Arthur Hayes: The AI Bubble Could Crash Bitcoin in 2026

No, because you're assuming that everyone just buys because I say that I'm buying. So there's three problems right now with the AI story.

Not as if if AI goes down 50%, people have capital with which to invest in Bitcoin. And so I think Bitcoin gets thrown out with the bathwater. So that's why I'm, you know, bearish on pretty much all risk assets outside of, uh, large energy producers.

So I think the first response is that I'm not a money manager for anyone other than myself. I'm not a financial advisor. I never claimed to be. I think if you read my posts, I tell people what I am doing with my portfolio and there are some people who think I'm retarded. And then there are some people who agree with me and the market goes up or goes down and I'm right or I'm wrong. 75 to 90% of the time. Um, and I've worked very, very few percentage points of the time in terms of my calls. But, you know, I make money because I bet bigger on things I have more confidence in than the things that I lose money on. So I don't know where this ethics controversy actually is. I'm not your money manager. I never said I was your money manager. This is what I'm doing. You can choose to listen to it or not. You can choose to do your own research and make your own decisions or not. But I am not telling anyone what they should or should not do and never have done. And so I challenge Zach Zpt to find language on any of my posts where I tell people, 'You should do this or you should not do this,' or 'You should buy this or should not buy this.' I tell people I am doing this. What you choose to do with that information is up to you.

No, because you're assuming that everyone just buys because I say that I'm buying. And if you read some of the comments, a lot of the comments are, 'Oh, Arthur, you're an idiot. Oh, Arthur, you're the Jim Cramer of crypto. I'm going to confiscate everything you do.' So, I mean, I think that's very flattering that Zach Zpt thinks that everyone just blindly does what I do just because I post a chart and a funny sentence about what I'm doing with my portfolio. But I would challenge it again to prove that anyone is actually just blindly following me. Or there's not a large or maybe even larger percentage of people saying, 'Ah, there's an idiot. I think I know more than him. The fact that he's actually posting this means that it's going to go down.' So again, I am just posting what I'm doing. What people choose to do with their own portfolio is their decision. And I don't know how many people follow me by doing what I'm doing, or people say, 'Hey, that must mean that I should do the opposite because Arthur is saying it.' So again, if you don't know that.

Because at the end of the day, I'm in the attention business. And so my goal is to bring attention to a thesis. Whether you agree with me or disagree with me is not the point. The point is to bring attention. And I believe that over time, attention, because humans are generally primed to be optimistic about things, people will read a thesis. 'Oh, maybe that makes a lot of sense. Maybe I'll do my own research. Maybe I'll come to the same conclusion as Arthur.' Or maybe I won't. But at the end of the day, I am trading in attention. And if you take a look at the business of Maelstrom, we're in the business of giving attention to products and projects. Now, you could say that this product is good or bad. My goal is not for the project to say that this thing is going to go up or go down with 100% certainty. It's, 'Hey, look at this project. Make your own decision. At least you're paying attention.' The worst thing in crypto or any asset class is to be ignored. And that's my job to bring attention to something.

So, you know, thankfully we started acquiring hype in sort of early February of this year. Same with Nir Zcash. We've been acquiring since October of last year. So we came into sort of the Iran war with a pretty solid setup of a portfolio that didn't require much adjustment. And, you know, once there were so many questions around the macro and what was going to happen with liquidity, and now you have this war like, 'Okay, we're in these positions. We like these positions, and we're just going to sit here and do nothing.' And it was only recently when I sort of got the jitters and had a change of heart in terms of what I think was going to happen with liquidity and politics and the oil price that I decided that we had to dump a lot of stuff on the books.

