Loading summary
Transcript67 lines
- [00:07]
Brandon Bailey
You've had a dynamic where money's become freer than free. If you talk about a Fed just gone nuts, all.
- [00:14]
Host of TFTC Podcast
All the central banks going nuts.
- [00:16]
Brandon Bailey
So it's all acting like safe haven. I believe that in a world where central bankers are tripping over themselves to devalue their currency, bitcoin wins. In the world of fiat currencies, bitcoin is the victor. I mean, that's part of the bull case for bitcoin. If you're not paying attention, you probably should be. Probably should be. Probably should be.
- [00:36]
Host of TFTC Podcast
Brandon Bailey, welcome to the show, sir.
- [00:39]
Brandon Bailey
Thanks for having me. Excited to be here.
- [00:43]
Host of TFTC Podcast
Excited to have you, dude. I mean, we've been. I mean, I've known you for what, probably like seven or eight years now at this point in the bitcoin mining space. And, I mean, you've done an incredible job of knowing what's happening inside and out, starting a galaxy. Moving on to Nakamoto. You just launched Dimetrics AI to help people grasp what's going on with the bitcoin mining and AI compute themes for colliding with each other. And I'm incredibly excited for this conversation because I think, despite what many bitcoiners will get butthurt about, is that AI is taking the wind out of the sails of bitcoin. I'm incredibly excited about what's happening in AI and I actually think it's going to be massively beneficial for bitcoin overall. And so having you here to help get a lay of the land, of what's happening on the compute build outside, I think is going to be incredibly valuable.
- [01:44]
Brandon Bailey
Absolutely. I'm really excited to dig into the conversation. There's a lot to talk about.
- [01:49]
Host of TFTC Podcast
Well, a lot to talk about. Where do you think the best place to start is the history of this emergence of AI Compute? I think maybe starting there, the intersection of bitcoin and energy infrastructure that really exploded in 2021. What the Bitcoin miners were able to do successfully in terms of locking down power and really ingraining themselves in energy systems. And then the appearance of AI Compute really, I don't want to say throwing a wrench, but really accelerating things on that front as well.
- [02:27]
Brandon Bailey
Yeah, let's jump into it, because that history there, especially in the 2021 kind of 2022ish era, is ultimately what led a lot of these bitcoin mining companies to kind of stumble onto this sort of gold mine of the power portfolios that they ultimately amassed. And so it was really around 2021, shortly after we had the China mining ban, which resulted in as much as 30% of network hash rate coming offline. And this redondillation, hopefully that's word of bitcoin mining hash rate relocating to the us so bitcoin mining was largely dominated by China. That's where a lot of the activity occurred. After the China mining ban, we saw this sort of gold rush of reestablishing bitcoin mining hash rate and compute in North America, which led to some of the first publicly traded bitcoin mining companies. So that's where you had companies like Marathon and Riot and Clean Spark really start to emerge in that timeframe. And so we were kind of seeing the institutionalization of bitcoin mining during that era, which led to a lot of these companies acquiring massive amounts of power so that they could ultimately have a pipeline to build out these 100, 200 megawatt data centers that were all going to be allocated towards bitcoin mining, which made a lot of sense at the time. Bitcoin ran from, I want to say the lows. Coming off the lows of last cycle, I think bitcoin touched maybe around 4k like in the bear market post 2017, and then from that point ran up to the highs of 60k. So the margins on bitcoin mining were just, I mean, incredible during that 2021 bull cycle. So all of the investment that was pouring in from these public companies acquiring that land and power made a lot of sense given the economics at the time. And so that's what ultimately allowed these companies to be sitting on effectively gigawatts of power. And then as we now seen, AI really blow up, all kind of starting with the advances that were made with ChatGPT and et cetera. You've seen obviously an explosion in just demand for AI and we're kind of going through this massive capex bull cycle. All of these bitcoin miners were super well positioned because they already had all of this energized power. That is one of the biggest bottlenecks right now when it comes to just trying to stand up more compute.
- [05:23]
Host of TFTC Podcast
Yeah, and I think it's, it's been funny watching it all play out because you're seeing the, I mean, going back to the 2021 era, having been involved in bitcoin mining back then, and still to this day, the mad dash, the lessons that you had to learn to acquire power were pretty massive. It was a steep learning curve. And it's interesting watching the AI world get into this obviously out of necessity and the amount of Power that's needed is almost incomprehensible and it's bringing a different caliber of capital to it because you have these hyperscalers, you have Fortune 100 companies really leaning into this, putting tens of billions of dollars of capex, trillions of dollars estimated the next couple of years. And it's, it's interesting to see how they're entering this market and really trying to, I don't want to say bully, but just like brute force their way in and really coming to understand the limits of the energy infrastructure that exists in America that bitcoin miners have become very intimate with over the last 10 years or more specifically the last seven years. And I actually think it's good overall because it's making the public aware that hey, energy systems are important and want to barrel into this AI future. We really need to figure this out first. This is like layer zero of what we're building here.
- [06:58]
Brandon Bailey
Totally. I also would add to that as well. I think it also kind of is touching on the fact that I think that there's a lot more education that's required right now when it comes to understanding energy policy, how the grids work, the need for more generation, which has been a talking point for a long, long time. And I think that there's a lot of FUD right now, which is really just a result of, I think, miseducation when it comes to, you know, the, the amount of power demand that is out there to ultimately serve this compute and like what it ultimately means for local communities. I mean, we saw that with bitcoin mining. It's kind of interesting because it's like history repeats or history rhymes to a certain degree. So a lot of the criticisms we saw of bitcoin miners entering certain communities with respect to the power demands, you're seeing a lot of those same kind of arguments being made with respect to these AI data centers. And so, you know, I just think like the, the tensions and the challenges you're talking about with respect to just getting like, you know, your, your actual site energized or getting an interconnection agreement there. Those are really, really difficult challenges that you can't just always brute force your way through or just throw more money at the problem to solve it.
- [08:21]
Host of TFTC Podcast
Yeah. And so with that in mind, let's, let's walk through this. What I think we can do sort of like order of operations of how this actually scales, starting with bitcoin miners that lock down land and power over the last five years. What are the decision frameworks that they're operating now as the AI companies come to them and say, hey, you have these power deals, we have a need for power. What are the sort of economic decisions that are being made by the miners right now? What's the opportunity cost? And how do you think they should be approaching the question of should I mine Bitcoin or should I put GPUs in a data center with the power that I own?
