On The Brink with Castle Island - Weekly Roundup 09/04/26 (HuggingFace hack breaks containment, Hyperliquid onshoring, RSA260 factored, is the stablecoin sandwich cheaper?) (EP.738)

Episode Date: September 4, 2026

Matt and Nic are back for another week of news and deals. In this episode:  We determine the winner of the bad boys theme song competition Dwarkesh ignites controversy with his OpenAI hack recap What actually happened with OpenAI and Huggingface? Gwern's and Matt Huang's short story competition Will AI make us dumb? 200m Americans KYC data has been hacked Prediction markets lawsuit heading to SCOTUS? Robinhood Chain is taking off Strategy is buying Bitcoin again Should Members of Congress be paid more? Cronos suffers a defi hack Revolut gets a bank charter Is the stablecoin sandwich cheaper for cross border? RSA260 factored - panic? Content mentioned in the episode: GPUworld.com Wenxin Du, Three Myths About Payment Innovation  

Matt Walsh and Nick Carter are partners at Castle Island Ventures. All Vees expressed by them or the guests on this podcast are solely their opinions and do not reflect the opinions of Castle Island Ventures. Guest and hosts may maintain positions in the assets discussed in this podcast. You should not treat any opinion expressed by anyone on this podcast as a specific inducement to make a particular investment or follow a particular strategy but only as an expression of their personal opinion. This podcast is on the brink podcast. Welcome to On the Brink. I'm Matt Walsh. And I'm Nick Carter. I'm tired today.

I was on a five and a half hour flight delay yesterday. Flying just seems to be harder these days. I feel like the peak of flying was in like 2010 and it's only been worse. It's awful. We were delayed because no one picked up the pilots at their hotel. Isn't it great when it's like, yeah, totally fixable administrative issue? And it's not even like weather. Yeah, that is awful.

But a lot going on. Should we address the winner of the bad boys competition? There was a shocker. Yeah, so there were two entrants. I think really only two submissions. And we received votes through a variety of different methods. So I don't know how much integrity the election had. Yeah.

There were emails. We didn't actually really reconcile or deduplicate. There were DMs. A lot of DMs. Messages delivered in person. But I think there is a clear winner. Yeah, the clear winner is definitely number two. Song two.

Shout out to listener Ian who put together song two. Ian. You know, some people were saying that we didn't play enough of the good parts of one. And I think one is very catchy. I like them both, to be clear. I'm not picking favorites. One's good from start to finish. So I can understand that the one supporters are a little grieved.

But we're just acknowledging democracy here. Ian has, I believe, given us the license to use the song. We do have a message from Ian saying that we can use it anytime we want. We need some bad boys now. But people have been good mostly lately. People have been mostly good lately, yeah. But I think we're going to have some bad boys here at some point in the next year, don't you? Yeah, I mean, it's not like we look forward to it because that would imply something really bad has happened.

Scam or hack or, you know, credit crunch. but I mean it's human nature so it probably happened So I think we should roll Ian's song towards the end Maybe at the very end of this episode Yeah for those of you who are just turning in And miss last week's episode

Here's 15 seconds of Ian's song The official winner and new bad boys anthem You have been out there Be a bad boy You have been a belly, belly, belly, really, really, very, really, All right, I mean, I think that speaks for itself. Yeah, I mean, does what it says on the tent.

We have the best listeners. Didn't expect that result, actually. No, no, wasn't expecting that. Pretty busy week here. I mean, where do you want to go with this? There was Castle Island content. We had Guy Young from Athena talking to Wyatt about on-chain financial markets, tons of deals.

Dwar Keshe is just, I don't know, Dwarkesh is lighting the AI complex on fire. He's got politicians tweeting about his blog post. There's a lot to talk about. Dorcasch has actually united the world. And I was an original fan of the Dorcasch podcast even before he was super famous. And I like him a lot.

I think his post was awful, actually. Like really reprehensible and super bad. I think it's the worst thing he's ever done. Yeah, so his post, for those of you who have not read it, we'll put it in our newsletter this week. But he wrote a narrative on the AI Hugging Face incident where agents ran rogue at OpenAI, hacked into hugging face,

I guess hacked into a bunch of stuff at Open AI itself, but he really added a lot of drama to the story. I think he's been reading too much sci-fi, like us. He's been reading too much sci-fi lately.

It didn't have to be as dramatic. He could have written it in a very dry and technical sense, and the report still totally is comprehensible if you strip out all of the drama. It still makes total sense. Yeah.

