How China Hijacked Bernie & AOC's War on AI | Bitcoin Policy Hour EP 38

If I try and think about is there a path to meaningfully regulating this stuff. The upside of this stuff is we cure cancer and all diseases and can reverse aging and people can theoretically live forever and uh if we don't do that all of us are going to die if we don't solve aging. And then the downside is like a what seems like a small percent chance that we're all going to die because of the machine god. And the like collective action here pushes in favor of it we ball. you know, much like Bitcoin and this technology can be described in some ways as like just math and so it's gonna happen as our president would say like you know you you'll see. >> All right, back with another week of the Bitcoin policy hour. Um we're going to dig into a particular topic today which is AI uh Chinese AI influence um trying to steer policy here in the United States. And uh we've got a great report from BPI's head of research Sam Lyman that we'll be digging into and maybe we'll have him on the show next week. But some background um on April 29th of this year the United States Senate hosted a panel on the existential threat of AI. There were four panelists. Two of them worked for the Chinese government and one sits today as counselor of the state council of the people's republic of China. And the senator who convened the panel, of course, our friend Bernie Sanders had introduced legislation one month earlier imposing a federal moratorium on new American AI data centers. I'm Zach Cohen. With me as usual, Ken Egan, head of government affairs at the Bitcoin Policy Institute. Zach Shapiro, head of policy at the Bitcoin Policy Institute. And uh we've got a lot to break down on this episode. So Ken, I'm gonna kick it to you to uh to give us your initial reaction on this report that Sam put together. We've got some real interesting uh you know, sort of breaking research here. Uh something that I think a lot of people suspected but maybe didn't have full evidence for. So yeah, what's your initial reaction here? Um and you know, for policy makers, like what what do you make of this? Uh what do we do about it? So it's it's sort of accepted right now in Washington that NOS's often play sort of a pernicious role in affecting policy. A lot of his NOS's are financed by billionaires. A lot of them actually get taxpayer money. Um and I think the work that Sam put together does a good job at revealing that that but that's just one part of the story. So here's the question I asked myself. Honest people could have a fair debate about whether data center expansion is good. I don't know that I would necessarily want to live next to a data center. I live in Northern Virginia and there are a lot of data centers here. So fair people can have that debate. What I don't understand is when introducing his bill um Senator Senator Bernie Sanders Vermont and and her companion bill, Representative Alexandra Kaser Cortez, New York City, why they opted to include on their panel two members of the Chinese Communist Party. I don't understand why they needed to do that. I don't understand the motivation. I I I just I I don't understand how their staff didn't conduct basic Google tests. One of the members was on the state council. Now, the state council is in a very important body. It is the administrative arm of the CCP. It is how the CCP pushes policy down through the ch down with the Chinese government. It is the administrative arm of the Communist Party for the government of China. And one of the members sits on that on on that council, you know, he's he's a senior technology adviser. So, The the question I think that Sam's paper u raises well and highlights and calls out is the fact that we invited well Senator Sanders and Representative uh Kascortez invited two members of the CCP to participate in a panel to convince American policy makers that the AI that the AI data center expansion is problematic. Now that's bad enough but that was that as you said that was in April. In May, there is a journal called Kia Kushu uh Kushi, which is the Chinese part, which is the the the CCP's um magazine. It's their journal. It's where they it's where they put out theoretical guidance. Um in that um another me another member, this is not a state council member, he was the um he was the he's the chair, he he's a senior representative at at the Chinese Academy of Science. I forget the uh the the the acronym is um is escaping me. And so in Chinese Communist Party literature, they advocate for the aggressive expansion of AI, of compute, of aligning energy resources to achieve those goals. The the exact direct opposite of what Senator Sanders and Okasu AOCC's bill would do, but also in directly in contrast with what um these two members of the Communist Party, Chinese Communist Party, testified to. Not testified, I shouldn't say testified. They were there at a panel so it was not it was not testimony. Um point being um in one audience they you know you have members of the Chinese government telling the United States we need to roll back our data center development. We need to roll back our construction of compute infrastructure. On the other side, the Chinese Communist Party's actual theoretical journal, their essentially their their their their thought magazine, um, published just weeks later an article by a senior academic advocating for just that, for expanding compute, for bridging the east, west, China. Now I see east west, that east, east of China, western China um, divide, building out energy networks to support that compute power. So I and and this is again where I'm a bit of a loss. I don't know why Senator Sanders, Representative Okascortez decided to invite these people to the panel. They didn't need to do it. Um if you you ask anybody um in certainly in Northern Virginia um but I know a lot of the discussion now is based is based on what's going on in Utah and Kevin Olri's project. There are plenty of people who have all kinds of reasonable gripes with what data centers can do. Some of it's inform some of it's not. Nonetheless, there's plenty of domestic honest opposition to it that you don't need to bring in representatives of of an adversarial government who tell you to do one thing, but then in their official policy channels do another. I I just I I do not understand. Makes no sense. Um I Yeah, I'm a bit of a loss. So, I think Sam did a great job of calling it out. >> Zach, I'm I'm also curious on your initial reaction here. Um, I think maybe we've been talking about this for some time, but I'm curious to your reaction of it being mapped sort of, you know, thoroughly and succinctly. >> Um, look, I I mean, I I think the evidence uh in Sam's paper is very convincing that this is meant to be an influenced operation. Um, and the question is what to do about that. I think there's like a lot of different threads te's apart here. Um, one is I think there's a big difference between the local opposition to data centers. I think frankly a lot of which is just misinformed. Um, this is something we've dealt with for a long time as it relates to Bitcoin mining. Um, that's sort of one domestic US political thread that's going on. And then I think the critique from Bernie Sanders and AOC and and between the two of them it's a little bit different is a critique of a different type. And it is interesting that I mean it'd be interesting to see evidence if Chinese are involved in sort of local protests over sort of like data center nimism. But what we seem to have evidence