Loading summary
Transcript99 lines
- [00:00]
Michael
It's the thing that sucked the air out of almost everything. And whether it's finance, digital assets, or tech is what's happening on the AI front. I think it's. It's going to be interesting to see how fast it catches up. But you can have all the technological innovations, but you still have to coordinate economic activity, whether it's via the different agents or ultimately, how do you like these tools are downstream of what are the goods and services that the market's delivering. Whether it's financial services, plumbing, you know, somebody still has to go market that, get to the market. The point being is that the winners aren't going to necessarily be the people just holding Bitcoin or developing Bitcoin services or developing the best API tools. It's going to be how do you merge the best form of money along with AI?
- [00:49]
Liam
It all comes down to computers communicating.
- [00:53]
Michael
The information superhighway can be a confusing mix of on ramps and off ramps. Bitcoin is worthless artificial gold.
- [01:03]
Liam
Is it still rat poison?
- [01:04]
Michael
Probably rat poison squared.
- [01:06]
Host
We need to get into the world of.
- [01:08]
Liam
Okay, this is actually foundational technology.
- [01:11]
Michael
What the Internet of money does is it creates a single network which can do a micro transaction to a giga transaction. The Internet is going to be one of the major forces for reducing the roll of government. The one thing that's missing that will soon be developed is a relia ekish.
- [01:39]
Host
Alrighty, gentlemen, welcome back to another episode of Final Settlement. Today is Monday, June 22nd, 10:40am Eastern Standard Time. Gentlemen, icks are world champions. The United States is good at soccer. Anthropic released mythos fable and then unreleased it within a matter of days. The Iranian conflict is, I think, still going on the Schrodinger straight. Is it open or closed? We're not really sure. How could you be? Gentlemen, how's it going? Michael, Happy Father's Day. I hope yesterday was nice.
- [02:15]
Michael
Thank you. Happy apology to you as well, Liam.
- [02:17]
Liam
Soon.
- [02:18]
Michael
Soon Liam will be a daddy. He's somebody's daddy somewhere. Anthropic. I can't put my finger on it. Is it a show or is it really, like just a fumble?
- [02:31]
Host
No, there's no way that that wasn't orchestrated. Like, release it, have it be out for, you know, a couple days, and then be like, oh, the NSA says it broke all of its system. We've got to hide this thing.
- [02:44]
Michael
So. So it could be a show, but let's work down the sequencing because there was a few different pods, like triangulating, and you guys can clear up If I'm, if I'm missing anything is they rolled out. You know, they had this model, I think, at least six months before.
- [02:59]
Host
Right.
- [02:59]
Michael
Because I was recently listening to that, that pod we were talking about, Dylan Patel, and I think that was an older pod, like three to five months ago where they're bringing up Mythos. It was when 4, 6 or 4.7 came out. I think it was 4.7. So that was a while ago. And then that was discussed. And then let's call it 60 days ago, they released it to roughly 50 important companies. And one of those companies is reported that was like, tied to China. And so they were already kind of in negative light of like, hey, you know, you, you potentially leaked this, whatever. So then that's in the process. And then Fable gets rolled out with the guardrails. And the thing I didn't necessarily know is that Amazon is one of the largest backers, like, like seed or early investors, along with obviously the Data Partnership or the, the. The AWS infrastructure. So they have every interest in them succeeding from a game theory investment perspective. They try to call him. He doesn't answer, doesn't. They can't get, get a hold of them. So then they go to Besant and then that's where the whole thing, you know, happens. And then they put the export control, and then I think he. He comes out and has some blog. The point in bringing that up is like, that's a lot of, like, sequencing and orchestration for it to be a full thing. I mean, unless they're, you know, any Jassy and those guys at Amazon are like, oh, yeah, this can actually know. Pump the bags. Because we can construct this narrative that this thing has otherworldly powers. I don't necessarily know, but I just want to break that down because that's important. If we're going to go and say that this is a huge, you know, deal, which I'm usually on the conspiratorial side, but sometimes, like, the most obvious is just the most obvious thing. So anyway, with that, Liam, where do
- [04:38]
Host
you stand on it?
- [04:41]
Liam
Yeah, I mean, I, I can't even figure out Dario, just because he really has the worst messaging that they can possibly have. He calls it a cyber weapon, and then he releases the product after telling the government that it can't be released to the public, it's too dangerous. He gets cleared for 50 companies and then he releases it to. I think it was SK Hynix, which is the Chinese company that is attached to it. It's pretty clear that Amazon is trying to cover their ass because Dario says that this could be a potential cyber weapon, and giving it outside of the guardrails would open it up to massive liability. And so I think that with all of this, it's even more clear that while Dario is just kind of positioning himself as putting it as a cyber weapon, this is. These things are just code. They're going to proliferate and these open source technologies are going to get out there. He's just trying to stifle as much open source and bring as much potential liability and concern around it in order to let his model proliferate and gain as much share into the IPO as they can, is my view.
- [05:54]
Michael
All right, let's.
- [05:55]
Host
I think the other. The only other element to that would be, I think there's a play, and this is what Gurley was talking about a couple weeks ago, where you put this out into the world. You create basically fear around the power of it. And, you know, this would speak to the credence of, like, there is some orchestration here in terms of releasing it and then unreleasing it, because that creates the perception that it is so powerful, that it is a national security threat, all these things. And I think part of the goal would be or could be in that scenario that they want regulators to come in and say, we need to ring fence these things and basically give anthropic, you know, regulatory ring fence and allow them to stomp out competition by having the regulators basically bless them as okay to use but within, you know, a security parameter effectively, and that would help stifle any competition. I think you're right, Liam, that it also has to do with the open source stuff, which, which we'll get to as well, and some of the other headlines here. But. Go ahead, Mike.
- [06:58]
Michael
Yeah, I mean, in that respect I could see, but this is like, I think, not that you were positing this, but like what you were referencing about the, you know, from a marketing narrative, I think is like the 2D version. Remember we talked about it, like personally or behind the scenes. Like, there's a 2D and there's a 3D. There's a 2D is like what everyone, they want you to see.
- [07:18]
Host
So it's like, oh, this is a marketing stuff.
