Anthropic's IPO, IREN’s $3.65B GPU Financing, Strategy Sells BTC, DMG’s 50 MW AI Data Center

Summary

Blockspace: AI & Bitcoin — Ep. “Saylor Sells, DMG 50MW Deal, IREN GPU Financing”

Hosts: Charlie Spears, Colin Harper
Guests: Francis Corvino (Lygos Finance), Sheldon & Steve (DMG Blockchain)
Date: June 1, 2026

---

Episode Overview

In this episode, the Blockspace team breaks down a major week at the intersection of AI, Bitcoin, and digital infrastructure. They discuss Anthropic's confidential IPO filing, MicroStrategy's (Strategy’s) signal-sending Bitcoin sale, the emerging shakeup in Bitcoin treasury preferreds, DMG Blockchain’s notable $50M AI data center deal in Canada, IREN’s landmark $3.65B GPU financing, and the regulatory and political pushback facing US data centers—most notably, a new Utah executive order and the nationalization chatter around AI companies.

---

Key Discussion Segments & Timestamps

1. Anthropic Files for IPO — “Market Rocket Ship”

[00:05–08:51]

• Breaking News: Anthropic has confidentially filed for an IPO, racing OpenAI and SpaceX.
  • “Anthropic is one of like the big three IPOs of Anthropic, OpenAI and SpaceX which are anticipated this summer and later through 2026. In recent Anthropic news... $965 billion valuation, which eclipsed OpenAI’s value for the first time.” — Charlie [02:25]
• Confidential filing interpreted as a competitive move—potentially to conceal less favorable financials versus OpenAI until closer to IPO.
• Revenue Surprises: Anthropic’s reported ARR jumped to $47B from $10B, surprising skeptics of the LLM business model.
  • “That’s been one of the key criticisms... are you actually going to bring in the revenue to justify the compute? And Anthropic coming out and saying, well, actually, yeah, probably.” — Colin [04:52]
• Fast user adoption and a “Claude moment” (in November 2025 with Opus 4.5) marked a turning point for competition with OpenAI.
  • “There absolutely was a Claude moment last winter around November... users saw a huge evolution change.” — Charlie [07:45]
• Healthy competition praised—multiple players hot on Anthropic/OpenAI’s heels: Deepseek, Kimmi, open-source challengers.

---

2. MicroStrategy (Strategy) Sells Bitcoin — “Sending a Message?”

[08:51–15:34]

• Headline: MicroStrategy (Strategy) made a tiny sale: 32 BTC ($2.5M), the first since 2022.
• Hosts question whether the move, long telegraphed by Michael Saylor, is about risk management, market inoculation, or simply testing the process.
  • “He very clearly said, quote, maybe we will sell some to inoculate the market. Like, to me, this is exactly what he said he was going to do.” — Charlie [11:07]
  • “If you think you’re bad at timing tops, don’t feel bad guys. Even the pros with billions of dollars are doing it.” — Colin [10:02]
• Audience and experts debate: Was this sale “meaningful”? Potential for a “gradual tolerance” approach if larger selloffs come.
• Notable Quotes:
  • “I don’t anticipate Saylor selling a significant, material amount unless forced to.” — Charlie [12:18]
  • “Anyone who would have been freaked out about Saylor selling bitcoin already freaked out two weeks ago.” — Charlie [12:18]
  • “Never sell bitcoin. Okay, we’re going to sell a little bitcoin. Okay, we’re selling a medium amount... and then full capitulation. I can imagine these memes in my head.” — Charlie [13:39]
• Twitter Takes: Adam Back speculates on tax loss harvesting — hosts discount the seriousness of that for such a minuscule amount.

---

3. Francis Corvino on Saylor, Stretch vs. SATA, and Treasury Competition

[16:13–31:04]

• Industry Insight:
  Corvino (Lygos) explains this is more “sending a message” than true liability management.
  • “It’s definitely about sending a message... I think it would be way more concerning if Michael Saylor was unwilling to do this.” — Francis [17:01]
• Saylor now experiences “reverse reflexivity”—selling Bitcoin can paradoxically harm both share price and Bitcoin price, unlike prior cycles.
• On Bitcoin Preferreds:
  • Jeff Walton (Strive, SATA) emerging as the “anti-Saylor” and a more relatable public face for Bitcoin risk instruments.
  • “He looks like a sort of normal guy you could have a beer with. Michael Saylor doesn’t present himself well as the type of guy you could go have a beer with. George Bush won a presidency off this.” — Francis [21:04]
  • Stretch is struggling to maintain its peg, while SATA is gaining favor—though in absolute terms, it’s “peanuts to elephants.”
• Instrument Complexity:
  • Simpler structures like SATA may outcompete more complicated ones like Stretch: “As an investor you sometimes do value simplicity, particularly on a product which is supposed to again, act like a catastrophe bond…” — Francis [28:32]
• Memorable Moment: Francis takes a swipe at Saylor’s “shitty AI videos” as part of the PR challenge. [18:35]

---

4. DMG Blockchain’s 50MW AI Data Center Deal in Canada

[31:14–48:23]

• Major Announcement: DMG secures a 12-year, 50MW deal for a single-tenant AI data center at Christina Lake, BC.
  • “Going from tier nothing to tier three is a lot of work... but we were approached by a party that took a look at our design for bitcoin mining and turned around and said, you know, a lot of your design works well for AI compute.” — Sheldon [32:48, 33:18]
• Transitioning Infra: Bitcoin’s robust electrical and power infrastructure can, with upgrades, serve AI—but it’s not turnkey. HVAC, fiber, and redundancy are complex hurdles.
  • “Our existing infrastructure... can essentially stay the same... but we have to add in UPSs, bus bars, things you need for GPUs versus ASIC miners.” — Sheldon [35:59]
• Market Dynamics: There’s a “scramble for megawatts,” and modular, regional sites (sub-100MW) are seen as the next growth area.
  • “These smaller, modular AI data centers are... going to have a huge chunk of this market because there just isn’t enough infrastructure to go around.” — Colin [38:23]
  • “There are companies much bigger than us still trying to get AI deals done... I really think it’s hard to turn a bitcoin mining operation into an AI operation.” — Sheldon [41:37]
• Operational Insight: “You can’t give up 10 minutes of downtime” in AI, unlike Bitcoin. Data center design for AI is far more stringent.