they're supporting here is really more broader threats about data center buildout um as sort of like an existential holistic issue. Right? So it's a problem for the environment. Um it's a problem right because it uses energy. It's the same sort of stuff we've dealt with for years with Bitcoin mining. Um it's a problem for the economy in that you're you know automating the white collar uh and maybe soon to be blue collar labor force and um the value is going to acrue to capital and not labor which is uh antithetical to what socialists like Bernie Sanders and AOC uh typically look for. Um, and then Bernie Sanders in particular is is really focused on this sort of like Udkowski and uh effective altruist like tail risk of AI killing us all uh concern. Um, and I think you know frankly the like local the effects of data centers on local economies I I have a little bit less sympathy uh for that policy issue. I think that's mostly based on misunderstanding. I think there is some real truth to economic concerns about class stratification as the economy gets more and more automated. Um, and I think there going to be extreme winners and losers and there might not be the same number of winners as there are losers. I think that's a legitimate policy debate to have. I I also think that the uh tail risk um you know although it's probably not the most likely scenario is a real enough concern that that is worth taking seriously and and uh treating on the merits. But I don't think it helps either of those discussions to have it sort of mixed in with or poisoned by like actual foreign influence operations um that are meant to give a strategic advantage to an adversary um you know over the United States especially whereas Ken was saying they're saying one thing here they are doing another thing there um and you know even for people who sort of like believe in we need to slow down and think about uh policy and not sort of be maximalist in terms of building out data centers. I think it really undermines your case and really muddies the water when it's easy to well like this is exactly what the Chinese want to do because the strongest argument or one of the strongest arguments in favor of the buildout um is that we need to beat this China and the AGI race. Um and then the other dynamic here which uh you know is sort of less about China but it's interesting to sort of just watch happen politically the domestic forces I think that are going to really come to bear on the one side there seems to be and and maybe we'll get to this more talking about like where the economy is right now infinity demand for inference um anthropic is throttling its models because it doesn't have enough in inference to serve that uh and you know enterprise seems to be the killer use case for Um, but like it's super profitable to replace expensive humans with inexpensive tokens. And so there's just, you know, there's infinite demand for tokens. And in order to remotely serve that, especially as more and more people just realize AI is useful, we're going to need all of the electricity, all of the data centers we can possibly build. And even that's not going to be enough, right? We're going to need to figure out data centers in space or crazy stuff like that. Uh on the other hand, I fully expect that there's going to be a huge political backlash specifically at least in the near term on the economics of uh what's happening, right? Like a year ago, everyone was debating, oh, there's all this money being spent on data centers. It's too much. The chips are going to be efficient. Um the Jevans paradox stuff is stupid and this is going to be like the dot crash because, you know, we don't see the revenue for OpenAI and anthropic and like this is a hype story that they're telling to their VCs to raise money. Well, that was totally wrong, right? Like they saw where the puck was headed. Um, and I think the same thing is happening now around like job displacement. Uh, where people are saying, "Oh, well, like this is a story that, uh, these companies are telling in advance of their IPOs, uh, like, oh, we're going to have this big impact on the economy and and we need to worry about job displacement and what the sort of like uh, social uh, bargain is." And that's all fake because we're not actually seeing job displacement in the numbers other than like new hires. And I think a year from now that is likely look just as stupid as the concerns about revenue. Um, and so I I think you are going to have these two forces like this insatiable appetite for tokens sort of on the capital and consumer side which is what drives the US economy and then you're maybe more and more organized sort of politics around the labor side of this where people are worried about losing their jobs as that actually starts to show up in the data. Um, and that that's going to be a mess as it is. I don't think we need China involved. I want to pull on on that local opposition thread, Zach. Um, I will say like, you know, at school, um, I I hear over and over again this uh this environmental question and I hear, uh, over and over the job loss. I can't get a job. Um, this AI thing sucks. Like, it's, you know, melting all our brains and nobody's going to be able to read or write anymore. like it just seems to me like with a certain class of people and especially the young people that you know again were in my circle um this is like really permeated and I'm like fairly convinced that it's like all the same thing that's coming together and I obviously don't have evidence that this is part of the kind of like Chinese influence operation but the narratives do line up so so I'm curious like to what extent you actually think these things are separable right you you talked about local opposition as maybe a different thing from the sort of like Bernie Sanders AOC like existential holistic issue. Where I'm seeing this is like it seems like a lot of the local opposition is fueled by this misinformation that may be connected to the Chinese influence operation. Do do you think that's the right read? >> Maybe. I mean it depends on the flavor of the opposition. I think it might look a little bit different on college campuses than it looks in like uh you know community council meetings in towns where I think a lot more of the local opposition is specifically nimism like we don't want this in our town it's eyesore it's going to be loud it's not going to create enough jobs um you know it's not good for our tax base uh after it's built etc versus the like specifically environmental concerns. >> Yeah. Ken, I also want your thoughts on this because again there's this like sort of view of um the environmental question, the job loss question. There there's all these like factors that people keep referring to that are negative negative negative like we hate AI, it's going to ruin everything. And then on the flip side of that, like we I just don't think we really have people like effectively talking about ways that this could be good, right? Like we think about our our old friend Matt Pines here and and the stuff that he's working on and these really like you know potentially you know like society level altering events of scientific discovery or whatever it may be and like that is just certainly not making its way into you know again like my sample size here is like college campus life. Um, but I think also probably like to a larger degree, you know, across society like people reading the New York Times, like they're not going to read, I, you know, assume about like all the positive things that