The Bull Case For GLXY With Duncan & Rittenhouse Research

I think it could be worth like $100 a share in the next few years. So that's personally how I'm thinking about it. It's definitely a long-term investment. I think as they execute on the data center business, if they continue to execute there, then like I'm really happy to hold this for a very long time. I'm bullish on crypto, I'm bullish on AI, I'm bullish on the management team. So that's kind of how I think about it long-term.

This episode is brought to you by SKALE, the next evolution in Layer 1 blockchains. Explore the gas-free ecosystem at scale.space/skale. /eco and follow their journey on X @ScaleNetwork. Thanks to SCaLE for making today's episode possible. Nothing said on Empire is a recommendation to buy or sell any investments or products. This podcast is for informational purposes only, and the views expressed by anyone on the show are solely their opinions, not financial advice or necessarily the views of Blockworks. Our hosts, guests, and the Blockworks team may hold positions in the companies, funds, or projects discussed.

All right, everyone, welcome back to Empire. Very excited about this one. We were supposed to have Novogratz on the show. We had to push that until after their earnings. Their earnings— Galaxy earnings are coming up August 5th, I think it is. These guys can correct me if I'm wrong, but still wanted to— still wanted to do something on Galaxy. Full disclosure, I own 3 stocks. They are Coinbase, Robinhood, and Galaxy stock. So I do just want to give a disclosure there that I'm long Galaxy.

Wanted to—

we're still going to have Novogratz on the podcast, but wanted to have 2 guys, Matt from Rittenhouse Research and Duncan from Flood Capital, on. They are probably the 2 leading voices in the more crypto-native world talking about Galaxy as an underpriced stock. So Duncan, I forget if you've been on the show before, but welcome. Welcome or welcome back. And then, yeah, Matt, good to have you here too.

Yeah, it's great to be here.

Yeah.

Yeah. No, and I guess Matt and I both own the stock as well. Just making that clear.

I figured you guys did. I figured you guys did.

Very long.

Yeah, yeah.

So, um, the, the other thing, just quick housekeeping, is we have Digital Asset Summit coming up. It's October 13th through 15th. Uh, code EMPIRE100 gets you a ticket, uh, discounted ticket, get $100 off your ticket. And the other thing is, if you're listening on Spotify or Apple, you probably want to watch this one on YouTube. We're going to do a lot of screen sharing, showing some different models, stuff like that. If you're not watching on YouTube, that's fine. Okay, let's jump in with Galaxy is a very complicated business, actually. I spent a lot of time prepping for this podcast and just reading about Galaxy as a business over the last couple of weeks. And it's a very misunderstood business, I would say. Like, I think if you look at something like Coinbase's business or Robinhood's business or, I don't know, Facebook's business, like these are very understood businesses. Galaxy is very misunderstood because they have a crypto business that is like, I think people know, especially people who are listening to the show and they think about Galaxy as a business, you've got the OTC desk and the asset management business and the funds, but it's kind of like, what does Galaxy really do over there? So I think that'd probably be the best place to start. Maybe Duncan, I will pick on you. Can you just give a— like, how do you think about the components of Galaxy? Like, walk me through their business.

So I basically break down Galaxy into 3 main buckets. They have their balance sheet. So the market cap today, it's trading around just under $30, $28. So that's around like a $10 billion market cap. And they have a crypto balance sheet which consists of— so as of March 31st, $1.3 billion in kind of like Bitcoin, ETH, SOL. I think they bought some HYPE recently. I know they have some SUI. Crypto is up like 40% since then. So let's call that like $1.8 billion. In liquid crypto, then they have a billion in stablecoins, and then they earned— or they just did a raise for $500 million in cash. So you're looking at, you know, roughly a $3.3-ish billion balance sheet depending on, you know, how their crypto performance did. I know that they like, you know, took a big Hype position, which has outperformed. They've done a lot of these treasury companies recently which have outperformed. So, you know, let's say like $3.3 to $3.5 billion balance sheet. And then that leaves you kind of with like $6.5 to $7.5 billion of like equity value for all their crypto business lines, which we can run through. They have a bunch of like joint ventures within crypto, and then also they have this like massive data center business that's just starting to get off the ground right now.

Okay, so we've got a $10 billion business, we've got the balance sheet, we've got the crypto business lines and the data center business. Matt, is that roughly how you think about the business?

Yeah, I think that's, that's about right. And I think on top of that $3.3 billion that Duncan mentioned in like liquid assets, which is like cash, stablecoins, um, Bitcoin and ETH, they also have a couple hundred million dollars of like illiquid venture investments, um, that are also incremental to that $3.3 billion.

Yeah, like there could be some sleepers in that. They don't really like break out the balance sheet very well, but you know, they've been investing invested in companies in crypto for a long time. So an example of one that they've kind of monetized in the past was they were an early investor in Fireblocks. So they've sold some of their stake. I don't know if it's their whole stake. They're also an investor— well, is it BitGo or Bullish who's going to IPO soon? But like—

Bullish. Bullish.

Yeah.

So they're an investor in that. They have kind of these sleepers in that $200 million that actually could be pretty material in the future.

Let's talk about what I personally find to be the least interesting part of those 3 components. So you've got the 3 components: the balance sheet, the crypto business, and the data center business. I would say the, the most exciting by far is the data business. What I find the least exciting is the balance sheet. The balance sheet. But I think it's actually important to start with that because once you— so we've got this $10 billion market cap, $3 billion balance sheet. It kind of contextualizes it. It helps you understand if the crypto plus data business is overvalued or undervalued because you're valuing the other things at $7 billion, obviously. So walk me through, let's get into the balance sheet a little bit more.

Yeah, so I would basically think of the balance sheet as enabling them to have this liquid crypto exposure, make these venture investments and infrastructure investments off the balance sheet as well. But it also provides a really strong capital base to expand their crypto business lines. And also the data center business line. So for example, with the data center businesses, a lot of these Bitcoin miners are trying to convert into like AI data centers, but they, you know, historically like very poorly run companies don't have much cash on the balance sheet. And so it's very difficult for them to like build out a data center. It's very capital intensive, right? So Galaxy's in a completely different position where they have like, you know, a very strong balance sheet, and they actually have the best, uh, deal economics out of like the few other deals that are out there with like Wolf, um, Core Scientific, and ALPD, because Galaxy's funding the entire buildout themselves, where like those other companies have to, you know, look at different sources of financing. So that actually gives them like a big edge, um, in the data center business side, and also just like they can enable a lot of their assets for their different crypto business lines, right? And, uh, ETFs, like providing liquidity and such, because they're also like a big ETF provider.

Yeah, and just to piggyback on the point about having a stronger balance sheet versus some of the other Bitcoin miners, just to like contextualize, Galaxy's like $3.3 billion of like liquid assets available is like well north of what anyone else who's who's a Bitcoin miner has when they're trying to do these data center conversions. So most of these other businesses, like if you're looking at Core Scientific and, and TerraWolf, and these numbers are as at the end of March, they all have under a billion dollars of cash. And then Duncan, you sort of touched on this, but the Bitcoin mining business has pretty much been a horrible business to be in historically. For a variety of reasons, and we can jump into those. But these, these miners have basically been burning tons of cash and diluting their shareholders to seemingly no end, whereas Galaxy has historically ran a pretty profitable crypto operating business. So their trading business, their investment banking business, their asset management, the staking, that's like the core of the digital assets operations. I think that's generated over $2 billion of cumulative profits since, since Galaxy was founded. Whereas all these other Bitcoin miners have been losing money hand over fist.

Yeah. And one point I'd like to highlight here is if you look on the right in 3-year change in common shares outstanding, so TerraWolf, they've tripled their share count. That's the exact same story. Irene looks very similar to that. Mara, Riot, they've all had insane dilution. So if you're looking for a long-term investment, these Bitcoin miners are very hard to hold cycle to cycle because you know, when Bitcoin goes down, they're unprofitable. In order to stave off bankruptcy, they need to issue more shares. And, you know, Galaxy's there. This is before they did their raise, but, you know, you're talking over the last 3 years, like 10-15%. And Matt might have the details on this better than me, but they also, like, were buying back stock when Galaxy was like $2-3 in the last bear market, right?

Yeah, that, that's right. And I think, like, it probably makes sense just to sort of level set, um, a bit on, like, comparing Galaxy to some of these Bitcoin miners, because that's honestly, one of the— what sort of got me interested in Galaxy to begin with is Core Scientific was the first of these Bitcoin miners to, to strike up a deal with an AI cloud compute company. So they, um, basically signed a lease to lease their data center capacity to, to, to CoreWeave, um, back in June of, of 2024, so a little over a year ago. And then they continued to, um, increase the amount of, of that lease. So Over the past year or so, Core Scientific has basically slowly converted all of their Bitcoin mining operations into an AI HPC data center business. So they were sort of the pioneers of these Bitcoin miners in converting and pivoting to AI. And their stock price has done really well. It was a couple of bucks at the beginning of 2024 when they came out of bankruptcy, and I think it's around $13 today. So when I started looking at some of these other Bitcoin mining businesses, um, to, to sort of think about it as a catch-up trade in, in that they should do well if they're able to successfully pivot their business to AI, just like Core Scientific did, um, what was really attractive about Galaxy is that at its core, they weren't really a Bitcoin miner. They had this profitable digital assets business, um, and that's really a massive difference when you're, when you're looking at Galaxy versus these Bitcoin miners.

