GLXY 7.30.26
Analyst 2
That’s helpful feedback. I’d consider Coinbase the granddaddy of this industry, and knowing that your VC friends place Galaxy in the number 2 spot after Coinbase is a pretty decent achievement. Regarding the liquid part of the book, we understand the cash, stablecoins, and tokens they hold. But it was more around the digital operating business that they’re trying to build. The crypto exchanges, the GalaxyOne platforms, etc. That’s what I was struggling to value because, for cash and stablecoins, etc., I'm ascribing it somewhere in the $5 to $7 per share range. That's straightforward. It's just that it's more of the operating business, which I’m struggling to value. Like, should I not value the business at all? Could it be worth $5? Could it be a $15 value per share? That's where I was getting stuck.
Analyst 1
So all I can say on the digital business is, I think they've got a real trading business and digital business there. I hear the GalaxyOne stuff is great. It’s very difficult to value because I think it earned a ton of money in Q3 of last year. So if you value it on anything that's got Q3 in the LTM, then it looks like it's worth a ton, right? And if you don't value it there, then it doesn't look like it's worth much. So I treat it as an option value. In my model, I had it at $600 million to keep it really low. That might sound low because that’s what they made in Q3, but they’re nowhere close to that now. And I'm not a big crypto guy. I don't particularly love it. You could go either way, but that’s what I valued it at. I could believe that I’m 4x too low. I could also accept that crypto is past the peak, all these dApps are unwinding, and $600 million is too high. However, $600 million doesn’t move the needle much here.
Analyst 2
So $600, for example, if you're assuming roughly 200 million shares, it's $3 per share in value.
Analyst 1
There are 400 shares outstanding, right? I think we're using just the Class A and ignoring Mike Novogratz's Class B. So yeah, I've got 400. So that's why I said $1.50.
Analyst 2
But I think the model had everything. I was just looking at this screen, and it showed 200 for Class A. But yeah, makes sense. So, not ascribing a lot of value to that business, that’s what I understood.
Analyst 1
And look, I'd love to be proven wrong, because if I'm proven wrong, it's probably going to be on the upside of my number, not the downside. So yeah, if that happens, that'd be awesome. You'd rather this contract be directly with Meta or Microsoft than CoreWeave, though I think they say this is CoreWeave supporting a specific hyperscaler who'd probably step in if they happen. I guess to start, you said, if I remember correctly, you value the CoreWeave project at a piece of this at $20 per share.
Analyst 2
I’m not sure if you’re using $20 per share. I’m using approximately $392 million total shares outstanding.
Analyst 1
I meant after the term loan and the project financing, all that sort of stuff. I'm not sure, but I'd love to know how you're getting to $20 and, particularly, what you're valuing, how much of that is on the CoreWeave contract versus the assumption on the terminal value.
Analyst 2
Yeah, so what I did was essentially build a really long model, which essentially has the value from 2026 onwards to 2043. Starting with Phase 1 this year, I’m assuming two quarters of rent following the July 6 energization date and handover. I’m assuming around $150 million in rent, which would translate to approximately $130 million in NOI. Starting in 2026, and then this lasts all the way to 2041. Then Phase 2, obviously scaling up and becoming around $250 million of rent recognized from Phase 1, starting next year, which will increase by 3% or so every year with annual rent escalation. That's what I'm building in. For Phase 2, starting next year, I'm assuming that you're only 30% of the annual rent you’re getting. So you're getting around $150 million from Phase 2 starting in 2027.
Analyst 1
For Phase 2, on July 22nd, they published the post-construction illustrative annual financial summary. Do your numbers differ at all from the financial summary that they published?
Analyst 2
Let me look at that. Do you have that handy?
Analyst 1
It's the 8-K that they filed on July 22nd.
Analyst 2
OK, so 8-K, they have the financing agreements. Do they have the rent information in there? I had fed all this information to Claude. Let me see.
Analyst 1
It's the July 22nd 8-K. It's a little investor presentation on July 22nd. And if you scroll all the way down, they've got the annual financial summary for Phase 2.
Analyst 2
OK, got it. I was looking at the wrong one. It had that illustrative financial information of the project.
Analyst 1
So, do you differ from any of these numbers? I guess you haven't read their 8-Ks, but were you making hugely different assumptions? Because I've just been working on this now.
Analyst 2
I don't think the assumptions will be very different, but what are the assumptions that you have right now?
Analyst 1
When I look at this and DCF it out, my worry is that this is a big $3.5 billion project. The numbers make sense. They’re doing project financing and not putting a ton of equity in, but the CoreWeave contract through 2043 pretty much covers the debt. You're not making much of a return, so all of your return comes in the terminal value of this asset, which is fine for an infrastructure asset.
But I guess that goes to the thing I'm thinking about with Galaxy, which is twofold. They've got another 2 gigs or so to lease up. Where’s the market for leasing assets right now? So if they announce new leases, will they be better than the CoreWeave lease or worse? And then, if we fast-forward to 2043, it makes sense, right? If you did a big infrastructure project, the first 10 to 15 years would pay back the CapEx. You then have a terminal value asset with another 30 years on it. And that would make sense here, except I'm wondering, the power market in 2043. The power market right now is the hottest we’ve ever had. If you went back to 2016 and needed 1 gigawatt of power, you wouldn’t have had trouble finding it. How do you underwrite the power market in 2043? How do you think about the economics? Do you agree that you’re really betting on the terminal value here?
Analyst 2
Yeah. The market now, I’m getting the same result in my model. When calculating the data center EBIT, I get approximately $13 billion in NOI. The total debt service, including interest and principal, is also around $13 billion. For Phase 2, the total NOI or segment EBITDA is approximately $130 million in 2026, $330 million in 2027, and $716 million in 2028. This translates into negative EBIT in 2026, positive EBIT in 2027, and $400 million in 2028. But yeah, the point is the same. That’s how Nebius and similar contracts are structured. During the initial term of the contract, you’re only recovering the depreciation on your GPUs and not making money. You're essentially making money on the asset's terminal value.
Analyst 1
I guess I don't disagree. You're taking the same bet at all of them, and I think the Galaxy camp is pretty nice, but how are you thinking about that terminal value?
Analyst 2
Yeah, so, for example, for companies which are in the colocation space, that in that the terminal value is sort of de-risked in the sense that this is a power data center shell, and you essentially have to believe that if the data center total demand or the compute demand remains at the level where you are right now and doesn't grow at all, you’d still need these powered shells. So from that perspective, actually going into the colocation players and thinking about the terminal value and the assets for that, the buildings, the shells still have a lot of value, and the land also has a lot of value from that. So I’m actually not very worried about that; this asset can’t generate terminal value. The worry is more around the neoclouds, what happens to the depreciation of the GPUs? Will you be able to write it off? But you don't take the GPU depreciation risk in the colocation players.