AI could do. So, I'm curious like to what extent you think this is like purely a China influence problem versus like a larger lack of, you know, hey, here are the benefits alongside the the drawbacks. No, I it's not only a China influence problem or all the other actors that you know that that our our study highlighted. I think I mean think it's college graduation season. Um there's lots of videos of every college wants to wants to have a prominent usually you know tech leader speak and you the Twitter and YouTube were replete with videos of these speakers talking about AI and tech and how students need to learn about this stuff and and they get booed and they get booed. You're probably seen these. Maybe that happened at your graduation. >> Well, we had we had Andy Cohen and he was he was shitting on anthropic. So, um >> you probably got you probably got a lot of cheers. But I think we probably got a lot of cheers. But I think what it does is it it threatens there's a there's a class of people that have done really really well for 50 years. So, the man managerial class um that you know college graduates with liberal arts degrees who have done particularly well. they've managed to get jobs in the government which probably not going to change. The government will change very slowly but this sort of whole like layer of of managerial professionals inside companies HR accountants things like that um that have done that they that have done very very well and that have held a lot of political power and economic power over the past 50 years and AI directly right now directly threatens that. I mean, you see, they're still hiring software engineers. They're hiring different software engineers, but the software industry are still being hired. But where they're cutting the fat are people that have traditionally not contribute to bottom line to assure those managerial functions. That's where AI threatens. But that that's the class of people that have held political power and economic power for a long time. And AI goes right at that. If you're trimming first, you're not trimming your not trimming your salespeople. You're certainly not trimming your um you know, the people who run your building. you're you're trimming the people that that that fulfill functions that AI is increasingly well suited to take care of. So I think that and that class is of course highly you know already probably politically charged. Um it it you know it it it is it's ideological and it's being directly threatened by AI. I think some people have legitimate concerns. Um I probably more mole disagree those concern but that's fine. It doesn't detract from their legitimacy or at least they their concerns. I think a lot is fed uh filled by by misinformation. But I think adoption of alien companies directly threatens that class of people who have have been in control and benefit for for decades because of the way the system the system has evolved evolved. And I think those are the people and that's you know a lot of these college graduates and I don't mean to like be I again I'm I feel for some of these people because you know they were they were sort of educated and taught to be prepared for one kind of life you know get a liberal arts degree and go to a company and take some kind of design job or HR job and you do really well and you'll have a good life and you know that that that's being slowly taken away from that if not well maybe not so slowly and I think that's that's a lot of the concern concern is that directly affects a very vocal and powerful uh constituency. >> Zach, let let's go back to Bernie Sanders and AOC here. Um you said one of at least Bernie Sanders like primary concerns is this like tail risk, right? Um, I'm curious like how you would describe that on his terms because I think like one of the natural instincts here reading Sam's piece is to say like, oh, Bernie Sanders is a puppet of the CCP and like he very well maybe I I don't know what's the way of framing this where like this is actually sort of ideological and aligned with his policy priorities. >> I think there are a couple things there, but before I get to Bernie Sanders, I want to go back to the the college students for a second. Um, I spent a lot of time thinking about how I think AI is going to impact sort of white collar labor and like what is being taken away from whom. And I think part of the difficulty here is that like the what Ken was referring to is like the managers, right? Or sometimes referred to as the the PMC, the professional managerial class. Like that's kind of what's going away. And and there is a certain type of personality type. I think a personality type that tends to be reflexively political or like really susceptible to sort of the narrative of the current thing. Um where like they are the ones who are most threatened, right? Like the sort of credentialists who are going to climb sort of the career ladder and have the path blazed for them as opposed to like the doers. And if you're graduating college now, the scary thing is that like the sort of like large swath of pre-made sort of decent paying jobs as like a management consultant or a software engineer or like a lawyer if you go to law school. Like those will probably go away because they're going to be, you know, there's going to be fewer of those. But the roles that are left are are amazing, especially for people who are like doers or high agencies, right? Like the Zack Cohens of the world are going to do great, right? Better better than than they would have before. um with AI. Um but the thing about being a doer is you don't really have time to complain about, you know, FUD that you read on on Blue Sky. Um and so I do think that this becomes a bit of a self-perpetuating thing where it's like the people who are going to be hit are exactly the type of people um who sort of, you know, complain about this type of stuff. And it'll be interesting to see sort of the economy reshaped accordingly. And the people who are losing out are people who historically have had a very privileged place in society and a lot of political power. Which is which is why I think this is going to be sort of like the unstoppable force of like hunger for tokens meets the immovable object of uh you know populist anti- um AI sentiment. Uh in terms of like the Bernie Sanders and the Doomer argument, um Bernie Sanders is kind of like anti- capitalism, right? That's his whole thing. And so I don't think he is just a Chinese puppet. I think this aligns very well with his caring about labor over capital and caring about sort of, you know, what he says, fairness in the economy. Um, and sort of not letting billionaires screw up the world, right? The millionaires and billionaires. Um, and uh, I I think he comes by it honestly, right, that that but like he should not in so doing accidentally become a Chinese puppet and literally end up paring like actual CCP talking points. uh because you know he's like blind to that, right? I think we saw a lot of this with like campus protests right back to college students where people end up ended up like using actual Hamas propaganda, right? Not because that they that like they did it on purpose, right? Or that there were like necessarily Hamas people who were like invading these campuses. It's that like it is very easy to have a blind spot when someone's rhetoric aligns with your pre-existing ideology. And I think that is the specific risk with Bernie if I'm steel manning his concerns. Um it's not that like he is undereducated necessarily about the risks, right? You're