The Bitcoin Treasury Machine | Harry Sudock & Rory Murray

Bitcoin mining and AI are both technological waves that are running into atoms problems. Bitcoin has always existed at the edges. It adapts to the new era and the new narrative and the new thing. What you have with Bitcoin is you have a liquid global permissionless asset that trades 24/7 that has upside optionality. You use that appreciating currency to borrow in a depreciating currency. You use that depreciating currency to go get assets that are going to appreciate. We've been through multiple liquidation cycles. We have institutional trading infrastructure. We have banks that are getting involved. The liquidity profile of the collateral liquidation mechanism is as close to seamless and lossless as we've ever seen any market for credit ever.

So the reason that you're seeing a proliferation of AI as part of the mining strategy is because there are actually megawatts that are better suited for AI compute workloads and there are other megawatts that are better suited for mining and there are a third category of megawatts that I think deserve to have a combination of both workloads installed against them. So that's from a very operational perspective how I think the sector is thinking about it. The other is that from a durability of cash flow and therefore multiple expansion, the market is willing to value equities that have AI revenues at a significantly higher premium than mining revenues.

Well, I just think it's wrong. I don't think AI is competing for the same energy story that Bitcoin mining is. I think that they run even a narrative basis. The way I would characterize the AI energy story is that we are not generating enough power. The way that I would characterize the Bitcoin mining story is that we are not consuming power nearly as efficiently as we could be. And to me those are similar in that they both represent higher utilization and higher production of electrons but the operationalization of each of those strategies looks quite different.

Yeah, for sure. I think that the concentration of hash rate, forget what corporate structure it lives in, but the concentration of hash rate geographically will get decentralized because the large concentration of energy-backed compute is going to get pushed through an AI workload because it's the highest enterprise value activity for that megawatt concentration. And then there's very likely to be a squishy layer of Bitcoin mining attached to it to make it flexible and responsive. And then you're also going to see mining move deeper across the geographic frontier, the interruptible frontier, and the cost frontier, which in a lot of cases looks like jurisdictional arbitrage. But it also looks like geographical and runtime arbitrage as well. I mean the entire history of Bitcoin is Bitcoin quite frankly being a honey badger or more succinctly quite frankly a cockroach right and I think that if anything this AI story is one of one I believe continued American dominance across energy and financial and capital markets and secondly, you know, again just to your point is that they are two completely different spaces for the particularly for the training side you want these big mega campuses you're sucking down lots of power you want to be 100 miles outside of an NFL city, you want that great transmission. Bitcoin has always existed at the edges. And I think the criticism of Bitcoin often has been, oh, the narrative changes and evolves. And that's been the transformational part of it over time is that it adapts to the new era and the new narrative and the new thing. And so now it's going to move back, I think, from a decentralization perspective. Whether it's public miners, I think we're finding some what there we would call them smaller sites now as we're kind of scaling up, but to us five years ago, three years ago, they would have been midsize or kind of regular sites. And so you're moving to this kind of hub and not only do you have that squishy layer kind of around the mega campuses, but you're going to move to this hub and spoke model where smaller campuses and in kind of less dense jurisdictions with less built out transmission necessarily into population centers are going to be fantastic places to continue to do Bitcoin mining because those are going to be competitive power prices. And then you're going to have this other part. So I again I just think it's going to be a more is more story.

100%. And so when you live in the physical world you just run into a different set of constraints than a software company would run into. So like you find a pocket of power that's really well developed but it's going to take 2 to 3 years for fiber to get run out to it. We've run triple-digit megawatts against a Starlink mining very successfully. And so great we because you need no bandwidth. You need so much less bandwidth than AI. You know when you think about it you need multiple fiber runs, redundancies, all the spine and fabric that's required. Just talk about the inter data center networking and kind of all of the what's going on there all the competition there. I mean, this is crazy, right? Versus this again, which is just you're out in the middle of North Dakota or Wyoming, you know, and so what you find is that this power is a great utilization for revenue on a short to medium-term basis while infrastructure catches up and you know, it's going to take time to get that fiber laid out there. And so we as a miner who has significant AI ambition but also mining prowess and appetite for growth or portfolio reconstruction, we are able to be the player that takes advantage of that time arbitrage on the physical infrastructure. So great, we're thrilled to run mining for x number of years while the catchup is happening. We'll run it against a Starlink. We put in modular deployments. We're able to put down and pick up and not cannibalize the electrical infrastructure because if it's great for mining, we have every opportunity to build AI behind it. And so we're able to take advantage of some of those buildout waves that lets us lead with mining in the front and then slot into AI from the back.