- [07:20]
Michael
The 3D version is similar to Facebook, Google, Amazon or Tether USDC X stablecoin. That there's this illusion of free market, but it's really public private partnership. And that's effectively where Dario and Sam and these, these guys look like they're against each other. When in reality you're just getting pushed into either model, you're giving up all your information. And in that world, and AWS is, you know, obviously a large infrastructure player there, that it is, this orchestrated view of these things are insane. We have to regulate them, we have to guard rail them. And then like what are the second and third order effects from that? In my view, while that sounds like the craziest version of all this, that's actually probably the most likely because it's just the status of like the escape velocity of these tools. And that was the interesting thing with the, the Dylan Patel, his most recent one he was referencing. Like, you know, eventually it makes complete sense for Ken Griffin or whoever to go to like Anthropic and say, here's a billion dollars of forward revenue, let me have the model first. Then you can just take that demand. And so then you have like the largest people with the most capital to be able to get access to the latest models that you're just going to start to see this like bifurcate from like what everyone else has to what these largest firms have. And so I can, I. I don't know. It's. It's a very interesting time, but I can definitely see how there is a growing narrative. And then there's the other aspect that you have to like tie into this, which is the public sentiment because I'm not as close to it, but you hear the people that talk about it and it's like more than I forget what the, the way to contextualize or contrast it was. But it was like pick your worst thing that exists from a public sentiment. And like AI is even worse than that because everyone, these people just keep saying that it's going to take their jobs.
- [09:14]
Host
It's worth being critical. It's worse than crypto, I think.
- [09:17]
Michael
I mean, I think crypto is so bad because nobody even cares anymore. It's like, it is, you know, the worst thing you do is irrelevant or nobody cares. These people actually like. But anyway, yeah, so there is, it's super fascinating. I don't know if we want to go anywhere further here than go into the open source stuff because that's another dimension to all of this, which is going to be fascinating to see how
- [09:38]
Liam
it all plays out before we do that. The one thing that is particularly interesting on the 2D versus 3D angle is you rarely hear Democrats and Republicans agree publicly. But the one thing that both David Sachs, Bernie Sanders and J.D. vance, while they haven't come out and outright said it they've all said it makes sense for the US to get stakes in these large AI companies. And it does make sense when you think about everything from while they were likely collecting all the data on people before this and just storing every prompt and every usage of your token previously. Now Claude specifically says that they'll hold this for 30 days and you can see where it is a national security concern for the US government to least position that they need to be able to see how people are using the tokens in a area like this in order to get the data to make sure that nobody's using it for various purposes. And so I think that that is a very likely consideration on how all these frontier models end up to both with government ownership and the feedback of information directly to the government or through some data broker in the as an intermediary.
- [10:52]
Michael
Yeah, there's, there's so many dimensions and I mean part of like adding this is this is the most relevant thing happening right now. And we, we do you know just in general like AI will be covering it more and we have a lot of stuff planned but I don't know if it's for this pod, but it's like when you look at just the dynamic of China versus the west, like China has the vertically integrated stack and there's a lot to discuss around like the constraints they have with compute and so they had to do more with less. And that's where the DC deep seeking these other models versus like how we've progressed. But I think where a lot of this goes we can probably mirror this for the past hundred years. But the most analogous right now that most people listening and we know is like ETFs versus like self custody. And you naturally it's a little bit inverted because AI started centralized and I think will naturally go the other way where Bitcoin started decentralized just because of the nation, the notion of how people would secure it and digital bearer asset, et cetera, et cetera. But the point being is that I think there's always and there always has been and there always will be individuals that are going to take the most consumer friendly, easiest path passive. But then on the other side of that is the same version of setting up, going out on the curve, buying actual spot Bitcoin, figuring out the self custody, running a no whatever it is is this version of open source figuring that part out. But I think it ends up barbell because all the inference providers and people in the middle might not be able to like sell and distribute from a commercial perspective that if this like ends up happening where you put these guardrails out. And so I don't like actually I think everything's up for grabs right now. I don't think anybody knows how this ends up because the cat's out of the bag. And we'll talk about some of the open source models that are released. But like this stuff's there and I think that what was it yesterday there was from whoever really said GLM about like Musk said like the same level of like Mythos will be out by end of year. And then the, the founder of that company was like before then. And so it's like this stuff's out no matter what. So then it opens the question if it's out no matter what, like what is the real ang here? Is it something else or is it just trying to delay the inevitable?
- [13:04]
Host
Yeah, one, just one tidbit on what you described there in terms of like I thought Chamath had an interesting anecdote that he said he basically bought a bunch of land to build data centers on like two or three years ago with the thought that he would just flip it in a few years. But now he's realizing basically like the vast majority of the infrastructure is supporting these closed source models and if we want to have a future where we can actually have real competition and for on the open source side, like you're going to need to throw infrastructure at that side of things or else it'll sort of like peter out potentially. And so now he's basically saying like, yeah, I'm not going to sell it. I'm just going to build infrastructure myself that'll support open source models. So I think it's not a foregone conclusion that there will be this like healthy competition. I think there's a lot of variables at play there. But moving to some of the other headlines that we wanted to cover here, this was last week Microsoft eyes Deepseek for enterprise AI. So Microsoft is moving copilot to usage based pricing as it expands the enterprise AI tool and it's now considering a Microsoft hosted version of deepseek. So they have a cheaper model option there. Why it matters. You know, Deep Seq is a Chinese AI company and so this caught some, some attention last week. And then Liam, I wanted to share a tweet that you put out last week related to all of this just around sort of open source and what this move for Microsoft means. So maybe if you can just sort of walk through what you were describing here.
- [14:41]
Liam
Yeah, for sure. So Deep seq raised almost $7.5 billion this week for the first time. Now, just like Zai, which is a GLM 5.2 model, they're public and they just hit like a trillion dollar Hong Kong valuation, which I think is like 750 billion US. And these are doing pretty minuscule revenue. But you can see a world in which by proliferating the AI models and ecosystems that people will pay for the API, especially when it's token costs that continue to go down. But on the Microsoft side and essentially frontier versus open source, the main thing that is important to know is that the using the frontier models for everything is incredibly inefficient from a token perspective and it can't be subsidized forever. And so what Google essentially is saying is that allowing folks to use Copilot Cowork for unlimited cases will allow some users who are power users that will use way more tokens than anticipated. And using that only for the best models is just way too expensive for Microsoft. And so what we're going to see is essentially we've had the great experimentation for the past three years and now we're moving to the optimization phase. And so that's not only just for rationalizing costs, but essentially re engineering the endpoints in order to go towards more efficient models for different tasks and moving not just towards the hyperscalers hosting the best largest frontier models, but they're going to increasingly create the systems to host open source models as well. That's essentially the only way that things can go. Otherwise it's, it's going to get way too expensive too quickly.