- [09:08]
Brandon Bailey
Yeah, that's a great question. So I think it really starts first with sort of the easiest way when you're thinking about this decision, you're walking through this calculus is you can just look at market comps first and foremost. So one of the things that I've been looking at just as an investor in this space is that most bitcoin mining companies maybe trade within a range between 4 to 6 times EBITDA effectively. And you have more traditional data center players like a digital realty trust or Equinix that trade closer to 20, to call it 24 times EBITDA, EBITDA. And so there's a massive discount that is being applied to the valuation multiple for using bitcoin mining as an offtake, as an off taker of that power capacity relative to more traditional data centers. Right, as an off taker of that power capacity. And so when you look at that relative arb there, there's a huge potential for these miners to just be re rated by changing effectively the off taker of that power or moving to AI versus Bitcoin mining just purely from a multiple expansion standpoint. One of the reasons why there's such a large variance between the multiples for those businesses is because with the data center space and when you look at the economic terms of these agreements, you know you're getting a 10 to 15 year lease effectively that's guaranteed cash flow. If you look at historically the income stream of bitcoin mining, right, it's very volatile. You have things like the having which are going to cut your economics in half. And then you also have the fluctuations of bitcoin price dynamics. And so you have periods of really, really impressive gross margins in mining. But you also have these longer sort of lulls of more challenging economics. And then you also have more challenges with respect to the economic useful life of the asics. And so it really just boils down to the fact that you've got 10 to 15 years of locked in pretty much guaranteed cash flow. Your tenant effectively in these leases are the most profitable, largest companies in the world. You've got a Google or you've got an Amazon, you've got a Microsoft effectively as your tenant. You know, these companies generate billions of dollars of cash flow a year. So that lease is essentially viewed as being as good as gold. Or you know, if you have a parent guarantee from one of those companies, that lease is pretty much as good as gold. And then the other big thing there as well is like you actually can get financing to actually do the build out. One of the challenges we had in the mining industry was, you know, how do you actually establish a credit market to ultimately support that industry. So you had a lot of these companies had to rely solely on equity dilution to ultimately be able to fund the build out. And then you get into these massive questions of returns on capital. Are we actually creating value here? Given what's happening with the economics and the dilution that investors are taking on with these data centers? It's the opposite. You know, you're seeing companies able to finance the bulk of the capex cost and to be able to do that at an incredibly favorable interest rate. So that's a lot of what these companies are looking at. And when you compare those two, it's, it's, you know, it's almost like a no brainer, you know, especially if you can land one of those hyperscaler tenants. It's, you know, you would do that
- [12:51]
Host of TFTC Podcast
deal all day in your mind who's done this the best so far? The transition from bitcoin mining to hbc.
- [13:01]
Brandon Bailey
You know there's, I think, I think, I think for all of the ones that have signed leases, I think that they've done a really, really excellent job. So you know, those companies are Core Scientific, Hut 8, Cypher, Galaxy, Terawulf and Applied Digital. I think you really have to commend Core Scientific because honestly they're kind of the pioneer in this. They were the first ones to really make this pivot. They kind of started the wave for these bitcoin mining companies and kind of helped to enlighten the market on the potential of this transition. And I think they have 590 megawatts of crit IT capacity all leased with core weave and they've already begun delivering that capacity. So I think they've done a great job to date. I think Cypher has done very well. Terrawulf, hut 8 as well as over there. I think all of these companies are doing an excellent job kind of carrying the flag and helping to pave the way for other bitcoin mining companies to demonstrate the civility right now. Because the way that I would also Characterize this is there's a couple different phases as we go through this transition. Right. The first phase is can you actually sign a lease? Which is very challenging. The sort of process of ultimately courting one of these hyperscalers or large Neo clouds is very, it's a very tedious, arguous process. It's very difficult just because of the level of demands and expectations that they have. Right. And if you haven't already delivered one of these data centers before, right. They're going to spend an incredible amount of time diligencing you. They've got to feel comfortable that you can actually deliver. So phase one is really can you get the lease sign, can you get to a definitive agreement? Then phase two is can you actually execute on delivering this project on time? And that's kind of where we're at with a lot of these companies. Core Scientific is the first to actually
- [15:16]
Sponsor Announcer
deliver
- [15:18]
Brandon Bailey
a couple energized buildings. And a lot of the other companies are coming up on that. That's really the other big risk vector which is can you actually deliver the construction? And then after that it's really just maintaining and operating the data center. But the upside from an investor standpoint is really on phase one, can this company actually sign a lease? And then two, can they actually deliver it on time? And like to me that's where I think that that's the biggest window of an opportunity for investors that are looking at these companies for their potential to ultimately transition that compute or that power capacity.
- [15:57]
Host of TFTC Podcast
And for phase two. Correct me if I'm wrong, but it's a bit different than bitcoin mining where a lot of the bitcoin miners, when they would lock down power, say okay, I'm going to build a 200 megawatt facility. A lot of them would, I mean do a lot of home baked solutions to actually manifest that bitcoin mining operation. But from what I understand with the AI, Compute is that a lot of these providers already sort of have out of the box solutions or maybe like design specs that they'll just hand over and say, hey, here's how we operate, build this. And it's not really, you're not really dependent on getting creative and coming up with your own solution. I feel like everybody in bitcoin mining in the 2021-today was sort of learning on the fly of like how to actually build a data center. But I think to your point about the demands and the diligence is necessary to get these, these hyperscalers comfortable with, with partnering with you. It Comes down to can you actually build to spec what we need?
- [17:01]
Brandon Bailey
And here's the design that's absolutely right. It's really a whole different kind of philosophy and approach. Bitcoin mining is like how can we stand up this infrastructure the fastest and in the most cost effective way possible? Which in bitcoin mining you could actually be rewarded in a sense for cutting corners or finding ways to. How do I just get this infrastructure still up in a, in a cheap way so that you can maximize that payback period. And, and you wanted, you wanted to do that in bitcoin mining because you have difficulty. Right? Like it's a race against the clock. You know, more power capacity is going to be coming online or more hash rate is going to be coming online and that's going to impact your margin. So the faster you could get your machines online and energized, that was only going to help you with high performance computing. And these data centers, it's completely opposite. Right. It is much more about can you build to the spec, can we make sure that we have all the redundant systems? Because uptime is vital here. It's just a very different approach and design philosophy. And these data centers are much more complex than bitcoin mining data centers and have significantly more redundant systems, the cooling systems, the intricacies of all of those components. It's just a whole different sort of ball game when you're kind of comparing those two worlds.
- [18:36]
Host of TFTC Podcast
Yeah. What are your thoughts on the scale of the demand for compute, the individual sites and the spectrum of scale that may emerge? Because I think obviously you hear the headlines, Colossus 2 going to be a gigawatt. Colossus 1 was what, 300 megawatts. You have many paper scalers going out there and saying we're going to go build a gigawatt multi gigawatt facility. And I think we're beginning to see as the supply gets tapped on these large scale operations, people beginning to look at smaller scale, like 20 to 50 megawatt and trying to figure out are these viable for, for AI compute data centers. And that's I think a part of the market that many people are trying to explore. And what are your thoughts on that spectrum of scale?