Anthropic IPO Impact:

• Both DMG execs see the IPO as a clear tailwind for AI infra, noting the surprise of Anthropic forecasting profitability.
  • “Anthropic is forecasting profitability, which really shows it, kind of has the naysayers said there’s no profit to be made... Well maybe that’s not necessarily the case.” — Steve [45:34]

---

5. IREN’s $3.65B GPU Financing for Microsoft

[49:25–58:14]

• Deal Structure: IREN (Iron) closes multi-billion financing at ~6% blended rate—well below historical miner rates (which were closer to 10% during the ASIC boom).
  • “The TLDR for the deal is this financing comprises $2.10B in private placement... and another $1.55B delayed draw term loan at SOFR +225.” — Colin [49:25]
• Hedging, prepayments, and tie-ups with Microsoft underpin the high credit rating (A by Finch).
• Major Shift: Bitcoin miners never secured such favorable mainstream lending—sign of the times as AI capex eclipses old mining.
  • “The story that bitcoin miners just could not... crack how to finance your rigs... and yet here we are, the AI HPC run, and lenders love this.” — Charlie [54:21]
• Economic Stakes:
  • Quoting Jensen Huang’s recent bullish prediction:
    • “AI data centers could cost, quote, $80B to $100B per gigawatt in the future.” — Charlie [57:16]
  • Current economics for AI data centers now estimated at $10M–$20M per MW just in raw infra, and $30M–$50M per MW all-in if including GPUs.

---

6. Utah Data Center Backlash & The Politicization of AI Infra

[60:52–75:41]

• Backdrop: Utah’s governor signs an executive order establishing higher environmental standards for data centers following a public uproar over Kevin O’Leary’s 40,000-acre “Stratos” AI campus.
  • “Utahns deserve confidence that water resources, air quality, utility rates, wildlife will be protected. In Utah, environmental stewardship and economic opportunity go hand in hand.” — Governor Cox (quoted) [60:52]
• Key Takeaways:
  • The order is largely symbolic, setting agency guidance rather than blanket bans, but tightens “large data center” from 50,000 to 10,000 square feet.
  • Real battle is between local/county regulations and state/federal executive agency mandates.
    • “This reflects a kind of, I think, disease in the federal bureaucracy where we have moved so much of the legislative process to the executive branch via agencies which we don’t vote for.” — Colin [66:15]
  • Polls show public concern centers on electric and water usage.

Wider Political Context:

• Seize the Means of Compute?
  • Bernie Sanders’ proposal for a mandatory 50% public stake in major AI firms sparks debate.
  • Hosts argue for some kind of “sovereign wealth fund” to avoid permanent wealth stratification from AI, but criticize the approach as “populist economic gobbledygook.”
  • “Keep that communism out of my data center is my final take.” — Colin [75:41]
  • “Seize the means of compute. That’s absolutely going to be a line sometime soon.” — Charlie [75:41]

---

Memorable Quotes

• “If you think you’re bad at timing tops, don’t feel bad guys. Even the pros with billions of dollars are doing it.” — Colin [10:02]
• “He posts these really shitty AI videos. I’m really tired of seeing those AI videos... If I were him, I’d be tired of that level of scrutiny.” — Francis [18:35]
• “It’s peanuts to elephants right now in the grand scheme of size here.” — Francis [24:01]
• “These AI factories… can we just make a petition to just excise that term... Like, a factory assumes you’re building something. I get it, you’re generating compute, but..." — Colin [58:46]
• “Keep that communism out of my data center is my final take.” — Colin [75:41]
• “Seize the means of compute. That’s absolutely going to be a line sometime soon.” — Charlie [75:41]

---

Final Segment Timestamps

• [00:05–08:51] Anthropic IPO and context
• [08:51–15:34] Saylor’s Bitcoin sale
• [16:13–31:04] Francis Corvino on Saylor, preferreds, Stretch vs SATA
• [31:14–48:23] DMG Blockchain’s AI deal deep dive
• [49:25–58:14] IREN’s $3.65B GPU financing for Microsoft
• [60:52–75:41] Utah executive order, political response, sovereign wealth fund debate

---

Key Takeaways

• AI & Markets:
  • Anthropic files for IPO before OpenAI, confirms AI arms race in valuation, scale, and capitalization.
  • Market increasingly views AI infrastructure as a more lucrative, bankable, and lower-risk play than legacy Bitcoin mining.
• Bitcoin Treasuries:
  • Even “never sell” maximalists like Saylor adapt—though in cautious, symbolic steps—affirming that corporate treasury use cases will evolve as instruments, rates, and risks change.
  • New preferred products (SATA vs Stretch) signal intensifying competition post-bull market.
• Infra Pivot:
  • Bitcoin mining’s technical advantages (cheap, robust power) are enabling rapid pivots, but the AI colocation game is far more demanding on redundancy, latency, and vendor relationships.
• Regulation & Politics:
  • Local pushback against “AI gigafactories” may presage a new era of regulatory hurdles in North America.
  • Heavy-handed state interventions (like Utah, or Bernie Sanders’ proposal) are generating fierce debate about how the proceeds and externalities of the AI boom should be distributed.

---

Listenership Note

• Blockspace is departing CoinDesk. Subscribe directly to their feeds for continued, in-depth coverage at the frontier of AI, Bitcoin, and digital infrastructure.

---

Transcript

[Colin]

Foreign. What's up y'?

[Sheldon]

All?