- [16:41]
Michael
I think that's somewhat right. I don't necessarily. I think there's a few other factors that come into it. Like one of the big ones is whatever the model is, whether it's the number of parameters or as it grows, specifically the older model, they're able to produce the number of tokens at anywhere between 10 to 100x cheaper like month quarter over quarter just by optimizations and efficiency. So I think like you're gonna see it's too early. I don't even know if like the games actually started in all this because as you like start to peel back these layers, you realize holy crap. Like one I think you from just pattern recognition or just the way the Internet's evolved, you end up with the early winners from just like being first to market like AOL was for so long that most people don't know how to use this. It's going to take an insane amount of like diffusion to get people to even understand why they would need like larger models. So you'll have the people at the edge using them. But then you can imagine as people are using ChatGPT or Claude, there's think about how much friction it is to go do any of this stuff, to go and like run through optimizations or run through, or even go and go to perplexity, know why you would use it. And so I think as you do that these firms are also going to be throttling different tokens for different use cases on different models to keep you within that and lower their cost. The other thing that again, I'm taking this at what I heard, but it makes sense, is that it's my understanding that the researchers have been the most valuable people. That's where you saw for the past two years, you know, people like Meta paying a million, $10 million for a researcher. Because what's the value if a researcher takes all that knowledge and they can fine tune a model or fine tune a different algorithm to leverage less compute for more output? So it's valuable there. But the idea that engineers historically were the like not the best engineers playing with AI, but now what's happening in the acceleration, now you're starting to see engineers step in to play with these open source models around the optimization of what's happening there. So we don't really even know what will like be created, the opportunities that sit in between that. And so I think, I mean, I think where all of this goes, it's, it's actually eerily similar to Bitcoin in the sense of like you want to be able to own your trajectory or basically die by your sword or like, you know, the, the idea that specifically if you're a company that if you have proprietary information which is effectively what makes all these things valuable because all the Internet's data has already been, you know, indexed. And then I forgot who was telling me. But it's like the reason part, this is a thesis though, part of the reasons a lot of these outputs are getting worse is because now there's so much AI slop on the Internet that these models are being trained by more of the slop. So it's like less of that independent if that's true or not, that we know that at the edges where the context can be driven from the company is how you can really fine tune the underlying model to give you more of that output. It's like one plus one equals 10. Because you're taking your proprietary information, your company's brain and then layering it into these algorithms and models. Point being is if you continue to give all of that up, it's just rational. Whether they say they're keeping that data or not, that it will get leaked, they will build things on it and others will be able to take that context and data and then build other things. And so the rational decision is between that along with 10 other things like you can just be turned off and clouds just turned a dog recently. Like everything about it, whether it's the outputs to just being down down that you are going to want to host these open source models, you're going to want to be able to set up the parameters to throttle on basic like you know, the average, when you're just trying to like do a cron job is going to be on a lower tier model versus something that super like deep research on a, on a, on a newer model. It's so crazy because it, there's so much here and it's accelerating that the net net of all of it is anybody on the planet earth, from an individual to any size business is able to just do like 100x if they're able to wrap their heads around this today than they could before.
- [20:48]
Host
Yeah, before we move on from all this stuff, there was a few charts that I just wanted to run through that are related to everything we're talking about here. So this one Michael, I think you shared was basically showing hyperscaler Capex is on trend to outpace their cash flows by the end of 2026. And related to this is another chart that's just showing what we referenced around Chinese AI models rapidly gaining traction and they tend to skew more open source. And so this was an article in the Financial Times basically describing that Goldman Sachs is predicting that the use of AI agents would result in a 24 fold increase in token consumption by 2030. And that huge rise in demand would exacerbate a shortage of chips over the next 12 to 18 months. And basically while there's speculative fear about how a single LLM could rise to dominance and what that means, the bigger concern to them is that basically relative model parity and also closed versus open source could compromise the pricing power of a lot of these more of the closed source ones effectively. And then one other one that I thought was pretty interesting was just a breakdown of basically how tokens are spent and used. And you know, this is basically estimating that only around 20 cents of each dollar reaches real users. So you can see the breakdown here. Like the vast majority of tokens are spent on fixing AI generated bugs or rewriting code. And so, you know, there's some, I think as you know, these will get more efficient. But like right now this is not a great sort of breakdown in terms of how these tokens are actually being used and then the implications of that for a lot of these companies.
- [22:42]
Liam
Basically.
- [22:43]
Michael
Yeah, working backwards, I don't necessarily. Not that it's not fair what's shown here, but if you took this graph and you put amount of kilowatts to power a computer when the Internet first came out and how much value you were able to discern, it's like, well, people were just figuring out how the do you search on Google to find what you need and then how much noise was there and you ended up on gambling sites and porn, et cetera, et cetera. And so I think like there's just a natural version of we all don't know how to use these things. So you're just plugging stuff in and then you're refining in that 20%. It actually kind of makes sense. That's the proto distribution of people that got to the other side of it and use it for something valuable. I go back to that Patel pod, the recent one where he was referencing. They were doing like a analysis where I'll butcher this, but they went to like almost every power generation firm and were able to. They use like the schematics and they use the satellites to find like power sources and they'll go into the power sources and like back, go back and get all the different like contracts. They do all this stuff and one of their clients came to them and they're like, how did you guys do this? Or how many people? And it took them like, let's call it, I don't know, five to 10 people. I think it was actually one person that put it all together because he had went from like, I don't know, a couple thousand dollars to in like two months this year, 5 million to $7 million in token spend for his firm. But the way he likens it is he's growing too fast and it's like, do I had headcount or do I had this? But the point was that what he delivered on trying to do analysis on infrastructure, build out another firm had worked on it. It took them a hundred people and they still hadn't got it to this level or scale. But that's a firm that's super sophisticated. They understand how all this shit works and then they've like invested in making it happen. So I Think it's just we're still early there. I forgot there was something else. What did you start with before this?
- [24:41]
Host
It was this chart just around cash flows being. Yeah. Outpaced.