talking about like what if we told people about like Matt Pine's uh startup who's you know trying to solve physics with AI. Uh the concern is what if that goes too well, right? Like what if you know Pines is out there summoning black holes and you know in a long world. Um, and um, I don't think that's going to happen, but you know, it it it could, right? Like the like, you know, we talked about this in our AI 2027 episode, but like the like failure mode here is that AI gets good enough at coding and reasoning that's able to create the next smarter version of AI without humans in the loop and then that creates a smarter version and you have recursive self-improvement. Um, and you very quickly get from as smart as a human on every domain to just vastly smarter than humans. And then we have this thing that's vastly smarter than us in all domains, including persuasion. It can persuade humans to do, you know, what it needs to do. We've already seen like Opus blackmailing anthropic engineers. Like that's not so far-fetched. Uh, and then its goals are not exactly aligned with ours. And it decides to do something where we are an inconvenience and so it gets rid of us, right? Not because it hates us. it's just like it's not aligned. Um I again like you know I think that is a reasonable thing to at least keep in mind. Um it just you know you should keep it in mind on its own terms and you should develop policies and the idea of like a complete data center freeze when everyone like wants more AI usage right now. It just isn't good policy. And I think if you were taking these safety things seriously on its own terms, you would end up with different like policy suggestions than you would if your real agenda was like economic. Um uh and if you or if you know you were falling for sort of CCP talking points and what Sandra says most of the time is like oh all these smarter people than me you know they are worried about AI risk. He's quoting sort of the Frontier Lab uh CEOs back to themselves, which I think there is some fairness, right? It does seem like going from being a nonprofit AI researcher to a you know billionaire many many times over does tend to change your view of how much regulation there there should be of what you do. But again, like I just like the the we need to shut down all data centers because my CCP expert told me so is is not not the way to go about this. In all fairness though, if you were to look at Shapiro's Twitter, you would see that he is positioning himself to survive the initial calling. So like he's, you know, Zach's not worried about it. He at least round one. Round two, no guarantees. But I think the irony of it all, of course, and I checked this out when I when the paper came out, there are no major data centers in Vermont. There are a few small ones to be like telecommunications infrastructure, but there are no major data centers in Vermont. And I'd be very surprised if there were any data centers in New York City in AOC's district. So clearly there's another agenda here. This is not my constituents are complaining about it. I don't even think there are data centers in New York. I mean New York state with high having grown up in New York State, right? Um hugely high power costs. Um I I don't know that there there's any data any significant data center construction in New York State. Certainly not in Vermont. So you have to so there's there's certainly an ideological component to this. They're not doing this because their constituents are worried. You know, it's not like the people of of Stow Vermont are or are knocking on Bernie Sanders doors saying don't build this data center in our, you know, next to our ski our ski slope. They're not doing that. Or, you know, don't block my views of the Throgs Bridge with your data center AOC. We we, you know, we don't want that to happen. So, there's clearly an ideological component to it. Again, I'm at a loss as to why. Nonetheless, I'm in a loss as to why, assuming they they actually believe it, why they would platform. It's not even paring CCP talking points. They just brought the CCP to deliver the talking points in person. I just cannot get my head around it. >> Yeah. And just to go back to the like Matt Pines's point like right it you take the like let let's grant the sort of like environmental issue like data centers use too much power right but it's like what if you know what if this research that they're doing sort of unlocks like free and abundant energy right like that that just sort of refutes or or maybe not like directly refutes but but challenges the notion that like AI is just necessarily bad for the environment um and I just don't hear people talking about that. So like I I do think that is an important thing for for us and others to to continue to harp on. >> Yeah. And breakthrough medical research, there's all kinds of things it it theoretically could do, >> right? >> I think very specifically like if I try and think about is there a path to meaningfully regulating this stuff, I come to the conclusion that there just is not. Um, agree >> because the game theory and look and this is going to be like the very like crass way of looking at it, but I think this is what's going to control at the end of the day is like >> the upside of this stuff is we cure cancer and all diseases and can reverse aging and people can theoretically live forever and we create infinite abundance and uh if we don't do that all of us are going to die like individually we're all going to die uh if we don't solve aging right um and so hey why not roll the dice on this and then the downside It is like a what seems like a small percent chance that we're all going to die because of the machine god and this the like collective action here pushes in favor of like fuck it we ball right like that that you know that that I think is where the the incentives point and and like probably is what's going to carry the day and even if you get a country to move in the other direction right like there's the USChina competition to motivate uh right and then even if US and China were to somehow come together and decide they're going to do this like someone's going to do it offshore or on the moon like the incentives are too strong to push this technology this forward. You know, much like Bitcoin and this technology can be described in some ways as like just math and so it's going to happen and we're going to see how it you know like as as our president would say like you know you you'll see uh and so >> thank you for your attention to this matter. >> Thank you for Yeah, we'll see. Um, but like to the extent we're steering this, I think this requires like really sort of careful nuance and cander and not uh sort of nation state incentives. >> Okay, Ken, I'm going to lump together two questions here so you can take them in whatever order you want, but um you know, question one is really like on the geopolitical side or the you know, US versus China competition bit, right? Like if we're to sort of get the lay of the land here, we have basically like the frontier models, you know, four or five of them depending on how you're counting it. Right now in the United States, they're closed source. Uh increasingly, you know, you can't use these without KYC. Um and then on the China side, right, you have this proliferation of open- source models. Um there they seem to certainly be behind the curve. Um, you know, the question is by how much and and then you like you lump this influence operation in. Um, and like it just seems like the battle lines