On The Brink with Castle Island - Weekly Roundup 06/12/26 (Strategy survives, Zcash Orchard bug, the thin model hypothesis) (EP.725)

Episode Date: June 12, 2026

Matt and Nic are back with a new week of news and deals. In this episode:  Spanish soccer club Osasuna hedges their relegation on Kalshi Strategy shocks people by buying the dip Is Strategy sacrificing MSTR to save STRC? Saylor bolsters his cash position What factors are dragging on the BTC price? Zcash fixes a scary inflation bug SBF formally applies for a Presidential pardon The CFTC proposes banning a subset of prediction market contracts Polymarket thinks that Kalshi is spying on them European fans in America for the World Cup are discovering Buc-ees The DATs are struggling Hester Peirce gives her farewell address Japanese banks are launching a joint stablecoin Will there be offshore interest-bearing USD stablecoins? Anthropic drops Fable  The thin model hypothesis What could pop the AI rally?

Matt Walsh and Nick Carter are partners at Castle Island Ventures. All of these expressed by them or the guests on this podcast are solely their opinions and do not reflect the opinions of Castle Island Ventures. Guests and host 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 for informational purposes only. Brought down by bad mortgage investments, Lehman, which has 25,000 employees, will be liquidated.

The federal government loans American International Group, AI, $85 billion. This is a different kind of market, and the Fed is asleep. The federal government is stepping it to stabilize Fannie Mae and Freddie Mac, the two mortgage giants that have been threatened by the housing crisis. The Bank of England has pumped 75 billion pounds more into Britain's ailing economy with a new round of quantitative easing. You print a couple trillion dollars, and all of a sudden, people start to worry.

So out of this worry, we have something called a Bitcoin. Welcome to On the Brink. I'm Matt Walsh. And I'm Nick Carter. And you wanted to start this one out by talking about relegation. Yeah, so we don't have, it's not really a concept in American sports, right? No, I think it should be. I like the idea. If you do really badly, you go down to the lower league.

Yeah. It could work in baseball. You were saying it could work in baseball. Oh, yeah. I mean, I think the Red Sox are a minor league franchise right now. So they should be down there. So in English soccer, there is a whole pyramid. So it's like the Premier League and then the championship and then League one and league two.

That's the order. which doesn't make sense. The naming needs to be worked on. And if you're in the bottom three teams in the Premier League, you go down. And if you're in the same vice versa from the... And it makes a huge difference. If you get promoted from the championship,

you make something like $100 million more the next year because you get TV money. So all this is to say in Spain, they have the same thing. So there's a La Liga club, which we think is Osasuna. They were facing potential relegation they placed a bet against themselves on Kalshi, apparently.

I think it was through an intermediary. So they were hedging the financial hit of dropping down to the lower tier. Wow. So this is an instance of something we've thought should happen, which is people starting to use prediction markets for actual hedging. Yeah. Which they kind of need. They need corporate buyers to start using it as a hedge,

not just have speculative usage. Here's the kicker. they didn't get relegated. They didn't. So they just bought the insurance, basically. Payed up for expensive insurance, and they survived. That's an interesting use case. You're right.

That is like an actual economic utility for these prediction markets in a way that a lot of this is not. I wonder if that's legal. Is there any law against that in sports? Apparently, they bought it through an insurance provider, and the insurance provider bought basically reinsurance. on Kalshi. Oh, interesting. I could definitely see that. I mean, buying it through an insurance provider seems run of the mill. Apparently it was Susquehanna that did it. So this is actually what polymocrate and Kalshi need is corporates starting to hedge as opposed to just retail trading

on things and speculatively. It's interesting that it was Suscohanna on the other side of it. I do think you're going to see this growing market of companies like Susquehanna, these market makers. and prop trading firms that go out to brokerages and directly engage them for over-the-counter of prediction markets and don't actually settle it back to Kalshi and Polymarket. Yeah, that would be, I guess that's a risk factor is that they go off exchange. But I agree. The OTC market's pretty big. I mean, a lot of that will hit exchanges at some point, but I think you'll see these desks

just warehouse a bunch of that risk. So there's a lot of other interesting things to talk about, but maybe first, let's do the deals. Deals of the week. First one up is a big one. Morpho, the on-chain lending protocol. They raised $175 million from Paradigmund, Indrecent crypto, and Ribbett Capital. Vinyl equity is a tokenization-focused transfer agent. They raised $20 million from Jump Capital, Mitsubishi, and index ventures. Transfer agents are hot right now. In a world of tokenized security, the TAs are critical. So we'll see if the existing TAs move into that space,

and I expect we'll continue to see net new TAs started. Next one up is TVL Capital. This is an on-chain platform for structured products. They raised $5 million from framework and flow traders. Then we have Kiavi. That's an AI-powered lending platform for real estate. They're acquired by Figure for $717 million. Interesting deal there for Figure.