- [24:47]
Michael
This one's honestly probably the most fascinating to me and I don't know how to. I'd be curious lame on some of the research you've done of like it's my understanding you initially we talked about it maybe last week or two weeks ago is you have this like build out that's happening and independent if it's right or wrong. I think the consumers are going to end up holding the bag here because all these companies went from spitting out free cash flows to building on infrastructure to taking on debt. And like I think Larry Fink even came out and says it's like your pension that's going to be, you know, funding it. And I think we all know there's some level of bubble here. Where it's interesting is because it's my understanding in the US we don't have the power, but we have the GPUs. We have Nvidia. They're at the frontier of being able to get the most tokens per kilowatt because we needed to, because we didn't have the energy. Now in China they have the energy but they don't necessarily have the GPUs because of export controls and other things. And they're not even really using the energy, they're just doing more or less. And so there's a notion of it's less even about like ARN usage. That's one thing on tokens. But it's like how do you go and train? If you're going to get the models larger and you're going to input more data and there's more to develop, you have to be able to continue to run these training cycles. And so it's like which one is actually right and it. And where does the end state go? Especially because as these models come out you can effectively. I forgot what the term is. You can distill them and like effectively clone them. And so are you always going to be at the frontier? And then maybe you. That that was part of the thesis of like. Well then you let Ken Griffin and others use it because if you only have a certain amount of people leveraging it, then you can get that capital without having to have it in the wild and let people, you know, distill it into their own proprietary model or open source it.
- [26:37]
Liam
I don't understand why a citadel with as much money as they have wouldn't just take an open source model, recruit some of the best engineers and researchers from some of these other labs and do it themselves in house. I think that they may. But in addition to that, I think that there's two angles here. One is just the models for many use cases can get smaller because a lot of times people don't need to be on the actual frontier for 95% of their tasks. They can use smaller open models because even the open models, as they get better will see token costs come down by like 100x to do the same type of tasks. But as you kind of mentioned, I do think that there will be most Fortune 500 ish size companies that will have their own proprietary models that may either be forced from the frontier or try to bring in some of their own frontier folks and or use the knowledge that they can get from kind of siphoning off all the information from these proprietary and open models in order to kind of fork and develop their own models internally as well as all the data that they have and kind of orchestrate and provide the models their own internal tools.
- [27:55]
Michael
Yeah, I think that makes sense. It was, it's my understanding that like for whatever reason there's a number of them, but SaaS didn't proliferate in like the east the way it proliferated here. I don't, I think there could be a cultural aspect of like owning the data, but it's ultimately like how we went from on Prem to SaaS. And what you're describing is going back to on Prem and you have to like learn that and then manage that. That in the east, specifically in China, the firms never got to the scale of SaaS. But it's because most of these businesses, they may not have like all the tooling and that flywheel you would get with like a salesforce, but they develop their own infrastructure and so they're like very well positioned to be able to do what you're describing. But it, it really made me think of like how we are in for rude awakening. And, and it probably makes sense that these stocks have taken a complete hit because we've seen this anybody list if you've played around like, I mean Brian, on the content side, like there's no shortage of you can just develop the quote unquote proprietary use case for software and what you need and then why are you going out? And then you can start to not only daisy chain it to other applications like a CRM that you can use, but then you're able in real time to just update the software where you can't do that if you're using Intercom or whatever it is. And so I, I do think that makes sense that you're going to go back to more on prem managing your data. But also the notion of SaaS and what does that look like? Imagine being an investor at this point, like in something like that, especially when distribution is such a hard thing in this world of AI and then the customer acquisition cost of that. It's going to be very interesting on that dynamic where Silicon Valley and US as a tech hub has been proliferated on this SaaS notion. That's kind of moving away from that, I think.
- [29:43]
Host
Yeah, I did want to pull up just something I saw this morning. I don't know if you guys had a chance to look into this, but Japan getting to the game. Sakana Fugu launched I guess late last night. And this is an orchestration, multi agent orchestration system accessible via a single model API. And so they're claiming, you know, shoulder to shoulder with models like Fable and Mythos in terms of capability. Sakana Fugu itself is an LLM trained to call various LLMs in an agent pool. Any thoughts on this or have you guys not had a chance to look into this?
- [30:23]
Michael
No, I just saw it come out this morning.
- [30:26]
Host
Okay.
- [30:27]
Liam
I haven't had a chance to use it yet, but I do think that it's pretty interesting and essentially almost what we're doing seeing like from the parallel is Metcalfe's law of the number of nodes of different endpoints of AI is only increasing from here and thus the orchestration and value to the consumer by being able to pinpoint a bunch of different endpoints that can optimize your token usage is going to just deliver more and more value to the consumer at a later stage. So I thought it was interesting. I haven't used it yet, but excited to.
- [31:04]
Sponsor Representative
All right.
- [31:06]
Host
This was I guess over the weekend yesterday. Just wanted to cover this quickly. I don't know if you guys have thoughts, but it kind of relates to the open source conversation we're having and you know, owning, owning your own data, self hosting things. This was Matt corral is saying GitHub has decided to open source. Has decided our open source project has been permanently banned with no explanation and no option to appeal, pointing to a terms of service that clearly does not cover anything we've ever done. I guess it's time for Bitcoin projects to leave GitHub. Any thoughts on this?
- [31:37]
Michael
No, but I mean maybe this is a good transition to, you know, why we're focused here. We're talking about it like obviously it's the thing that sucked the air out of almost everything. And whether it's finance, digital assets or tech is what's happening on the AI front I think it's, it's going to be interesting to see how fast it catches up. But in the you can have all the technological innovations but you still have to coordinate economic activity whether it's via the different agents or ultimately how do you like these tools are downstream of what are the goods and services that the market's delivering. Whether it's financial services, whether it's plumbing, you know, somebody still has to go market that get to the market robotics will play in. So point mean is that the winners aren't going to necessarily be the people just holding Bitcoin or developing Bitcoin services or developing the best API tools. It's going to be how do you merge the best form of money along with AI? And so that's part of where I think we're taking a personal interest is for our own building. You can get ahead of the market but then when you look at entrepreneurs it's pretty clear now doing more with less is the mode, especially in the bitcoin world where you can spend less Bitcoin, hold on to more of it. And if your companies you're investing in are so I think like this is a easy example if you're going to invest in a company, sure you can. If GitHub has the hosting, your engineers are used to using it, the tooling. But then the reality is you want redundancies and fault tolerance. So you should be able to house this in either an open source or your own on prem environment where that code base is sitting there as well that you can't get cut off and then not have access to it.