- [19:32]
Brandon Bailey
I think the demands for power capacity and compute are enormous. And I think that you're going to see sort of all parts of the spectrum be in demand. To your point, the early innings of this, it was all about the mega sites, the 1 GW site, hundreds of megawatts of scale. And what's really fascinating about this is you kind of have to really think about the undercurrent and what's driving it. And it's really, to me, the way that I would describe it. It's the game theory that's at play here. And you kind of have this game theory playing out on two different levels. There's the game theory of nation states. Understand that AI is this massive revolutionary technology and that we need to be a front leader or we need to be the leader in this new technology. And so the US is largely in an arms race with China on trying to be the world leader in AI technology, right? Which means that the US needs to foster inside of its borders an environment that can allow this industry to flourish and ultimately make sure that it has the resources that make sure that the entrepreneurs inside of our borders have the resources and the access to the raw input such that they can go build and create this to try to win that. So you kind of have this, you know, that game theory playing out, the US versus China in this race. And then you also have the hyperscalers themselves competing with one another, right? Because these businesses had been, you know, more sleepy, less growth focused, just printing cash. The cloud business is just crushing at the margins, right, of like a Google cloud, Azure, aws, just crushing it. And those are massive cash cow businesses for those companies. But with AI, there's a massive risk now to some of the moat that they've established. And so you're even seeing these companies competing with one another. So you have Nvidia competing with Google, Google, Amazon, Meta. They all recognize that this is a massive opportunity for them to grab market share and also protect the moats that they've already established. So each of them are incentivized to spend a tremendous amount of money on capex, right, to make sure that they can protect their moat and also try to gain market share. And so those two dynamics to me are what are really driving the enormous demand for compute. And of course you have SpaceX now as well. But all of those companies are competing and what they can afford to have is to say we didn't invest enough money and we allowed Google to have more access to data centers, which gives them just an unfair advantage because they didn't invest enough capex. So to me that's what's driving a lot of the undercurrent of the demand for power. So that's one component and that's where we saw the demand for these 1 gigawatt sites, et cetera. But now that you're seeing a lot More pushback, right? Like the new trend has been the social climate and the political climate around data center development. You're seeing a lot more pushback in communities around wanting these massive scale projects. And we've even seen some projects more recently be canceled or paused because of just the pushback in local communities. So that's where I think that there's more of an opportunity for these smaller scale sites and where it's like maybe we don't go 1 gigawatt, 1 gigawatt, let's go 50 megawatts, let's go 20 megawatts, right? It's a little bit easier to get through and get approved. So that's where I think you're starting to see more demand or increasing demand for those sites. I think inference training is also driving demand for those smaller sites. So I think that there's a massive opportunity on that front right now for some of the smaller players. And I think that you're just going to see more overall demand across the spectrum. But I think it's going to be easier for people to get things stood up in the smaller sites. And then the last thing I'll say is that I think that because of the level of difficulty with respect to just getting power energized, I think that you're also going to see the data center world takes some lessons learned from the bitcoin mining space with respect to these modular data centers, chicken coop style data centers and using those as a way to just to try to get compute stood up more quickly. So there's a lot of interesting early trends that we're seeing across this space. But I ultimately think that you're going to see the demand across the board.
- [24:45]
Sponsor Announcer
Sup freaks? This rip was brought to you by our good friends at BitKey.
- [24:47]
Brandon Bailey
Guys.
- [24:47]
Sponsor Announcer
Things happen, things get lost, you need to replace your phone. Life happens. And bitkey has been designed with this in mind. So if you're looking for the easiest way to secure your bitcoin off the exchange, go get a bitkey doesn't come with any seed phrases. There's no single point of failure. It's a two or three multisig. You have your device, you have the mobile app and block stores a key in the server. So you have collaborative multisig there. On top of that, the new bit key has a screen so you can actually see and verify what you're approving. You can look at transactions, addresses, account changes. The difference between trusting and knowing has been built into the new bit key with the screen. They've changed code Delegation and other advanced features like inheritance recovery. If you're looking to get your bitcoin off exchange, you haven't done so you're looking for the easiest way to do that. Go get a bit key. You can use our code today TFTC to get 10% off the new bit key. So go download the bit key app, use the code TFTC to get 10% off the new bit key and start your self custody journey today. This episode has been sponsored by our good friends at bitkey. Sup freaks? This rip is brought to you by good friends at Lygos Finance. Celsius blockfi ftx. They took your bitcoin and gambled it away. Lygos can't because they never hold it. Non custodial lending on Bitcoin's base layer. Your keys, your collateral, it's verifiable on chain. Go to lygos.finance, tell them that TFTC sent you.
- [26:05]
Host of TFTC Podcast
Now to your point about the, the hyperscalers, the Googles of the world, it was funny how quickly. Not funny, but just interesting to observe how quickly they went from like using their cash flows to buy back stock to the capex expenditure to the point where Google, what do they do? 84 and a half billion in an equity raise that the Berkshire Hathaway and others participated in. So literally issuing stock to invest in this capex build out and they're in full growth mode. And I think that's a big question on everybody's mind, is this a bubble? And I know I had my thoughts, which I don't think it is. I think there could be bubbly aspects of some parts of the market. But when you look at the demand for tokens today as it stands with the agentic economy being what, five, six months old in earnest. And then you think of robotics coming down the line and you think of the demand for tokens that exist today and you just project forward as adoption continues to increase at the individual and enterprise level. And then you get into robotics and I don't even think we've seen the tip of the iceberg in terms of how many tokens we're going to need to effectuate this agentic and AI driven economy.
- [27:23]
Brandon Bailey
I totally agree with you. My line of thinking is pretty much aligned with you, which is like it's not exactly the same as 1999. I mean you're already seeing the demand, right? From like a usage perspective, you're already seeing the revenue. You're already seeing like the demand side of the equation from a profitability Standpoint, just looking at, you know, anthropic revenues, if you look at like Google's revenues, even for like Gemini, like you're seeing the signs when these companies announce their earnings that the profitability element of this is real. And then I also agree with you that there are several different waves of this right now. There are more people that are not using AI than are using AI. We're still very, very early in getting people to just be using LLMs in some form or fashion on a daily basis. So I still think the adoption curve of using this technology is still very early. It's early from an enterprise level, it's early from just an everyday consumer perspect. So I still think we have a lot more Runway from that perspective. And then exactly like you said, you also have sort of the next phase, which is like, once we've built the intelligence layer right now, how do we start putting that intelligence into things like robots? And the demand for that is also going to be pretty insane. And when you think about that too, it's like we're going to need a lot more memory, we're going to need a lot more CPUs, GPU. There are so many elements of this overall AI trade that are going to have, you know, tailwinds behind it from my perspective, for many, many years to come.
- [29:17]
Host of TFTC Podcast
Yeah, and I guess that's the. I mean, maybe. I think obviously one of the big beams the last couple of months is your token usage has been subsidized by anthropic in OpenAI and they're going to need to really push that, the cost that they're bearing onto the end user, which is beginning to happen. You have companies like Uber saying we blew through our budget in the first quarter, first quarter and a half of this year, first four months of this year. And I think that's one thing I'm still trying to, to wrap my mind around. It's like, okay, what is the actual cost of a token, particularly from a frontier model versus an open source, open weights model, and how does that affect the demand for power in the long run? I'm trying to. Does token usage get so efficient and so cheap that the return on the capital invested in the infrastructure compresses over time? Or is the demand for tokens, even if they are becoming more efficient and cheaper, just going to be so massive that it doesn't really matter? How are you thinking of that? That calculus?