Yeah, maybe if I— the way that I understand it, um, Matt, you had this line in your, on your research report, which we can link to on, uh, on page 6. Um, I can, I can pull it up right here, but you said, uh, let's see, let me pull this up so I can make, make sure I get the language right. You said, while other Bitcoin miners have talked at length about their AI data center ambitions, they do not have the balance sheets nor credibility to execute deals with hyperscaler tenants. So the way that I understand this is basically Basically, like Duncan was saying, the Bitcoin mining business is really, really tough. I think a lot of these Bitcoin miners see this massive opportunity, which we'll get into, and I want to spend the majority of this podcast actually talking about that, is there's this huge opportunity actually with these hyperscalers, because they need a bunch of power. Hey, maybe the Bitcoin miners, they're like, all right, we have the power. But unfortunately, they don't have the balance sheets that someone like a Galaxy does to actually pull off this pivot. Roughly the right way to understand it?

That's, that's definitely, I think, in my opinion, a huge factor as to why Galaxy's been successful in their pivot to AI is just the strength of their balance sheet. Um, I guess like if you think about it, right, if you're CoreWeaver, if you're Oracle, if you're Google, um, you're making basically a massive investment into the NVIDIA GPUs, um, that are going to go inside one of these data centers. So you need to really have a ton of confidence in the developer of the data center that they're going to have the financial resources to, to be able to complete these massive construction jobs on time. Um, you obviously don't want to invest a ton in these chips and then have them sitting idle because the developer, um, is behind schedule. And obviously none of these Bitcoin miners have built AI data centers before, so there's just a, a ton of risk associated with a hyperscaler like Oracle or Google going to a Bitcoin miner Um, and I think Galaxy's balance sheet gives them a lot of credibility, um, in those conversations.

Yeah.

Anything else on balance sheet that you guys think is important?

No, I think we covered the, the main points there. And even if you think about like some versus some of these treasury companies, right? Like, you know, these treasury companies are trading at historically like you know, around 2x NAV. So like, you want to put— Galaxy's not going to be issuing stock to buy more crypto, but like, just putting that in context, a $10 billion business minus the $3.5 billion balance sheet, you know, you got, uh, $6.5 billion, uh, for their entire crypto business line and data center. Like, it's, it's very cheap in my opinion versus these treasury companies.

Yeah, let's talk about, let's talk about how we get to that math. So the crypto, okay, the crypto business, this is where it really starts to get interesting. How do you, so I'm on their website right now and I'm pulling up, they've got data centers and digital assets. Under digital assets, you've got trading, lending, derivatives, investment banking, asset management, infrastructure solutions, technology products. And for anyone who's been, yeah, in, in or around the industry, Galaxy has always kind of sat at the center of some of a lot of the things in the industry, right? They bought GK8, so they have, um, uh, custody. They're, they're, they're always one of the biggest lending, lending players. They've got, um, the OTC desk is really impressive. They've got their funds, they do, you know, trading and staking and all that kind of stuff. How do you guys think about— what's the, um, analogy for their crypto business? Like, is there— how do you think about just the— what they really do?

I'd say for sure, and maybe we can just throw up my screen. Because I have some data points on the crypto business. But basically, for some context, so this has been Galaxy's main focus since the beginning, and then they just kind of came upon the data center business. But we can talk on that later. So they have, I think they have around 800 employees, 650 of which are in the crypto business. So by far it's historically been their main focus. And yeah, just running through some very high-level stats. Of an overview of what they do. It's like extremely extensive. I call them like the Goldman Sachs of crypto or like the Berkshire Hathaway of crypto because they just touch literally every element of crypto. Like Galaxy is involved at some point.

I think that's the right framework, by the way. I think the Galaxy of crypto— I don't know if Berkshire Hathaway is the right, uh, but I, I think, I think Galaxy is the right mental model, especially because they don't really have the retail— like Coinbase has more of the retail business, right? So that could be like a JP Morgan with their Chase business or like a Bank of America. But, uh, yeah, Galaxy, I do think because they're only an institutional business, I do think that is the right framework.

This episode is brought to you by SKALE, the next evolution in Layer 1 blockchains. With a gas-free, invisible user experience, instant finality, and robust security, SKALE allows for limitless scalability and has already saved its 50 million users over $11 billion in gas fees. SKALE's high throughput, zero transaction costs, and powerful Web3 infrastructure make it the ideal platform for gaming, AI, and other high-throughput apps. To explore the gas-free ecosystem, visit scale.space/ecosystem and follow their journey on X @skale_network. That's Skale spelled with a K. Thanks to Skale for making today's episode possible. Shout out to today's sponsor, Katana. Katana is a DeFi-first chain built for deep liquidity, and high yield. No empty emissions, just real yield and sequencer fees routed back to DeFi users. Incubated by Polygon Labs and GSR, Katana is launching with notable ecosystem of primitives, assets, and partners like Morpho, Chainlink, and more. Katana's 1 billion KAT campaign is live. Bridge and deposit directly into vaults in one simple click and start earning immediately. On your Ether, Bitcoin, USDC, and more. Follow the link in today's show notes to get started. As always, investments in blockchain technology involve risk. Terms and conditions apply.

So walk me through, walk me through the different product lines.

Yeah, so, you know, they have basically like a massive trading business, which they say, like, so derivatives, lending, trading, structured products, um, This is of the, the last quarter, so I think probably March 31st. It's like nearly a $900 million loan book. So that's like the second or third largest CeFi lender in all of crypto. And if you think of something comparable, like, you know, I think Maple has, you know, maybe a bit bigger but a similar type size loan book, and that's like a $500 million market cap in crypto just on that like one business line. Um, you know, nearly 1,400 trading counterparties with, you know, over $10 billion in trading volume per quarter. An interesting thing that we saw just a few weeks ago is that like massive Bitcoin who had like 80,000 BTC, chose Galaxy to execute the trade. So that was like a $9 billion trade just on that like one customer. And Galaxy seems to be like the go-to place for these like big institutional players to like, you know, execute trades of that magnitude. Um, they have an investment banking business. So, you know, they're, for instance, earlier this year, like they advised Bitstamp on their sale to Robinhood. Uh, they were on the CoreWeave IPO. They've been involved in treasury deals like S-Bat and BMNR, like Tom Lee's ETH play. In terms of that, like, when you think about the product offering they can bring to these companies, it's like, hey, okay, we're going to invest in you, you know, we could help you raise money, we could execute your trades, we could do your custody, we could do your staking, we could help you utilize your assets in DeFi. Like, they have just such a broad suite of product offering that it makes a lot of sense for like these companies to partner with them. Matt had like a great little post off his Twitter account on the SBET deal where Galaxy is managing like half of, um, the ETH under SBET, and they're going to be charging between like a 50 basis point fee to like a 1.2, uh, 125 basis points, plus, um, you know, take like staking the ETH. So Matt, I forget how much, like what, what your kind of math was on that, but that's just like a good context on, you know, their business there.

Yeah, and I think, um, if you just go back up real quick to the, the trading part, um, the franchise trading. So just to— it's, it's really hard to value this type of business, um, just because there's, I guess, there's a lot of volatility in these volumes, um, and you would expect that as more institutions are onboarded into crypto, they would pick up over time, but I think just like as a framework to think about valuing this, there have been a couple of prime brokerages. One is Hidden Road. They got acquired by Ripple Labs for $1.25 billion. That was earlier this year. So that's sort of one data point. And then FalconX, which is another prime broker focused specifically in crypto, they raised around at an $8 billion valuation in the summer of 2022. So obviously there's a ton of noise going on in the market. It's tough to know whether that valuation was impacted by what was going on at that time, but those are just like two data points that I sort of think about in terms of what the franchise trading business might be worth.