I would never tell any of our technology or operations people that that's how we would characterize it, but you get what I'm saying. Like it's not that far off relative to AI. That is correct.

Yeah. So, one of the things that we believe pretty strongly is that retrofitting a mining building for AI workloads is not the way to efficiently develop customer centric solutions. So, I would take us to Sandersville as a good case study for us on how we're thinking about this. Two plots of land, 50 acres where we've got 250 megawatts of Bitcoin mining operational. We've got 122 acres right around next to it. That's going to be a Greenfield AI development plot. And so when the time comes for that AI workload to be ready for service, we're going to cut the power over to that new plot and then be able to shift the power, the monetization, the utilization of that power contract over to what looks operationally like a green field development even though it sits directly next to what would be mining infrastructure that's been fully deployed for years.

So I think it's important to take a step back and understand what is the AI playbook, right? There's four steps that's required to be successful building, operating, monetizing power through an AI lens. The first step is the acquisition of power and land. It's what CleanSpark is phenomenal at. It's what the sector has been phenomenal at historically. The second is the leasing activity. So how do you come to a commercial agreement with somebody who wants to take advantage of that rack space and bring their own GPUs but sign a lease for the infrastructure to be able to power that compute workload. The third is how do you pay for it? The capital intensity of AI data centers is dramatically higher than Bitcoin mining infrastructure buildout. And so we're seeing a lot of activity in the debt markets. We're seeing a lot of activity in the capital markets more broadly because the how do you pay for it marries very directly into the lease because when you think about the collateralization of these projects there's actually a few components to it. The first is the land and the power that gets contributed into the collateral package. The second is all of the capex spend that goes into the development of these much higher dollar value data center environments. And then the last is the revenue associated with it which is the firm lease agreement with the tenant. And so you take those three pieces of collateral, those are super underwritable because at the end of the day, what is the bond market betting? The bond market is taking the position that power and land is going to continue to be valuable. Well-built data centers are going to continue to be valuable and investment grade tenants are going to pay their bills on time.

I mean it's not like doing hard things at 10x the scale over 3 to 5x the timeline is necessarily that much easier is what I would say. I think that what you know I don't think you trip fall land in Bitcoin mining energy and power asset generation, kind of all of these things. I think you have to have a little bit of a mindset of wanting to be at the tip of the spear from the tech, capital, digital asset technology and energy infrastructure in the US. I mean, I'm going to kick that back to you, I guess, but these are hard things at a larger scale. And so yeah, I think what it does is that it's a potential opportunity to engage with a different segment in the market of more of this kind of investment grade, but to me it's a maturation of who you're dealing with. And it's always where we were going to go as Bitcoin miners because Bitcoin is going to be somewhere at the base layer of the settlement layer for monetary technology in the United States and globally. And that means that if it's going to be backed by energy, then you're going to go expand into the energy market. So, I just think it's a maturation.