- [33:17]
Host
Yeah, okay. Unless there's anything else on that. Let's move to some stablecoin headlines from last week. The first was Fidelity moving to manage stablecoin reserves. They want to get into the game of actually managing basically the money markets that back stablecoins. And so it says they're seeking to carve out a role for itself in the fastest growing area of digital finance, managing the reserves that backs stablecoins. And then pretty much simultaneously State street had a very similar announcement that they want to also get into that game. So thoughts on this on ramp Finance
- [33:56]
Sponsor Representative
is live and the Genesis program just launched. Over half the spots are already gone and once they are claimed they are Gone for good. The first 210 signups get one year free of multi institution custody Highest on ramp rewards tier earn up to 5% 1 1/2% cash back on every swipe assigned copy of gradually then Suddenly by Parker Lewis 21,000 sats deposited into your account upon activation. Sign up in 5 minutes on ramp bitcoin.com use the code TLT and do not wait Once the sponsor claimed the Genesis program closes for good.
- [34:30]
Michael
Yeah I'd be curious Liam's like I didn't fully. I think I went down this rabbit hole when either Van Ack or somebody else in that same maybe it was bitwise rolled out their own I don't know. I think it was an ET was for stablecoin reserves and they couldn't really understand it at at face value because it was like well why wouldn't somebody just go to Treasuries? But I think it. It's the understanding of like a access to Treasuries and then access to Treasuries in size when you're thinking about hundreds of million dollar clips or billions of dollars and then the liquidity is. That's the idea that this is like a scale game and then it's bips that you're able to get in and out. And then I think there's the other piece which I'm less familiar with. But it's just this whole notion of liquidity being the real flywheel or vector you want to plan in in money movement and that Fidelity just wants to be closest to that capital because then it allows you to do other things. So that's my way. I've been able to discern why these firms would want to get closest to how do you just park investors assets in these money market funds that are. That are providing stablecoin issuers to access them to issue stable coins.
- [35:40]
Liam
Yep, agreed. Not too much to add there other than there will be a ton of these. And I think Circle had six partners at the time when SBB blew up and they almost were caught in a very bad position until they were bailed out. So would anticipate that there can be room for a lot of these different players managing stablecoin reserves. Just because folks don't want to be too dependent on one single bank in the event that there is a bank run.
- [36:11]
Host
Makes sense. I thought this was interesting. This was Franklin templeton filing for two ETFs that automatically invest dividends from stock holdings into Bitcoin. So it's basically targeting a 95% split between stock and a BTC allocation savvy way is described by Nate Garazi here. Savvy way to introduce bitcoin into portfolios with zero friction. Pretty interesting. I don't know if you guys have thoughts on this, but I thought this was pretty unique.
- [36:39]
Michael
Yeah, I probably thought this was the most interesting thing from the financial side or digital asset side this past week. I don't necessarily know what stocks they're investing in that are giving the dividends. And then also is it via. I'm assuming it's probably via an etf, but I think that this is a much more realistic way for adoption. I think that's the big problem in the narrative right now with stretching DATs is that nobody's giving up the actual ways institutions will allocate to bitcoin. We saw this firsthand back in the day when we're working with credit funds that couldn't put bitcoin on their balance sheet, but they, for whatever reason could put the excess dividends from the yield in the credit fund into btc. And so I think this is a nice way to take that excess profit, revenue, whatever you're calling it, move it into bitcoin from an exposure perspective. And it reduces a lot of friction from having to size it, what percentage. You're just able to give it to an asset manager to move it over, over and then you can, you know, gain your confidence, conviction, education through something like that. So I thought this was super interesting and obviously coming from a larger firm versus a smaller that has like already the distribution and opportunity to market it.
- [37:50]
Host
Yeah, the, the other thing I would say is like, you know, well, first of all, Franklin Templeton I think is pretty, pretty forward thinking on, on everything digital assets and bitcoin. So this isn't necessarily a surprise that it came from them. But what this actually reminds me of is like, so back in the day when I was still at Brown Brothers, you know, a lot of my day was spent talking to fund managers and you know, across the asset class spectrum, but spoke to a lot of, you know, long only equity fund managers that typically had like a pretty sizable cash, cash position in their portfolio, anywhere from like, you know, 5 to sometimes 20% just sitting in cash. And I would often, you know, talk to them and more so personally think about like, well, what if you just took even a portion of that cash and held it in bitcoin? And this is kind of a flavor of that in the sense that like, you know, it's 95% stock. It's these, you know, predominantly dividend paying companies and then it's just converting some percentage of that dividend into bitcoin and sort of building a more of a bitcoin based cash reserve in that portfolio. So it reminds me of a thought that I had many years ago which is like, you know, I understand why fund managers have sort of these buffer cash positions. In one part it's, you know, if they're wary about valuations in their, in and around their portfolio, they want to have basically some cash buffer to add to positions if you know, there's a drawdown in names that they hold. But it's like, you know, at the same time that the, those large cash positions create a drag. Right. So if they're measuring against their benchmark, whether it's the S and P or something else like that cash position on their monthly attribution report is a drag against their relative performance because you know, The S&P 500 is obviously fully allocated. There's not a cash position in those indices benchmarks. And so if you had a cash position that was basically working for you in the sense that, you know, part of it or at least part of it is in bitcoin, I think that makes a lot of sense. So this is very interesting to see. Liam, you have any thoughts on this?
- [39:54]
Michael
Yeah, real quick before Liam's, I think that's a really great example from use cases.
- [40:02]
Host
We'll see.
- [40:02]
Michael
And it's like one of the many different use cases for on ramp finance because there's an idea that you can take everyone, it's very similar. Micro finance is the same as like macro. It's just different scale sometimes different duration, risk profiles, but ultimately like individual. If they're keeping 10 to 50 to $100,000 in cash for a rainy day emergency fund, they're still a drag because they could be allocated to X, Y or Z if they're able to have any rewards and then sweep those rewards into BTC while keeping the principal whole. That's a nice use case. Especially if you're looking at, well, bitcoin's a flyer that is going to appreciate much faster than anything else I could have adopted. While minimizing your downside, I think we're just going to see more and more of this. We have to educate, market it more, provide materials and we're also kind of in this weird bear. So there's not like a lot of people trying to figure that out. But that will I think be a huge use case for individuals that think there's something interesting in bitcoin. But not ready to put principal at risk yet. They can just take those excess rewards.