Yeah, for sure. And then going into the asset management, so this is as of March 31st, I'm pretty sure. So they have like over $7 billion of assets on the platform. So that's between like their ETFs and their, their various funds like venture funds and such. You know, they've won really big contracts being kind of like the Goldman in the room, the adult in the room, like the whole FTX estate deal, right? Like they were in charge of custodying those assets. They have over $3 billion in staked assets, which I think this is going to, you know, tick up a lot in the next few quarters with all these treasury plays. You know, they're the 4th largest validator on Solana. You know, they're an ETH validator. They're rumored to becoming a HYPE validator. Um, and then they just announced like a big venture fund of $175 million. So they are just like, you know, every aspect of crypto like comes up again, all the asset management side. And then this last point, which I think is pretty interesting and could be like a big growth avenue from in the future, even though they probably aren't generating a ton of revenues here now, is this like GK8 platform, which is like an institutional-grade custody and tokenization platform. You know, they want to tokenize equities. They want to, um, partner with— so Galaxy's also like a 33% owner in AllUnity Stable, which is basically like the first fully regulated euro stablecoin. Um, so like GK, it's doing the tokenization for the euro stablecoin. And eventually I think their goals are like, once they get like European stocks on chain, like GK will be the custody provider for that and like help settle between euros and and those equities. So like, these are like massive kind of like sleeping parts of the business that could be huge in the future, right? You know, Circle at like a $50 billion valuation, you know, could AllUnity be the Circle of Europe at like a 10%, uh, you know, market cap of that? That, that could be huge. Um, so that's kind of what I would kind of summarize the crypto business as. I think Matt's points are excellent in terms of They have historically, you know, from the Goldman report here, let me find— yeah, so from, you know, 2022 to 2024, their expenses on the digital asset business have been like rather fixed, and then their revenues have fluctuated a lot, but they're kind of set up to really scale this business. They've invested like a lot of capital into building out the infrastructure, building out the teams, hiring new people. And now that crypto is getting more positive regulation, you know, the Genius Act got signed into law, the Clarity Act passed the House, you know, the SEC just put out a really bullish thing on market structure and tokenization that hopefully they can keep their costs relatively fixed and really scale their revenue here. And we should see them become— this business line become way more profitable. And I think Galaxy, like how I think of the crypto business is Instead of picking like one specific altcoin or one specific crypto stock, GALAXY definitely feels like a holistic way to play just like the success of everything in crypto, right? Like more institutional adoption, stablecoins, the Euro stablecoin, tokenization, increase in lending. And that's kind of how I think about it. And then if you think from like a Goldman Sachs point of view or like a, uh, like a BlackRock's point of view, like what would it be worth to them to acquire Galaxy for all of these business lines, right? Instead of having to build them out in-house. So that could be like, they could be willing to pay billions of dollars to just kind of absorb all these business lines from Galaxy and not have to build it from scratch in-house, especially if this market's getting more competitive.

Yeah, I think I think that's really a huge point there. There's like a lot of strategic value in the digital assets business of Galaxy just because they're the only like full platform that has this comprehensive set of offerings that's tailored specifically for institutions and companies operating within the crypto ecosystem. Like there's, there's really no other comparable business in that sense. And then the only, the only thing I'd add is also what I think makes this platform valuable and unique is that with that strong balance sheet now, they, they have a lot of optionality in terms of other smaller crypto companies or products that they can acquire and then immediately sell those products into their institutional customer base that they already have. I think, yeah, I think it's over 1,000 trading counterparties. So those are institutions that they're, they're trading with. Um, so any sort of new products that come out that could be sold into those types of businesses, I think it would make sense for Galaxy to acquire and immediately be able to cross-sell those products.

Yeah. And then just on the valuation front, so this is like Goldman put out a report a couple of days ago and this is just like how they kind of value the digital asset business. So they think it's worth around like $6.4 billion.

What are the comps here? Can you talk us through this? Okay. Average adjusted PE is 20. 28.

Yeah, so 30, the multiple they're using, 35x. So I think they comp it to different, uh, investment— I think I have that pulled up here. They comp it to different—

so they, they must be comping it against something like a Coinbase, right? Because the— I think the PE of something like a Goldman or a Morgan Stanley is 15 to 15, 16, 17 roughly. Uh, so they— it sounds like they're comping it more to something like a coin— a crypto business, an eToro, a Coinbase, a Robinhood.

Yeah, so I think they're doing like a blend of like kind of investment banks and then these more growthy crypto businesses. Uh, so like right there, you know, on their— on Goldman's projections, $6.4 billion in market cap, which is, you know, quite a lot. Um, and yeah, I think this— their, you know, their, their balance sheet estimates are low given like the movement that's happened in crypto and stuff.

Okay, wait, so hold on. So for people who aren't watching on YouTube, so they've got the— zoom in a little bit, Duncan. So the digital assets segment, you've got the adjusted net income is $183 million, and then they've taken this PE of 35, and that gets an implied market cap of $6.4 billion for the— just the digital assets business.

Yeah, yeah, exactly. And then I have been—

is there a counterargument here though that, that they shouldn't be comping it against something like Coinbase or a Robinhood, and instead you should really just comp it against something like a Goldman, which has a PE of— yeah, like, I just looked it up— 15.5 instead of the 35 that they put out.

Yeah, I think, I think that is a valid argument, um, just because in terms of like the actual business model of providing financial services to institutions, that's— it's more aligned with a Goldman than it is something like Coinbase because that's obviously a retail-facing business. But, um, it— on the other hand, like, Galaxy's book value, if you look at the book value per share, has compounded from— I think it was a dollar back in 2018 to close to $7 today. So they basically been growing that book value per share at a ridiculous clip and, and well above what Goldman's been growing at. So to, to that end, I think you would, you would argue that it should trade at a premium to, to where Goldman trades, just because of how much faster they're growing. Yeah.

And also just like the crypto business, like I think that those volumes are going to ramp a lot. Like it's more, you know, Goldman's business is more mature, whereas like these crypto businesses are probably going to be a bit more growthy. Yeah. Nice.

What's the most— what's the most interesting part of this business to you guys? So, right, you have the franchise trading, right? Which is you've got this massive loan book, you've got like a billion-dollar loan book almost. You've got like over 1,000 counterparties. So you've got this huge OTC desk, like big trading operation. You have the investment bank, um, you have the asset management business, you have the infrastructure solutions. Like, what do you guys get re— what, what's the most exciting part of all this to you, Matt? Maybe if you want to start.

I mean, I guess this will probably be kind of a cop-out answer, but I think what makes it unique is that all these different offerings are packaged into, to one business. Um, but I guess if I had to pick, I think the infrastructure solutions around staking and custody and tokenization is probably the most exciting just because of stuff like the AllUnity stablecoin, which, which Duncan touched on a little bit earlier. But it's basically going to be one of the first fully regulated euro stablecoins, and Galaxy owns 33% of the business behind that stablecoin, which I think it just launched. Either yesterday or the day before. So obviously with Circle ripping to $40 billion market cap, you can do some, uh, moon, moon boy math and get to a pretty crazy value of what that could potentially be worth down the line.

Yeah, I'm gonna agree with Matt because I think that this stuff is just like not priced in at all. And if Galaxy is able to become like a leading player in, you know, tokenized equities or, you know, in the euro stablecoin market, like those could be worth billions of dollars on their own. And it's kind of like a sleeping asset. Like it's a bit of a sleeper on the balance sheet. Like not many people know about this. You know, Galaxy has so many different business lines, it kind of just like falls between the cracks. Like it kind of gets lost.

Yeah. Well, I guess the reason it probably gets lost is, I mean, it's just not a big business at all relative to the rest of the business. And also custody, it's really a custody platform. Like, they— I, you know, I know GK pretty well. Like, I think they bought them for around $40-50 million. They were doing roughly, let's call it, $10-20 million in revenue. Like, custody is just not that big of a business. And you can see, like, Anchorage, BitGo, uh, Copper, all these guys, like, they, they, they started as custodians and then they had to expand out and do other things. It's funny, all these companies are now kind of building the exact same platforms. Like, Anchorage is now doing lending and trading, and BitGo is now doing staking and lending and trading. And so they're all actually in a weird way competing against each other and building the same thing.

Yeah, I just think like, yeah, we'll have to see because the tokenization like rules aren't clear yet, but that's obviously a potentially massive market in the future, right? You bring all equities on-chain, like that could be huge. And then that coupled with the other synergies of Galaxy, like could, could be really big. But yeah, I guess we'll just have to see.

Yeah. Okay. Anything else on the crypto business that you guys want to touch on?

No, I, I think we basically covered it. I think this year and next year we'll really start to see it become a lot more profitable. Like I've said, like the costs seem relatively fixed and now you're getting this increased regulatory clarity. Like, you know, Bitcoin's doing really well, more volumes, and hopefully they can keep those costs relatively fixed and ramp revenue significantly. So I'm pretty excited for that.

Let's get into what I would— I'm going to self-describe as the most interesting part of the conversation, which is the most misunderstood I would say, definitely, which is the data center business. So, Matt, I'm going to tee you up, I think, for a conversation here. Duncan, before I'm going to pause actually the screen share and just actually talk more holistically about just like— I don't actually— let's, let's not even talk about Galaxy for a second here. Matt, can you just tee us up for a conversation around like what is happening? Macro-level hyperscalers, OpenAI, Anthropic, data centers, electricity, compute? Give us the, like, lay of the land of what's happening right now.

Yeah, so I mean, I think just in the past week alone, Meta yesterday came out and said they're going to spend over $100 billion on, on CapEx next year. Microsoft, similar amounts. Um, and Google and Amazon are also raising their CapEx guides as well, because what all of these large hyperscalers are really doing is, is investing in data center capacities. So that's like the physical building itself, and then also the NVIDIA chips that go inside of the buildings to train their various AI models and to build compute capacity so that their customers, such as like OpenAI as a customer of, of Google and Microsoft, so that OpenAI can, can train their AI models. Um, so if you think about like AI workloads, they're extremely power intensive because these models are processing enormous quantities of data. So for these AI data centers, you really want as much power as possible at a single location because then you can basically maximize the amount of GPUs that you have together in one cluster and train the largest model possible. And then what, what also makes these AI training data centers a little bit or actually a lot more different than the current data centers that exist today is they don't have to be near the end users. So just some examples of this, Meta came out and said that they're going to build a 2 gigawatt data center in Louisiana that's going to be roughly the size of Manhattan. XAI is building a 1 gigawatt data center.