- [30:31]
Brandon Bailey
Yeah, I mean, I'm also trying to wrap my head around that efficiency dynamic with respect to tokens per watt or however you want to sort of characterize that Efficiency metric. Efficiency metric. I think it kind of circles back to, I believe it's Javon's paradox that people talk about, which is like, you know, the cheaper something gets, the more you'll want to consume it. I feel with my own personal just usage of LLMs, that's kind of where I'm at. I'm like, the more tokens you're willing to give me, the more tokens that I'm going to consume or I'm going to look at it as like, okay, well, I might have been restricted from doing these other things I want to do because I've hit my usage limit. If you give me more tokens now, there's more things that I can do. So I think that, you know, as things get more efficient, you will also have probably step changes in the overall consumption and demand that probably offset a lot of the efficiency gain and continue to keep the overall demand level or the need from a power sort of perspective elevated. Or there's. Right. And you still have, like I said, you still have sort of that undercurrent of, you know, the game theory of, you know, I'm, you know, Meta is competing with Microsoft, is competing with Google, is competing with Amazon and they're all going to want to make sure that they have enough Runway such that they can protect their moat, protect their margins. And then there's going to be all kinds of other things that I think that we haven't imagined as well that are just going to allow for more consumption, more demand for using the tokens for other use cases, other applications. So that's kind of where my current thinking is on it as we think about that. But it's going to be very fascinating to kind of track that trend over time. Yeah.
- [32:30]
Host of TFTC Podcast
And bring this back to energy infrastructure. I mean, it seems like, I feel like in the next year all the nooks and crannies of available power on the grid that's willing to either that's either available to be acquired by companies looking to train or run inference on these models, or that is willing to transition from a use case that is not AI compute, we're going to hit that limit of where like, all right, we've sucked all the energy, all the available capacity out of the grid and now we need to expand generation. How do you, how do you view the generation expansion playing out?
- [33:14]
Brandon Bailey
I think that that's going to be another big wave which is like the, you know, bring your own power behind the meter power at a lot of these campuses. So you know, I think to your point, the generation piece is, is going to be interesting to see how we, we ultimately solve that. You know, obviously there are plans for like nuclear projects and things like that to try to bring more baseload. But the, one of the big challenges we're already observing with some of that is the timing mismatch, right? Like how long does it take to actually build out or stand up? You know, at least for a nuclear, for a nuclear plant it could take five years or longer. The demand is here. We need, we can't wait for that. We need this demand now. But I think like the next phase of it is like how do you bring more like gas beaker plants or other things online or whether it's solar or what have you, all different forms of generation to these data centers and you're seeing some early elements of that. But I think that that's going to be the next sort of wave just because I think the grid interconnection piece is going to become more challenged. And then I also think if you are able to tell the narrative of having brought your own power to a community, then you're not necessarily taking anything, you can make this point that you're not taking anything away from the local community. So I just think that that's also just going to be viewed more favorably from a social and political perspective. But that's also a way that you can increase the amount of capacity at where you already have some of these existing campuses. So I fully, fully expect that to be another big wave. And we're already seeing companies kind of hint at this piece of. I've seen Tyler Page on various talks really talk about the behind the meter power opportunity at many of their existing campuses. Core Scientific and Adam Sullivan have also talked about that. From an investor standpoint, I actually find that that's an element that I don't really think is getting enough attention, which is what is the incremental power capacity potential from bringing additional or behind the meter power solutions to an existing site to expand the potential leasable capacity at an existing company's sites or campuses.
- [35:44]
Host of TFTC Podcast
Yeah, it's so fascinating. How big is this opportunity? What is happening right now in terms of just broadcast implement implications on humanity and markets specifically. Again, going back to everybody drawing parallels to the dot com bubble, I don't think the parallels are as clear as many people think they are. I think they're actually completely different and there's a lot of doomers out there who are like it's a bubble it's going to pop. It's going to pop. Look at SpaceX. Yeah, it may be overvalued right now, but again you look at what Mid Journey launched last night with that MRI competitor where you can do scanning, image scanning of bodies in a minute. And it's like this technology is valuable. If it works as advertised, you could replace every MRI in the country, in the world in the next few years and have a better product that saves millions of lives. And so there's a there, there I use. That's another thing we can get to eventually is like we're talking about the infrastructure build up and on the application usage. Like I know you, you have built a product using this and we'll talk about diametrics, but I think just first trying to paint the picture of, of what's happening here from an economic disruption standpoint and the opportunity of like wealth creation that exists.
- [37:10]
Brandon Bailey
It's, it's, it's remarkable. You know, I think it's a generational wealth creation opportunity for people. I think it's, it's already largely been that if you just sort of look at how some of the, you know, the, the various names across the, the broader AI sort of bottleneck thesis or bottleneck trade have performed. So I'm talking about, you know, look at Nvidia over the past few years, Micron, SanDisk, AMD, intel, right? Like the returns on some of these names, I mean you're up 3 to 10x across a lot of these. You could have almost just picked any sort of semiconductor stock and had generational return type of performance on some of these names. And so from an individual investor standpoint or even if you're an institutional investor, you know, it's, it's been incredible market performance just on the back of what's happening here with respect to the, the amount of capex that's being poured into the space. But then when you talk about, you know, so what does this mean for like everyday, you know, people's everyday lives? I mean it's, it's, it's incredible and it's, it's still hard to almost fathom or imagine some of the possibilities. I mean, we'll talk about it, but like DI Metrics is for me personally one of those things. I've always wanted to build an application of some sort. I have no coding experience, knowledge whatsoever, but leveraging these tools I was able to actually take something, an idea of mine, and actually turn it into some kind of tangible product that other people could use or some, some other sort of application. So when I think of the power of this technology, it's, it's a, it's a massive force multiplier for people. Right? Just the extension of the amount of additional knowledge that it can help provide an individual. You can basically learn anything now and get that information instantly. Prior to the LLMs, you could always go to Google and do a search, but that wasn't always effective. You can basically learn about any industry, any incredibly complex topic or subject and get an incredibly accurate or reasonable response. You could have it provide you direct sources to like, you know, peer reviewed papers, what have you and even have the system explain it to you, like give you an ell 5. Right. Which is remarkable. So like I think the tool really empowers people to learn about whatever they want. I mean, which is massive. I mean I just think that that's massive for people. And it's really hard to kind of quantify like if you want to put this into like a GDP perspective, like what does that actually mean for productivity? It's sort of hard to actually quantify that at this moment in time. But you know that it actually has massive implications for society just in terms of leveling the playing field. Right? Intelligence is now a commodity that's, you know, you can't say enough about the importance of that ultimately for the world and for civilization.
- [40:35]
Sponsor Announcer
All right, freaks, you know me, you know I don't take sponsor money from products I wouldn't use myself.
- [40:40]
Host of TFTC Podcast
So listen up.
- [40:41]
Sponsor Announcer
The Aven Bitcoin Visa card is one of the most interesting things I've seen in the bitcoin lending space in a long time. Here's the deal. You can get a line of credit up to a million dollars backed by your Bitcoin without selling a single sat. No gains, no annual fees, no minimum draws. And your Bitcoin is custody by Bitgo, which is one of the most trusted names in digital asset security. Even never lends it out. There's no rehypothecation. You stay in control and guess what? You can lock in a fixed rate for up to 10 years. That's 10 times longer than most lenders out there. Or go interest only for up to five years. Rates start at 7.99% APR. For a product that lets you keep your stack and still access liquidity, it's hard to beat. I mean the duration in the rates is the best I've seen in the market to date. You also get 2% unlimited cash back every time you use the card. Spend fiat. Keep your Bitcoin the whole game if you've been stacking for years and you need liquidity without triggering the taxable event, this is worth a serious look. Go to aven.combitcoin that's AVN.com bitcoin check it out. Freaks. Look at me. I'm glowing. I've got like an angel's halo going around me. You know why that is? I feel good. I feel taken care of.