So none of the editorializing was necessary. And now it's everyone into a tizzy from Bernie Sanders to my barber brought it up when he was giving me haircut yesterday. How does that even come up with your barber? I mean, I would think that your barber would be talking about like sports. Mostly we were talking about his olive grove actually. He actually sold me a bottle of olive oil during the haircut, a single source. And then we were talking about whether he was immune to AI risks, like as being that he's a barber. Yeah. I would think he is.

So we were discussing whether a humanoid robot could cut hair and he made the point that well, it just couldn't provide the same level of conversation, which is true. You know, no robot has the charisma of a Greek barber as of right now. But is that what you're looking for? Because I'm of the mind that if I could just fall asleep

for 20 to 30 minutes, I would rather do that. Yeah, no, I mean, I'm pretty socially awkward, so I actually hate to talk to my barber, but I have to. So,

but I would have to. actually love it if I could arrange for kind of a silent haircut. I think it's really the Chinese Communist Party hacking into the humanoid robot while I'm getting my haircut that I would fear. There is an additional tail risk, I think. I told him that I thought his job was secure as an olive oil merchant and barber. He has probably some of the most AI insulated jobs. And I said I was screwed probably. But then he brought up the hugging face incident and he said we're all screwed because of it. He heard about it from this guy called Dwar Cush. I wish I was making this up. That's wild to me. This is a Greek barber in Miami. I don't think he owns a laptop.

And he's aware of this. So this thing is really cross the chasm. Everybody knows about it. That story is really broken contain. I mean, I think the problem with the post from where I said is that it's describing a lot of human characteristics to things that were programmatically put into the wild and functioned almost as intended if you look at the original you know design yeah i mean you know that meme where it's like it's i can't describe a meme on a podcast it's stupid but you know someone's like asking the i to say that like they're an evil scary villain and the i repeats it back to them and then the person is afraid right that's actually what happened because opening i created a fleet of 50 000 roughly agents

that's first of all that's crazy they had them do a hacking challenge and they turned off the part that says stop if you fail so just keep going maximum persistence what do you think is going to happen yeah these are sophisticated it's obviously a sophisticated model we know it can do hacking

they told it to go do whatever it takes to pass the test the test half of the answers were unpassable right so what do you do if you're a really super smart agent you you've been given a test, and there's one way to solve the test, which is not by solving it, but it's by going and finding the scan tron key. You're going to go find it. Right. So from my perspective, what happened is exactly what Open AI set out to happen and aided by the fact that

they're pretty bad at operational security practices on top of all this. It's also just an incentive problem, don't you think? If you're at Open AI right now, same thing for Anthropic. You're incentivized to say how powerful this is and that we need some regulatory guardrails here so that you can have a moat and so that you can counteract the Chinese open source models that are pushing so powerfully into the enterprise here in the U.S. market. So I'm skeptical of just the intentions of some of these big labs. Certainly. And I think it was just a badly designed test. And you don't need to ascribe any real agency or human desire to these agents. to understand what happened.

I think it's actually very comprehensible based on the objective reading. So I think it was a huge mistake by the AI safety community and now you have people like Bernie Sanders exploiting it. Bernie Sanders had a tweet today. It leads with pause AI development now and it references this Dwarkesh post at length.

I mean, it's a very long tweet. I'm not going to read it. But I mean, this is being used by the politicians. Now this broke contains. this blog post. I mean, if you find yourself giving grist to Bernie Sanders, something's gone wrong. Also today, New York City Mayor Zoran Mandani has banned generative AI for one year in New York public schools for preschool through eighth grade. I'm actually not against that, frankly. I mean, the kids are not going to learn. Yeah. The kids will not be able to learn if they're using

AI for their homework. Yeah, you know, I guess in high school, you definitely would need this, but I don't know. I can see the argument that maybe no kid should have access to it, but certainly you'd want to ban social media first, I would think. So we're going to get into the deals, I promise. But before that, this story makes me think of the story competition. Have you come across this? I had not heard of this until you mentioned it right before we started recording. All right. So I have recently become. an amateur fiction author. I don't know how to put it,

but I have now discovered that I enjoy writing fiction of late. And amazingly, coincidentally, this short story competition has come out by Matt Huang, paradigm, Guern, the Gwern, and Neil Stevenson

on top of all that. And the idea, it's actually a huge prize pool. The idea is to write a short story, imagining a world where AI stops getting better, but everybody gets AI. So there's total equality of access, but no more progress. And so what does that world look like? What's the point, I guess, of this? I think the point they're trying to make is you can still get abundance from AI, even if it doesn't get any better. And we're at a time of fear around AI and hatred and anti-data-center activism. And I think what they want to do

is encourage people to dream. Even if AI never got one I-O-to-better, it would still make the world a better place. That's my guess of what they're after. Yeah, I would agree with that. I mean, just imagine if everyone in the world had super intelligence in their pocket. So I...