Yes and no. I think it gets a little bit more nuanced. So I went back and I challenged myself, 'What did I get wrong?' Here I go on stage repeatedly, I come on podcasts like yours, and I talk about liquidity, and I say, 'As long as there's fiat liquidity increasing, then Bitcoin should go up.' Now obviously that's not happened. So October 2025 last year Bitcoin at 125,000. Today it's, you know, 60 to 50% lower. Yet there's been more money created in that period of time. So what about my mental model is wrong? And I came to this conclusion that I was a bit lazy. I usually only forecast, 'Okay, there's going to be more dollars, euros, yen, yuan created in the future.' But I don't actually look at where those units of fiat go. And so if everything is going towards AI, how much debt has been issued since ChatGPT was commercialized on November 30th, 2022, to the present, versus my rough estimate of how many dollars were created. So I went on Perplexity, as I do, asked some questions, and it came up with an estimate of about $1.5 trillion of AI-related debt was issued between 2022 and 2026, estimate, with 75 to 80% of that debt being issued between 2025 and the present. So Bitcoin was able to rally off the FTX lows to 2025 because the liquidity wasn't really being sucked up to a large extent by AI. But as the CapEx and spending on actual cash flows for the hyperscalers and the lending by the various financial institutions picked up late 2024 into 2025, it became almost impossible for Bitcoin to receive that liquidity because everyone is just doing AI. Now they're financing it, they're trading the stocks, and that's why Bitcoin peaked and has struggled. Whereas AI and, you know, pick your supply component of the CapEx bill that has gone up, you know, five, ten, 25, 50 times in, you know, 12 to 18 months. And so that is what's happened with all this liquidity: it's gone to AI. And then, you know, we have these three mega IPOs coming up. Obviously, people have to sell something to afford these things. I think yes, on the margin, people have probably liquidated some crypto, but the main story is a lot of dollars were created and they all went to finance AI, and they didn't make their way to Bitcoin. And that's why we suffered from October 2025 to the present.

So there's three problems right now with the AI story. Number one, the most vexing problem, is the price of energy. So the longer that Trump and the Iranians don't come to some sort of deal to resume flows through the Straits of Hormuz to a closer level to prewar, the more catch-up the market has to do in terms of rebuilding inventories for oil and natural gas and other commodities. And so once the deal is announced, you know, maybe energy sells off. But at the end of the day, people have to rebuild their national and commercial inventories that were run down from February until today. That's the reason why the price of oil hasn't really gone up that much, given the size of disruption. And then it comes down to just-in-case hoarding. So nations were caught out. They didn't have enough stored hydrocarbons. They're not going to make the same mistake again. They're going to say, 'Okay, I had 30 days. Maybe I need to have 200 days of imports covered,' and they'll go out and buy it. And that's going to cause, at least in my opinion, the price of oil and other hydrocarbons to increase into the future. And the problem with that is essentially all AI: as you convert oil and hydrocarbons, nuclear, renewables, whatever, into energy that powers a turbine, that creates electricity. The electricity goes over a silicon chip, does things, and out comes intelligence. If energy is more expensive, then dollars per token has to be more expensive as well. Which means that either the usage of tokens declines or the growth of token usage decelerates. Both are bad for the story of AI, because we have to assume accelerating use of tokens, which leads to accelerating CapEx build-out, which leads to accelerating orders and earnings for all these companies that are in the supply chain. That's what we're investing in. If it starts decelerating or falling in terms of usage, then that destroys the whole narrative, and that's truly predicated on energy. The second is politics, especially in the United States. And I took a look at a very simple thought experiment: if the Republicans led by Trump are going to get their asses kicked, as Polymarket says, looking at the numbers, is there a way that they could win? Is there a path to victory here? And so if you take a look at the two issues that animate voters in America, it's affordability. That's it. That's inflation pretty much directly related to the start of this war and the increase in costs of gas and food and other things. But the problem with inflation is it's very hard to like, turn on a dime. Even if the war ended today, it's not as if inflation would drop 10% lower and people would go, 'Oh my God, it's still better.' It's almost baked in. There's not really much Trump can do about that in the short term between now and November. The other issue, which I think people are more afraid of, is that yes, inflation is bad because I pay 5% more or 10% more for something, but at least I have a job. But what if this AI thing is as transformative as all the AI tech bros say it's going to be, and I no longer have a job in 2 or 3 months? Then I go, '10% inflation is bad, but 0% income is worse.' And so you have this visceral fear from people of all income strata and both political parties in the United States about AI taking their jobs. Also, they don't like all these data centers being built in their backyards that are using all these materials, that's creating inflation. And they feel very poor because essentially the AI story goes to, you know, a handful of men and their 1 or 2,000 employees across most of these companies. That's who's getting rich off of AI. And if you don't have a lot of financial assets, you can't participate in the AI stock bubble. So you have essentially a broad swath of both Democrat and Republican America who does not have a favorable view of AI, and the Democrats are starting to tap into this angst. You've seen proposals from both Warren, from Bernie Sanders. You've seen AOC talk about how AI is unfair and all these things. And I think that that is an issue that could flip seats for the Republicans if Trump, through the AI tech, throws it under the bus, at least rhetorically. He could come out and say he's going to do something about it. This is now his number one issue. He believes that data center buildout should be slowed down, that there should be some sort of redistribution of these amazing profits and productivity that the AI tech growth is generating to the American people. Now, as any good politician, Trump can say one thing and do another. The goal is for him to win the election. So as a populist, he should tap into the public anger, take the talking points from the Democrats, make that his issue, and take his team over the finish line. And so I think that there is a very existential risk that Trump flips the narrative on AI and starts preaching taxes, regulation, moratorium on data center buildup, which would completely destroy the bullish narrative for this. But the market is very scared of taxes and redistribution, because that completely destroys the reason for paying these ridiculous forward multiples for all these companies. And in my analysis, I link this as a reflexive loop between the higher the oil prices go, the less maneuvering room Trump has on the affordability issue, the more he's got to lean into another issue. And I think the only other issue that can get enough votes for him to keep the Republicans in power is this AI thing. And so I'm afraid that Trump flips the script and says, 'Okay, I can't do anything about the Iraq war. I'm not going to do a bad deal or whatever. However, he's going to frame it. The oil price is what it is, but I know that people are more scared about AI and I can become their AI savior. And if I created the AI tech bros, I can destroy the AI tech bros.' And that's how he can win votes in November, which is very bad for the AI story. So that's why I'm bearish on AI. The setup isn't there. And the third thing is these mega IPOs: the SpaceX IPO at 100 times sales. You know, if it's a low float FTX shitcoin, right? 4% is being floated at some massive valuation. As shares start to unlock in August, September, you have a massive deluge of selling at the same time. Anthropic and OpenAI are scheduled to list their IPOs in September at multi-trillion dollar valuations. And so it's going to be very tough for the market, I believe, to absorb all this. And to me, that points to the AI story, at least in this time period from now until November, peaking sometime this summer. And because everyone's invested in these things, not as if if AI goes down 50%, people have capital with which to invest in Bitcoin. And so I think Bitcoin gets thrown out with the bathwater as well, as a correlation one fell off. And so that's why I'm bearish on pretty much all risk assets outside of, uh, large energy producers. And as I came to this realization, you know, last week over the weekend when I was listening to some of the macro people that I follow and doing some of my own independent thinking, that's when I told myself, 'Okay, I just got to get out.' Because even if it rallies past where I sold it, so what? This situation here, these risks are there. And the last thing you want to do is have to sell into a falling market. So basically that is the main reason why you sold the world coin, for example. That's why I sold all hype. That's why I sold near. That's why I sold real coin. Zcash is a bit of a different story, but those three, that's why I sold it.