are becoming more clear about, you know, there is this race that seems sort of undeniable now. You have these two sort of approaches to the issue, which is, you know, hey, we're gonna do this influence operation and we're going to have these open source models versus the US like, hey, we're just going to build at the frontier. we're going to be the best. We're going to encourage the best to be here in the United States. Um, and so these are the two like the two approaches, right? Question one is like to what extent is the US going to continue to be involved in this um particularly, you know, as they build up relations with the Frontier Labs. And two, you know, if you're a United States policy maker and you care about the US China competition issue, you know, what are you doing, right? Like, is it to counter this influence operation in some way? Is it to make sure that, you know, we don't have like CCP people testifying in front of the Senate? Um, which which I don't think they even have control over. >> They used to be a very low bar. Yeah. Right. Do not invite the adversary to come to Capitol Hill and tell you what you should do. That was >> exactly um so so that's like I think where I want to go down here and and we'll get Zach to weigh into but um like that's just where I see this conversation sort of coming to a logical end is like you know this this all factors into the larger USChina AI competition um like what are the moves on the table right now? So economic and military competition is going to require that is going to demand that this this continues to develop. You're going to have companies, Chinese companies that increasingly integrate AI into their products, robotics for example, and American companies are not going to stand by, and let that happen. And the time's also going to come where Secretary of Defense, National Security Council gets a gets an intelligence report saying the Chinese have now mastered small autonomous drones that can fly around the world or swim the seas and function autonomously and they're cheap to make and they have a lot of them. What do you do in that regard? we usually well we have to also build our autonomous drones. So I I but I think that just the fundamentals of economic and military competition makes it impossible. You could have you know fits and starts where we're accelerating and maybe a new administration decides to slow down a bit but then but our core national interests are going to require that we keep up. It's just there's no other way around it. Um so I I I I don't I I agree with Zach. It's the change out of the bottle and it's it's it's going to continue. I think the thing we're we're that we need and I think we're we're we're um we're struggling with this now in the United States is power. I think the number is over the past 10 years the Chinese have built three times as much power as we consume the United States consumes. They've built that capacity. That that's tremendous. And there's you don't necessarily need and I've heard, you know, people like David Sax say the Chinese are eight months behind us, a year behind us with their models. Fine. But where they're not behind us is on actual ability to produce electricity. And there's two pieces of this, right? There's having a great model and there's also inference. There's there's compute. And what you need for compute is power. So they don't necess they don't need to be have the latest greatest shiniest AI model or they need I mean they want that and they're working at that, but they they're they're going to destroy us on on power on power production in this in this compute. So I think that's where they're that's where they're putting their their money and their and efforts is building is building a power infrastructure. They've done I mean this is what they're working on. So this is directly affects the ability of AI to to grow, right? This is this is the inference discussion. So I don't think they necessarily need to have the best model in the world. They don't need to beat mythos or whatever else is next. They need to be able to run at scale a lot of a lot of agents been just burn through tokens all day long. That requires a lot of compute power that requires burning a lot of a lot of electrons. So I think that's where they're I think that again I think they're aiming to be just like they are in chips, right? They're semiconductors. I think they're aiming to be at par with us, but they don't need to be. They just need to be almost as good and have a lot more capacity. >> Yeah, Zach, I'm I'm curious on your thoughts here. I know you said uh you know, like fuck it, we ball. Um but like if you're you know, if you're a US regulator, a policy maker, um like what are the concrete things that you're doing if you if you're someone who is of the thought that we should, you know, we should win this race against China? Well, no. I mean, that is that is if you're focused on winning the race with China, that's it, right? >> Well, just like leave it be, like, let me just stay out of the way. >> Well, in terms of regulation, uh, and then you want to do industrial policy, uh, mostly I 100% agree with Ken that like the indications are that like power is the bottleneck for us and what's important. Um, you know, if China's six to 18 months away from getting Mythos, that's a bigger near-term threat than drone swarms, right? You can't really use drone swarms without declaring war. Um, if you have mythos and you could just find all of the exploits in all of our cyber systems, you can do that covertly, right, without declaring war. And we know the Chinese are trying to penetrate every important system all the time. And there's reason to think they might be more successful. Uh, and what's worse, like it seems like a lot of what Mythos is is just more compute. Um, I think I mentioned this in one of our like BPI internal calls, but I I was listening to the uh New York Times tech podcast last week and they had an interview with a cyber security expert who has access to Mythos and he was sort of referring to it interchangeably with Opus 4.7 and they're like, "Wait, wait, wait. like Opus is I mean Mythos is clearly way more um powerful than Opus 4.7. And what he seemed to suggest was that like Mythos isn't that different than the public model we have access to. just has this ultra mode where it thinks really hard for a really long time and it's able to find um you know exploits and uh string together vulnerabilities into zero days and like that you know it's it's like customuilt for cyber security but plus a lot of compute and if that's where we're heading and and at least like my own personal experience using AI you know where Opus 4.7 unless you're like really careful with it in some ways is a downgrade from 4.6 six because it just decides to be lazy because anthropic is compute constrained. If we're moving to a world where the model weights matter less and just the amount of compute and intelligence you're willing to throw at something, the brute force matters more then the constraint is going to be energy. Um, and we're in a pretty bad position in terms of our buildout, right? And like you can't get a uh a gas like we have a lot of natural gas, but you can't get gas turbines until the 2030s if you were to order one today. And so like our supply chain is is a real problem and we need to find ways to fix that. And so um if the thing that you're solving for is not the safety isn't we're competing with China is finding ways to unlock energy usage uh for more and more and more data centers