- [41:00]
Liam
Yeah, I think it just takes a lot of emotion out of this, which is especially in bitcoin, just given its historical volatility, is something that just keeps folks on the sideline either waiting for a better time when they think bitcoin's going down and they can get in at a cheaper price, or just when bitcoin's ripping and they think that they missed the boat. And so I think that these types of products are, are just great to kind of DCA and without putting principal at risk.
- [41:27]
Host
Yeah, that's a good point too. And then the other thing I would say is I like it from the perspective of like it's a, it's analogous to a cash position from a portfolio construction perspective. So just from like a perception angle, it's less, I guess it's more palatable to someone to introduce bitcoin exposure to them. Not as like a one off flyer thing. It's like, no, this is just sort of like supercharging your cash position in this other portfolio of predominantly stocks. So I think it's a pretty elegant way to introduce bitcoin exposure. Moving on. This was making some rounds last week. This was Illinois governor, this gentleman here, Mr. Pritzker, put forth this bill that's basically a 0.2% tax on crypto transactions that includes transfers between personal wallets. So this is pretty outrageous. The crypto council called it the most punitive digital asset tax in the country. You know, I think this is interesting just from the context of, you know, you obviously have certain regulators, certain jurisdictions moving in one direction and then you have sort of outliers that are moving in the opposite direction in terms of these punitive measures. But I think at a higher level, who knows what's going to happen with the Clarity Act. But the broader sentiment from a regulatory perspective seems to be very positive. So this is to me sort of an outlier, but did make a lot of noise last week around this. Thoughts?
- [42:57]
Liam
I think that just to Clarify, this is 0.2% tax on both buying, selling, spending, bitcoin as well as just holding it. And that's a big distinction. It's akin to a wealth asset seizure tax, which is obviously unconstitutional as it just given its private property. It looks like it's going to go into effect as of January 2027. Just because this was slipped in there at the last second. Not great to see. I imagine significant pushback both from digital asset native folks as well as I know that this is going to get a lot of national coverage too as the time goes on. Just given everything that's going on in California and asset seizure taxes there. So I don't think this is going to be the end of it. I think that we'll probably see more, not just digital asset wealth taxes like this, but everywhere. And it just clarifies even more like why you want to be in friendly jurisdictions with, with friendly rules is these regulations continue to bifurcate.
- [44:02]
Michael
Yeah, that's all I got. I mean it ties back to all the stuff we talked about. Owning your data, where you live is increasingly going to matter. I think it just came out that Morgan Stanley's looking at setting up their office down in Dallas. I don't know if it's a satellite or another office, but we've seen like every, every large specifically financial service firm. It'll be interesting to see what the big firms, whether it's like CME or specifically Cumberland and Jump, who are focused predominantly on digital assets do. I don't necessarily know you can, if you can stay in a, in a state with this kind of tax as a business. So it'll see, it'll be interesting to
- [44:39]
Host
see how it plays out.
- [44:40]
Liam
Even though Citadel isn't like that much into digital assets, I'm sure that they trade it just given how often they trade assets and just try to make money off of either way. I would imagine that this is kind of the writing on the wall that gets them to even further exit their, their position in Chicago and Illinois in specific. Not great to see
- [45:04]
Sponsor Representative
whether you've been in Bitcoin for 10 months or 10 years. Managing your bitcoin wealth has been a fragmented experience. One place to buy, another to custody. A patchwork of providers for your ira, your loans, your estate plan and everything in between. Every additional platform is another point of failure. And that is why we built on Ramp Finance, a unified platform for your entire bitcoin life. Enhanced brokerage cash bearing accounts with earn rates up to 5%. A card that earns cash back that you can use to stack more bitcoin. Multi institution custody IRAs and built in inheritance planning. All under one roof. The genesis program is live and spots are filling up fast. You can sign up in five minutes at onrampbitcoin.com and use the code TLT to lock in. One year free of multi institution custody and our highest earn rate.
- [45:56]
Host
Another headline from last week. More on the regulatory front. We kind of talked about this a few weeks ago in terms of, you know, the genius act now exists, it's law, but there's still steps towards actual implementation of this stuff. And so this was Federal Reserve rolling out proposing rulemaking requiring stablecoin issuers to maintain customer identification program. So 130 page rulemaking to implement the genius act effectively. Five members voted to approve the proposed rule and Kevin Warsh abstained. Thoughts or implications of this gentleman?
- [46:34]
Michael
I don't have much outside of. I mean we see a lot of this behind the scenes and it's, it's kind of bullish for the digital assets like in whole Bitcoin that you're seeing increased regulation. That's what was going to come. It's the same thing. If you're going to tax these assets, you're going to make them into just how you think about everything that exists in the traditional banking system. BSA not saying it's good or bad, it's just what comes with government oversight. So it makes sense that this is coming. It also makes sense. This is effectively how you end up with this like quasi CBDC because you're just attaching all the information. You're going to have more information attached to your identity. You're going to be able to get access to these. I think we all know there'll be some level of UBI coming at some point and it's all going to be tied to this. So it is what it is. I think that's really part of all the stuff we talk about is the education that like you can use these tools and you can get the efficiencies in the same way you can get efficiencies from a frontier model. Just know that it's not your info and data. And ideally, if you want to, you know, protect yourself, you have to be able to get smart of all. I think the way you get smart here is by sweeping your funds into Bitcoin versus holding large cash balances or relying on a, you know, CBDC slash stablecoin system.
- [47:47]
Liam
Exactly. It'll be interesting to see how the stablecoin world bifurcates in the US where there is KYC versus kind of offshore, where a lot of these regulations and rules are a lot less defined. And if and when this US gets applied to other jurisdictions as well, or if tether kind of continues to operate like pirates or turns into the Navy.
- [48:11]
Host
Well said.