Kyavi, I think they cater to making loans to folks that want to flip houses. That's kind of where they had their initial wedge. Yeah. Next one up is MNX. This is a AI-focused futures exchange built on the mega-eath blockchain. They raised $6.4 million from Village Global, Cambrian, and North Island Ventures. And lastly, we have neuro-robotics. That's a humanoid robot manufacturer.

There's $1.4 billion in around led by Tether. Who plants integrate its wallet software into Neuros robotic systems. Wow. Wow. Is that a thing? We're going to have robots with stablecoin wallets attach them? I mean, these robots are going to be trundling around and they're going to have to find a way to pay for things.

Why not tell them? Who do you think stands to lose the most from robotics? I think the landscapers are probably going to be in a tough spot. Yeah, I was thinking. The obvious one is like delivery drivers. I mean, that's not humanoid, but that's the near-term one with Waymo and all these cocoa robotics running around and cabbies and truckers. I think that's the near-term one.

But it's funny that people think AI is this purely white-collar thing, but it threatens all labor. Oh, definitely. Yeah. I mean, is shoveling snow, do you think it'll just be the job of humanoid robots at some point in the future? But there's a kind of beauty in shoveling snow, you know, I think it's quite an enjoyable task. Maybe not when you live in Boston. Yeah, within reason.

A few inches is good. I actually don't think I would get a robot just to shovel my snow, but I probably would get a robot to mow my lawn. So there were a couple big, big news items this week. First one, everybody was speculating about strategy. People thought strategy had sold a lot of Bitcoin. People thought if it came out on Monday that they had sold, that would have actually been very positive.

because it would have explained Bitcoin's collapse last week. That was not the case. They did not sell any additional Bitcoin. But they did top up their reserve slightly. Yes. Through common stock sales. And then they bought more Bitcoin. So they did the opposite of what everyone thought they were going to do.

They did buy 1,500 BTC. Have you heard anyone that's come up with a good explanation for why Sailor paid off the convertible debt, the $1.5 billion, three years early. Yeah, I've seen some convoluted explanations, but I wouldn't do them justice. And unfortunately, I'm not enough of a sailor whisperer. Yeah, I mean, I guess I'll caveat that I don't hold this security and I don't really follow it day to day. But, I mean, he took his cash position from two years of cash to service the preferreds.

Cut that into like six months or something by paying off the convert. early. Then he sold Bitcoin a couple weeks ago, spook the market. And I was of the view that, you know, if over the weekend he had sold 1.5 to 2 billion of Bitcoin and had enough cash to service a couple of years of the prefers, then that would have been really positive for the market. He kind of did the exact opposite. So he sold more common, sold over 100 million of common, bought some Bitcoin and kept some of it in cash. So seems to be really, uh, doing things at the expense of the common holders. So actually, when this drama was going down,

I fed Claude all the information. I'm like, what do you think Sailor is going to do? And Claude said he'll protect STRC at all costs. Really? Digital credit is the whole franchise. That's how they've been buying Bitcoin over the last year. MSTR is the loser. You know, there's only two ways to fund that, really.

You can issue common. They are slightly above their version of, MNAV. And so in theory, they can still issue non, non, well, it's dilutive, but non-destructively. Or they could sell Bitcoin, but that's, you know, dangerous. So it seems to me that this is the direction of travel. It's protecting STRC, keeping that dividend alive at the expense of the common shareholders. Kind of some scary downward moves in the price of Bitcoin, especially knowing now what we know that strategy wasn't selling. I mean, that was pretty violent few days at the end of last week. Yeah. So there's this bigger seller. Maybe everyone is selling. Maybe this is a SpaceX IPO funding.