[Colin]

Welcome back to Block Space Live presented by Clean Spark Charlie. It's happening, it's happening. Anthropic is going public. They filed a confidential prospectus right as we were prepping for this show. So we're going to start kick off the show with that. There's not much information to cover because it is confidential but we are going to touch on it. Then we are going to be moving to to the Gigachad doing what he said he would. That is strategy, selling a very small amount of bitcoin potentially to send a message. Following that we have Francis Corvino of Ligos Finance on to talk about this sale and also what's going on with Stretch Vis a vis strives SATA, which is also a preferred stock and how the competition between preferreds within the Bitcoin treasuries landscape is heating up. Following that we have the DMG team on to talk about their new 60 megawatt AI deal in Canada. Looking forward to that one. We had, we, we had the Cathedral team on recently to Talk about Now Sphere 3D to talk about some of these opportunities for smaller AI sites. So I'm really interested to see what DMG has to say on this market. And then we will jump into Iron's new financing deal, multi billion dollar financing for its GPU rollout for its Microsoft contract and then we will end with a little data center hate, or rather I should say guardrails. I shouldn't be too strict here, but Utah's governor just issued an executive order to ensure responsible rollouts of data centers in the state. Obviously this follows a few weeks of kerfuffle over a massive data center spearheaded by Kevin o' Leary in Box Elder County, Utah. Charlie that's right.

[Charlie]

Block Space goes live Monday, Wednesday, Friday at 1pm Eastern featuring quick hits on the latest in AI emerging tech, a little bit of bitcoin and bitcoin mining and markets. Make sure to like and subscribe. Hit the notification bell. If you're on YouTube, get that push notification to your phone. While you are transiting back and forth from work as you race the New York Stock Exchange, hit the big red button to buy that stuff. Anthropic ipo. And if you're still on Coindesk, we are leaving Coindesk very soon. Go over to the Block Space feed. Yes, the Block Space RSS or podcast feed. Wherever you you listen to podcasts and subscribe there because we're leaving Coindesk soon. This show is brought to you by Cleanspark Nasdaq listed ticker CLSK. More on CleanSpark later on the show. Let's kick it off with some Anthropic news. This just drops. So if you are listening later, like this was literally 15 minutes ago pushed on Bloomberg Terminal. I saw it first where Joe Wiesenthal, Bloomberg affiliated, Joe Weisenthal, you know, tweeted it's happening and it is. Let me share this. Here we go. Here's the news on Bloomberg. Anthropic files confidentially for IPO in race with OpenAI anthropic has confidentially submitted draft paperwork for a public listing as it races longtime rival OpenAI to make a Wall street debut as soon as this fall. No details on pricing or number of shares. This is kind of this Anthropic is one of like the big three IPOs of Anthropic, OpenAI and SpaceX which are anticipated this summer and later through 2026. In recent Anthropic news, Anthropic recently raised 65 billion in funding round at a $965 billion valuation which eclipsed OpenAI's value for the first time. So Anthropic on a rocket ship, not a SpaceX rocket ship, a market rocket ship, surpassing OpenAI in valuation. So Anthropic is the first to confidentially file. It is kind of a race to the market, Colin, 100%.

[Colin]

And I wonder if part of that race is why they file confidentially. Right. It's not uncommon. But usually you, I would assume you file confidentially if you're not totally thrilled about the financial metrics you would be reporting. I could be totally wrong here.

[Charlie]

But, but when do financial, when do the financials matter? They, they don't matter anymore.

[Colin]

Well, they, I think they matter with respect to if you file before OpenAI and your numbers are worse than theirs, then that creates a competitive disadvantage for you when both of these things come to market because that's really their primary competitor and that's people are going to weigh if I'm going to pick one, which one am I going to pick? Right. And so perhaps that has something to do with it. That being said, we don't have solid revenue information because it's confidential. But Anthropic did report in May that the, that its revenue run rate and that's, that's ARR. So it's not the specific revenue they booked over a certain quarter had ballooned to 47 billion, up from 10 billion the last time they reported. And this shocked a lot of people. Actually there was a Just a huge surge in revenue and people weren't expecting this because that's been one of the key criticisms of these LLMs is like, are you actually going to be able to bring in the revenue to justify the spend on the compute? And Anthropic coming out and saying, well, actually, yeah, probably, but super exciting to see. And I think that now all eyes are on OpenAI with regards to their own path to public markets. I mean, this is pretty amazing, right? We got the SpaceX IPO. Now XAI is a division of SpaceX, correct?

[Francis Corvino]

Right. It is.

[Charlie]

It isn't.

[Colin]

So you've got the three, maybe we won't say the three leading because Google's Gemini is probably used more than grok. But you know, XAI also has a deal with Anthropic and they're kind of

[Charlie]

playing Both for the Colossus 1 data center.

[Francis Corvino]

Right.

[Colin]

So they're kind of playing a few sides of the, of the game here. Right. So, but you know, I have, I

[Charlie]

have a couple of charts here. One which is, this is from last February from Epic AI, kind of tracking Anthropic's revenue rising much faster than OpenAI. And they anticipated that it would. That they would flip OpenAI's revenue in mid-2026, which, like, what a call, because that was pretty dead on. And then while I was, while I was trying to get this, I was trying to find a chart to show the different revenues or the different valuations between OpenAI and SpaceX. So I had chat GBT generate me a chart of the two of the valuations showing that Anthropic did, in fact, flip this red line here. Flip OpenAI open AI.

[Colin]

Yeah, look at that hockey stick, man.

[Charlie]

Look at that.

[Colin]

I mean, this, this chart really illustrates, I think, what a lot of people who use a number of these tools have felt over the last year. I mean, I remember the first time I was like defaulting to Chad GPT. The first time I used Claude, I was amazed. It was, it, it, it was much faster, the responses were cleaner, it was in more depth. And I just felt like it was a, there was almost like a overnight, you know, leap. Almost like the way that parents will talk about, like, oh, my baby's face looks different today, or like they grew an inch overnight. I mean, seriously, it's like there was this like, whoa, we just got a huge gain here. Seemingly nothing.