This relates back to the New York thing because I have figured out what I want to write in my story. I won't spoil it. My theory is that AI will make us incapable of learning anything and just really dumb overall. Oh, so you're not going to be having an optimistic story here. Well, you know me. So I think we're going to go for the dystopian.

You know, they're really cynical. But I mean, is anyone, I mean, 20 years from now, no one's going to ever write an essay from scratch, right? They won't know how to do it. Five years from now, everybody's going to have an AI recorder recording every conversation. They're not going to have to remember anything. So our brains are going to atrophy. so it's the size of a walnut.

You're going to have things on LinkedIn. Like, I actually know how to use Excel pre-LLMs. I mean, think about handwriting. Today, handwriting is kind of quaint, right? Like knowing how to write cursive. Yeah. It's going to be like that, but with like typing out an essay from scratch with a keyboard.

No, it'll just be voice or neural interface. The five-paragraph essay. Kids aren't going to know how to actually write one. They'll never have to learn, right? So it'd be like, well, doing long division manually or something, we'll just use the calculator. Yeah. All right. So you're going to win that?

I don't know. I've tried the essay first. All right. Let's hop into the deals of the week. We actually had two deals in the Castle Island portfolio that went public with raises this week. The first one is Felix Pago, which is a stable coin remittance and financial services company. They've raised a combined $200 million in equity in debt from Andresen, QED, us at Castle Island, switch ventures, and general cattle.

list. Big congrats to Manny and Bernardo over at Felix. Really excited about that one. And Groma, of course, another one of our portcose, a real estate tokenization platform. There is 17.5 million in a round with Coinbase Ventures, F Prime, Valor Equity Partners, us again, and L1D. So huge congrats to Seth and the whole team over there, Groma. Moving on, RQD clearing, which is a clearing and custody infrastructure provider. they raised 74 million from Bain Capital and Nica. Next up we have Firelight. They're a crypto cybersecurity company.

There was 8 million from Gumi Cryptos, Maven 11, and Mennel Aventures. Here's a rumor here, but Pauly Market, there's a report out this week that they're planning to raise a billion dollars at a $21 billion valuation, led by a $300 million investment from Donald Trump Jr.'s 1789 Capital. I think that's a, the last round there was, what, 20? so that's flat to the last round.

Yeah. Have you noticed that Kalshi is actually getting market share on Polymarket? Yeah, definitely has a better infrastructure at this point, it seems. And also they have the U.S. market from a regulatory perspective. They're just far ahead, it seems. I just think they're out executing Polymarket. Well, Polymarket just needs to do some market reforms. I mean, we talk about this a lot.

The UMA system has to go. Yeah, that's a relative. of the past. They've got to get rid of that. I think they should jettison the blockchain stuff, go for full KYC. I mean, speaking of KYC, wow, I don't even know if we have that on our roster to talk about today. Big hack

this week. So two-thirds of all Americans' K-YC documents have been hacked. It's unbelievable. I mean, yeah, it's over 200 million Americans, is that right? Yeah. So let's just retire that approach to KYC while we're at it. Just unbelievable. Next up we have Bicko.

They acquired Nydig's institutional trading business. Next one is diameter pay. This is a stable coin payments infrastructure company. There is $10 million from CMT digital and light speed faction. And lastly, open reserve, they're a stable coin base bank. They raise $25 million from A6 and Z crypto, jump capital, coin base ventures, and others. Let's stay on the prediction markets. New Jersey has asked the Supreme Court to decide whether states can regulate sports betting on prediction markets. I think we've been talking about this every week, I feel like. But we have a split circuit right now. So the Ninth Circuit, I think, identified that sports events contracts are gambling. There's another

circuit. I forget which one that said that they're not. So maybe that was the Third Circuit. So right now we have the Ninth Circuit saying something different from the Third Circuit. This is textbook why you have a Supreme Court. I think we just need to get this over with, don't I think? It's interesting that being in crypto this long turns you into like a constitutional law analysts. Yeah, yeah. But it does. But that's exactly right. Split circuit means

that it's much more likely you're going to get a I guess cert. CERT. CERT. Oh, a writ of certiorari. Seri-R-R-R-A. Going back to AP government. I love that class. Yeah, I mean, I think we've actually been saying we do need some kind of a

ruling from the top court. Now, let's say that it comes down and it's it says that the federal government does not have ultimate oversight on sports betting and that the states can make their own call. How bad is that for Calshan polymarket do you think? Well, I think it hurts them a lot in most of the most important states, don't you think? I think New York goes off the table. I don't know about California. I mean, I guess it really, it would, you'd have to just look at draft kings, right?