I mean, I'm eight years into mining professionally. You know, my fun fact is that I've mined through three halving epochs at this point. Which is brutally, brutally hard. The competitive environment that Bitcoin mining pushes up against corporate strategy is very very intense. It requires a lot of discipline and long-range thinking. It also requires getting a lot of things right because each thing you can get wrong is disproportionately painful and punitive. So I think when you look at our story coming from grid and then integrating into CleanSpark when you look at CleanSpark's heritage of how we've built our power portfolio and our mining business it's required a tremendous amount of non-consensus viewpoints and countercyclical investment and so that meant when Bitcoin went from 20 to 60 saying this is too rich and exiting some of the Bitcoin position in favor for more hash rate and more infrastructure. It also meant taking a view on having a more distributed portfolio of smaller sites in order to grow faster during that period of time. Then it also required the intellectual flexibility to say okay this smaller site profile is not as attractive to the AI client base as the larger concentrated site profile. So we reacted appropriately. We added a 285 megawatt site to the portfolio. We added a site with capacity up to 600 megawatt 50 miles away from it. And so, what you've seen from us and why we've been successful and had real durability is because we've had the appetite to invest countercyclically, we've also had a tremendous amount of mental flexibility, which is that when you get new information, it's very appropriate to make new decisions. And so having that as a cultural touchstone for how we develop and enact strategy as a corporation is something that I think our shareholders and the industry at large has come to expect from us and really value about us. But there's, not because we're here to shill the company, but just because there's a tremendous amount of satisfaction that I get from working in an intellectually stimulating environment where the stakes are high, the dollars are big, and the impact that we get to have on the market broadly is very very large. Because if we are able to play a critical role in enabling some of these AI players to be successful, the American economy is going to thrive and we as professionals and our shareholders that come along with us, our employees that come along with us, they're going to thrive along with us and that is very very motivating and very interesting and it lets all of the time and the hard work that's required to be successful, we have stamina to do it because of how exciting and valuable that work is. So I just want to address it head-on a little bit as well which is that you have not heard the words pivot from us. You've heard the words expansion and that is very very distinct I think in our business strategy, capital strategy, operating strategy and approach to this market.

So, we were sitting in the basement of this resort hotel and casino about two months ago at our executive leadership retreat. And we're kind of, we've made no bones about kind of this expansion to AIHBC. And while my background is in Bitcoin, my passion is that my role is Bitcoin Treasury, I felt like it was my role to really in that room make sure to really challenge our priors and make sure that what we're doing that just because we've done something in the past doesn't mean we should carry that into the future again with that intellectual flexibility if it's not going to serve the business, the shareholders and ultimately and we believe this the stakeholders in the American economy. And so, we challenged, okay, great. We've been Bitcoin miners. That's been our access to land and power. That's funded our business. That has driven shareholder value. That's created the opportunity set to continue to grow and expand into this kind of adjacent energy market that has this entirely new kind of profile. So, why do we want to stay dedicated to Bitcoin? But one, it's that we've seen a lot of others that have wound down perhaps maybe a little bit prematurely their Bitcoin mining operations and are going to be in this very long middle where there's two or three years between when revenue starts to kind of come in from the AHPC side. And so I can understand kind of bringing in capital ahead of time and maybe your operations are not as profitable as you kind of paint them and so you don't want that drag and maybe it's management cycles, but fine. But they've wound that down. So we have a profitable scaled mining operation that can fund us during that expansion period, right? But the second piece is why hold Bitcoin on the balance sheet. Okay, great. So you hold USD on the balance sheet or you hold kind of classic kind of you look at the revenue lines. What's your current assets? What are your total assets on? You pull up a 10K or a 10Q. Great. You can borrow against that. You can go tap the markets, etc. But particularly for people still in this kind of nascent space, the scaled access to those capital markets is not as mature as it's going to be once we kind of start to face off against some of these IG counterparties. And so what you have with Bitcoin is you have a liquid global permissionless asset that trades 24/7 that has upside optionality that you can borrow dollars against. So you take an asset that we agree maybe over one week or even a month or a year might have some volatility in it, but we believe is going to go up. This kind of technology adoption curve we think number goes up over time. So we think that that has upside optionality. It's an appreciating currency. You use that appreciating currency to borrow in a depreciating currency. You use that depreciating currency to go get assets that are going to appreciate. Those assets that are going to appreciate are going to bring revenues back onto that balance sheet and kind of give you that flywheel. So, one, it's upside optionality. It's the ability to use an appreciating asset to borrow in a depreciating asset. And then it's kind of what we touched on originally, which is that the Bitcoin doesn't just come in and get immediately monetized into USD. What we're really pioneering here is how do we take that and drive incremental margin out of that Bitcoin, whether we're turning it into dollars or whether we're kind of turning it into additional Bitcoin on that balance sheet again to drive that flywheel. And I'm happy to get into kind of the specifics of that. But this is a game of inches, right? And so no, I'm not scared. I'm actually happy about the expansion because we always had this view that energy is the critical underlying asset of the critical settlement asset of the greatest economy in the world. And I think we're seeing that really be realized. But what I think it is is that the difficulty adjustment, if you're at the top of the difficulty adjustment, if you have the most efficient machines and globally competitive power prices, the rise in difficulty doesn't scare you because it's going to knock off kind of less efficient operators. And so to me, it's the same thing with what we're expanding into on the digital asset management side is that if we can drive 5 or 10% more revenue per Bitcoin, per kilowatt hour out of that Bitcoin that we turn into, that's a competitive advantage that compounds in almost unfathomable ways over three or five years. And I would be curious kind of your if you think that's fair or you have anything to kind of expand or.