Not really, because all he has to do is change the rhetoric. I don't think he's going to pass a single bill to do what he says, but he's going to sell the American people that he understands the issue and he will do something about it after November. And then once we get to after November, he'll do whatever he wants to do. And it may not include doing anything that he told people he's going to do during the campaign trail. That's politics 101: tap into the anger, say you're going to do something about it, when November comes, backtrack. 'It's too hard. There's nuances, this and that,' all the reasons why he didn't. He doesn't need to actually fulfill this promise of extraordinary taxes on AI, more regulation, data center buildup moratoriums, all this sort of stuff.

Yes, because the markets take all these politicians literally rather than seriously. Every time Trump says, 'Oh, I'm going to bomb Iran to the Stone Age,' or, you know, 'I'm going to do this or that,' the markets freak out, right? Or 'I'm so close to doing a deal. It's right on my... you know, I'm about to sign it. I'm really about this... about to sign it.' Markets dump 15% and then nothing's ever signed. Right. So it's this schizophrenia of markets where they just take the politicians at their word when in actual fact, if you look at the follow-through on the rhetoric, it's almost always not the case. So markets are going to say, 'Oh shit, the party that was for AI that pumped my bags, that made this a national security issue, that was taking government money and investing in these companies and giving them preferential treatment on contracts, is now saying that there should be some sort of redistribution to the people.' 'Oh, fuck. I don't need to be in this trade anymore.' And they say, 'Well, the risk is too great to bet that after November, Trump changes course. I should just take my profit and sell now, right? I'm up so much. Why not take some chips off the table? I'll just wait. I'll wait until after the election. And if Trump goes back to AI, then maybe I'll get back in.' Right. Two, three. I think that's the critical period. You have three mega IPOs launching. You have desperate politicians trying to find talking points that are going to resonate with undecided voters. And you have peak expectations for earnings growth in the AI story. And so I think something has to give here.

