:18:49Speaker 06

  You can't undo everything in life, but you can undo prediabetes. More than 2 in 5 adults in the U.S. have prediabetes. Most don't know they have it. But with early diagnosis and action, you can delay or even prevent type 2 diabetes. Visit doihaveprediabetes.org to take the 1-minute risk test and start undoing the things you can. Brought to you by the Ad Council and the Centers for Disease Control and Prevention.