Yeah, I mean I think that the way that I would digest a lot of what you're saying because I agree with all of it is that we have a philosophy of organizational design around why we've built the business in a way that's able to realize the vast majority of that value but also the value across the rest of the things that we do which is that we didn't wake up as a big Bitcoin mining company. We have a strong and deep heritage across energy markets. We were building micro grids long before we ever realized what an ASIC was at CleanSpark. And so, we spent a lot of time developing energy native expertise. That meant building micro grids at military bases. It meant building the software that ran demand response protocols. It meant getting called out to Atlanta, Georgia to a data center on a sales call. That was where we found our first Bitcoin mine was a sales call to sell them a micro grid to help them drive down energy costs. We saw that opportunity and rather than sell them our product, we bought their data center and that started this cascade of scaling our Bitcoin mining, scaling our direct energy and land ownership profile. That was 2019, 2020. We bought that asset at the end of 2020 and that kicked off what was really a 5-year hyper-growth cycle to become the largest domestic producer of hash rate. And so over the course of that scaling journey, we became great at power and land acquisition, not just energy development and management. And over that period of time, as we scaled, we also realized that the Electron is going to have more utility across more types of compute than we'd anticipated 5, 6, 7, 10 years ago. That brought us directly to the crux of this opportunity to expand into a secondary form of compute that is likely to grow to be the majority of our business over time from a revenue perspective, but not from a strategic footprint. So, talk about the organizational design. We think that we're going to be great at developing digital infrastructure for AI use cases. We think we're going to be great at integrating Bitcoin mining into energy systems, some of which are going to be in the data center, some of which are going to be at the utility level. We also think that interacting with energy more directly is going to be something that we're going to be good at into the future as well. Maybe that means behind the meter generation. Maybe that means more intelligent power sourcing on the power markets. We don't know what that's going to look like yet, but we know that we have a thesis about how energy management generation and consumption is an opportunity for us to drive shareholder value in the business. And then the last category which is what Rory's covered a lot of is CleanSpark Capital is a concept that we've been kicking around for many many many months.

So very softly. And so the first step towards that is by taking a fundamentally differentiated view on treasury management. We are not a DAT but we hold a lot of Bitcoin and we do so in a variety of productive ways. What's really important about that productivity is it's not just about what are we able to extract from our holdings or our balance sheet. It's how does our capital activities integrate into a scaled operating business to make the operating business stronger and open the door to more types of opportunities across our balance sheet utilization and management profile. And so, I think of companies like GE, where GE Capital was a huge component of what they did that had nothing to do with how they built turbines or washers and dryers or all these other types of hard asset activities. It had to do with the financialization layer of everything that they did as well. I also look back to things like Berkshire Hathaway where they took a position where they bought out GEICO and utilized the GEICO insurance premium model to take a front of the locomotive capital engine and drive the incremental float from those insurance products to be able to unlock a bunch of different operating businesses elsewhere across their portfolio. And so the marriage of operating business in the physical world and the financial business in the markets world marry together to create differentiated and hugely outsized outcomes.