I mean, so there's either two situations, right? Trump and whoever, you know, they all freak out. They just pump so much fucking money back into the system that people forget about all these things, and AI keeps going up again. I don't think they can really do that until after the election in November, or you know, we find that a lot of people lent too much money to, you know, second- or third-rate AI stories. They don't have the cash flows to represent solvency at particular levels. And there's a financial crisis or the threat of one. Do we get more money printing? And that'll benefit Bitcoin over AI because people will be done with AI. 'Oh, it went up. It had a massive correction, whatever that was. There needs to be a new story. I'm not willing to pay 100 times sales for an AI company again. I want something else.' And I think in that scenario, buffered by more money, Bitcoin can outperform.

I mean, I think that's a great story if you could make it stick. Let's go. I believe Hype is one of the best products ever made in crypto. It still is. It has great product market fit, and it's one of the only projects that makes money and gives that money back to token holders in terms of buybacks. The question is what you're going to pay for those earnings. And so I think when I got back into Hype in January and February, you know, we were paying like nine times earnings. And I'm talking about the circulating supply pipe, not the fully diluted. And now I think it's around 20-something like that. Coinbase trades at like 60 times earnings. Robinhood at like 40-45 or something like that. And CME at around like 25. So on a pure multiples basis, Hype is undervalued. And I think as people recognize that story, then the Hype multiple expansion is real. And you know, maybe the TAM is the entire trading universe. Ultimately, the reason why I sold it is, you know, I just think that short-term liquidity situation and positioning in the market just means that we're not going to be able to get people to believe in that story yet. But again, I think the story was believable in the near future. And I still believe that Hype is one of the best projects in crypto.

Yes. That's a precursor: clear revenue generating metrics, actual clients using those revenue generating metrics, and you're seeing that revenue accretive to the protocol. And this is the most important, which a lot of projects fail at for various reasons: you make money as a protocol and you give it back to token holders via buybacks and burn emissions, however you do it. You need to give your money back to the token holders. If you can accomplish all those things, you'll be like Hyperliquid and one of the best performing altcoins in a bear market.

No. I think privacy coins are definitely a needed thing. And I was prepared to hold our Zcash position through this macro turmoil. It was a core position. I wanted to get to a certain percentage of our holdings of Bitcoin and Zcash as the privacy alternative. But I cannot hold that in good conscience if there's no formal verification that a bug like this did not result in some sort of minting. Now, I don't believe it actually happened. I believe exactly what the researcher at whichever the labs discovered the bug and patched it. I believe they did everything. There's nothing funny. But that's not the point. We have to be perfect. And I'm not going to hold the amount of money that we were holding in Zcash if my belief of perfection has been violated. So I am still on the lookout, you know, all saying that Zcash could fix this problem. And I could be buying back at much higher prices, right? But again, this is not about the price. I sold this thing after it was dumped. I woke up in the morning, it was down like 30 or 40%. And I still was like, I cannot hold this based on how I view this position and why I hold it. The core tenet of belief in this coin, if that's violated, then I have to sell because I'm always worried about capital preservation first and then capital appreciation.

I realize that, but I think that a lot of these things, like how much in terms of fees are they generating for a particular project or protocol? Okay, you've got all this adoption. Let's take it for that. And I'm not really sure how much adoption it really is that's paying real fees. How much is that subsidized? And at the end of the day, does it accrue back to me as a token holder? That's the game, right? You're investing in these tokens to make money. You make money because this product does something valuable for people and they're paying for it, and then you receive some of that as profit and that profit comes back to you. But if we're just spinning our wheels, no one's actually paying for anything and no one's actually making any real money, then we have adoption—who gives a fuck? No one's making any money. And then whatever that token is connected to that project, it's not going to do very well.

Thank you.