A lot of people think that's it. There's a lot of questions. I mean, you have confluence of factors. What would be your kind of top three to five factors on why Bitcoin's not going well? And I'll go first, maybe. I'll say there's, you know, there's strategy. There's just kind of this looming question around a huge holder of Bitcoin and whether or not he's going to have to start selling. So that's one. Two is, I think, quantum. I think people that are close to the ground on quantum see how fast it's moving and see that there's not a roadmap. Three, I would say, is just macro. I mean, hot CPI print. I think there's just a lot of uncertainty on the inflation side. So those would probably be my top three. Yeah, my list is very similar. Number one, I think, by,

far is just sailor, you know, he was basically the only big buyer in the market that stopped. It's not even so much the fear that he would sell. I think it's just the lack of him buying. So I think that's the number one feature by far. Number two, I agree is quantum still. That's kind of a slow bleed. But you see Zcash and we're told, talk about them in a minute they have their quantum roadmap. Ethereum has it. Salana has it. Bitcoin really is behind. relative to everyone else. And there's a lot happening on the quantum front, actually. There's this ECDSA. fail challenge

where people are optimizing Google's circuit using AI. AI is clearly and demonstrably increasing the pace of mathematical discovery. So I think that is a very material fear. And then, yeah, I think macro, the Iran war heating up again. And certainly I think the triple IPO threat here, I mean, I think that is real. I think it is sucking a bit of liquidity out of virtually every other asset class. Yeah, that's a good point. I didn't say that one is I think there's just hot money going into other sectors right now.

Maybe another one I'll add is just I think a lot of just the retail fast money is probably in the prediction market game as opposed to buying cryptocurrencies right now. So, C-Cash had a crazy week. AI is pretty good, as it turns out, and security researcher using, I think it was Claude, found a critical bug in Zcash's orchard shielded pool. The problem with this is that inflation bugs are particularly bad in privacy coin contexts. And Zcash has had them in the past. Zcash had one notable one that was patched and Manero had one as well. And in a privacy coin, you're basically trading off auditability for privacy.

So while no one thinks this bug was exploited, we have no way of knowing right now if it was. I think you could reasonably guess that it wasn't, but we can't know for sure that this bug wasn't exploited. Yeah, it's scary stuff. I mean, what year was the last one? It was 2018 or 2019. I remember we talked about it a lot at the time.

Yeah, and they used some kind of turnstall, I think approach to prove that it wasn't, hadn't caused inflation. I mean, this is the same. I think Zcash has the same plan now is basically pull the funds out of the shielded pool. And if the right amount of funds come out of the pool, then you know that there wasn't inflation. And I believe that you can't spend, you know, to an exchange directly out of the shielded pool. Right.

The attacker wouldn't necessarily presumptive attack or wouldn't necessarily been able to monetize. it directly. So I think it's reasonable to say that probably this wasn't exploited, but it is scary. It's a reminder that, A, these AI tools are making these attacks much easier for regular folks or even for security researchers. And B, in a privacy coin, there's always a little bit of a question mark. So I'd had a material impact on the prices of Zcash over the past week, understandably. In other news, SBF, he has formally applied for a presidential

pardon. You think he's going to get it? Yeah, I mean, I think a lot of people saw this coming with his pro-Trump tweets despite him being a massive Democrat donor. And I mean, he was as liberal as they come,

from what I remember of him back in the day. He was a vegan, effective altruist, consequentialist. about their screamed Maga at all. He also gave a billion dollars to Joe Biden, didn't he? Yeah, I mean, he was the biggest donor in the 22 midterm. So his turn to being pro-Trump, I think, was pretty cynical.

We'll see if it pays off. I don't think it will. I don't think he's sorry is kind of the, don't you have to be sorry for what you did to get a part of him? Because he hasn't really taken responsibility, right? No. No. I mean, he still tries to talk around what he did even today, years later. This was an interesting thing this week.

So Janice Henderson, one of their subsidiaries, took a stake in Athena's ENA token. And they're going to use staked USDA for cash management. Really interesting development there to see Janice Henderson getting in the game aggressively with Athena. Yeah, and full disclosure, we are holders of Athena. Athena seems to be moderating the strategy. a little bit, doing less of the basis trade and diversifying their asset management a little bit. Coinbase, I think, also took a stake in Athena recently. Yeah, that's right.