[Charlie]

There absolutely was a Claude moment last winter around November when Opus 4. 5 came out and there was absolutely Claude moment. And that's kind of when the users saw There's a huge evolution change, but since then it has been a neck and neck race with OpenAI dropping 5 GPT 5.5. I think GPT 5.6 comes out this week is what's anticipated. And they, they've just kind of been taking, you know, they've been trading places for who's got the best model and it's obviously the hot discourse online. In addition to that, I will just comment that I'm happy that there's two competitors or two horses in this race for the, you know, most for the leading frontier model and then there's multiple horses in the pack behind them. Deepseek, Kimmy, you got some open source models coming behind them. So yeah, very exciting. I'm very glad it's not just one company. We don't have like a late 90s Microsoft situation on our hands rather.

[Colin]

Speaking of micros. Yeah, speaking of micros, formerly known as MicroStrategy Strategy out here doing what they said could never be done. Affecting a little Tucker Carlson cadence there. But that aside, strategy has sold some bitcoin. And Charlie, it was a very small amount of bitcoin and we're going to have Francis Corvino on here in a second to talk about this, but they made their first bitcoin sale since 2022. I don't really recall that sale in 2022, but it makes sense considering bitcoin was super stressed that year, right? Like 80% drawdown from all time high, I think after the FTX fallout. But they filed an 8K on June 1st disclosing they sold 32 Bitcoin worth 2.5 million during the week of May 26th through May 31st for an average price of $77,135 per coin. Now the ironic and funny part about this is they bought like 20 something thousand odd worth bitcoin just like a week ago for 80,000. That's the average price. If you think you're bad at timing, tops. Don't feel bad guys. Even the pros with billions of dollars are doing.

[Charlie]

Is Sailor a pro? Because it's a meme now by how poorly he executes. Because think about it like bitcoin's down to 71.5 right now. So technically Saylor's sell wasn't even that great of a sell yet.

[Colin]

Right? So all that being said, kind of interesting and I just wanted to bring up. A few notes here from the Twitter sphere. This is Kyle Reed head, one of the owners of Milk Road, saying this man bought 850,000 BTC over the last six years. True. He spent more than $60 billion buying this Bitcoin. Also true. Give or take a few billion. Maybe he decides to sell 32 of those Bitcoin for 2.5 million to remove some risk. Now, I think that to be fair, he kind of elucidates what he means here. I don't know how much of this actually reduces risk and is more of a test drive, right? Which maybe is the same thing. But obviously

[Charlie]

he very clearly said, quote, maybe we will sell some to inoculate the market. Like, to me, this is exactly what he said he was going to do without explicitly saying it.

[Colin]

Right. I guess, though, Mike, question is, it's a matter of scale. I would say is why I bring this up because, like, obviously shedding 2.5 million does nothing for your balance sheet risk here. That's. That's not even a drop in the bucket. That's like barely, barely spittle, right? But, you know, to your point, Charlie, maybe there's an argument to made that like, oh, hey, we did it and we can do it again. But there's a much. There's a bigger difference between strategy selling 2.5 million versus 25 million versus 250 million, right? Like, if they start moving up and maybe it's kind of like, you know, building up your tolerance or something or micro dosing, it's like, maybe they'll get to the point where they. They're just like slowly moving up to those higher numbers to get the market ready for it. But I don't Even know if 2.5 million sends a message other than we did this, right? There's not really. No one's gonna freak out about that. But 250 million plus, you know, a larger number might be. Might spook some people.

[Charlie]

I would say that anyone who would have been freaked out about Saylor selling bitcoin already freaked out two weeks ago when he all but explicitly said he was going to do this. So this is just the most gentle way of him just following through and saying, okay, and here's the sale. I. I don't expect this to really change much. I don't anticipate Saylor selling, like, a significant material amount unless forced to. That we've already done. We've already had a few takes on, like, why this was like a pretty intuitive idea. This makes sense. We get it. You know, this, you know, be a smart play. It's. It is a break. Sailor unquestioningly said never do this and that they would never do this. And he is doing this. So I almost imagine, you know, one of those like, almost like 4chan Reddit memes where it's like, never sell bitcoin. Another side bracket which is like, okay, we're going to sell a little bitcoin and then another line, okay, we're selling a medium amount of bitcoin, but it's not going to actually hurt us. And then full capitulation. I can imagine these memes in my head already. And that is, I think what

[Francis Corvino]

could

[Charlie]

scare people is seeing the trajectory of like rolling back statements. So now when Saylor says we will never do something, do you believe him?

[Colin]

Yeah, probably not. I mean, do as I say, not as I do. Right. And I think that anyone who believed like, never, like they're never going to sell their bitcoin was probably a little delusional anyway because, like, why would you ever. Why? This is the, like, this is the bulk of your balance sheet. Like, not in saying as the bulk isn't even like doing it justice. It is your balance sheet. And so they're going to leverage it at some point. It makes sense. And I think that anyone who was looking at, especially the preferreds, knew that this was going to happen eventually. We have a few more takes here on. On Twitter, Blockstream CEO Adam Back says in response to why would he do this? A tweet from George Gammon says, tax loss harvesting. I mean, no, like 2.5 million dollar law. You know what I mean? Like, that's nice.

[Charlie]

Oh my God, Adam, this is the, this is such a dumb take. Nobody can imagine tax harvesting the.001% of your stack.

[Colin]

Yeah. And maybe to be fair to him, he's talking about like selling in larger volume. Right. And then Lynn Alden chimes in here, says, yeah, I believe on a recent learnings call he said he was going to sell maybe something you can do to service liabilities and such. But again, this does not. This, this is not a meaningful enough amount to actually do much with it. I mean, you know, you can put that 2.5 million to the, to the preferreds, but that's like what, like a. Not even a day's worth of having to pay out what they need to pay out for these preferreds.