You'd have to figure out where they operate and that's where you can ultimately run one of these things. I think it would be pretty devastating blow. Certainly the TAM would be a lot smaller. There's also just the question of will it just apply the sports betting? That's my guess, right? I think that having prediction markets at a federal level that are outcome-based prediction markets that are not tied to sports betting, that's probably not going to get rolled into this, is my guess. You know, an adverse regulatory decision would be a great thing to hedge using a prediction market. But then, I don't know, you got some settlement questions.

I don't know if they're hosting these markets. When does it go into effect? Yeah, will we be shut down federally? Have you been following Robin Hood chain? They are the center of the action right now in crypto. Seems like there's a new meme coin frenzy going on. So I think Robin Hood chain eclipsed $4 million in daily fees this week. As far as I can tell, it's a lot of meme coins trading against tokenized stuff.

stocks and some of these liquidity pools. But I guess you have to tip your cap to Robin Hood. I mean, I don't know if this is a flash in the pen or not, but the chain is being used. I will admit, this is one of those things that normally I'm on top of this kind of thing in crypto, but I haven't participated in this one at all. I know that meme coins are going to be around forever and they'll ebb and flow. It's just hard to really get yourself to pay attention to a meme coin frenzy when it's happening. Speaking of things that are hard to pay attention to, strategy has begun buying Bitcoin.

They've resumed. They bought this week 4,603 Bitcoin. That's $370 million. First purchase since late June following months of rebuilding the cash reserve and defending their preferred shares. Yeah, that's interesting. I mean, a lot of hate on the timeline around, okay, you sold Bitcoin, then you bought it back higher. But I don't know. to be too critical of sailors funding all of this with the sale of micro strategy common equity.

It's just unbelievable to me. And right now, as we speak, actually, the price of Bitcoin's up five and a half percent over the past 24 hours. You wonder if that's strategy buying some more. So we should have led with this, actually. This is just breaking as we go to press here, courtesy of Alex Thorne. The House Republican leadership has canceled the votes, including the clarity vote. So I believe there is no vote slated for clarity prior to the men terms as of right now. Well, just to clarify that, so the House recess, so to speak, has been changed. And so the September session is only going to be one week, but clarity is in the Senate right now. So I guess Alex's point is that if it gets passed in the Senate,

then there's not enough time in the House to get it done and that you'd probably have to get done in the blame duck. Yeah, I don't know exactly the significance of this, but it doesn't on its face seem very positive for clarity. I wonder whether these House members just can't work a little bit more.

So they're only going to work for one week in September. What are they doing? Just campaigning? I mean, this is a weird thing to say, but I feel like they don't get paid enough, actually, for how I think that's a great point. I don't think any politicians get paid enough. And I think that's why you have

corruption. Imagine if you, as a House member, you made a million dollars a year, and if you were in the Senate, you made three. Don't you think we get better quality people? You get better people. There would be less of an opportunity cost for being an office, and there'd be less of a need to monetize being an office later on by being a lobbyist. Or just monetize via insider trading, which seems to be the de facto way to make money. Yeah. I think that's right. I mean, I don't know what someone in the House makes right now, but you can almost guarantee. that you could make more if you just went and were like a middle market executive at a financial services company, do you think? Without a doubt. I think it's in the, if I'd guess, in $200,000 range. Yeah, so a rank and file member of Congress makes $174,000 per year. Yeah, I mean, that's just barely going to cover your rent. I mean, in Washington, D.C., because you've got to figure you've got to have somewhere to stay

down there. That's why a lot of these guys end up sleeping in their offices, right? Yeah. I don't know. They should be paid a lot more. I agree with you. In other news this week, the SEC has released a proposal to overhaul transfer agent rules. So it would be the first overhaul in 40 years.