Morgan Stanley has announced that they're going to refer wealth clients to Galaxy, allowing them to lend their crypto in exchange for shares of spot crypto ETPs, including Morgan Stanley's own Bitcoin trust. You're seeing more and more of this where you have folks that own Bitcoin in a spot capacity, turning it into the ETPs. I think the interpretation from a lot of these firms is that that is not a taxable transaction to take Bitcoin and turn into an ETF. Again, I'm not a tax advisor, so look into that.

But a lot of people are treating it as if it's not taxable. And it just seems like that's a popular thing right now. And I think part of that is the wrench attacks going on. Part of it is just people don't want to have to worry about inheritance planning and how your assets are going to get to your next. kin, but definitely something to keep an eye on because I think most of the banks will probably end up facilitating this at some point. There's a lot of prediction market news this week.

The CFTC has said they have a rulebook coming. These proposed rules would let regulators reject certain contracts that are vulnerable to manipulation, potentially banning any contracts that have to do with war, terrorism, or assassination, and a subset of sports contracts that are potentially. easily manipulable too. Hmm. That's interesting. So they're going to put out rules here.

And then I think these rules will get challenged. And I continue to think we're going to end up in the Supreme Court here on the sports betting. So, I mean, it kind of looks like carte blanche, frankly, on the sports betting, aside from some niche contracts. And then the thing that I actually am uneasy with is the war contracts. I think that is one of the main things you want to hedge probably. Yes. Right. And so I know we've had cases where, you know, members of the military have bet on, you know, they've basically improperly monetized their information that belongs to the government doesn't belong to them, bet on military operations. I don't think they should throw the baby out with a bathwater, though. I think you just should crack down more aggressively through existing methods, crack down on insider trading or members of the military.

trading on these contracts, but keep them going. Because these are actually some of the most useful contracts out there. They really are. I mean, I think that's the highest signal in terms of anything on prediction markets outside of maybe political elections. It's just, will we be at war worth a certain country? Polymarket has sued Minnesota and New Mexico has sued Kalshi. Illinois wants a tax per wager.

So there's the state legal warfare. over these things continues. This animosity between Kalshi and Polly Market, do you feel like it's at a fever pitch right now? So the New York Post reported that Polymarket believes they're being spied on by Kalshi. And they apparently keep a dossier that documents a dozen incidents that it believes show a suspicious similarity

between the two firms. Of course, it is true. I mean, Kalshi and Polly Market are imitating each other with various product announcements and marketing stunts. I mean, I think they were both doing a free grocery thing in New York at one point. So apparently, Polymarket thinks that there is a mole in the office who is leaking to Calci. Wouldn't actually surprise me if that was the case. I have zero evidence, but I completely believe it.

I mean, there's a lot of chicanery going on. The PR stuff is just crazy because, you know, one week it's a wave of stories bashing Polymarket, then you got a bunch against Kalshi. It's just, I wish these guys would just settle down a little bit. So Kalshi for their part called the stories the claim, sad and borderline delusional. Oh, okay. I was watching the, have you been watching any of the Stanley Cup finals? No, there's just so much sports going on right now.

There's the basketball. Everyone seems very interested in the basketball. I'm not a basketball fan. I don't know what's going on. There's been some good games. I was watching the Stanley Cup the other day, and it was Kalshi banner ads on the, you know how they,

it's white boards now, so they don't actually put up physical, you know, billboards anymore. But it was Kalshi, and then like five minutes later, it was all polymarket. Those guys are spending a ton of money on customer acquisition.

It's like the most intense period of sports ever right now. Now, the World Cup starts tomorrow. The, probably the biggest UFC card ever and its history happens on Sunday, the one at the White House. Oh, yeah, that's on Sunday? The French Open was last week. The Stanley Cup is going on. It's the NBA finals. And again, the World Cup is starting.