[Charlie]

Yeah.

[Sheldon]

Right.

[Colin]

So ultimately it's, it's really just an extremely small amount. And maybe it was just a test drive just to be like, okay, here we, here's the process. We did this and like, like you were saying, signaling to the market.

[Charlie]

Look at that take Underneath from account cure 8 standing 5033, quote, who tax lost harvests in June.

[Colin]

I mean, I don't know what I can. I don't know what to tell you. Cure eight, like, you could tax, loss, harvest at any point and maybe bitcoin is hopefully ripping by November or December and you don't have the opportunity.

[Charlie]

What about.

[Colin]

But even then they had buys up to like 120,000.

[Steve]

So, like.

[Charlie]

Yeah, yeah, I know a lot of room here.

[Colin]

All right, well, we'll table that.

[Charlie]

I'm looking forward to being able to tax gain, harvest, you know what I'm saying?

[Colin]

We're all waiting for it, brother.

[Charlie]

All right.

[Colin]

But with that out of the way, we have Francis Corvino of Lygos on to unpack this news and also talk a little bit about the preferred landscape with Stretch trading under a hundred dollars

[Charlie]

again this week on up. Francis, welcome back to the show.

[Francis Corvino]

How are you?

[Charlie]

Fantastic.

[Francis Corvino]

Awesome. Awesome.

[Colin]

All right, Francis. So square, square circle for us here. Everyone's talking about, you know, sailor doing this for liability, seller doing this to, like, show the market's not going to get spooked. It was 2.5 million. I mean, what do you take for why it was so small this time? Do you actually think it has the effect that the market can see they can do this and no one will get freaked out? Or do you think this is just, you know, like the Joker from A Dark Knight? It's not about making money, it's about sending a message.

[Francis Corvino]

Oh, no, I mean, I think it's definitely about, like, sending a message. Right. He has levers ultimately, at the end of the day, in terms of how he manages his treasury versus the bitcoin that he has and increasing his USD treasury and, like, we would expect him to use them. So I don't think this is surprising at all. I think it would be, like, way, way more concerning if Michael Saylor was unwilling to do this. If Saylor is willing to slowly sell his bitcoin, there is sort of this unfortunate reverse reflexivity here where he used to be able to issue bitcoin, the share price would go up, or, sorry, he used to be able to issue equity, he'd buy more bitcoin and the share price would go up. Now it's sort of the opposite, right? Where Michael Saylor sells bitcoin, his share price goes down more than the price of bitcoin goes down, and he pushes the price of bitcoin down even more as a result of selling the bitcoin. So being a market mover has its advantages and its disadvantages. But I think in this case, Saylor's kind of learning some of the downsides of the reverse reflexivity of being able to use your equity in such a way.

[Colin]

Just to double tap that question, though, Francis, because one of the things I wonder, it's like, clearly it's selling a message, but in terms of showing how the market will really, truly react to this, I feel like 2.5 million is much different than if you were selling multiples. More than that. Like, what would we be seeing right now? If it's 250? Is this, like, I used in the segment, the analogy of, like, micro dosing? Is he, like, trying to build the market's tolerance to larger sales? Maybe start with a small one?

[Francis Corvino]

I don't think so. I think ultimately he's just trying to show that he is not too stubborn to sell a portion of his bitcoin so that the wider market can, like, believe in him. He's using, you know, a very small percentage of his overall, overall treasury that he's selling here. So I don't think. I don't think it's worth being too concerned at this stage. Like, it just seems like a normal. Honestly, the market is like beating this to death, right? Like. And that's why he's doing it, because ultimately Saylor is not great from a PR standpoint in many capacities. Right. He posts these, like, really shitty AI videos. I'm really tired of seeing those AI videos. I think most people in the industry are. He sells like a little 21 Bitcoin. Excuse me? He sells 21 Bitcoin and we all have to hear about it for a long period of time. If I were him, I would be tired of that level of scrutiny. Ultimately, what it comes down to, I think, is something just much larger and more philosophical in that bitcoin has completely lost all of the people that it looks to. In order to fill the news cycles, there was a period of time where you had Trace Mayer, who was the digital gold guy. He was a foundation of what people look to as a bitcoin financial mind, right? And we could look at him and we could look at the cycles and we could move things forward. Now we're in an era where we really, really just predominantly have Michael Saylor. He is the largest fixture of bitcoin and he's the thing, he's the person where something's happening. You know, Jack Dorsey is still a big deal in the bitcoin community. He's still active, he's still doing stuff, but he's not making Headlines every two weeks. Like Michael Saylor is. So we're sort of forced to really, really pay attention to Michael Saylor and he is not a person that even obviously likes to be in this position, you know, So I, I think that this is probably just a way for him to alleviate a little bit of the scrutiny of this and also to take advantage of the, the market cycle. Because people are talking about it, right? Like this is a bit of a nothing burger in some senses, but people are talking.

[Colin]

Yeah, that's kind of part of my point is I feel like this, you know, doesn't even like if the point is to show that you're like, okay with reducing risk, you're showing that you're selling it, but like it's such a small amount that it doesn't matter. And so I don't even know if it's like when people are saying like, oh, this is sending a signal, you know, like this is, you know, the market might react poorly to this. It's like it's peanuts in comparison.

[Francis Corvino]

And you can look at how his competition is sort of like beginning to position themselves as well.

[Colin]

Right.

[Francis Corvino]

There's a couple different ways in which I think Jeff Walton is trying to position himself as the anti Michael Saylor. Right. I personally think that Michael Saylor is a diligent, practiced risk manager who's run a public company for a very long time and is actually more of the traditional finance world than the bitcoin world. But at the same time he projects forward again this AI slop buy bitcoin cyber hornets mentality. And that's what the public sees. And I think Jeff Walton has seen this as an opportunity to go be good at social media, not AI slop videos. He goes on one of the largest financial investigation YouTube channels in the world in Coffeezilla, and he makes a strong pro principled argument about sort of like, you know, comparing this type of instrument to the insurance industry and just taking a very principled risk first approach. He's very likable. He looks like a sort of normal guy that you could go have a beer with. Michael Saylor doesn't present himself well as the type of guy that you could go have a beer with. George Bush won a presidency off this. Right. So I do think that we're starting to see for the first time Michael Saylor have a little bit of competition in terms of like, who is the focal point, because again, we have a growth here and it would be really exciting to see someone compete with microstrategy in a big way. And I think that that's a bigger story than just the 21 bitcoin.