I mean I don't think you need that much IQ but the explanation was I mean I just and to me this is the biggest it's the biggest misnomer. We're one of the top 10 corporate holders of Bitcoin in the world. We're the largest domestic hash rate producer and it's been really hard to talk about that in certain ways because we are a company with 13,500 Bitcoin on our balance sheet but we are not a Bitcoin treasury company right and we generate yield on our Bitcoin but we don't generate whatever the BTC yield metric is and sats per share is that's not how we measure ourselves. We measure ourselves on shareholder return. We measure ourselves on stakeholder return. We measure ourselves on kind of total capital assets that we drive over time. And so, I think it's exactly right, but yes, not to jump in too much again.

So this is a critical piece. We believe that our obligation is to take the productivity of our business lines and drive them into the highest growth, highest value return opportunity in front of us. So let's say that we were able to take a lot of dollars in from our first AI deployment. I think that right now the biggest opportunity in front of us would be to plow those dollars into the next AI deployment. I think over time as businesses move from their hyper-growth phase into more of a cash flow and compounding phase, the opportunity to think more about the opportunity cost of each incremental net income dollar is going to get more competitive. But right now it's about growing to become a significant player in the AI compute world in the same way that we had to grow very aggressively to become a significant player in the Bitcoin mining landscape.

Yeah. If you don't know where the yield comes from, you are the yield. No. And let's be very clear, and this has been a point of frustration for me because, again, I'm not here to wear my arm out patting ourselves on the back necessarily, but we take it very seriously to be very transparent about our strategies and what we do. Be very specific about what is possible, what isn't possible, what are the risks and possibilities there, and to take diligence and analytical rigor to everything that we do. So where does the yield come from? Well, first and foremost, you can't generate sustainable yield if it's just return on capital.