In the U.S. Open golf is next week. And that, and that too. So it's an amazing time to be a sports fan. Now, you haven't reconsidered. your no World Cup posture. Well, yeah, I guess I should report on that. So I was invited to the World Cup. And unfortunately, I think I have a conflict,

but I would actually like to go. I've reconsidered going. Yeah, I mean, if you were free, it's just a great experience. Although I think FIFA is kind of, they've soured a lot of people's views. People feel that they're nickel and diming them on the tickets. And certainly they're squeezing every.

ounce of juice out of the fans this time around. Yeah. It just seems like a, I don't know, probably not once in a lifetime, but the World Cup probably won't be back here anytime soon, right? No. I mean, we had it, I think, in 94 or 96 maybe, and that was right. It has been interesting to see all these European fans visiting random American cities for the first time. and they're kind of discovering, I think a lot of them haven't been to America before. And so they're discovering like Costco and Buckees and Target. And they're,

some of them are, you know, reporting kind of a childlike joy at seeing like a gas station that's on 50 acres of land. That sells pizza. Yeah. I mean, they just don't have that in Europe. No, they doubt. So it's been quite,

entertaining to see them discover, you know, what America has to offer. Yeah, I mean, can you imagine walking into a Costco at the age of 25 and, you know, never having seen that before, you'd probably be aghast. It's like, which was the Soviet leader that went to an American grocery store and realized that communism had failed. Oh, yeah, yeah. Who was that? Was that Gorbachev? Might have been Gorbachev. Do you have any updated views on, uh, on, on who? who's likely to win the World Cup? I'm still pulling for the U.S. Yeah.

We could. I think we'll make it out of the group stage. My official pick is in. My official pick is Spain. It's very boring. Spain. Oh, interesting. It's going to win it.

Yeah. I could see that winning 1-0, something like that. Yeah. In other news, DAT news, I guess you would call this company a DAT fold, a publicly traded Bitcoin company. They announced that they have sold

$45 million worth of Bitcoin. They have paid down some of their debt. It looks like they took out a U.S. dollar loan against some of this Bitcoin. That's always a tough thing to do. We're going to shore up the balance sheet, it sounds like. Yeah, I mean, Fold is also an app. I mean, they're not a pure dat in that they have a product and users and things like that. But the poor Dats, I mean, the DAT sector is just down terrible right now.

Down real bad. And I don't know how this resolves itself because I don't see activists coming into the Dats. aggressively. I think there's the only thing is the Dats, they're all trading the discount. They sell the coins and they buy back their shares. So they just shrink, basically. But then what are you? Yeah. I mean, then there's no reason for it. But I mean, in a world of ETFs, abundant ETFs for any coin you want, why do we need a debt? A bunch of these dads, and I haven't seen this widely covered, are getting notifications from the exchanges that they're not hitting the minimum listing requirement. some of them are not filing

SEC updates in a timely manner either so I think from just a corporate governance perspective you're going to see a bunch of these things just really drift from what is required yeah I mean you've already seen reverse splits in some cases and

you will certainly see delistings here too yeah so the humanity protocol that is I hadn't really heard of them they are a palm scanning identity project they were exploited for $32 million via compromised private keys, and the token fell 80% on the news.

Yikes. That doesn't sound good. This just came across, and I guess I should say we're recording this on a Wednesday, but Hester Purse has just put out a speech, which is her farewell address. So we'll have to talk about her tenure, when we have more time to digest this.

But what a great run she's had. Yeah. I mean, she went from kind of the protest vote at the SEC under Gensler's tenure. And she was a rare voice of sanity and, you know, defended the crypto space when it really wasn't popular in Washington. And then, you know, was able to actually effectuate a lot of the changes she wanted to make and had been arguing for over the years. So she has just done a tremendous job. Yeah, those are big shoes to fill. It's actually, it's interesting that we haven't seen any Democrats nominated to the SEC yet. It's kind of odd to me.

Crenshaw was never replaced. So is the SEC short? Yeah, it's short commissioners right now. Oh. And so the CFTC is too. I mean, C-LIG's the only one over at the CFTC right now. Quite odd. This also just dropped today.

Japan's three largest banks, MUFG, SMBC, and Mizuo are forming a council. look for joint stable coin issuance by March. Interesting. The U.S. is not the only stable coin country out there. Now, do you think that will be dollar stable coins, or will that be yen stable coins? I have to imagine it's yen. I think there's an interesting play for non-U.S. banks to issue stable coins in a complete disregard to genius and just create a euro dollar market for stable coins,

have them pay interest. And that just, that would have to be offshore. they wouldn't be able to list with the U.S. exchanges or anything. Yeah, exactly. But you want to see interest-bearing stable coins take off. You just have really reputable firms launch them outside the U.S., and they will almost certainly take off. Yeah, I think that's right.