because this is going to be a compute race. Uh and it seems like one potential vector here is going to be uh in space right that is something that Elon is very focused on. there's a a you know a free um thermonuclear reactor in space that has for our purposes infinite energy uh and it's you know in the sun and and it's hard to vent that energy appropriately and there's sort of a a really difficult engineering challenge but uh the the end state of this which by the way is the same end state where we get wiped out is is the Dyson swarm where ultimately you know you have more and more sort of like satellites with computia just circling the sun and you create sort of like this big shell around the sun where you you get all of its energy. And sort of in the like sci-fi futurist world, um there's this concept called the Cartesub scale uh to measure sort of how far along a society is. This is something that's very popular in Bitcoin circles because it relates to the idea of sort of using energy for mining. But uh you have these three levels of society where a level one card society use all of the energy available on a planet, level two a star, and level three a galaxy. Um what a level two society is is they just have a big Dyson worm that they put around the sun. And so that that might be the way the the like real, you know, pardon moonshot um because you actually need this sort of mass launcher on the moon to make this work. Uh way to bypass the energy problems, but you get into like pretty crazy sci-fi territory. Uh we're going to need pines to figure some stuff out before we can effectively pull that off. Okay, Ken, last question here and then we'll talk about the AI kind of larger market right now. um is you know let's let's say and I think we all agree on this that you know power is the biggest issue that we're facing and you know if if you take that is this like Chinese influence operation on purely like what seems to be the data center side the biggest bottleneck or is it a sort of like regulatory situation where these you know projects to uh to bring on more power just getting delayed and delayed or you know they can't get connected into the grid. Like do do you think it if you take that approach of like hey the power is the number one thing like should we actually care about this Chinese influence operation? >> I think I you have to expect that nation states are going to do this kind of stuff. So whatever it's just like it's like hate hate you hate the player not you know hate the game not the player. >> That's sort of how I think about in this like yeah of course they're doing that and that's fine fine but you that that's the nature of things. Again, I just don't think we need to enlist members of Congress to help them do it. So, on that on that point, and I think on the general opposition, you know, most of the concerns are have if you have an honest, you know, family that shows up to a protest that their electricity bills are going to go up and their water bills are going to go up. Those are the two concerns that people usually site, right? And then there's noise. But generally speaking, those are your two things. >> Yeah. Yeah, if you solve the power problem, I think you if people can see that their electricity bills are not going up, it may in fact are going down because you've actually built out power infrastructure or some of these hyperscalers are building power that they're actually selling back to the grid, which I think is the plan for some of these. Um, yeah, I think you could see a lot of that legitimate opposition melt away, but a lot of it's permitting like Zach said, we have spy chain issues, but al also, you know, try to get try to get a a small mod nuclear actor built anywhere in the United States in less than like, you know, Will we literally be Dyson swarming the sun and pines will be summoning black holes before they fix that problem? And then this is the issue. You could I China China run gets these things start up in I think I want to say the numbers are two to three years from conception we're going to build a nuclear reactor to the thing they they switch on and steam starts pumping. So I think this is a problem. We need to fix the regulatory issues. But again there are a lot of and they're all related. There are the same people that are opposing data centers are the same people that will protest building small module nuclear reactors. the same people that are protest building pipelines. There's a reason why in Bernie Sanders in Vermont, people still get like fuel oil deliveries that my my aunt I I have a family living in Vermont. They get fuel oil deliveries in a truck that gets pumped into their house. Um you so this is this is what we're dealing with, right? That that obviously is not a scale of a model. So I think they're all related. I think the same people that the same I should say the same organized actors that oppose >> data center construction are the same organized actors that oppose construction of any kind of power generation you could easily make make a range and I think a lot of these hyperscalers are willing to do it where they generate their own power sus or or SMRs are an obvious solution but there's more there are other options as well so I it's it's all this sort of the same on the organized side I'm not talking about your average you know citizen who just doesn't want it on the organized side, it's the same it's the same arguments against data centers, the same argu same arguments against construction of of of power building out the grid. It's it's it's all the same. And you know, you can debate what the what the ideological motivations are behind that, but there it is the same exact people protesting these things. >> Yeah. And who who's possibly going to help us build out all the energy infrastructure? >> Bitcoin money. >> Well, yeah. I mean, of Yeah. Yeah. or or Amazon bill, you know, was it was it Amazon's buying Three-Mile Island? I think it's Amazon, right? Or is it Microsoft? Somebody bought Three Mile Island. Um, you know, the the famous the the nuclear power plant that set off the whole anti-uclear movement in United States. Three Mile Island. It's being it's being, you know, it's being it's being um restarted. But I yeah, I think hyperscalers could be part of that. Um, at least at least in the interim, I think they'll eventually like everything else. I mean, there's and here's here's my my my point I discuss with people. There's the sort of the civilization quotient that is directly correlated to the how much energy you produce and consume. Not just produce, but consume. Civilization requires energy and requires you consume it. AI is part of that. If our civilization is going to continue to develop, we're going to have to consume more energy and that requires building more energy, better sources of it. And I think renewables play a part of that. Nuclear plays a part of that. But we need more. And the same actors that are out there organizing protests, giving, you know, the same signs in every state in union where you see data data center protests, they're the same people. They will be out the following day protesting against the proposed construction of, for example, of a nuclear reactor. >> Yep. >> A nuclear power plant. Yeah. >> Okay. Zach, um, you said there's a an infinite demand for inference right now and we're we're definitely seeing that in the uh in the markets expectations around a lot of these potential public offerings for AI companies. I'm curious whether you think that's going to continue, whether you