- [41:51]
Host of TFTC Podcast
I feel blessed, healthy, happy.
- [41:53]
Sponsor Announcer
That is because I'm a crowd health member. My family and I have been crowd health members for five years now. Literally this month, five years ago, we joined crowd health. We've had two babies, we've had multiple health events, and we're never going back to health insurance. Crowd health is crowdfunded healthcare. So you, you sign up for crowd health, you pay a monthly fee, you
- [42:10]
Host of TFTC Podcast
help out with other people's bills, and
- [42:13]
Sponsor Announcer
it's significantly cheaper than health insurance. We were on cobra. It's a family of three when I, when I left my last job before I went full time to tftc, went on the crowd health. Now as a family of five, we pay, I believe, $700 a month. It's significantly cheaper. They're going to negotiate prices lower for you. They've consistently negotiated health care prices as much as 50, 60, 80% in many cases. They help out with babies. If you have a pregnancy, you pay the first $3,000 and the crowd covers the rest. If you have a regular health event, you pay $500 and the crowd pays the rest. Go to joincrowdhealth.com, sign up today. Use the code TFTC. Opt out of health insurance. I'm uninsured and I love it. Use the code tftc@joincrowdhealth.com you'll get 99amonth for the first three months that you're on the crowd health platform in the community. Bitcoiners, you found sovereign money, now find sovereign health and sovereign healthcare.
- [43:07]
Host of TFTC Podcast
And so many people are sleeping on it right now. That's what I keep wondering. What is the adoption curve look like or adoption timeline look like? And is it the adoption just forced on people? Is just looking at how some companies are implementing it and experiencing the same productivity gains, others obviously implementing it. I think that's maybe that's something we can talk about. That's observing particularly the headlines you see out of Uber, like we're blowing through our token budget and the trend of token maxing the CEO coming out like, you're not spending tokens, like you're, you're not doing your job. And I think one of the things that really stood out to me as somebody's been implementing this at our business here at TFTC is like the implementation details are very, very important. Yes, you can token max. If you're token maxing liberally without any intent or thought about what the actual end goal or product will be, which should be to produce something that makes you more efficient and adds to your bottom line, then you're just going to be blowing money. And I think that's the phase we're in right now is people sort of trying to map the territory, particularly on the implementation side, to make sure they're actually getting value out of these LLMs. And I think that's a massive arbitrage right now that exists is those who understand what these things can do and how to implement them to do your job better, and those who are simply using them as like a new Google search.
- [44:37]
Brandon Bailey
I totally agree. I think that what you just laid out is sort of maybe the counterpoint to the. I kind of view it as a little bit of like the counterpoint to the narrative that AI is going to take all the jobs, which is like using LLMs is a skill. Like you have to. You have to learn how to maximize the output of the system. It's not something that you can just like go to and say, like, you know, I'm just going to give it a prompt and it's going to just do all the things like you, you actually have to work on your prompt engineering. You actually. It's one of those things where you just have to kind of spend time understanding the system and how it works in order to get better at it. And I think that the people that invest that time or token maxing, so to speak, you kind of almost have to go through a phase of token maxing so that you can learn how to use this system. What are the LLMs good at? What are they not good at? What additional context do I need to include in my prompt to make sure that I can. One shot. This output versus going through many iterations and back and forth with the LLM. So it's a learned skill. And I think that that's where we're at because we're so early in this inning. People are still trying to figure out how to use the system, how to get the most value out of the system. But I think the people that are investing the time to learn that right now are going to set themselves up in such a way that they're going to be Some of the most valuable people for enterprises because companies are going to implement strict budgets. Everyone gets a certain amount of tokens and then it becomes that efficiency gain. How do I get the most output for the least amount of tokens possible that I can spend? And that's going to be one of the. I think that's going to be the primary way that enterprises start evaluating individuals or employees, which is like, how much output can you derive from leveraging this tool? And to your point about being forced to actually use it, if you're one of those people that are anti the technology or a laggard, you're just really setting yourself up to be at a tremendous disadvantage relative to potential some of your other employees internally. And so it's kind of an interesting incentive mechanism where you're doing yourself a massive disservice if you choose to not use the tool. I mean, same with the Internet age. Same with the Internet age. I think it's really no different in this instance here.
- [47:20]
Host of TFTC Podcast
What's your personal journey been like? Mapping the territory and getting used to these tools?
- [47:24]
Brandon Bailey
Man, it's like a roller coaster. It started out as like, wow, this is amazing. It can do all these things. This changes everything to wow, that's how it started. And then the more I used it, I was like, wow, it's missing things. I have to provide additional context. You can't like, I guess the way that I would describe it is like the LLMs aren't great at assuming, like, they're not great at seeing around corners. I'll put it that way. Right. Like, unless you give an LLM, like the context of I am trying to build this system. We need to build a really strong foundation. These are all of the potential pitfalls I'm trying to avoid. Here's why I'm trying to avoid it. Can you help me think of a way to craft the best possible system that can avoid these potential edge cases? Right. If you don't give it that context, it'll just give you an answer without thinking ahead to where that thing could break down, which leads to a significant amount of iteration and back and forth and having to rebuild. And so that's where I think, to me, that's where it kind of comes back to that token maxing thing. It's like I had to go through a lot of that to get to a point where I understood, like, what things do I need to provide this LLM to make sure that I don't have to spend a lot of time and a lot of tokens doing this back and forth. So for me it's really been about like, how do I create the right processes effectively when I'm trying to build something to make sure that I'm setting myself up for success without burning a ton of tokens. And now that I've kind of learned how to do that a little bit better, I'm kind of like back on the upswing of this is amazing. And you also kind of work through the evolution of the different models, so you kind of see the different step changes as the models improve and that just kind of reinforces everything as well. And the last point I'll leave you with, which is like, when you think about where we're at with the improvements to the models, think about how much compute like opus 4.6 was trained on or 5.5 was trained on and you think about how much more compute is now being energized. Right? The bulk of what we've been talking about from a power capacity and utilization standpoint, like models aren't even training on the power capacity that we're talking about that's coming online. So when you really think about the future step changes of these models and what they're going to be capable of with all the new compute and power that's being thrown at them, you got to be really excited for what could become possible in the near future here.
- [50:24]
Host of TFTC Podcast
Yeah,
- [50:27]
Sponsor Announcer
yeah.
- [50:28]
Host of TFTC Podcast
It's insane to your last point there, it's like because I played with Fable 5 before it got behind the by the US government and it was like, holy crap, this thing is really good to step up opus4.6 for the written word for opus4.6 is the brain of our open claw agentic system that we've built out here at tftc. And it does very well. I can't imagine what happens when you have two more step function improvements at a similar cost to what Opus 4.6 is today. It's mind boggling to think of, okay, how much better is this going to get over the next year or two? To your point about that learning curve too, I think that's one thing that over the last two months is really going back to the context, making sure that you're prompting it the right way. And I think having sort of persistent memory for context for your agent is. Is imperative. And we've been in the process of building out our company brain for, for the last two months and it is insane how just building knowledge graphs and semantic search functions within the server that our agent runs so that it can ping first before it burns Tokens. So it gets all the context from the knowledge graph and the semantic search system that we set up and it's just able to one shot things in our voice and it knows our business inside and out. And I think once you understand the tooling that you plug into these LLMs, that's when you really unlock the superpowers.