[Charlie]

Yeah. I can't imagine what's. The only sailor reference in the Epstein files is where he's, like, just talking about his yacht and going to cans and being, like, boring dude at a party.

[Francis Corvino]

Yes, exactly, exactly. Like, there is a lack of relatability there. And he's still been able to create this just massive character in the bitcoin community with these, like, very corny cyber hornets quotes.

[Colin]

Right.

[Francis Corvino]

He literally is described in a leaked document as being so insufferable that a person who is paid to improve your life.

[Colin]

Creepy.

[Francis Corvino]

Possibly. Possibly improve your lifestyle.

[Colin]

Epstein, Sailor. Creepy. No, no, it wasn't Epstein. It was one of Epstein's publishers. Publishers, but a publisher who was close to him. This woman sailor's been following me around this gala all day. He's a creep. He sponsored it anyway. Brutal. We won't go into that, but I want to double tap on Jeff for a second. So he's. He's. He's at Strive. He what. What's his title? Chief Risk Officer at Strive. And so I wanted to kind of tie this back into Strive and their preferred SATA, because there's this. There are rumblings that maybe SATA is sucking market share away from Stretch because Stretch has kind of had trouble maintaining its peg. Do you think that's valid? Do you think that we're. You talked about it on a few segments back when we had you on about the digital credit landscape, as they call it, is really starting to heat up with now you have, you know, the front runner in Stretch. But this SATA Preferred has become very attractive to some of the retail holders who love these instruments for the income. So do you think there's some merit to the fact that maybe SATA is giving Stretch a run for its money right now?

[Francis Corvino]

I mean, we're talking like peanuts to elephants right now in the grand scheme of size here. I think Sada's treasuries like 500 million, and then Saylor's had, like, close to 8. 8, bill. So we're not in the same ballpark quite yet, but in terms of, like, real innovative competition. Absolutely. I think that there's really four key components here. One is you have, again, like, don't want to too much beat this one to death, but, like, there's that backwards reflexivity where the market is used to Michael Saylor being able to issue equity to buy Bitcoin, which is cheaper than his equity, which pushes the price of bitcoin up, which pushes the price of his equity up even more. So he can just keep this cycle going. Now Michael Saylor is faced with the conundrum of being on the downside of that again, where he has to sell Bitcoin because the bitcoin is valued at more than the equity that he owns. So he has to sort of do the worst, the opposite thing where he has to sell his Bitcoin, which pushes the value of his equity down more than the price of Bitcoin. So getting them, you know, farther away from being able to take advantage of the cycle in the right direction. So I think Saylor's sort of lost the tides of the market to some extent. The capital stack is also much more complicated when working with Stretch than with SATA. There is no debt that is senior to you at SATA. It is just SATA. That's all that's there if there was ever a sort of catastrophe. Because ultimately these things work like catastrophe bomb bonds, right? Like if Bitcoin is unable to increase in value at anything like 13.85% or so per year for SATA and like 11% for. For stretch, then the overall value of the equity will diminish to zero because it can't keep up with payments and it'll go to zero. But our expectation, right, is that, you know, Bitcoin is going to appreciate at a greater than that rate. And the dividend that are being paid by Stretch and being paid by SATA are equal payment for the risk of Bitcoin not appreciating that amount on average over the next eight to ten years or so. These are catastrophe bonds. And I would rather have a simple catastrophe bond that pays me every single day and has a very simple debt structure to it and has a person at the head of it whose decision making I can put myself in the shoes of versus just much more complicated, complicated instrument in Stretch with distributions which happen 14 times less frequently, has a significantly more complicated cap table and has not advertised the type of risk management that I may want in a catastrophe bond like product.

[Colin]

So. Sorry, Charlie, I've got a. Unless you have a. I have a closing question.

[Charlie]

I, I was just. This might be a little bit of a. I know. Do you. 15 seconds. Francis, have you seen this? How do we resolve the poly market conversation around strategy selling Bitcoin? Okay, maybe, maybe this is another.

[Steve]

We'll.

[Charlie]

We'll circle back on this later. But I think.

[Colin]

But apparently like the oracle for it's messed up like they sold and the. The market hasn't resolved yeah, so the market needs.

[Charlie]

Yeah, it's like within the challenge window

[Francis Corvino]

for the poly market.

[Charlie]

Anyway. This is another thing we probably, we could circle.

[Sheldon]

Yeah.

[Francis Corvino]

So the problem is truly kind of unbelievable because ultimately every single financial product that ever exists had to elect back to someone reporting an event. There is literally no way around that. So yeah, the Oracle problem will show its face. No matter what kind of financial contract it is, there's always going to be an Oracle problem. That's my take on that.

[Colin]

Francis, one last question here. I want to get this up. We had a segment last week where, where we covered this tweet from Jeff Dorman of arca, basically making the argument that, you know, the wheels, not the wheels are falling off. But he says here that strategy story has gotten out of hand and he makes the argument here that sailors should have just stuck with issuing common stock, issuing converts, buying bitcoin, and not get into all of this preferred nonsense. You kind of mentioned that strategy seems to be losing some of its foothold with, with a competitor like Strive obviously stole the, the lion's share of the market. But what do you make of this? Do you agree with what Jeff is saying here? Do you think strategy kind of over engineered this thing or.