- [48:12]
Michael
On that note, somebody had sent me tether. I think this just ties into these themes because I don't think they're on the list is Binance. I think I kicked out of Mika because they didn't you know it's actually interesting because you hear Binance on face value they look like they're getting you know, super buttoned up, et cetera, et cetera. But like the way that it came out for Mika was like they did none of that. They just pretended to and they thought their size would allow them in Europe was like get out. And then I saw last night that the same gentleman, he had shared it was a tether. Ran into something as well with M or something in Europe. But the point being is like all this stuff, I think if you're going to operate you're effectively going to have to bend the knee or you're just going to get run out of town. And I don't necessarily know if there's a way around it. I think this also goes to back back to like the self hosting and providing services for open source compute and you're going to be able to I do think like end state. That's why it will always end up to Bitcoin because you can't censor and you can, you can you know, move around those servers but it's not really commercial businesses. I don't think that you're like the Venus stuff, I think Yusuko brought it up last week on the pod and you know, didn't call them out on it. But it's like this notion you're gonna have a token attached to compute and it's a regulated company and now you're gonna like how are you planning to do that if you get.
- [49:37]
Liam
Yeah, just on that specifically too. It looks like something like 80% of digital asset firms face immediate expulsion or forced exit from the EU market on June 30. I would imagine that just given the scale of that that there is some sort of delay or pushback but pretty wild to see the amount of non compliance. And I haven't talked to many firms over there but would imagine that there must be some miscommunication if it's that many firms that would be forced to be kicked out soon.
- [50:08]
Michael
It could also just be the insane oversight. We're seeing this in this in the US with certain states asking for certain information to get licensed there similar to how the BIT license was. And there's huge capital requirements and things involved with it if you're doing certain financial service activity. And so if you're in the, in Europe and we've seen you know, how they fumble the bag in multiple angles. One of the big ones that they're obviously concerned with is capital flight and so capital Flight, keeping it in. I think I don't know how much you guys have seen. If you go to NDK's Twitter you can see, you know, being, I don't think he's in Canada anymore, but being in, he was in Canada for a while. He's closest to just how insane and draconian those, those controls are that these firms that have had like these valuations increase if they leave. They have to exercise effectively. It's like they're selling a portion of the, like an exit tax for the company but they don't have to sell it, but they ultimately own the taxes on that so they can't leave. And so these countries are just starting to put this so it makes sense. Like again, without knowing more about where the requirement is and then what's the delta between those firms that are potentially not falling within the, the, the requirements. But I would imagine that there's just onerous. There's onerous. Like regulations and things that people need to do and a lot of these firms never got to scale or they had, you know, no KYC wallets if they didn't think they were a custodian. So you just end up with. I think there's just going to be more and more consolidation and, and like just closing down of companies because it just doesn't make viable sense to be able to either you can't raise to be able to pay just the regulatory fees.
- [51:58]
Liam
Yeah.
- [51:59]
Host
All right, we have a few minutes left. We had a bunch more on the list, but where do you guys want to go? We had a few deals. We had some Coinbase tokenized stock stuff. We had SpaceX buying cursor. Where should we go?
- [52:13]
Michael
We need the Coinbase one real quick.
- [52:15]
Liam
Just.
- [52:16]
Host
Yeah. So this was a write up from Alex Thorne. Coinbase touts tokenized stocks with true equity ownership. But how. Michael, maybe I'll hand this to you.
- [52:27]
Michael
I think the main thing is tokenized stocks are coming. Coinbase released a bunch of interesting things last week related to AI native Ria. I think the interesting part about Alex's research was referencing that like not all tokenized equities are, are considered equal and ultimately that these products that Coinbase is rolling out and I think others have as well similar like Robinhood a year ago when they were tokenizing secondaries for like OpenAI and maybe it was SpaceX or some another Elon company was that these are like effectively SPVs. So they're like second or third order proxies for the underlying equity and then they're, they're issued, not sanctioned by the company and then they're only tradable outside of United States. I think he referenced it's closest to X stocks versus and I think he had incentive to write this because I believe Galaxy has done formal partnerships with the right broker dealer coupled with the company. So they are effectively one to one equity proxies that are tokenized. Either way, I don't necessarily know where it all goes. I think it ends up with a hyper velocity, lots of losses, more people speculating on all these assets, but it's coming no matter what, that they're going to tokenize. And, and part of it we skipped over, but they're rumoring like even NASDAQ and the other stock exchanges are going to be moving to 24, seven markets, which is insane and kind of crazy to see how you're going to manage that in itself.
- [53:57]
Host
Yeah, I mean this is all coming. We've been talking about it for what seems to be like over a year now, but this does seem to be materializing. Liam, any thoughts on that or anything on SpaceX? I mean we, we didn't really get a chance to cover that. IPO went live I think around 135 and then it sort of actually started trading around 175, ran up to like 225. I think it's back down now. But then last week they bought cursor for 60 billion. Any thoughts on that?
- [54:27]
Liam
I think it's just amazing how quickly they were able to get to. I think it took them like two years to get to like 4 billion arrows. 300 people and exiting for 60 billion in valuation in just under three years. Pretty fantastic outcome and just really shows how efficient and how possible things are in the age of AI, how you can really speed run amazing projects with small teams and it's just a great outcome for the team. SpaceX is a generational company, but it doesn't necessarily mean that the valuation will continue to be priced this way over the next year or two. There could be some bumps along the road, but amazing to see what both Cursor and SpaceX have built.
- [55:14]
Michael
Yeah, I thought the most fascinating or interesting take was if you listen to this week's the Empire Pod, those guys are pretty smart. But there's this joke I've been sharing about the crypto people. Always could spot the Ponzi or Ponzinomics. And that's where like you see this all over Twitter. It doesn't really get talked about enough because the crypto people got burned so much and they've seen every Ponzi. So they're the first to be out there talking about all the craziness around like digital credit and like how is this sustainable? Whatever, whatever. But these guys are also talking about the low float and it's like this is the SVF, like FTX playbook, not only from the value, but then how you're able to leverage that balance sheet to go make these acquisitions. So I thought that was just interesting of like how Elon's been able to financial engineer not only the pump in going post ipo, but then how he's going to go and pick up these like valuable assets and then figure out, you know, we'll see where it trades in the next like 30 to 12 months once their lockup's done.
- [56:15]
Host
Yeah, I think the first lockups are in August. In August, I believe. And I think it's, it is like 20% and then it goes to 30% if the share price is like over 175 for a certain amount of trading days. So yeah, it is definitely some severe pumponomics going on right now. I think only 5% of it is floating and Elon owns the vast majority of it. So it could trade pretty crazily for the next few months. But I think back half of the year maybe comes back to endearth.