It would rewrite registration and recordkeeping for the 270 firms out there that keep track of securities ownership. And it would explicitly accommodate blockchain-based records. So they're inviting comment on this, talking about distributed ledger tech, talking about uncertificated, uncertificated security. that's a multiple. So very positive development. I think this is a, you know, definitely a catalyst to get more assets on chain here is just giving clarity for the TAs here. In less happy news, Kronos, which is the crypto.com backed L1, they halted the chain after an attacker drained $75 million from their largest lending protocol.

I'm not up on Kronos, I have to admit. There's just only so many things you can pay attention to. That might be the first time it's ever. come up on the show. Yeah, I don't know. The fact that a hacker was even aware that Kronos existed, you got to tip your cap. Elsewhere in DFI, Athena, has launched Athena Pay. That's a self-custody payments app on Avalanche in beta across 48 countries,

letting users hold balances in the stable coin or in yield, spend at merchants. Another DFI protocol moving into more of the full stack game. Yeah, and definitely listen to. to the podcast that Wyatt did with Guy Young, the founder of Athena this week on our podcast, definitely worth a listen. He goes into some detail on this. Yeah, Athena has been through a big change. I mean, the entire structure and the core of the protocol has dramatically shifted over the last year. Revolute was in the news this week. They won conditional OCC approval for a U.S. National Bank Charter. It's their second attempt. I didn't realize they tried for a state license in

2021 in California, but they never got approved. But what an amazing company Revolut has turned out to be, huh? Yeah, 2021. It was really a different world, you know. I mean, things have opened up so much in terms of charters now, both at the state and the federal level. So speaking of the Fed, actually, after we were in Jackson Hole, there was another group of individuals in Jackson and hole shortly thereafter, the central bankers. And there's a speech that I want to discuss from Wangshendu of Harvard Business School. She's an economist. And I think she's ex-New York Fed. And the speech is three myths about payment innovation. And actually the first two myths were, I think she did a good job of debunking them. They were real-time gross settlement is the future wholesale payments.

retail payment frictions are mainly a legacy technology problem. I actually don't disagree with her at all. The third one, the third myth that she's busting is stable coins are efficient for cross-border payments. So that's a myth. She doesn't think they're efficient? No, and it's interesting because she did a case study on a certain company name of Felix Pago for Harvard Business School. Oh, I've heard of that. I think that she would actually know a little thing or two about,

stable coin payments being efficient for cross-border. So here's her argument. And you give me your thoughts. Quote, using a stable coin for fiat-to-fiat cross-border payment requires constructing what we call a stable-coin sandwich. The sender begins with local fiat currency and then converts it to the stable coins. Because 99% of stable coins are denominated in US dollars, this on-ramp can entail a fiat-to-stable coin FX conversion. The stable coin then travels across the public blockchain at the destination. it must be exchanged for the recipient's local currency

and off-ramped into bank account or digital wallet. The blockchain is only the filling. The two slices of bread are the conversions between fiat currencies and stable coins. These conversions add intermediaries on an off-ramp fees subject to FX liquidity constraints. So basically your argument is, and I think this part is just factually true for the most part,

the blockchain element is only part of the sandwich. Yeah, that part's true. But she then sort of uses this to dismiss the efficiency of cross-border stable coin payments. Yeah. So I actually agree with some of what she's saying, but I think the market is moving really fast. And I think what you're seeing now is that, yes, there is a lot of conversion into local currency. I think there's going to be more of a propensity to save in dollars in some of these geographies. So I think you're going to see more of that where a recipient will receive the dollar.

and then just spend dollars. And we've talked about this in the past around just spontaneous dollarization. I think that'll happen. But the other thing that I think is likely to happen is that local stable coins, although they won't be a dominant force in the market, will be a necessary lubricant, where you can imagine sending a U.S. dollar stable coin into Mexico, having that converted perhaps into a, you know, peso stable coin. And then that peso stable coin goes into the bank and is fungible one to one.

for bank deposits. I think that'll happen. All that being said, I think it's already way more efficient than sending this the traditional way transfer-wise and Western Union. Because what that analysis, I don't think, takes into effect is just the enormous capital cost that is born at these legacy remittance companies, where they need to just tie up capital in so many of these geographies. And so maybe it's not necessarily the retail person that is absorbing the friction, but the friction exists in the system. So they're facilitating a quote-unquote real-time payment, but it's not actually in real-time. It's just a shifting of omnibus accounts. Yeah, she does mention wise and remitly as examples of why you don't need stable coins to do cross-border, which is true, but I think she's