[Francis Corvino]

I mean, it's definitely, it's definitely fancy. Right? Like I think for every additional percent that stretch has to increase the dividend BY it's like $100 million per year in payments. Like that's, that's, that's significant. Yeah. And when you start to have sort of all of these reflexive instruments around that as an investor you're unclear how he, like it's a bit surprising, right, that, that maybe he cleared away all the convertibles which were at 0% while still having outstanding debt that he has to pay monthly at, you know, around like 11%. So yeah, I do find it as an investor you sometimes do value simplicity, particularly on a product which is supposed to again, act like a catastrophe bond for a worst case scenario. I want to understand what the worst case scenario looks like. If there are all these levers that are available to the person who I'm entrusting with the potential downside risk here it becomes very challenging to want to put a lot of my money into that if I can't understand it.

[Colin]

Yeah, maybe a little more simplicity and fewer slideshows with rocket boosters about the strategy capital.

[Charlie]

How is bitcoin supposed to go up if you don't put rocket boosters on the slides? Colin, SpaceX IPOs around the corner.

[Francis Corvino]

Well, if I could just say with the last thing. And I really want to just harp on this Jeff Walton guy because I'm so impressed and I'm so excited to see just like a new face in the bitcoin community. And I really hope that Jeff Walton will come on this show because I think he would be great here. Jeff, I saw that you have a Chambers Bay little poster in the back of your room. We'll take you to Chambers Bay on Lygos if you come on this on block space. But yeah, I'm really excited. He's a new face. He went on Coffeezilla, which is just a massive agency or a massive YouTube channel that goes far beyond just the bitcoin and cryptocurrency communities that will actually grow and show the sort of non bitcoin world that we continue to be serious. Right. And I'm really excited to see that rather than sort of these like rocket

[Colin]

AI baloney, it would be really, really nice to have some more mature discussion around.

[Charlie]

Yeah, he made coffee zill look dumb. I'm excited to get a Jeff Walton. Maybe, you know, if you want some old Jeff Walton lore, maybe you could commit to becoming a lazy lion. His OG NFT collection that everybody totally forgot about that I'm still keeping the receipts for. So Francis, thank you, producer.

[Francis Corvino]

We're cutting that.

[Charlie]

Yeah. Anyway, I'm a huge fan. Francis, thanks for coming on the show. We'll see you again very soon.

[Francis Corvino]

See you guys. Bye.

[Colin]

See you, Francis.

[Charlie]

All right, we have next Sheldon and Steve from dmg Blockchain CEO and COO talking about the big deal they just announced. But before that, a word from our sponsor, CleanSpark.

[Colin]

We are CleanSpark, America's Bitcoin miner. A publicly traded company with the largest operating hash rate powered entirely by self operated infrastructure across four states. This is our proof of work. We are setting the standard for what's next. Learn more about the intersection of energy and [e-mail] all right, before we bring on Sheldon and Steve, quick spark notes for this deal. 50 megawatts of critical IT load for DMG for an AI data center colocation service to a single tenant at its Christina Lake site in British Columbia, Canada. 12 year contract deal, first phase targeted by the end of the year sprinting into the contract. Love to hear more about that. Let's get the fellas up on the stage and talk about another AI deal.

[Charlie]

All right, bringing you guys up here. Sheldon, Steve, welcome to the show.

[Sheldon]

Well, thank you for having us.

[Steve]

Thank you.

[Colin]

Yeah, thanks so much for coming On Last Minute. Really appreciate it. You know, this hit the wire for us this morning. Y' all made time, really appreciate y' all coming on to talk about it. So set the stage for our listeners. Can you just give us a little more background about this deal and how y' all landed it?

[Sheldon]

Yeah, well, I don't know if there's any magics to landing it. Just a lot of work, you know. DMG for your viewers, if you haven't heard from us, we are a Canadian company, one of the first bitcoin mining companies we've been operating almost close to a decade. It's our 10 year anniversary as a company. But on this site, Christina lake, it's about 500 meters from the US border, about two and a half hours north of Spokane. And we've been operating a bitcoin mine there for eight years or so. Obviously with what's going on with AI started taking a look about a year ago at converting this site into an AI data center. Going from tier nothing to tier three is a lot of work to do. But we were approached by a party that took a look at our design for bitcoin mining and turned around and said, you know, a lot of your design works well the way we originally put it in for bitcoin mining with our design of how we would run our AI computer from a power point of view. And they're still cooling and networking, which is why we sort of looked at it and said, you know, we might be able to get a portion of this 50 megawatts up and running this calendar year, which is part of what we're working towards to get the chicken the egg. Right. First we have to get the definitive agreement. But you know, we've already started working on the definitive agreement. So it's really just about how fast can we engineer, procure and install.

[Charlie]

So.

[Colin]

Oh, go ahead, go ahead.

[Steve]

I mean, we announced in December time frame that we were gonna start on this journey and here we are six months later. So this, in terms of the amount of time, as Sheldon was saying, it wasn't magic. It didn't just necessarily appear. We did a lot of work. We've done a lot of prep and a lot of outreach, outreach to potential clients, outreach to the supply chain, really lining things up. So we have the possibility of being able to do this deal. And as you know, there are specific windows that the off take, call it the offtake community, Neo Clouds hyperscalers are targeting. We are able to fit into this window that they're looking for. So in this case the stars align for us to be able to get this deal.