Sorry, sorry, Meta is building a data center in Louisiana the size of Manhattan.

I think Zuckerberg compared the size of the Instagram campus to Manhattan. Um, and just like another—

I actually, sorry, I listened to a weird podcast with Sam Altman and Theo Vaughn, which was the crossover you didn't know that you needed, but you, you did actually. Um, and he's, uh, Sam's like, yeah, you know, pretty likely that, uh, the entire world will just be covered in data centers and that's the only thing that will And, you know, 50 years, the only thing that will build physically is data centers. Also, you're muted, but yeah, that's turning into a must-watch, must-watch podcast.

But yeah, so like the other thing is you want as much cheap electricity as you can get as well. So basically bringing that all together it somehow basically Galaxy bought this before getting it.

Before getting into Galaxy, you had a slide in your research which talks about the value of a power contract. And if you're watching on YouTube, you can see this. If you're just listening, maybe, Matt, you can walk us through— can you walk us through the value of a power contract? Yeah. So it flows like, you know, this person flows to this person, flows to this person, goes back to the utility provider.

Yeah, for sure. So like your, your AI apps, which is obviously at the top of the stack, that's like your ChatGPT, um, and that's drawing on compute from the underlying model. So in that case, it would be from OpenAI's model, um, and then OpenAI is running that compute, so they don't actually own the, the GPUs. They're renting GPUs from Microsoft, uh, from CoreWeave and from Google. So those are basically— those companies are the hyperscalers. So that's your Amazon, Microsoft, Google, Oracle, and CoreWeave. So they're buying these NVIDIA GPUs, putting them into these massive data centers, and then they're renting out the compute capacity from those GPUs to their various customers. Um, And these hyperscalers, they both build their own data centers. So they have pretty big in-house real estate development teams that can go out and, and build these things greenfield, like basically from scratch. And then they're also renting third-party data center capacity. So that, that's where Galaxy would come in with Helios is a hyperscaler needs data center capacity. So they, they, they go out to data center developers. and try to procure that capacity and move their GPUs into the, the developer's data center. Um, what, what's really happened is that until the emergence of AI with ChatGPT, the power demand in the US has been pretty flat for the past couple of decades. Like the manufacturing and factories have basically moved offshore. So you're like moving offshore these really energy-intensive workloads. And then also like the energy, various like systems, like think about your HVAC, it's got way more energy efficient over time. So like as a result of those couple of factors, there really hasn't been much expansion of, of the grid or the energy supply in the US up until recently with this AI boom. Um, and now there's like various forecasts out there. So there's Goldman Sachs and Anthropic both put out reports in the past couple weeks that basically said that the US data center capacity is going to need to double over the next couple of years. So basically you have all these hyperscalers who are scrambling to increase their data center capacity and they're all sort of in an arms race to develop the best model. And that's sort of where Helios comes in because Helios, when Galaxy acquired it— Wait, wait, sorry.

I know I keep interrupting. I just want to make sure I fully understand it. So basically you've got ChatGPT, ChatGPT is kind of, is connected to OpenAI and using the model. The model is using— leverages the hyperscalers like Microsoft and Amazon and stuff like that. The hyperscalers are purchasing these GPUs from NVIDIA. But really the important part for this conversation is that the hyperscalers need, need their, need rent or they need data center space. And they do that in two ways. One is they actually own data center space themselves. And like, you know, you might have the Amazon data center, self-built data centers and stuff like that. You might have, you know, Meta building their thing in Louisiana that's the size of Manhattan, but also to accelerate it and to speed it up, they rent data center space from third-party data center owners. Is that— that's—

that's exactly right.

And where Galaxy comes in here is they own this thing called Helios, which we'll talk about in a second. And that would be that, that would be a third-party data center owner.

Yeah, that's right. And then there's one other, there's one other component here which, um, is definitely worth mentioning, is that CoreWeave is obviously like a very new, uh, cloud compute provider, and they don't build their— like, as of today, they're not building their own data centers. They were— they rely entirely on third-party developers. And then Oracle is the same way. So Oracle's been recently gaining a lot of share in this AI cloud compute market, but historically they haven't built their own data centers. So you have these two sort of emerging cloud providers, neither of whom is building their own data centers that they're all trying to rent from third parties. So that's basically created this massive demand tailwind for somebody like Galaxy because they have the ability to deliver that third-party capacity.

Yeah, and, you know, could you share my screen quickly? I think this gives like a good context just how important the energy side is. So I guess this is from a former Meta employee, like, saying that energy is the biggest bottleneck. Like, a lot of people have been focusing on chips being the biggest bottleneck, but energy is actually a huge bottleneck. Um, and so that's why it's like, you know, these Bitcoin miners that have these huge energy contracts that they've secured, you know, pre the AI boom, are now converting into AI data centers because it's like way more lucrative. So, you know, even if you have the money, you can't even spend it because—

so what— so for people who can't see this tweet, what it says is even if Meta— or, uh, scroll up a little. Okay, an interesting comment from a former Meta employee: energy is the biggest bottleneck right now. Even if Meta— uh, oh wait, I lost it. Even if Meta wants to spend $100 to $150 billion on CapEx for AI infrastructure, they can't. It's not just NVIDIA. Transformers, power equipment, cooling equipment, and the availability of power is all limited right now. Schneider Electric is completely booked until 2030. Even if you have the money, you can't spend it. So it's— yeah, it's funny. It's like you are starting to realize that kind of the capital in the world feels infinite right now and it feels like there's just an unlimited amount of money to go into these things. But the bottleneck is actually the physical centers.

Yeah.

Yeah. And you you can't, you can't even start construction on a new data center until you have all these various different permits and approvals, um, which can take years to get because these utilities have to do all different types of studies to make sure that it's not going to negatively impact the grid by bringing these new power loads online. So that, that can take basically years just to get approved. And then once you are approved, you have to get all those very long lead, uh, power infrastructure items like transformers, for example, it's going to take 24 months probably from the time you order it to the time it gets delivered.

Yeah. And then this is a cool chart that just shows like energy production in China versus the US. And like, as Matt said, like US energy production has like stagnated like pretty significantly. And now with AI, like, you know, there's a lot of directives from you know, the current administration, like, we need to get this power and like this show on the road. So just very positive tailwinds for people like in this AI data center space.

Yeah, China needs to catch—

it's like a national security issue that like the US needs to catch up here.

This episode is brought to you by SKALE, the next evolution in layer 1 blockchains with a gas-free, invisible user experience, instant finality, and robust security. SKALE allows for limitless scalability and has already saved its 50 million users over $11 billion in gas fees. SKALE's high throughput, zero transaction costs, and powerful Web3 infrastructure make it the ideal platform for gaming, AI, and other high-throughput apps. To explore the gas-free ecosystem, visit scale.space/ecosystem and follow their journey on X @skale_network. That's scale spelt with a K. Thanks to Skale for making today's episode possible. Shout out to today's sponsor, Katana. Katana is a DeFi-first chain built for deep liquidity and high yield. No empty emissions, just real yield and sequencer fees routed back to DeFi users. Incubated by Polygon Labs and GSR, Katana is launching with notable ecosystem of primitives, assets, and partners like Morpho, Chainlink, and more. Katana's 1 billion KAT campaign is live. Bridge and deposit directly into vaults in one simple click and start earning immediately on your Ether, Bitcoin, USDC, and more. Follow the link in today's show notes to get started. As always, investments in blockchain technology involve risk. Terms and conditions apply.

Agreed. Agreed. Let's get into Galaxy. Matt, I interrupted you like 3 times when you were starting to talk about it.

Let's start.

There is, I think maybe take us back to 2022 or 2023 and tell us how Galaxy kind of came to have this mining business, miner, and then data center and the electricity contracts and walk us through the whole story here.

Yeah. So At the end of 2022, Galaxy basically bought the physical Helios data center facility, which was 200 megawatts, um, and it was used as a Bitcoin mining data center. And then they leased that space back to Argo, which is who they bought, um, the facility from. So basically, Argo was a Bitcoin miner. They were under extreme financial stress in 2022 just because of the bear market in Bitcoin. Um, so to, to, to provide them with liquidity, Galaxy said, okay, we'll buy your data center and then we'll lease it back to you so you could still continue mining, but we'll own the physical facility. And then basically since that time, in, in 2024, at the end of 2024, that, um, agreement to, to lease that space to Argo was, was terminated because that's when Galaxy decided they were going to lease the space instead to, to CoreWeave. Um, and CoreWeave really figured out, because they're basically like a scrappy new entrant into this AI cloud market, um, and not sure if you guys are aware of this, but they, they were an Ethereum miner way back when, so they had experience and expertise in operating the, the GPUs because those also used to be used for Ethereum mining. Um, And they, they basically were trying to essentially catch up to all those large hyperscalers in terms of providing AI cloud compute services. So they, you know, were really scrappy and really quick thinking, and they basically went to Core Scientific, who they had a small little relationship with. So they were renting out a small portion of Core Scientific's data center space and basically told Core Scientific, Hey, we'll, we'll pay you more than you're making from Bitcoin mining if you let us use your data center, um, for our AI cloud compute services. Um, and then that's sort of where Galaxy came in, is because they also had a relationship with CoreWeave, just because CoreWeave was operating within the crypto ecosystem. And like we've discussed, Galaxy touches pretty much everything within the crypto ecosystem. Um, and, and, and they're basically now taking that data center Helios, which is in West Texas.