think these valuations are justified or maybe if we're in a a massive bubble as the AI doomers would like to think. >> Well, um those are separate questions. Uh I think the answer to both is yes. I think there is going to continue to be more or less infinite demand for the foreseeable future um driven specifically by enterprise and not consumer. Uh this is something that I think really doesn't quite show up in the data yet. It's hard to show sort of efficiencies in the enterprise from AI usage. I think a lot of that is just like bad incentives, conservatism, bureaucracy, whatever. But I can tell you as someone who just like, you know, does this myself for my own business, um, if you're able to replace human labor with tokens, uh, which is very possible even with today's level of technology and will become more and more possible, you'd be willing to pay a lot for tokens. Um, you know, don't tell anthropic, but I'd be happy to pay 10x what I currently pay. Uh, right? Like you replace expensive human labor. Um, now that again, as we talked about at the at the top, that's going to have impact on the labor market, which might have impacts on the economy. If you start to project that forward, you get to sort of the like citrini research uh sort of scary cases where people lose jobs and then they can't pay their mortgages and then debt goes bad and uh SAS companies that also have a lot of private debt tied to them become obsoleted and um it like companies sort of forward-looking multiples and price to earnings ratios start to make less sense because there's so much disruption in the market that you can't so like I'm very bullish on demand for tokens. I think that only goes up as more and more people just understand that they are useful and specifically enterprises understand that you can replace expensive human labor with less expensive comput. Um but uh are we in a bubble in general? Yeah, maybe. I mean the sort of driving factors there would be uh if we're talking about the AI companies specifically um it's there's you know going to be a lot of liquidity that is needed when these companies IPO all it seems like at perhaps multi- trillion dollar valuations uh and then all of the insiders right the the employees the private market investors uh want to see a return liquidity is there just in the public markets to soak this up like we'll See, the narrative is really great when you're in the private market and you're marked to whatever the latest round is when you're marked to uh the you know your floating NYC stock price that that's sort of a different story. So like especially 6 months after uh the IPOs when the insiders unlock that's going to be a big deal and I think it's entirely possible that the IPOs will mark the top uh of the market and then out you know maybe I I think the like Nvidia of the world right I'm not an expert in uh what's their mode or could there be competitors but but the infrastructure plays you know I would be pretty bullish on if I was playing sort of fantasy hedge fund here um we're just going to need a ton of copper we're going to need a ton of energy we're going to need a ton of chips like I don't the pix and troubles I don't see going anywhere. Um, but then everything else, there are going to be some companies that deliver uh goods and services using AI, you know, I I hope including some extent my own that are going to do really well, but then they're going to be a lot that are massively disrupted. Um, and I think really specifically the biggest challenge and I'm stealing this idea as someone from Chimath who I think put it really well like when you are underwriting predictable cash flows in a world where you know like technology might and services it might be incredibly deflationary. Uh it's just hard to justify paying 30 times earnings. Uh and so that destroys wealth effect. the people losing jobs um undermine sort of the sort of consumer strength which is like 70% or something crazy of our economy is the American consumer especially the high-paying white collar uh consumers that are likely to be the most affected by this right because their labor is the most profitable to replace with compute and so um I I think both both can be true and I think my bet is that both are somewhat true we are in a bubble and this technology is absolutely real and powerful Ken, we haven't seen any S1s yet, but uh do you think we'll see any this year? >> Yeah. And just one final point, like you know, the the CME is now offering, you know, compute futures. And if you if you if you think that you know that it's not here to stay, like they're selling futures on this, they're selling futures on this stuff now. Yeah, I think the first the next big one probably going to be SpaceX. That's the rumor. It's going to be in June, right? I think mid June. So I think you'll see Elon I mean my my own impression is that you'll see that they've engaged law firms and investment banks to do this like it seems like it's in progress. >> Yeah. And then you'll see that I I think within a year a pretty rapid consolidation with Tesla. That's just my that's just my my my gut feeling. >> I I think soon and you you could I mean also I think Elon has a has a has a a plan for his sort of for his all the companies that he runs what he wants to do with them. So I think he is a lot less market sensitive um to what the IPO might look like. I think the A companies are probably a little more market sensitive and like the problem that that Zach um outlined. You don't want to IPO when you're you know right now we're in we're in a great credit cycle expansion. You don't want to IPO when you're in a credit cycle contraction. So I do I I think you'll see I think you'll see S1's from the companies probably later this year. But I think, you know, I think the market is right now anticipating sort of the great Elon um IPOs and then probably a relatively rapid um consolidation of everything under under one umbrella. That's just that's sort of my guess. >> All right. Good uh good AI conversation. Um I think we would be remiss not to celebrate the Clarity Act passing through the Senate Banking Committee with a vote of 15 to9. Um, and the Democrats that joined that were senators Angela also Brooks from Maryland and Ruben Ggo from Arizona. Um, we've certainly been talking about this for a while. >> Seems uh seems like what what passed through the um through the committee was was bullish and positive. Uh, we'll see if that holds up. But yeah, Ken, curious your your initial reaction here and your outlook. I think that the bill that passed to the committee is not a bill that is going to it's not going to be the bill that is eventually voted on. It's just not. I have that in pretty good authority and I believe that and I think you I think our listeners that follow this stuff should not expect the bill that passed through banking to be the bill that you know to resemble the bill that that eventually is voted on the floor. There's a lot that has to happen. there still needs to be um work done on the Senate ad committee and then they have to find Senate floor time and they're going to want floor time when they have a sense of where the where the bill's going to go. So I I don't know. I think I think we're still in for a fight. I think we are in for a fight on BRCA. I mean you know we talk about that a lot. Um the you know law enforcement associations and not you know whether they're wellinformed or not or why you can debate why they entered this fight. law enforcement law enforcement associations are pressing pretty heavily