missing out on the hidden cost there, which is balance sheet. It wise and remitly, they have huge balance sheets. They have to type capital in hundreds of geographies. Yeah. So someone's paying for the efficiency. I think she's suffering from a failure of imagination as well, because if at any point, a closed loop stablecoin native economy develops, then you've eliminated the need for some of that on and off ramping. I presume it will. And then I think she's also making a bit of a factual mistake in saying that she basically implies that you have to bootstrap stable coin, dollar stable coin to foreign FX liquidity from scratch. So she treats USC peso as, for instance, a brand new FX pair that has to be bootstrapped from zero. But we know that's not true. Like the dollar

side is treated as fungible. Yeah. It's genius. You know, USC or Tether, whatever, that's just treated as dollars. So really, we're just talking about the dollar peso market. And increasingly, stablecoin PSPs are able to get something close to the wholesale rate for that. You get banks that are now intermediating these transactions. It's not just a liquidity on exchange. So I think she's getting that wrong. You know, these markets are more integrated into wholesale effects than she thinks. Yeah.

It would, to be fair to her, it would be impossible to know that if you weren't actually a practitioner. Yeah, I mean, we're probably some of the best position people on the planet to understand. and that maybe it's not her fault that she hasn't figured that out. But maybe she'll listen to this podcast and write a paper about it. Yeah, maybe. I don't know. I didn't read it, but based on your description, I agree with some of it. I think it's a fair analysis.

There's some finer market structure pieces that she missed. So this happened this morning. RSA 260 has been factored by someone who works at cognition. So what happened here? this is an 862 bit. Is it 862 bits? Yeah. So remember decimal digits, 262 refers to the number of decimals, as in how many

numbers in a row. The number of bits is different. It's always really complicated the units. The bit is more of a measure of its strength. So yeah, 862 bits, semi-prime. It's been unfactored since the 90s, since RSA emerged. someone called Eric Liu posted one of the factors

of this semi-prime this is a big leap forward actually so it's a classical attack to be clear that we don't suspect to the quantum computers are involved this is a leap from 829 bits to 8662 bits

so it is a reminder to be humble when it comes to cryptography because the attacker side is also advancing fairly quickly that's crazy I mean was RSA 260

one of the outlawed um outlawed pieces of math here uh in the munitions uh wars here the crypto wars in the 90s? I think some variant of RSA was but do not be too concerned

because the RSA that we use has 2,048 bits that's why it's called that so this one's at only 862 and it's not like three times harder to solve 2048 as opposed to 862 it's like in scientific terms a guerrillion times harder so it's many orders of magnitude away yeah so we're actually safe

RSA is for now cognition is just one of the hottest companies in AI right now this should be the way they should kick off a fundraising process right now assuming they did this on their own infrastructure it's like hey look what we can do a quantum computer will get it within five years, I think. It will get what? RSA. Yeah, I mean, the bit strength of RSA isn't as much of an impediment. There's another report this week that Payword, the parent company of Cracken, is pushing

back its IPO to the second quarter of 2027. They've filed, I think, confidentially the last year. They're making headlines still, so they announced a partnership to offer hyperliquid perps through their CFTC license bitnomial exchange. change. So I guess that's kind of how hyperliquid comes to the U.S. market, right? Is it's just a regulated front end that uses hyperliquid on the back end? That kind of makes sense to me. Yeah, I got a call from a Bloomberg journalist this week, and she asked me the same thing, and that's what I told her as well. I thought it'd be the Binance U.S. strategy or the

polymarket U.S. strategy. I mean, a totally different pool of liquidity. No, I actually, I think it's going to be the same pool of liquidity, actually. going to be the same? I just don't know if hyperliquid is... I don't know. There's another headline this week, which CoinDesk reported that North Korean hackers are using hyperliquid now. So the DPRK has apparently moved tens of millions of dollars through the protocol. That's not surprising. I mean, I think you're getting different pools. I don't know. I mean, is it, you're going to have to bootstrap a lot of liquidity on the regulated side to actually be compelling, don't you think? I think it'll be very tricky. Yeah. And Binance, US, I

I guess that is the example, but we also found that there was some behind-the-scenes stuff going on there, that the liquidity was maybe not as fragmented as advertised. Yeah, I mean, that example doesn't necessarily inspire that much confidence. Yeah, there's a penalty there. All right. Well, I think that is it for the week. Should we play ourselves out with the bad boys, the new song? Sure.

Well, it's just called Song Two. Song Two? Song two. Song two. All right. Everybody have a safe and healthy weekend and we will see on Monday.