They're already—

this has already started. They, they basically removed all of the Bitcoin mining rigs from within the data center and basically gutted all the physical insides and are basically going to repurpose it, um, in terms of buying new electrical equipment and cooling equipment and make it such that CoreWeave can move their GPUs into that data center.

And this came because Galaxy was doing Bitcoin Bitcoin mining lending, right? And in 2020 or 2021, they had lent Argo a decent bit of money. Then Argo got on the wrong side of a bear market and maybe hadn't sold enough Bitcoin like a lot of the other Bitcoin miners did. And so my understanding is that the loan was backed— the kind of collateral of the loan maybe was actually the mining site. And then they— is this true or no?

I'm not honestly sure.

I believe that's the case. But so they paid something like $65 million for the site, I believe. But like, I was, you know, so they bought—

so, okay, so Argo owned Helios. Galaxy bought Helios from Argo. Galaxy had lent a bunch of money to Argo. Argo couldn't pay back the loans. They end up buying Helios from Argo. They, as a Bitcoin mining site, data centers end up ripping, uh, or excuse me, AI starts ripping. This data center goes from, hey, it could be a nice thing to, to, to mine some Bitcoins, to, oh my God, this is the hottest commodity in the entire world. So in they hit the jackpot. Love Novogratz. He does seem to get very lucky. I was listening to another podcast that he did and someone said he's like the luckiest, luckiest businessman alive right now. And yeah, so he stumbled upon this like incredible, incredibly scarce resource, which is Helios, this powerful data center and the scope of it or the size. Oh, excuse me. And then they got into they, they found a single tenant for that data center, which is CoreWeave, right? That's the— to round out that story. Okay.

Yeah. So they have, they have 800 megawatts of approved power. So like they've already went through all the approval processes for that 800 megawatts and 600 megawatts is already signed to be leased out to CoreWeave starting at the beginning of 2026. And CoreWeave has an option to take up that last 200 megawatts. They have an opt— they basically have a right of first refusal on that, which expires in September. And Novo said publicly that he expects them to take that option.

Yeah. Yeah.

Talk about the size of the data center and just the scope of how big this thing is.

Yeah. So like Mark Zuckerberg literally said two days ago that the 1 gigawatt data center that they're bringing online next year will be the largest data center in the world. So what Galaxy's building with this first 800 megawatts, like if it was fully completed today, it would be one of the largest data centers in the world.

Galaxy would be one of the largest in the world.

Like 800 megawatts would, would be up there in terms of one of the largest data centers in the world because, you know, Facebook is basically saying that their 1,000 megawatts is, is going to be the largest in the world. And just like more context behind how big 800 megawatts is, Digital Realty, which is a publicly traded data center REIT, they operate over 300 data centers across the world and they have basically around 3,000 megawatts or 3 gigawatts of total power capacity. And it's basically like close to $100 billion of enterprise value for that company. And, and Helios at 800 megawatts would be roughly a quarter of the size of Digital Realty. But, but Digital Realty's actual data center facilities are each around 10 megawatts because they have 3,000 megawatts spread across 300 data centers, whereas Galaxy is building 800 megawatts in one single location.

Um, so Galaxy, now to summarize where we're at, they've got this strong balance sheet, they've got this kind of like the Goldman of crypto, and now they've got the data centers to power, to to power AI. Get into maybe the numbers. I don't know, Matt, I'm picking on you a lot. Duncan, jump in if you want.

Or maybe quickly put on my screen just this will help contextualize how big the— it's a little ChatGPT work here. But so, you know, Galaxy has 800 megawatts approved, 1.6, 1.7 gigawatts under study. A really important part of this 1.7 gigawatts under study is that they, they applied for this back in 2021/2022. Still when they were mining Bitcoin. So like, if you think about the process, um, in Texas of approving new power, you know, it's kind of multifaceted. It's like, you know, who applied first, who's gonna actually be able to execute on this. So, you know, Galaxy is like executing with CoreWeave right now. They have the balance sheet to build it out. They have the other relationships with like hyperscalers. Uh, the head of Helios basically said that they've had like a bunch of these like hyperscalers tour the facility. So like, they have these relationships, they have the balance sheet, they got the— yeah, so they have the balance sheet, they have the relationships. They were one of the first people to apply for this in Texas, like ahead of these other Bitcoin miners or these new build data centers. And they have like a bunch of people who used to work at ERCOT that work at Galaxy. So like they have like— they understand how this system works, and management has essentially guided towards like this 1,700 megawatts being approved in two tranches, an 800-megawatt tranche and a 900-megawatt tranche. And they've said like they're expecting the 800-megawatt tranche to be approved, you know, in single-digit months or kind of like before end of year. And then if you just kind of compare that, so yeah, here's the Facebook one or the Meta one, you know, 1.5 gigawatt phase one scales to 5 gigawatts, 2.7 gigawatts for Amazon, you know, multiple Projects totaling 1 gigawatt for Microsoft. So like if you kind of put this into context, it's in the same order of magnitude as like the largest companies in the US in terms of them building data centers. So they're actually like, this is a huge site and like, you know, Galaxy could be a top player in this new space. And as Matt said, you know, Digital Realty has 3 gigawatts across 300 locations. That's kind of like data center 1.0. Now we have Data Center 2.0, which is like, you know, multiple gigawatts in one location. So it's like a whole new business line that's going to experience like really rapid growth because of AI. And Galaxy could be like a leader in this space.

So yeah.

Okay, Matt, can you walk us through the economics of this and talk, talk through some of the numbers, both on the revenue side and the cost side?

Yeah, for sure. So just on the 600-megawatt lease that Galaxy has signed today with CoreWeave, like excluding that option, um, the CoreWeave will basically paying Galaxy $900 million on average per year over the next 15 years. Um, and then the margins on that revenue are expected to be close to 90% because the lease is structured in a way whereby CoreWeave as the tenant is going to be paying for most of, if not all of the property expenses such as utilities. So basically you have, you know, $900 million of revenue, assuming a 90% margin, call it $800 million of EBITDA. That's just a stream of cash flows that is coming in over the next 15 years to Galaxy, and it should cost around $5 billion total. For, for Galaxy to build that facility. And based off of what Galaxy management has said around what type of financing they're going to use, they'll probably put in a billion dollars of equity from the balance sheet, and then they'll, they'll find a lender for the other $4 million or $4 billion. Um, but if you model out the returns on, on, on that $1 billion equity investment, it, it's pretty insane how attractive they are, just given you're going to be collecting $800 million a year over 15 years of just pure cash flow. Yeah.

And, you know, maybe you should share my screen again. I have like a little model.

Duncan's like, dude, just keep my screen up, man. You're killing me.

So yeah, this is exactly what Matt just walked us through. So they have these two phases, a 200-megawatt phase one, 400-megawatt phase two. These two are already contracted with CoreWave, which is live, then that option we're talking about for Phase 3 is an additional 200 megawatts. Um, so if you go over here into the adjusted EBITDA for Phase 1 and 2, which they already have contracted out, you know, you're looking at nearly $700 million in EBITDA in 2028. And basically how these contracts work is they have like an escalator to like adjust for inflation and such. So like The rent checks go up over time, over the 15-year lease period. If they get Phase 3, that's an extra $200 million in EBITDA for them. So in 2028, they could be generating nearly $900 million in EBITDA just off their current 800 megawatts of approved power. Goldman Sachs had a multiple of 27x And so that's an enterprise value of $24 billion in 2028. You know, you subtract the debt needed of around $5.6 billion. That means an equity value of $18.3 billion. You divide by the total shares outstanding of 395 million, you get $46 per share alone in 2028. And then if you discount that back to today using what Goldman said is a 12% discount rate, that's $30 a share in present value today, assuming they execute on this buildout, which is greater than the price of the stock right now on Helios alone. And then where things get like super— and I should just mention this, the Goldman Sachs numbers. So basically what they did I don't know why they did this. Maybe Matt has some more color into the equity research world, but they assume that Galaxy's not going to get to Phase 3 in their projections. And then they picked this weird midpoint in between 2026 and 2027 where even the Phase 1 and 2 revenues haven't fully scaled up because these tranches are going to be coming online over time. Expected full completion date for Phase 1 is the third quarter of 2026. And the third quarter of 2027 for Phase 2. So they kind of picked this like weird in-between place between 2026 and 2027 to get their $200 million EBITDA number. And then, you know, you do the same math and they say it's going to be worth $8.50 a share. But if you instead think that they're just going to get Phase 3 approved or like CoreWave is going to exercise their lease and then you discount back from like 2028 numbers, it looks a lot better. At this $30 a share. And then where things get super crazy is if you go into them getting, you know, either part of the 1.7 gigawatts approved or the full 1.7 gigawatts approved. And that's where things get like really silly. So same math. You're talking about in 2030 or 2032, you know, $50 billion. And— or sorry, $36 billion in equity value, $57 billion in equity value if they get both tranches, uh, divided by the same amount of shares outstanding, that's like $145 on Helios alone in 2032 if they can execute on this.