- [52:11]
Brandon Bailey
Totally, totally. It's like the LLM wiki or like Karpathi's concept. Massive. That's a huge one. The memory files. I agree with you. It's kind of like once you figure out that setup, right. It's like all the connectors, the skills, all of that stuff really unlock significantly more value for you and also allow you to reduce the amount of the token burn. But that's what you kind of have to go through the trial and error of figuring out. How do you map out the system? Right. It's like you kind of got to build your own system, map for how you're going to operate this system and then once you've kind of perfected that, you really start to see the fruits of that labor.
- [52:54]
Host of TFTC Podcast
Yeah. Wild times. So let's talk about DI metrics. I mean you mentioned it. This is, was an idea in your head for many years and now with the tools you're, you were able to build it. And I've been playing around with it this week and if you're a nerd on the forefront of following what's happening on the infrastructure build out this, this tool that you've built is incredibly powerful.
- [53:18]
Brandon Bailey
Yeah. So the way that I would, the simple one liner I would give for it is it's really a tool for kind of the individual investor that is interested in the digital infrastructure space. So market intelligence platform for digital infrastructure specifically. And the whole concept and project, it started as a spreadsheet from my days at Galaxy and working in bitcoin mining. I would track a lot of these companies sites and all of this, the financial information about how much hash rate they had, how much bitcoin they mined and all of this stuff literally just via spreadsheet. And as these companies were making the transition, I was trying to do a similar thing where I'm just trying to track, map out where do they own sites, how much power capacity they signed a lease. Kind of doing all the typical work that an analyst would do if you're looking to be an investor in this space or in these companies. And it was with the LLMs where I was like maybe I can automate this workflow. It takes an enormous amount of time to ultimately comb through SEC filings and to go through investor presentations and keep up with, you know, all of the, all of the information that is required for you to map this out. And with every new company going into this vertical, right it becomes even harder for you to ultimately be able to track and follow everything that's happening in the landscape. So for me I was like, I want to take what I'm doing in spreadsheets and see if I can automate that workflow into basically a dashboard effectively that I could just use to, to, to monitor and track what's happening in this space without having to commit so much time to, to doing the manual labor. And you know, what I ultimately started building out was like the back end database to support this. Started trying to leverage the LLM to, to ultimately learn how to read through the investor presentations, like basically map out this whole thing. And where I kind of see the real value proposition for this dashboard is when you think about other market intelligence tools like a FactSet, like a Bloomberg, there's Koi Fan, there are other newer ones as well that do an excellent job of what they provide you. But what it is is more high level information, right? They're giving you, here's the line items directly from the balance sheet, from the financial statement, here's what the total debt number is, here's what the cash bal, here's what the revenue is, here's what the capex is. What I wanted to build was something that went super granular and super deep into the individual components of specifically the data center space. So I wanted to build a market intelligence tool that was going to tell you what is every site that a company like Cipher Mining owns, right? What is every site they own, when is, you know, what's the gross power capacity of that site, what's the leasable power capacity of the site, when is that power going to be available? What's the energization timeline, where is it located, who's the utility provider? I wanted to go super, super deep into creating a tool that would give you the industry level granularity, right as it relates to these companies. So if you're an investor in this, you don't have to still comb through all of those filings to pull out this information. And so that's what the DI Metrics market intelligence platform ultimately tries to provide. And then it ultimately is trying to give you that level of information through the MPC connection. And so it's like how do you take that level of information and incorporate it into whatever agentic workflow or process that you already are familiar with working with. And that's what I think the real value add or the biggest feature is of the platform is being able to connect DI Metrics database to your Claude or Codex or what have you, and then tell it or ask it like, give me every lease that Cypher has signed, what's the lease rate? And then ask it like, how does that compare to Terra Wolf or Rank? All of these companies, show me the top 10 companies that have signed like the most profitable lease from a gross margin perspective, right? So that type of analysis would have taken many, many hours for you to map out as an individual. Now it's just a simple prompt and like that's the value proposition. That was something that I was ultimately just trying to work through on my own as an investor, as somebody that actively trades a lot of these names and ultimately wanted to provide that type of level of market intelligence to a broader group of people. So I wanted to do that. And then the other big thing that I wanted to do is again, when you think about the broader AI bottleneck trade, I think of groups like Centrini Research, Fundai Semianalysis that all do amazing, incredible, excellent level of work and research as it's related to this trade. But a lot of what they're mostly focused on is going super, super deep into the chip component, the semiconductors, the hardware, the compute. And they do a phenomenal job with the research that they do. But I feel like the data center space, and more specifically these miners convert their capacity are somewhat undercovered. They're not as strongly represented relative to the semiconductors, the microns of the world. And so that's something that I wanted to try to bring a little bit more light or shed a little bit more light on is the highlighting the opportunity that these companies present as a bucket within the broader AI bottleneck trademark and with the terminal element of DI metrics. I try to do exactly that, which is show you the performance of these Bitcoin miners relative to the semiconductors, relative to the AMDs, the hyperscalers, et cetera. And these companies have actually provided almost, not quite as good of a return as the memory companies, but actually better than a lot of the other components like storage, et cetera. And I feel like that is not talked about enough. And I still think that there is an incredible re rating opportunity for many of these names. And so I really wanted to just try to bring more attention to that as well as highlight exactly what is that opportunity and how do you think about it? How should you think about this?
- [60:31]
Host of TFTC Podcast
Yeah, I love it. And it highlights like on the application side, like somebody like you, an incredible analyst who's been analyzing the infrastructure built out of Bitcoin mining and now HPC compute or AI compute as well. I keep making that mistake because bitcoin mining is technically HPC as well. But the AI compute and you have this experience and this depth of knowledge that very few people have and you're able to leverage the tools to build a tool that other analysts can tap into and you can get paid for it. And it's like, that's like you have your knowledge graph of DI metrics that only somebody with your experience knows how to map the territory of this particular layer of the infrastructure build out the sub theme within the AI build out and you can leverage the tools to build a tool that others can leverage to incorporate. There's definitely like you mentioned, semi analysis focused on chips like they can then expand their, their depth of knowledge and the AI build out into the infrastructure layer as well. And on your point of rerating like again bringing back like the fears of the bubbles, like where would you say we are in the, the process of rewriting? These companies that are earnestly making this transition are going to execute and those who are execute who, excuse me, those who are in the process and if they do execute, what do the RE ratings look like from here?