- [56:50]
Liam
Yeah, it's the same thing almost with strike too and under and or stretch I guess, and understanding who your counterparties are with all of the defi and leverage that is in that token, similarly to understanding everybody who's in the SpaceX stock if you're going to be an investor there. Just given all the lockups that will occur and who will need to necessarily give capital back to investors just because while you may be taking a longer time horizon deal, it's not necessarily, you have to understand everybody else on the cap table and their incentives if, if you're going to be holding this for a long period of time or if you're going to be trading it too.
- [57:28]
Host
And yeah, a lot of those, a lot of those investors have been in this for you know, a decade plus.
- [57:32]
Michael
So yeah, I can't even get mad about like when you see some of the AI stuff recently, it's like, yeah, if I was focused or we were focused the past five years, you could see how people got in. It's kind of crazy, that circular group between like the cursors, anthropics, OpenAI to get in at seat or whatever. But like SpaceX, these people have held for 25, like if you can hold for 25 years. You deserve whatever is coming to you on this side. I know we're wrapping up a couple quick hits. I thought it was interesting let in putting on support for Tether's gold back Stable coin think makes rational sense for obvious reasons. We talked about where gold's going to play next to Bitcoin, especially digital gold. Also Tether I believe deprecated some of their. It was another Stablecoin related to Tether because it's my understanding that XAUT will be kind of their dominant platform which is backed physically. And I think you could take delivery of gold. I think that'll rationally be the way that anybody's going to allocate to digital gold will want to be able to know where it sits and be able to take delivery. And then the other random one, I thought this Carta deal was pretty interesting. It was 15 million in equity and then 135 or 125 in debt. And I want to dig more into it but it was ultimately Galaxy Led and it's this notion of where you have like high net worth individuals that are international that are looking for credit based services in the United States and that for whatever reason if their credit profile doesn't translate over they're able to use like existing metadata. And you've seen some of this stuff. But more on the micro side, there was companies back like decades ago now would give you like micro loans based on certain data on your phone. Well this is on the other angle. It's for high net worths that wouldn't necessarily have a US issue credit card, but it's not a credit check. So they're able to use all this other metadata up to a $200,000 credit line. And then they're doing a lot of this with stable coins. So I thought it was just an interesting use case for stable coins. You see like, you know, the same stuff. We had something on the list which was like Brazil and money movement but I think we were all kind of familiar there. I thought this was an interesting kind of newer model I want to dig more into but I thought it was worth calling out.
- [59:46]
Host
Yeah, that is interesting. All right, anything else? Good place to wrap.
- [59:55]
Michael
How much? How much stretch you got?
- [59:59]
Host
Dude, it's going back to par. Didn't you hear?
- [60:03]
Michael
We'll see.
- [60:05]
Host
We will see. And even if it does, it doesn't really change anything. I think that's what people miss. There's mechanisms for it to get back to par, sure. But that doesn't change the overall Fragility of the system. We didn't even bring up the headline, but Mr. Saylor added to his cash reserve this morning, diluting MSTR common holders yet again.
- [60:31]
Michael
Maybe we'll do this on a last trade because it's something like I don't think we talk about enough that a lot of times when individuals are. It sounds like it's charged because it has been a little charge of like, this makes zero sense from just. Why wouldn't you hold bitcoin, et cetera, et cetera, but from a pure mechanical perspective, like, they're in a very precarious spot. Objectively, when you think about like common shareholders or stretch and the dividends that they owe and then the underlying bitcoin. And this reflects a flywheel, like in this one notion that it gets propagated a lot, that they have like 32 years or 64 years. It's like that makes zero sense because that means that they're selling bitcoin. And then like, why would. So it's just.
- [61:09]
Host
And if they actually do that in size, that 32 years compresses rapidly. Like it makes no sense. But that always is what these people fall back on as like a reason why they're fine when it just doesn't make any sense. Like these. It was a bitcoin accumulation vehicle, effectively a bitcoin black hole. And now they are open to selling bitcoin and they're stacking dollars. So the strategy has changed, I think that's fair to say.
- [61:38]
Michael
Yeah. And there's. There's like. I don't know if anybody's listening to this anymore that is anywhere near or if they have these assets, they're at least objective to be like, okay, well, maybe these guys have some thoughts interesting outside of this. But.
- [61:49]
Host
But the.
- [61:52]
Michael
There's this angle out there that I think really people believe is there's like a coordinated attack either on this guy, this product and business or. Or in general digital credit. But the thing I think that everyone forgets to understand is like, the truth is just the truth. And so you don't have to coordinate. Then people all objectively come to the same realization. That's why you see everyone coming out. And it wasn't popular and it wasn't okay to do as the price was appreciating. Because there's a lot of that affinity layer of like, oh, you're attacking bitcoin, you're doing this. It's like, that's one of my favorite things. A lot of people, they conflate that if bitcoin fails if MicroStrategy sells. They're like mutually exclusive. They're not the same thing. And so people can be long and bullish Bitcoin while understanding that this product and solution probably doesn't have a permanent or persistent chance of existing with Bitcoin.
- [62:44]
Liam
Yeah, I would just say two things because I, I feel bad enough when, when people have been in this instrument because a lot of good people got burned. But I would do two things. One, like do your own math on DCA into Bitcoin versus over the time that micro strategy first bought bitcoin to now and and what you could have gotten versus what they got and run your own numbers. And then number two, just because obviously they have a better chance of issuing their stocks during bull markets when things go at premiums and then have to necessarily sell when Bitcoin's at a lower value. So you can probably have more Bitcoin if you just do it yourself. And the other one is, rather than just listening to tweets that go out late in the morning on Saturday nights or podcasts, I would recommend to actually look at the company disclosures rather than just guidance and, you know, what goes on on podcasts, because that's where the real information is.
- [63:45]
Michael
Very well said.
- [63:46]
Host
Very diplomatic. Good advice.
- [63:47]
Michael
All right, good stuff. Good stuff, guys.
- [63:49]
Liam
All right, later.
- [63:51]
Host
Thanks for listening to this week's episode of the show. If you found the information valuable, please share the episode with a friend or leave a rating on your favorite podcast app. All the links we discussed in today's show will be in the show notes inside your podcast app.
- [64:03]
Michael
App.
- [64:03]
Host
Before we finish, a quick reminder that Onramp Media is for informational and entertainment purposes only and nothing should be construed as investment or legal advice. Regardless of where you are on your bitcoin journey, we'd love to hear from you. Visit onrampbitcoin. Com Contact to schedule a consultation with one of our private client advisors.