Uh, so all right, Matt, take us back. So, okay, Duncan, moon boy math, good. I like it, I like it. Matt, take us back maybe into reality a little bit, or, or, or you agree with, uh, with with this model that Duncan has?

Yeah, I think like the way we thought about it in our report was just to sensitize how much, how many megawatts would get approved and then to run the EBITDA per megawatt based off of the existing lease that they have with CoreWeave. So basically if they had the same economics on their CoreWeave lease for the entire 2.5 gigawatts at Helios, like we're talking tens of billions of dollars of potential equity value. So I think there's, there's obviously various different ways you can try to estimate what this is worth. But I think what's been approved today and what's on lease today with CoreReef is probably worth high single digits to $10 billion alone. But there's just all this embedded upside in the potential for them to not only expand with CoreWeave for an additional 200 megawatts, but then if they also get part of the 1.7 gigawatts approved, you could be looking at some seriously big numbers. But I think it's important to just call out that it's really hard for a new data center developer to sign a lease with one of these massive cloud companies like Galaxy because they had a relationship with CoreWeave from way back when, when, when, when CoreWeave was a crypto company. That sort of got them, that got their foot in the door. And then it also just so happened that CoreWeave was willing to take a risk on someone like Galaxy building them an AI data center because they're trying to catch up and become one of the largest AI cloud compute providers. So they need to be as sort of thinking out of the box as possible.

How tough, Matt, how tough is this to do? Like if someone was like, hey, Jason, here's $1 billion, go build a data center, I'd be like, I mean, I got no clue where to start. Probably figure it out. But is that, is that kind of what Galaxy is doing or do they have this expertise or did they like, how do they know how to do this?

Well, it's really twofold, right? So you have to find a customer first. You have to find a tenant because you're not going to be able to get project financing to debt fund a large portion of the CapEx that's required without having that customer first. So like I was saying, like, Galaxy was fortunate in the fact that they had a relationship with CoreWeave and they basically were able to parlay that into this, this massive lease. But I think going forward in terms of attracting different tenants, like someone like an Oracle or a Google, Galaxy having that lease with CoreWeave is, is a massive stamp of credibility because all these other hyperscalers know who CoreWeave is because they're, they're taking market share from them. And, and basically says to, to all these other hyperscalers that Galaxy is reliable, it can be trusted for one of these massive, um, construction projects. But to, to your question, again, having, um, relationships with the best contractors, because obviously the developers are going to use various, um, contractors to, to actually build these data centers. And there's been some sleuthing online from a guy on Twitter, um, that goes by Stefan Vanderlux. I'm probably butchering his last name, but this guy's a must-follow on Galaxy.

This guy is a detective.

So he basically, he basically did some sleuthing on LinkedIn and, and found out that Galaxy is working with one of the best contractors in the data center development game. So I think the fact that Galaxy can go to somebody like an Oracle or a Google, if they get this additional power approved, and say like, hey, look, we have a massive $3 billion balance sheet. We have relationships with some of the best contractors in data center development, and we have a lease with CoreWeave, who is obviously everyone knows in the AI cloud compute world, like, we are very capable of building you also a massive data center at Helios. And I think that's a pretty credible pitch, whereas all these other Bitcoin miners are basically lacking all three of those things. So when these other Bitcoin miners are talking about AI, they don't even have— they don't have— they don't have the balance sheet to do it, they don't have the relationships with the contractors to do it, and they don't even have a lease So I don't even think you can really compare Galaxy to most of these other companies that are trying to do this pivot.

Can I try to describe, describe back what's happening here maybe, and like actually try to summarize Duncan, some of the developments that you were saying? I just want to check that I got this right. So right now they have this data center in Helios and that's what it's called, right? Helios.

Yeah, it's in West Texas, Dickens County.

Everything's in West Texas. Okay. So they have 800, 800 megawatts of approved power right now. They have a 600-megawatt contract signed with CoreWeave that is providing roughly— what did you say— $900 million a year in, in, in annual revenue. A lot of the costs of that flow back to CoreWeave, which means there's really attractive economics on this. And I think the— what I read was the EBITDA margins on something like that type of deal are like 80 or 90%, which is great. The first 800— this, this buildout they've already built or then they're going to— they've already built this build.

Yeah. So they, they are basically in the process of doing— so it's in two different phases, three phases of core. We've exercised the extra.

Oh, okay. This is the first tranche was the 800 megawatt. The second tranche is the 900 megawatt.

So, so basically they have 800 megawatts of approved power. They have an additional 1.7 megawatts under study, which—

oh, I see. I see.

This is what they've—

this is what they've applied for.

Yeah. Yeah. They applied for this back in 2021.

This is the ERCOT, the ERCOT power in 2021.

Exactly. Yeah.

Okay.

I get it.

Okay.

So they've got this $800 million with CoreWeave, $800 million, $600 million with CoreWeave, then they've got the, the next 1.7 gigawatts of power that Galaxy's applied for with Helios, and they applied back in 2021 or 2022, and they're expecting approvals in two tranches, an 800-megawatt tranche and a 900-megawatt tranche. They think the 800-megawatt tranche will come by end of year. It sounds like end of 2025, which means they will then have a total of 1.6 gigawatts, which is 800 plus 800, which then means they have an extra, uh, so 1.6 gigawatts, 600 gigawatts have already been, or megawatts have already been contracted to CoreWeave. Um, I'm trying to do all this mental math. Okay. That leaves them with a, with 1 gigawatt left basically. Um, which they then need to contract out to another hyperscaler. and they've got a good deal with CoreWeave going. But if they can contract this out to another hyperscaler on roughly similar terms to CoreWeave, that gives them an extra like roughly $2.5 billion a year on 90% margins, which means— I don't know the market cap there. I don't know the comps to use, but an extra $2.5 billion of revenue on 90% margins is good. Did I— did I— how did I do, guys?

Yes, that's pretty good. The one thing to add is so they have 800 megawatts approved, 600 with CoreWeave, and then CoreWeave has an an option for another 200 megawatts, which I think, given, you know, the demand for, for power right now and the existing relationship with Galaxy— like, Galaxy's building out like a bespoke data center for CoreWeave, um, in this Phase 1 and 2— I think it's likely that they'll take it in Phase 3. Could be announced as soon as earnings, which is on Tuesday, August 5th, or they have until September to choose to exercise that or not. But personally, I'm not very worried about them being able to find— either CoreWave is going to take that or they're going to be able to find another tenant.

Got it. Okay. Can they raise— they— okay. So to build, to do this buildout, they need $5 or $6 billion. And I think I read they plan to finance this with— it was 20% equity, 80% debt, and they have the equity component secured and they're now in the final stages of getting the debt secured. So 80% of $6 billion, $4.8— they need $4.8 billion in debt. Who gives someone $4.8 billion in debt for something like this?

There's a lot of the large bulge bracket banks that do these types of construction loans. And then there's also a variety of private credit lenders such as something like a Blue Owl that would honestly love to invest in a massive data center. There's like a lot of demand on the investing side to do this.

What about outside of Helios? I read somewhere that there are maybe in one of your guys' notes that they're— they have a pipeline of like 40 Bitcoin mining sites that they're evaluating to either acquire or partner with. Uh, we've been talking the whole time about the Helios Center, but what about the other 40 Bitcoin mining sites that they're looking at?

Yeah, so this, this was like the, the management team at Galaxy talked about this on the last earnings call, and I wasn't expecting them to already be active in looking at new greenfield sites, so I was really excited when they brought this up, but they basically have a team at Galaxy that's evaluating and gets shown deals for acquiring some of these greenfield or, or brownfield sites, some of which are Bitcoin mining data centers today that can be converted into AI data centers. So they basically have a pipeline of all these different sites that they're looking at, and if they can find a good site where they think they can get a tenant at that site, it seems like that's going to be something that they're actively pursuing. So I mean, if you start doing numbers on how much EBITDA they could be generating, if they're able to do some greenfield sites outside of Helios, it gets really exciting. And it sort of brings back to a point we touched on earlier briefly, but this is basically a new market. These massive 1-gigawatt-plus AI data centers don't exist today. And these, you know, quote-unquote legacy data center developers such as Digital Realty and Equinix, their expertise— and they've been around for decades, but they've been building these small 10-megawatt data centers that are basically completely different in terms of the actual cooling infrastructure and the electrical infrastructure and the rack design than, than an AI data center. So this is basically a, a massive greenfield opportunity for, for new data center developers to come in and basically build the infrastructure for this massive AI boom. And it seems like Galaxy is, is pretty well placed to, to take a piece of, of, of this market, just given that they're set up well now having that existing 600-megawatt lease with CoreWeave.

And to add on to that, some comments from Novogratz from the most recent Spaces and some podcasts I've listened to. So, uh, yeah, from the sleuthing online, Stefan seems to think that Clayco is the, the developer, which is like a, you know, tier 1 data center developer.

And contractor— being a contractor—

sorry, contractor. And, um, Novo basically told the story about how I guess the, the CEO of that company was in New York. He's like, do you want to grab dinner? And Novo was like, you're the most important person in my life. Yes, I'll get dinner with you.