- [62:07]
Brandon Bailey
Great question. So I kind of break it down into two separate buckets in terms of how I'm looking at these companies. There's the companies, the bitcoin miners, if you will, that have signed leases and those that have not signed leases. And so the way that the rerating process works, there's kind of two big phases. The first phase is can you sign an initial lease? We kind of touched on this. That's the biggest RE rating sort of upside you see. So like once a company has signed its first lease, that's where you see the largest sort of pop in a lot of these companies. That's been true for Cypher hut Terrawolf. That's why some of these companies are up as much as 5 to 7x right from the point that they had no lease to sign to signing that, you know, that first lease and where they are today. And a lot of it has to do with the fact that in the early days because it's changed a little bit. But just to give the context, the big opportunity was the fact that these companies were highly correlated with bitcoin. And so if you look at bitcoin's price, performance More recently, let's say over the last 12 months, 12 to 18 months, Bitcoin kind of hit a high and then it started to sell off. Many of these miners sold off with bitcoin despite the fact that they had announced that they were looking to pivot to HPC AI. And so the value of their power portfolio, the value of their megawatts was declining because bitcoin was declining. And you basically had the market not believe that they had any real potential of converting to AI compute, which created a massive opportunity, right, from a value perspective. So then once you had Core Scientific sign its first lease, you saw this massive rerating because the market was basically assigning like a 1% very low probability to that ever happening. And so once you saw that firstly skit sign, you saw a huge step up in the probability which caused an enormous sort of re rating or revaluation of the megawatts of those companies. So that was sort of the earlier opportunity that has resulted in pretty tremendous upside. But then even after a company has signed a lease, there's still a pretty large discount that gets applied to even names like a cypher, a hut 8, a core scientific. And that second component is the market assigning a probability of them being able to finance the build out of the compute and then them also being able to actually deliver on delivering the building. So the construction risk, right? And so once they raise the capital, which many of the companies have been able to do more recently, that de risk the project. And so there's a little bit of a rerating on signing a very attractive debt financing. And then there's a rerating, right? Again, once the company delivers a successful shell on time, that construction risk effectively goes to zero. And so the easiest way to look at it from my perspective is if you look at Digital Realty Trust or these traditional data center players, that's your new benchmark. And these are effectively real estate plays. And the way that you evaluate real estate is you can look at things through the lens of a cap rate. So you could say a stabilized cap rate is roughly 6%, right? A DLR in Equinix, they're being valued at a five and a half to a 6% cap rate. These miners at any given time could be valued at a 9 cap, right? Like cypher could be valued at an 8 and a half cap or a 9 cap, right? And the delta between where a cipher's trading and like a digital realty trust, that's all the market's discount on the probability of, that's the execution discount. Effectively, that should ultimately trend closer to parity once a company has delivered a powered shell, once they've delivered on a lease. And so what's really fascinating is that you can track that relative spread of the cap rate daily just looking at the market performance and how these things are trading. And so that is a potential market signal. So that's one of the things that I look at. And you could say, hey, these miners have sold off exponentially. We saw a lot of this with the volatility in the market, just with the Iran war and the back and forth with Trump's tweets. Every single time you see a blowout in that spread. Incredible buying opportunity. So that's been one of the ways that you could trade it. And now you're also seeing an emerging opportunity with the smaller names. We talked about this. So that's, you know, the smaller bitcoin miners that maybe have 50 megawatts of power capacity, they're kind of been an afterthought. There's still a massive rerating opportunity for them through the same type of cycle and phase. Right. And so, like, that's really what I'm trying to triangulate on and highlight as the opportunity. Right. And kind of talk through that sort of overall revaluation process or timeline. But I'll pause there. I know it's kind of a lot. If there's anything you want to dig into, I'm happy to more, but that's kind of how I see the opportunity.
- [67:34]
Host of TFTC Podcast
Yeah, well, I mean, it's a big land grab right now. I guess that's the next question I have, is how much leverage do these miners who have power and land have in these negotiations? And how much of this is a lotto ticket versus, yes, you may have the land in power, but you've gotta execute on the back end. And bringing back the fact that a lot of these AI compute, the people that are looking for infrastructure for their compute come with a spec and design. Is the execution risk significantly high or is it like, hey, you have land and power, you have the leverage. It's like, how hard is it to figure out how to get this thing, get a lease, get the financing, get it stood up and energized?
- [68:24]
Brandon Bailey
The execution risk is significant, and that's why you kind of have to. You have to bisect the field a little bit. Right? So if you're Talking about a HUD 8 or a core scientific at this point, because they've already done it, they've kind of proven it out, any incremental lease is going to be less risky than the prior one. Right. So there's a compounding effect to this. Right. Of course, for the smaller miners, yeah, the execution risk is even higher just because they have less financial resources. Maybe they have smaller teams. Right. But that's sort of your risk reward upside. But where it becomes really interesting is that the value of power that is already energized effectively or the power can already be drawn is so valuable right now because of the demand and the constraints and all of the other growing challenges that those companies could pursue. A JV style structure or other structures where they could potentially contribute that land and power at a significant markup to where it's currently being valued. So I mean, you have some of these smaller miners where their power portfolios are only valued at $200,000amegawatt to $400,000amegawatt. They might be able to contribute that land and power into a JV at a value that's a million dollars or more a megawatt. Right. And then lean on the balance sheet and financial strength of a JV partner to help de risk the execution component. Right. And then earn cash flow at whatever their pro rata share is through the jv. But even a structure like that is incredibly accretive for these smaller miners just given the literal markup that they can ascribe to the immediately available power that they have. And that's what people aren't seeing. And it's a little bit of a snowball effect. Like I'm saying, once you get that first deal done and you've proven that you can execute, it becomes a lot easier for you to get that second deal done. And you might not have to give up as much of the economics in order to get it done. So that just gives you a significant sort of tailwind and Runway for what these companies could ultimately become. And it's really a lot of these smaller names that have not been covered, right. That's like it's a sphere of 3D, it's a Greenwich, it's what was formerly Mawson, but a big digital energy. That's a DigiPowerX, that's a DMG Solutions, that's a Soluna, and there's many, many others. Right. The subset of these companies is much larger than just Terra, Wolf, core scientific cypher, hut8wolf. And that's what I want to shed light on. There's a large opportunity here and you can kind of pick and choose your spots across the risk spectrum, but the overall RE rating opportunity just in terms of how the power capacity is being valued to where you see a stabilized project being valued is tremendous, tremendous upside.
- [71:54]
Host of TFTC Podcast
I guess to wrap it up just thinking about not only the opportunities that exist for these individual companies, but I think for the localities too. Wrapping up with obviously ERCOT's leaned into this heavily with Bitcoin mining and now AI compute and there are some laggards across the country, some nimbyism going on. And I think to your point earlier about the narrative and the misconceptions around what's actually happening, like what's going to happen to the counties or states that say we don't want this here and what opportunity are they foregoing if, if they neglect to embrace this build out?
- [72:37]
Brandon Bailey
I think, I think similar to bitcoin mining, it's a bit of a missed opportunity for just job creation but more importantly to bring additional generation to your local county. To me the simplest way I can explain it's supply and demand. Everybody wants more generation, but generation is not going to come if there's no off taker. People have to be able to make money. And so if you're a county that's pro data centers, there's a way to kind of take advantage of, of this, this capex boom cycle and also find a way to bring incremental generation capacity. Right. I think to your county which could be net great. Right. It's also a way for you, depending on, you know, your, your locality or what grid, what have you. It's also a way for you to maybe bring more renewable energy or more baseload whatever it is. Just more generation in general is great but it's also an opportunity for you to kind of think about that mix the grid mix or the energy mix and also find ways to help to incentivize or shift that in ways that maybe align with whatever your view is and that county. So to me that's what I kind of think the missed opportunity is. But ultimately counties, states, they all have to compete. And so you know, I think it's, it's fine if a county doesn't want a data center. You're just choosing to not play in the game. But that may come at future consequences of just. You may struggle to get new generation and you may actually end up seeing power rates increase in those counties that were anti data center. I actually think that would be beautiful if you actually saw power rates drop in the counties that were pro data center versus the ones that were anti data center. But time will only tell.