So I think you'll see in the U.S., you'll see more complex structures like SOFi has the tokenized deposit plus stablecoin. And I think you're right. Overseas, you're going to, we're still going to have interest-bearing. stable coins dollar to nominated, but they will be walled off. I said there was something this week from someone at JP Morgan that was saying that tokenized deposits have not seen a lot of demand. And it was almost said as like an indictment of blockchain technology. It's like, yeah, no kidding, there's no demand for tokenized deposits. Why would you want a tokenized deposit when you could have a stable coin?

It's kind of funny that JP Morgan is saying this. I mean, they are some of the biggest deposit champions out there. Yeah, I mean, I don't think tokenized deposits are really going to be a thing. I mean, it's... Well, I think if you're a platform, you can funnel some of your users into them, and it might just be invisible to the end user. Yeah, I think that's probably right. But there doesn't seem to be that much utility outside the platform.

I don't know why. You'd see mutual acceptance of tokenized deposits. There's just not fungible, right? I mean, a tokenized deposit from JP Morgan ought to... not trade at the same parity as Deutsche Bank tokenized deposit. They have different risks. Well, I mean, in theory, you should hope that it would. I mean, you trust a wire from both institutions.

You're just, you know, a deposit is basically a deposit into an entity that is doing fractional reserve real estate lending. And so, like, that business at Deutsche Bank is a lot weaker than that business at JP Morgan. So it would lead me to believe that a, above the FDIC insurance threshold, those products should not trade a parity. Yeah, that's a good point. I mean, it's certainly nowhere near as secure as a stable coin.

I find it funny that people attack stable coins with these arguments from like five years ago, as if genius isn't the law of the land. Right. You know, they'll say, oh, these things should be regulated while they are. They'll say, oh, well, you don't know where you're, you don't know what's in there. Like it could be backed by anything. It can't. Right.

Or, oh, well, what if there's a run on the underlying assets or the underlying assets, you know, have liquidity issues? We're talking about short-dated treasuries. If treasuries have liquidity issues, then the world is ending, basically. Yeah, I mean, we have bigger problems. So when was the last time the U.S. defaulted on their debt? It's certainly not a long time. Yeah, not in the modern era. I'm not ever really. I mean, the Confederate states defaulted. I don't know if that counts. Yeah, I mean, we've shut down a central bank or two in the early days of the republic, but it's been quite a while. Well, I guess you could argue 1971 was a sort of default.

From a, it's more of just a, the composition of the dollar changed from gold to nothing. Yeah, I mean, I, I find it odd that the. critics don't engage with the reality of stable coins and they argue against a hypothetical world in which genius doesn't exist? I think it's an easier position to stake out as an incumbent that's fighting against this and just hope that people don't pay any attention. Speaking of lack of fungibility, I also struggle with that in the compute markets. So people talking about tokenizing inference and making that fungible. And there's a good episode actually with Brandon McBee from Corweave on Odd Lots talking about this a little bit.

But inference is just not, it's not a one-to-one comparison based on how you actually produce that inference where it's stored. So I really struggle to see how tokenized compute is going to pop on the scene. Yeah, that's a very good point because one provider is not the same as another. Different neoclouds have different capabilities. uptime differences, latency. They're in physically different places. So if you need low latency inference,

it really depends where you're sending it. And then the neoclouds have different networking. And then, of course, there's different models. So I agree than different GPUs. I mean, there's so many variables. It's hard to flatten that into a single commodity. I think the reason why you hear these big exchange groups, is talking about it is because there's a ton of demand to just be long inference compute.

But I don't think there's a lot of demand to be shorted. Do you? Right. Yeah, I don't know who's the seller there. So speaking of which, have you used Fable yet? I have. I think we're just burning through our credits over here. Yeah, I think they might be gone. I don't like how much safety there is. There's a lot of safety. There's a lot of safety. There's a lot of safety. There's a lot of Yeah, it doesn't teach you how to make a quantum computer. It doesn't do that. You can't synthesize novel pathogens. I mean, that's good.

But you can't do anything really. And it thinks that you're doing something wrong. It's like constantly accusing you of doing something wrong. I don't really like how much of a nanny state Anthropic has become. So I'm not thrilled with that. The thing I worry about is if they're just the sole winner, so then Anthropic decides what kind of. kinds of questions you're allowed to ask unilaterally.