I actually heard that. Yeah.

Yeah.

Other than my wife, you're the most important person in my life. Like, other than my wife and kids. So that is kind of a funny thing. And then also Novo on the most recent Spaces, I have a post on my Twitter. They did a couple months ago. I was able to ask a few questions to him directly and he was like, I would be very surprised and you know, the not too distant future if we're only talking about Helios as, uh, like the one site we're developing. So that's where things get like insanely crazy. Obviously, you know, the full— them getting the full 1.7 gigawatts of approvals in these two tranches isn't a sure thing, but if they are to get that, plus I think the Goldman report said like the average Bitcoin mining site that they were looking at was around 500 megawatts. In, um, in, in, in contracted power or approved power, then like, you know, you take this from, from 2.5 gigawatts of power to like 3 or 3.5 in this like massive new, like, you know, greenfield business that Matt's talking about. And that's where things could just get really crazy of like Galaxy as, you know, a really big player in this whole AI data center world. The other important point to note because they did, they issued some shares a couple months ago after they listed on the NASDAQ to help pay for potential acquisitions or build out Helios, is that as these tranches come online, like Tranche 1 and 2, and then eventually Tranche 3— I'm sorry, Phase 1 and 2, and eventually Phase 3— as those come online, they are generating huge rent checks each year, which then they can go and borrow against to fund the rest of the development. So they shouldn't need to, you know, uh, issue any more equity, which is really exciting, where it's like they can kind of self-fund basically like all this build-out as it starts cash flowing.

Ever wish you could access cash without selling your Bitcoin? Ledn, the global leader in Bitcoin-backed lending, makes that possible. Since 2018, they've issued over $9 billion in loans and not a single Satoshi of client funds has ever been lost. Here's what makes a Ledn loan different. With custody loans, collateral is not lent out to generate interest. No credit checks, funds in less than 12 hours, and no monthly payments. Repay whenever you want with zero penalties. And they don't just talk security, they prove it. Ledn offers proof of reserves reports verified by a top accounting firm every 6 months. Ledn is 100% focused on doing one thing better than anyone else: giving Bitcoin holders a secure, transparent way to unlock liquidity without selling.

Click the link in the show notes to find out more.

As always, investments in blockchain technology involve risk. Terms and conditions apply.

Maybe take some counterarguments here. One of them might be that They're in— they have one customer. If you were looking at anyone with one customer, you should discount that business heavily and think about like, I mean, there's key man risk and there's key customer risk, right? So, uh, like, I think there generally should be concern around anyone having one customer, and Galaxy shouldn't be any different. And specifically, I've read some fears about CoreWeave's ability to meet their rental payment obligations. Um, so, I mean, if your core customer starts failing to meet the rental payment obligations, that changes the entire scenario here. So Matt, how do you, how do you think about this?

Yeah, I think that's, that's definitely a valid risk, and in my mind, probably the biggest risk to the, this business overall is that you're extremely concentrated from a customer perspective in that all of the data center revenue today is coming from CoreWeave. Um, I guess first, I don't think a lot of people really understand CoreWeave's business model very well, just because it's a new business model, obviously. Um, and there's a lot of—

my screen up, he literally has a whole section of this in the report I was just going to— fears are overblown.

Yeah, it's like, I don't think people really understand Cory's business. Um, basically when they came to IPO, everyone saw a ton of debt on their balance sheet, and, and, you know, people love to doom post and talk about the dot-com bubble and how this AI CapEx build is reminiscent of the dot-com bubble. So when they see debt and they see AI, everyone wants to get really bearish. But the thing about CoreWeave's debt is that most of it is specifically earmarked to a customer contract that they have. So like the way CoreWeave's business works is they go out and they sign a customer contract with someone like OpenAI where they're going to rent their compute capacity from CoreWeave to OpenAI so that OpenAI can train their models. In order to provide that compute capacity to OpenAI, CoreWeave obviously has to buy GPUs. So they're drawing on this credit facility to finance the purchase of those GPUs. And the debt from drawing on that credit facility is paid off in the couple of years following the beginning of that customer contract. So they, they basically buy the GPUs from NVIDIA using debt from one of their credit facilities. They stand up that capacity, they rent it out to OpenAI, OpenAI pays them for that, and the cash from those payments goes to pay down debt. The debt's paid back in 2 years, and these are typically 4-year contracts. So in years 3 and 4, they're basically pocketing the difference as profit. So it's not as if all of this debt that's on Core Ridge's balance sheet is speculative in nature. It's specifically earmarked for a specific customer contract. I think that's, that's probably one of the main bear thesis on CoreWeave is around the debt.

There's a couple others that we can get into, but I mean, you had a couple of the risks that you wrote about in the research report that I thought were interesting. One is that there are these significant hurdles, right, in retrofitting traditional CPU-based data centers. For AI purposes, right? If you look at traditional data centers, they're oftentimes smaller, which means they're inadequate for LLM training. And they also like— if you look at, you know, like website hosting or SaaS applications, those are not that compute intensive. AI is super compute intensive. So traditional cooling systems and rack densities and things like that are kind of incompatible with modern like AI data center design. Sounds like you are not super concerned about that, but maybe that would drive up some of the costs if they're, if they're not factoring that in.

Well, I think that's, that's a benefit for Galaxy because, like, what you said is true. Like, if you have a small 10-megawatt data center, if you're Digital Realty or Equinix and you own these small data centers, it's pretty much going to be economically infeasible for you to convert that 10 megawatts into an AI data center because, like you mentioned, you need to retrofit all of the cooling equipment, the electrical equipment, and just the, the design layout of the data center itself, because the, the racks with, with the servers and the GPUs are all different. Um, but it just so turns out that these Bitcoin mining data centers are actually more readily convertible into an AI data center, just because if you think about Bitcoin mining, right, like you're, you're drawing on as much power as possible to do these massive computations to earn block rewards. So it's, it's, it's sort of similar in nature to what these AI models are doing. So it's easier to convert a Bitcoin mining data center than it is to convert one of these legacy smaller data centers. That's Digital Realty or Equinix owns.

Hmm. What about maybe the last one, which is, uh, uh, claims that NVIDIA GPUs have a useful life of only like 1 to 2 years.

Yeah, so that's something that's been popping up on Twitter a lot recently, but there's, there, there's people making an argument that because NVIDIA ships a new iteration of their GPUs every 1 to 2 years, that last year's model immediately becomes obsolete, um, because the new models are much more powerful. But this is just not grounded in reality at all. Like, Basically, Satya Nadella has gone on podcasts and basically said, yeah, you, you want the newest GPUs for training your models, and then you just move your older GPUs into the actual application usage or inference. So they're, they're still being used for 5, 6 years. It's not as if they're thrown out after the new ones are released. And then OpenAI has also said that only in 2024 they decommissioned some of these Volta V100s, which is the model from 2018. So I think there's a lot of empirical evidence that says these AI customers are using these, these chips for 5 to 6 years.

Let's, um, maybe as we start to think about rounding out this podcast, why, uh, I was trying to think of why— okay, I hear you guys talk, I read the reports, boom, very feels very bullish, Yeah, price goes up, data centers. Yeah, makes a lot of sense to me. Why hasn't the market understood this? That's what I was trying to think about. And one of the— probably the biggest thing that came to my mind is it's a very confusing story. It's not a clean story. It's very confusing. And having the crypto business combined with the AI business, that makes it very complicated. The crypto business in and of itself is very complicated. But you combine this complicated crypto business with this AI business and the AI business wasn't like a clean story from the beginning too. Um, what do you guys think about, uh, if Novogratz came to you tomorrow, he's like, Duncan, Matt, you two are deciding, should we split the companies or not? Should we split it out into two separate companies? What, what advice would you give him?

Yeah, I mean, I think, I think in the future there's definitely a high chance that they're going to spin the data center off as a separate company because like at maturity, the data center business is just a stream of cash flow. So you could put it into a REIT structure like Digital Realty or Equinix for tax benefit purposes, and you attract a whole different type of investor base. Um, basically people are looking for steady dividend payouts. It's, is, is a lot different than someone that wants to invest in a crypto-focused financial services company. So I think at some point in the future, it's definitely going to make sense to, to split them up. But for now, it probably makes sense just to keep them together because the data center business isn't generating revenue yet. It's not expected to start generating revenue into 2026. So I don't think a lot of value investors would be interested in, in buying a, a data center business that's not generating revenue yet. And then they also can leverage the, the balance sheet that Galaxy has now today to fund part of this buildout. I think having that big balance sheet, as we've discussed, is a really massive competitive advantage versus other data center developers. I think that $3 billion is more than Digital Realty or Equinix have on their balance sheet, and those are both, like I said, like $70, $80 billion companies. But I think down the line, it, it, it definitely makes sense to, to split them off. And if you look at their financial reporting, they've actually started this new segmenting where they're splitting up the digital assets business from the data center business. So I think that's a pretty clear indication that it's going to be split off down the line.