- [74:32]
Host of TFTC Podcast
You can see electricity dropping property Taxes maybe dropping, because that's another thing. I don't think people recognize the sales tax on the electricity, depending on where you are, it's just generating incredible revenue for these counties.
- [74:45]
Brandon Bailey
Proper conditions. That's a massive opportunity. I mean, you could mandate however you want to regulate it, but there's opportunities to say, hey, if you build a data center here, you need to contribute X amount of capital into like, education fund or whatever it is. But there are ways to take advantage of this growth and to capture some of the capital to invest back into the community in various ways. I think that's an absolutely excellent point.
- [75:13]
Host of TFTC Podcast
Yeah. All right, I lied. That wasn't the last question. The last question is where does. Where. Where do people mess up here? Because you're looking at, like, cues for interconnection, reaching tens to hundreds of gigawatts. It reminds me a lot of ERCOT in 21 and 22, where Bitcoin miners are coming and they're like. Or people saw the, the sort of gold rush of the exodus out of China, of hash rate to. To the us and they were like, yeah, I've got land, I've got power, but they're really just in the queue of ERCOT having to wait. And then they were going on the back end and selling the dream of, I'm going to get this much power in a year or two and we're going to be able to build this massive mining operation. Looking at what's happening with the AI build out, particularly as it pertains to locking down power, it seems like that's happening again at a scale that's an order of magnitude higher. How do you see that playing out?
- [76:11]
Brandon Bailey
I think you're absolutely right. I think there's a lot of sites that ultimately won't end up ever being built. And I think we saw this exactly like you said in bitcoin mining. But that's part of the challenge. As an analyst, if you're trying to invest in this space, which is figuring out what's real and what's not, when we talk about pipeline and a lot of companies talk about pipeline as being real, when exactly like you said, you just bought some land, there's no load study that's been conducted, have an interconnection agreement or any real visibility or line of sight into if you're ever going to actually be able to draw power at that site, and if so, to what degree or to what capacity, and you have a lot of people that are marketing these pipelines without having any of those things in place. And one of the challenges as an investor is understanding like what risk weight or what probability do I need to assign to a company's pipeline and why and what's the right probability given these various milestones that you could ultimately have in place. But I think that that's another element that makes me so bullish on the bitcoin miners, which is the grid interconnection queue or just the request for connectivity are enormous. You know, a large majority of those are never actually going to be permitted or they're certainly not going to be permitted on a timeline that is anytime soon. And so to me, I think it just reinforces the incredible value of having already available power, which all of these miners do. And I think that as that plays out and as more people realize that only 10 to 30% or some small percentage of the overall cue or request will actually be built, I think it's just going to reinforce the value per megawatt of the power portfolios of the companies that already have it. And then you throw in there the political risk as well. And that's just another sort of force multiplier on the value of the companies that already have the energized or approved zoned power capacity.
- [78:33]
Host of TFTC Podcast
All right, I lied twice, but I promise you it's the last time I lie. Last question. What does this mean for bitcoin mining moving forward?
- [78:40]
Brandon Bailey
Ah, this is great, great question. What does it mean for bitcoin mining? It actually kind of makes me pretty bullish on bitcoin mining in the more medium to long term. I actually think what we're seeing is just an acceleration to what many, many miners talked about for many years now, which is like bitcoin mining is going to gravitate towards the stranded forms of energy. And I think that the massive race for AI compute is just accelerating that. So a lot of the grid connected power capacity is likely going to be allocated towards AI the mega sites. We're moving away from these mega sites. And I think that bitcoin mining is going to gravitate towards those stranded forms of power and it's also going to be more distributed. It's going to be about your 5 megawatt site, your 1 megawatt site here and there that is sort of more of a stranded form of energy, which you could argue is actually great for the health of the network because it makes bitcoin mining a little bit more distributed and potentially decentralized, which I think is great. I also think that you could end up in a scenario where you just have less overall power capacity allocated to bitcoin mining. So if you have bitcoin price recover rapidly the power element, you're not going to have as many people able to quickly plug in asics. And so you could end up in this era where hash price is actually appreciating because people just don't have power availability to actually go chase those economics, which could lead to a very compelling golden era of bitcoin mining economics, which I certainly hope for. And I think that we should also end up in a scenario where we hopefully move away from the oversupply of asics. That's been one of the issues that has really played bitcoin mining which is just bitmain microbt bit deer. They're all trying to preserve their capacity at the foundries. They have to continue to mass produce these ASICs despite there not being demand. So you have this massive supply glut of machines effectively. And I'm hoping that we end up in a world where the supply demand between ASICs and the amount available power ends up in a more balanced and normalized regime which I think would make mining economics more durable over the long run. And so that's kind of some of my medium term to longer term thesis for bitcoin mining, but I'm pretty excited about it in the medium to long term.
- [81:18]
Host of TFTC Podcast
I agree there. I think it's overall bullish for the network. Many people, again, like I said in the beginning of the conversation, are like, oh no, AI is taking the wind out of the sales of bitcoin mining and that's bad. It's like well no, I mean everything's bullish for bitcoin. Everything's bullish Bitcoin. We're going to distribute hash rate. It's going to be geographically distributed. This will force ownership distribution to be more distributed as well. I think this is overall good. And then I think also for countries outside the United States that aren't taking advantage of this AI compute build out, there's going to be opportunities for bitcoin mining to land there as well. And while I do like my hash rate to be American made, it is good to make sure that it is geographically distributed around the globe in different jurisdictions. So. So I'm bullish as well. Brandon, this was an incredible conversation. Thank you for taking some time out of your day to do it. I think you're doing incredible work. And for anybody listening, make sure you go to DI metrics AI to check out what he's built. And if you're curious or if you're on on the beat of covering this infrastructure build out Brandon's built. Incredible tool.
- [82:37]
Brandon Bailey
Thank you. It was a pleasure to be able to come on the show. I really enjoyed the conversation as always.
- [82:45]
Host of TFTC Podcast
Awesome. We'll do it again at some point because I'm sure this theme is not going to slow down anytime soon.
- [82:50]
Brandon Bailey
Definitely not. Definitely not.
- [82:53]
Host of TFTC Podcast
All right. Peace and love, freaks.
- [82:55]
Sponsor Announcer
Thank you for listening to this episode of tftc. If you've made it this far, I imagine you got some value out of the episode. If so, please share it far and wide with your friends and family. We're looking to get the word out there. Also, wherever you're listening, whether that's YouTube, Apple, Spotify, make sure you like and subscribe to the show. And if you can leave a rating on the podcasting platforms, that goes a long way. Last but not least, if you want to get these episodes a day early and ad free, make sure you download the Fountain podcasting app. You can go to Fountain FM to find that $5 a month gets you every episode a day early ad free helps. The show gives you incredible value, so please consider subscribing via Fountain as well. Thank you for your time and until next time.