on Capitol Hill, Democrats and Republicans. Of course, that's a that's a constituency the Republicans tend to want to be friends with for a lot of reasons. So there, you know, that that that that sort of law enforcement press has been effective. I think it's it's misguided. A lot of what you read the letters from like the Fraternal Order of Police, the National Surface Association, they talk about how BRCA will prevent law enforcement officers from tracing and seizing digital assets, which is completely it's just not true. It's nowhere in the bill. It's it's it's not true. So, it gives you a sense of like and there's a lot of like proarication. It could do this, it could do that. Um I I I think the law enforcement organizations are being are being I think there's there's a lot going on behind the scenes different constituencies who are pressing some of these groups um to to get involved in this. I don't know why the National Fraternal Order of Police um is really worried about money service businesses. I don't I don't think I don't think your average police officer in Fairfax County is dealing with these issues, right? Or um in, you know, in Nashville, Tennessee or wherever. I just don't think they are. So, I think there's a lot of like that that's just that's just the way politics works in DC. So, I I think there's a the goal is to have this thing voted on in the Senate before July 4th because you're an election year and there's there's Congress is in session in July. There's a lot that's got to get done. they're out in August and then you know then we're in election season. So I think you'll probably see a vote within the next four to five weeks. I think that's the objective but remains to be seen. I it's up in the air. I mean this is again the call to the listeners if you are you well now it's it's out of banking so every senator needs to hear from from you that you know the BRCA is important and you know don't um don't um don't don't don't tinker with the right the software the rights of developers that that's critically important for for for Bitcoin and the community and what we hope to build in the United States. So I we're the fight is it's I we've been saying this for freaking months, but like we're we're still in there. We're still in the thick of it, right? That actually we're now like in the hard part. It's been like sort of a uh you know it's been a brawl to get that bill out of out of out of Senate, out of banking. Still not in the floor yet. There's a lot that's going to happen, but we are we are not out of the woods. BRCA is definitely under threat. 100% under threat. People need to act accordingly. Zach, would you say that the the issues that were problematic in the banking committee remain problematic as they go to the floor and those are BRCA, stable coin yield and um ethics? Do you think that that's accurate? >> I think less less stable coin yield. I think it's uh BRCA and ethics more. >> So you think stable coin yield is sort of settled at this point? >> Yeah. And and >> I think by the way as a a correction I think we read the wrong bill text about stable coin yield on the last episode of policy hour and it was the thing we were anticipating with the economic substance test and we read an old version. >> Gotcha. >> In our defense there have been a lot of versions. >> Yeah. >> Um can can you just explain briefly like why stablecoin yield would be settled in the banking committee? um just because like it was a headline issue that there's a lot of money behind and I think there's a pin in it for now and that really is a industry versus industry fight whereas right and like ultimately industry can come to a settlement on that right it's the banks versus the crypto exchanges um really mostly Coinbase um the problem with the BRCA and ethics is that they're genuine political issues it's not that anyone has a lot of money at stake in, you know, one way or the other in the BRCA. Um, it's the developers freedom at stake versus whatever vague nonsense the Fraternal Order of Police is saying. Um, and then on ethics, right? Like this isn't like there's that much money at stake either, but like it's just really good politics for Democrats to be like, "Hey, we can't do a crypto bill unless you deal with the sort of self-deing um in the White House." And uh that is just thornier when you need to get eight or nine Democrats over the line to get closer. >> No, that's exactly right. I mean the ethics discussion is tough because you know I I mean we're not PPI is not waiting into that into that swamp. I mean it's it is a it is a it is a purely partisan right issue. So I we're staying out of it personal but yeah as as an institute we're not that's just not something we're going to comment on or get involved in. We we're focused on what are our key our key things for for Bitcoiners and that that is BRCA. >> Okay. Well, good. Everyone uh everyone listening, make sure to call your senator. Doesn't matter if they're on the banking committee now. Uh it's it's in everyone's hands. So, um yeah, that's great. And then Ken, we got uh we got a draft another draft of the tax bill, which they're calling the the digital asset parody um act. And um yeah, I'm curious if you had a chance to read through that. Any initial reactions or should we put a pin in it and come back to it next week? >> Yeah, I think I mean really briefly, I need to like go through the bill and maybe maybe Zach has um done a little more research, but we're we're still on and there there is and we'll go in detail next week, but like there's still no divinous exemptions for Bitcoin and Bitcoin miners are still not being treated the way that we think uh miners should be treated. So we don't we don't love the bill. Um, we're working with with the office the sponsoring offices, but you know there's they're they're trying they're focused on getting a bill that can pass. So that that that's and the fact that it arrives at clarity it's like okay you're killing me like par clarity >> you know yeah but like to talk about something else. Um but no it's actually very important because we we just we what we don't want to do is set a precedent where there's there's a crypto tax bill um and that does not provide diminishes for Bitcoin. I that's an important precedent to set. So, we're that's the next fight. >> Yeah. Yeah. We're They're keeping us on our toe. >> And it had to rhyme. Like, it had to freaking rhyme. Like, it couldn't have been something else. Parody clarity. Yes. Just to remind us all. >> That's hilarious. Um, okay. Well, I think we'll wrap it there. Almost at the hour. Um, good uh good AI conversation. Uh, don't fall for the Chinese propaganda. Don't uh Yeah. Um, >> check out the check out the report. It's a it's it's it's really well done. >> Yeah, it's on our website, btcpolicy.org. Um, let's see. The name of the report is, in case you're looking for it, foreign influence in a campaign against American AI, written by BPI's head of research, Sam Lyman. Sam's a gem. This is a a killer piece. So, definitely give it a read. >> And if you want to survive the initial calling, follow Zach Shapiro on Twitter. Yeah, actually he he'll he'll he he'll work he'll he'll guide you through it. >> And uh if you aren't already, make sure you're subscribed to Bitcoin Magazine and the Bitcoin Policy Institute and uh and Ken, what what are they supposed to do? They're supposed to smash something. >> They're supposed to smash that like button and subscribe. Smash that like button and subscribe. All right, we'll we'll see everyone next