Yeah, I think Novo said on the spaces, you know, if the market's not giving us adequate credit in like a year from now, then we're going to consider that more, more closely. I think just in terms of like why, you know, we believe it's so cheap, or at least why I believe it's so cheap, is because like, you know, in this Goldman report they're not, you know, ignoring the fact that CoreWave could take this next phase, like exercise their 200-megawatt option, or they could get 1.7 gigawatts approved. It's just like hard for them to like bake that into their valuation, right? It's like these things aren't certain. So like we're not going to necessarily bake it into our valuation. But if you share my screen again, like they go through like the potential upside they see. As like CoreWave exercising its Phase 3 option, Galaxy, you know, getting additional approval and customers, and then Galaxy, you know, expanding their data center business. So, and they model that out here, like, you know, if they get Phase 1, 2, and 3, and then, you know, a portion of the 1.7 gigawatts approved or kind of all of it approved, then, you know, it comes out to very similar numbers to what's in Matt's report. And they actually use like a better multiple than what Matt has used. Like Matt used a 22.5x multiple as kind of like his base case here, and they're using like a 27x multiple. And if you look at like historical data center comps, they've traded at, you know, 26 to 35x multiples on the high end. So it's not like The market— like, it's not like the Goldman are kind of like— people who are starting to look at Galaxy are saying that this isn't going to happen. It's just like very hard for them to like assign the chance of it actually executing since this is like Galaxy's, you know, first time doing this. They don't have like a proven track record. But I think where there's like a real edge for people who spend a lot of time looking at the company is that like management has been very good at following through on everything. They've made like a ton of hires in the data center space. this is from Stefan— in the last 4 to 6 weeks, VP of construction, VP of inventory management, VP of procurement, head of data center finance, head of data center growth and corporate development. They're looking for a bunch more roles. They have this lease with CoreWeave. So you do definitely have to believe in the management that they're going to be able to execute here. And I think that's a massive reason why it's being discounted. The other obvious one that you said is just like, it's a fucking complicated business. They have the balance sheet component, they have the crypto component, they have the AI data center component. How do you really square all that together? And then like Matt said, there's two very different investor bases who would be interested in investing in the crypto and balance sheet versus the data center business. So I think that it trades at kind of like a discount to its fair market value, especially if they can execute on these next phases or next tranches or acquisitions. So that's kind of my take on it. It's also only been uplisted to the NASDAQ for a couple months now, right? Like, they were stuck on the TSX for 4 or 5 years, and the last SEC kept on blocking them from uplisting. So it's still a relatively, like, new company in the eyes of investors. And, you know, Goldman just dropped this report a couple days ago. Um, people are still just getting up to speed on Galaxy.

Yeah, I mean, it's real— it's really complex, right? You have the two separate businesses, and then one of the business in, in and of itself is super complex But I think once you sort of roll up your sleeves and dig on, dig into the numbers and what's going on and what they've already contracted with CoreWeave, it looks really undervalued. And I think it's just really a classic case of the market not fully understanding something. But as they sort of generate revenue at Helios beginning next year, I think the story will become a lot more cleaner. It should sort of re-rate accordingly.

So let's round out this podcast by just talking about the price of the stock, which is I think trading at around— what is it today? $26. I think it listed— it was, it was kind of $18 to $19 for a while, um, since it got on NASDAQ in mid-May, let's call it, uh, trading around like $18, $19, sitting— went up to $31, $32, let's call it, and now it's trading around $26. When is this I don't know if you guys agree or disagree on this, but like, what do you guys— how do you get out of that? Is this a trade? Is it a long-term investment? Are you— is there— do you have targets that you're going to start taking profits? How do you think about that?

Yeah. So just pulling up the chart and one more point on just like kind of how these things cannot make sense for a long time. I'm pretty sure like they signed the 600 megawatts with CoreWeave like here during the tariff meltdown stuff, and then it traded all the way down to like $8. Um, so even with like that contract being live, like it traded like way down, and then they list— they announced their NASDAQ listing, and you know, it started to reprice as like people are, are finding more out about it. And it's also done like just way more volume, way more liquid market at this point. How I think of it is, so if we go to the Goldman report, I have a good tweet on this. So like, I think if you get these 800 megawatts that Helios fully contracted, so they already have 600 contracted. If CoreWeave exercises the other 200 megawatts, you know, that was the, that was the sheet I showed here. That's like, you know, would be at Goldman's 27x multiple would be $46 in 2028 just on Helios alone. Discounting that back to today is roughly $30. So I think like a fair valuation for Helios right now is probably $30 a share just because that's just executing the 800 megawatts they have approved. Then you have these like upside cases of Tranche 1, Tranche 2, and then you have to add in, you know, the balance sheet and the and the crypto business. So the balance sheet is roughly like $8 or $9 a share, and then Goldman, you know, estimates the crypto business, um, you know, I think it was $6.5 billion, uh, so maybe like— let me do some quick math, uh, $300— so I would look at it as like a sum of the parts, right? So $6.5 billion $395. So that's another $16 a share. So let's say, you know, $9 a share for the balance sheet, $16 a share for the crypto business. That brings up $25 plus another $30 a share for Helios. So that's kind of what I think it should be worth today, $55. And then, you know, if you assume crypto is going to do well, they acquire another site or get approvals on different tranches. Like, I had this thread on Galaxy that really took off where I'm— where I say I think it could be worth like $100 a share in the next few years. So that's personally how I'm thinking about it. It's definitely a long-term investment. I think as they execute on the data center business, if they continue to execute there, then like I'm really happy to hold this for a very long time. I'm bullish on crypto, I'm bullish on AI, I'm bullish on the management team. So that's kind of how I think about it long term. Obviously, this is not financial advice. Go and do your own research. I've spent a lot of time on this. This is just my opinion and how I'm, you know, Galaxy's my largest investment right now.

Yeah. The only other thing I'd add to that, and I definitely agree, I think the way I'm treating is just sort of take it earnings call by earnings call because that's when they really go through all these major updates on the data center business. All of them have been extremely positive to date. Um, it's really just an execution story, and there hasn't been anything that I've seen so far that makes me believe that they're not going to execute. And the management team here, I think, is extremely strong. Like, Novogratz, I don't really need to talk about him, but if you think about building out this data center business, right, you need to develop relationships with with some of these massive technology companies. I don't think there's anybody better positioned in the crypto universe to develop those relationships than someone like Novogratz. And then there's also the, the chief investment officer of Galaxy, um, this guy Chris Ferraro. And if you look at his background, he came from a fund called HBS Investment Partners that I'm very familiar with from my past life, um, in finance and I think his experience at HBS Investment Partners is highly relevant, um, for, for what he's doing today and leading the build-out of that data center business. So you basically have these two execs, and I don't even think you could really compare them to any of the other execs, um, in the sort of broader crypto ecosystem. So that's something else I think is really bullish.

So Matt, when— what, do you have like a price that you're looking at for this or no?

No, I think it's just sort of a month-by-month type of evaluation. And but really, like, you're— as Duncan mentioned, you're getting exposure to two of the biggest trends in crypto and AI in this, the same stock. And if you do the sum of the parts, it looks undervalued here. So I, I definitely plan on holding for the—

plan truster.

Uh, Guys, anything else that we missed? This is a big conversation on Galaxy. Anything that you guys wanted to cover that we didn't?

No.

Maybe the last one cool point that I think is underappreciated and then we can wrap up unless Matt has any other comments is just like, you know, it trades at a discount because it's kind of this like conglomerate of a bunch of stuff. But if you look at like what's happened with them in the past, they've actually like generated a lot of synergies from this. Like we said, They were loaning to CoreWeave back when they were mining ETH. They had a relationship there. They were able to get on the CoreWeave IPO, and then they were able to lease the 600 megawatts out to CoreWeave. So that was like their crypto business actually influencing their data center business. Then you also had the Argo blockchain relationship where it's like they were lending to them and then they were able to buy Helios. So it's like since their tentacles are so far and wide within everything to do with crypto, and now the emergence of AI has come up and the power contracts are super valuable. I think that's a very underrated element of their business where they just have all these kind of synergies that are very difficult to value. But look what their crypto business allowed them to do on the AI side. And I think with stuff like GK8 or AllUnity, there could be other kind of green shoots of positive surprises where it's like, something in their crypto business allows them to make an acquisition or do a joint venture that adds like a ton of value to the company long term.

Yeah.

Matt, appreciate you. Duncan, appreciate you guys. We will link to a bunch of stuff in the show notes. So if you guys want to dig a little deeper, we can share some of Duncan's threads. And Matt wrote a great piece of research, so we can, we can share that.

Everyone follow Matt on Twitter and read this research piece. This is like when I read this, this is what gave me like the conviction during the tariff stuff. To really like jump in. And he's, uh, you know, definitely a leading voice on the whole like Bitcoin mining AI data center side. So definitely follow Matt here. I'll pull up his Twitter quickly, but he only has like 1,000 followers. Criminally—

Matt, what's your Twitter handle? Rittenhouse?

It's @rhouseresearch. Rittenhouse Research was taken.

rhouseresearch.

Criminally underfollowed.

Yeah, cool.

All right, well, thanks guys, uh, for, for doing this. Duncan and Matt, appreciate you guys, and thanks for listening, folks, and I'll see you Friday for the Roundup.

Cheers, bye!