You can't generate sustainable return on capital unless you have some type of operating business. And maybe that operating business is an investment business, but that would mean you have to have a process, you have to have controls, you have to have a view, you have to have risk management. So the return on capital that we generate starts with thinking about what is the operating profile of our core business, and our core business has been Bitcoin mining traditionally. Obviously, we're going to be expanding into—we'll talk about how we're kind of driving capital from the digital asset management side into the AI HPC and energy expansion side. But the first thing that we did in December of 2024 was start with a 60 to 90-day process to really decompose our entire business across what are the drivers, what are the risks, where are the expenses, where's the cost of capital, where's the cost of power, what is the timing of those cash flows? Because this is a really underappreciated point is that you might be doing something that makes a lot of sense in the aggregate but it doesn't make a lot of sense when you get upside down versus timing and asset liability mismatches. So that took us maybe 30 to 60 days to kind of get settled and really think through that. It took us another 30 to 60 days to really design the philosophy around that. And what we came to was essentially a two-pronged approach. So the first approach is that we're generating around 600 Bitcoin a month right now. That can go up and down depending on difficulty, operational hash rate, what we're doing. But let's call it 500 to 600 Bitcoin a month. So we know what our opex component of that is going to be. And the other thing that you can do is you can start to think about what are going to be your capex needs. So you can look out to September or December of this year or into next year and go, 'Okay, well, I need 35% of that right now for every dollar of adjusted Bitcoin that comes in that's worth a USD. Maybe 35% of that could go to opex on an ongoing basis, and maybe somewhere between another 35 to the other 65% of it might go into these capex opportunities right now quite frankly because there's so much opportunity to build out in this space.' So that's on a monthly basis. And the second part is that we have the Bitcoin, and look, I really do give Saylor a lot of credit in this, like volatility is vitality in the sense that you are taking an asset that—the other thing is that yield on any asset comes from moving it around, right? Moving it into hands where it's going to be—it might be inert on your balance sheet, it might be at rest on your balance sheet, but at movement on somebody else's, maybe they will pay you for the opportunity to do that. That's what the interest rates are. They go, 'Okay, I've accrued some capital, Vana has a great opportunity to build out a new AI HPC data center. He thinks he's going to get a 9% return on equity. He'll pay me 5% for the opportunity of using my capital in the meantime.' That's a good risk-adjusted return for me. I underwrite him and he gets his return. And that's what yield is, right? It's not this thing in the ethereal kind of out there that just comes from diluting down your business with no cash flows coming in and pushing that back onto a capital stack that doesn't have a self-reinforcing engine. Right? So going back to how we split it, we have the spot plus program which is what we can do is we can take because we have 500 to 600 Bitcoin coming in a month because that is our core operating business because it maps onto our exact needs and we need to monetize some portion of that. Look, we want to hodl as much Bitcoin as possible but at the end of the day we still have to pay salaries and insurance and power bills, all of these kind. So some version of that is going to get converted into dollars. So, I could do that in the spot market. We could go to the spot market and we can either do a market order or a limit order, right? And I'm not trying to—I'll be simplistic in a certain way, but I don't want to speak down to anybody who kind of understands these concepts at a core, but essentially you can just sell it market, which just says, 'Hey, I'm going to go take whatever the price is right now. I got 100 Bitcoin to sell. It's going to go down in the centralized limit order book probably down about two levels and I'm going to get an average price of—I'm going to get done within a couple cents or a couple caps, right?' You could do a limit order. You say, 'Hey, I'm going to sit here. I think that the natural volatility, what we've done is we've looked at kind of Bitcoin vol on a daily basis. It moves up or down. It might go up 1.5% and down 1.5% a day. That means you're going to have kind of a 3% range. I'm just going to set kind of consistent limit orders up 1.25%, 1.25% because eventually on a daily basis it's going to pop up and I'm going to get taken out.' That's one way to kind of try to outperform spot, right? And it's a very simple way to do it. Obviously the risks are if you're in a downtrending market if it doesn't do that if volatility compresses. So let's be clear-eyed about what that is, but that's one way to kind of think about how to manage some of the risk on the book. The other way is to go to the derivatives market and go, 'Okay, so it's kind of like when I go to Delta and I click on flexible dates, right?' We don't really—we know that we have to meet salaries and we talk about understanding the timing of cash flows, salaries are on the 7th and the 22nd, right? So somewhere between that, so I have 14 days in between that to come up with the dollars to do that. So I go, I don't really care if I sell Tuesday, Wednesday, Thursday, or next Monday, but what I can do is I can get paid for that optionality by the market. So there is volatility within Bitcoin. Bitcoin does not have an organic interest rate. But what you can find is that you can find that people will either want to take it on their balance sheet for trade financing or to do collateralization or to borrow or just for any of these other uses. So there's a natural volatility in that market. So we'll sell basically short-dated covered calls. Again, simple, not easy. And that's our whole philosophy. This should be extremely simple to understand. It should not be that complicated. The risk should be completely decomposed and understood. But it's hard to do it because of the institutional risk management and the consistency and the way that it maps onto operating business. So, we just sell short-dated covered calls, right? And so, what's interesting about short-dated covered calls is yeah, even as volatility has compressed, the kind of total all-in cash on cash or sats on sats return you're getting is incredibly attractive, right? And so what we've found is that by managing this in the way—and I can't give away the whole kit and caboodle on how we do it specifically—but we have a strategy where we go out x amount of days kind of knowing that we're going to have to meet these obligations, as we get closer to that we will increase what's known as our delta exposure and that delta exposure means basically you can think of delta as if I'm selling a 20 delta call that means one out of five times about 20% of the time on average you can expect to get taken out. Now there's going to be a lot of derivative heads listening to this who are going to talk about is can you really take a standard kind of curve on that and what are the real stochastic returns and should we use a jump diffusion index instead of a Black-Scholes-Merton model, kind of all that, but like long story short you can essentially think that if you're selling a 20 delta option about 20% of the time you expect to get called on that. What that means is if I'm selling a covered call that says that I have committed to selling you Bitcoin at $100,000 in seven days. If Bitcoin goes to $107,000, you pay me $100,000. But and you're paying me $1,000 in the meantime for the right but not the obligation to buy that from me. Bitcoin goes to $107,000. Great. You're in the money. You buy the Bitcoin from me at $100,000. You go sell back at $107,000. You're net up $6,000 on that $1,000 you paid me. All good for me.