Yeah, I think we're going to have, it's a big opportunity for Open AI to release their model with lower guardrails. I'm going to continue to believe, you know, critical queries and extremely long, lasting chain of thought queries, long duration, agentic tasks, those are go through the frontier models. But then the vast, vast majority, and I think 99.9.9. or more, we'll just go through cheaper open weight models. Yeah.

And I think as growth, the growth hasn't leveled off yet, but as it does, it must eventually. It surely will. They can't just keep getting exponentially smarter forever. It seems impossible. Then the open weight models will be competitive. I think they'll be more competitive, but there's just not enough software at that intersection to operate it, right? I mean, I have agents running tasks right now.

And if I wanted to switch those over to open source models, it's still pretty clunky. Yeah. So we need a routing layer. And I think right now, these concepts don't really exist. We don't really have intelligent model routing. And we don't have auditability into sort of like how many tokens we're using is the usage efficiency. So we're going to look back in five years and be like, that's crazy.

You didn't have routing and you didn't have auditability. Yeah. These are going to have to be built as layers between you and the inference. Yeah. I do think, though, that, I mean, look, we might be in a bubble for AI, who knows, but we were just talking about robots at the beginning of this. And if you have a world where, you know, there's robots going around making food deliveries, there's just going to be much more of a need for compute there to power some of these applications.

So you could see this growing, at least at the compute layer for a long time. Yeah, and of course, if compute gets cheaper, if a given unit of intelligence gets cheaper with an open-weight model, for instance, that doesn't mean we're going to buy less of the commodity in aggregate. Right. In fact, it's probably the opposite. We're going to have the Jevins effect, and we'll buy more.

Yeah, I think that's almost certainly what's going to happen. Yeah, people think that moving away from the frontier models is the thing that pricks the AI bubble. I don't know if I strictly believe it. I think it can dent the story for a little bit because the credit quality of the big labs might deteriorate if there's some commoditization. You see a couple headlines around, okay, we're not going to go do this build anymore. This data center's on pause.

The CDS on some of these hypers, maybe spikes up. There could be a little panic there in capital markets. Yeah. I mean, we're in the middle of one right now, just in the last kind of 48 hours. I think the real risk is political. I think it's you get these rallies in every city, every municipality, where there's a data center being contemplated. You get, you know, environmentalists shoot them down.

I mean, even non-environmentalists are skeptical. And then now there's this notion of sovereign ownership. So now even Trump is discussion with the labs. Should the taxpayer have an equity stake in the labs? I don't know how you effectuate that. They are kind of too big to fail, though, right? I mean, frontier model providers, you can't have them go bankrupt and sell for parts to China. That's not an option.

And also, it's politically unacceptable in a way for these companies to grow to 10 or 20 trillion in capitalization, putting a third of Americans out of work. That's not a tenable outcome. That outcome people talk about, it couldn't happen because it wouldn't be allowed to happen politically. Yeah. So how do you share it? the wealth. Maybe, yeah, a resource endowment for the taxpayer is the right thing. It's kind of seems like somewhat unprecedented, actually. It is. I'm sure there's parallels to other disruptive technologies, but it feels like the most disruptive technology that we've experienced in the U.S.

I think what's more likely is it ends up being heavily regulated and the margin gets regulated away. So the government ends up dictating to the model companies what their pricing is. And they're just legally not permitted to charge too much, the same as a utility or an airline. There's just so many other revenue streams these guys could have, though. I mean, they could take out Expedia tomorrow if they really wanted to. Right. But these horror scenarios where the model companies eat all economic activity and grow, enormously even from here whilst Americans are being put out of work, I don't buy it because I think

as that begins to happen, there's an enormous political backlash bipartisan against them. I think that's probably right. We didn't cover it at the outset, but there's some good Castle Island content this week. Wyatt sat down with Amid Malacan professor at Columbia Business School talking about stable coins and crypto in the geopolitical climate. I thought that was a really good pod. Yeah, Omead, Barry, erudite and clear explainer of blockchain's great episode. And I think that is it for the week. Again, we're recording this one on a Wednesday. So apologies if big news happens on Thursday.

We will cover it next week if that happens. Everybody have a safe and healthy weekend, and we will see on Monday.