[Colin]

So Sheldon, you mentioned a lot goes into going from tier one to tier three for a data center. And end of year 2026 seems like a very aggressive timeline for the phase one. I mean I don't know how many megawatts that is. Right. It's obviously not the whole whole site. But I'm just curious, y' all are, you know, sprinting into that phase one. It's did was it just like you said, kind of ready made and it was a good fit already. Did you already have the groundwork laid for making this transition? How are you going to speed

[Sheldon]

our existing infrastructure? It's around 35ish megawatts of power infrastructure that can essentially stay the same, you know, in the building. We have to change it around. We have to add in the UPS as we have to put in the bus bar, you know, things that you need for GPUs instead of ASIC miners. We already have 12 megawatts of cooling that we had previously purchased for hydro cooling that we haven't turned on. So it's just sitting there. So we have equipment that, you know, the lead at time for those can be quite long. So we were just lucky that we had this stuff, you know, originally for bitcoin mining. But it can be quickly repurposed into servicing GPUs. We already have a fiber line we'd already started a while ago a second fiber line discussion and timelines for that. So you know, two fiber lines are very important to have. The timelines to get those in are sort of and upgrade them to the speeds we need are coming in around the end of this year. So those kind of line up. So we have cooling available, we have power available. Our fiber upgrades seem to be on track, but the redundant power side isn't there. So we don't have a way to do redundant power. We can just rely on UPS as we add until we add in our own self generation on natural gas. We actually have a large transmission line of natural gas there. We've been working with a utility to get access to that which they've been telling us we can have that access. And so we're able and we won't have this done by by the end of this year, but we're able to add on around 50 megawatts of natural gas generation as our backup power. And to do that with our 50 megawatts of power, we have their utility. We actually have 75 megawatts of our utility to be able to add on 50 megawatts of natural gas for backup. Makes this site really interesting. Our off taker, we obviously disclosed that we won't have the natural gas up. But they're willing to work on the fact that with UPS's and the fact that our transmission line is rarely ever down for some type of emergency. The statistics are very good on a transmission line. They're willing to wait until we could get that gas generation up for the backup power.

[Colin]

So I think when a lot of listeners and just people watching this space, when they imagine these AI data centers, their immediate mind goes to some of these like, like an open, like this Project Stargate for example, right? These massive Hundreds of megawatts gigawatt scale. But we had Thomas Arrow on of Sphere 3D a while back and he was saying you got to watch this space because these smaller modular AI data centers are going to have a huge chunk of this market because there just straight up is not enough infrastructure to go around. You kind of mentioned that Steve. These Neo clouds are really just to make sure that they can meet their obligations. They're trying to make sure that they can get every single megawatt they can find. So can you speak to some of the opportunities for one of these sub 100 megawatt sites and why specifically these tenants might be looking at some of these smaller sites rather than going out and trying to get a big chunk like with Project Stargate or something like that?

[Steve]

Sure. Well when you look at kind of what the market wants, they're the large hyperscalers are going to want just huge amounts of power for these large training models, for these frontier models. For a lot of what the Neo clouds are looking towards is somewhat different applications. The applications are evolving and this 50 to 75 megawatt range that our site falls in is really everything that we've learned about the market is a sweet spot for what these Neo clouds are looking for. Ultimately as the market evolves you're going to have more inference, you're going to have the desire to have more sites spread out with smaller amounts of power that have low latency to those regions. If you take out things like high frequency trading where you want it essentially right near the actual trading site, we fit in nicely being let's say sub 5 milliseconds of latency to a large region in the position Pacific Northwest. And so we think this is how the market is going to evolve. One of the tracks that it's going to evolve. And so we think this site and other sites we're looking develop this is really going to fulfill a need that the mark where the market is going.

[Charlie]

So you guys. So the current tenant is under NDA so you can't disclose it. So rather than press you guys about that, I invite you to kind of like talk broadly about this trend in the industry. Because I'm seeing this a lot, a lot of announcements we have with, you know, hyperscaler or tenant can't disclose who they are. This is happening more and more and it, to me it signifies like this is an incredibly competitive landscape. Can you speak to just this trend in the industry overall and like why we're seeing it and yeah, yeah, I

[Sheldon]

think it is very competitive. When you look at the amount of power needed for what people are saying is needed in us, Canada, Europe, it doesn't matter where you look and you look at the infrastructure, it's obviously natural that bitcoin miners are moving to AI because we have large amounts of power. We also moved to AI because the return on that is better than bitcoin. And this is a guy doing Bitcoin for 15 years saying that. And we've never said we're going to stop bitcoin. We're just going to prioritize AI. But competitive wise, yes. There are companies much bigger than us still trying to get AI deals done, whether they're lois independent agreements. I really think that it's hard to turn a bitcoin mining operation into an AI operation. And there are so many parts that we had to go through in our company and see if we could tick these boxes. You know, nobody thinks about, you know, earthquakes. Nobody thinks about, you know, floodplains. Nobody thinks about the redundancy like we have to have. When you, when you're a bitcoin miner and you lose power, you lose the opportunity to make money for 10 minutes. And then when your power's back on, it comes back, but you're only losing 10 minutes of potential revenue. And so when you think about it that way, this is why we're so great to curtail. But we're horrible at AI. We can't be curtailed. We can't give up 10 minutes of operations when we're working with a third party that has a service level agreement with us. When we got into it, we found on the fiber side, like our ISPs don't even have the backbone required for where we need to be. So we're paying what will be a very large sum, just upgrade their backbone so that they can take this traffic the way that we need to. So a Lot of communities out there that may have legacy power that can become an AI site. It doesn't mean they have the other parts. And that's a really big problem is trying to find all the other parts. Power is so important, but your redundancy of power is even as important as the primary power. And then all the fiber and interconnection is super important. And then you get into the whole workforce part. We're in a remote area getting 60 or 80 electricians. This is not a trivial thing to do. There's a lot of work to do that we've spent the time to find the companies that can do that. We're lucky there's a lot of industrial work in the area. But I don't think that any bitcoin miner moving to AI is going to say this is easy. And I think it's really. How prepared are you to going into these conversations with an off taker? Because it's a different conversation than bitcoin mining where it's a very simple conversation. It becomes a very complex conversation about a lot of stuff we've never even heard or thought about before.

[Colin]

Yeah, it's incredible. I was going to use the analogy where you almost have to get like a hat trick or a grand slam for the right site. But it seems like, you know, you can even go up from 3, 4 to like 5, 6, 7, 8, 9, 10 things that you have to make sure are right. I mean you just listed a bunch of them there and I think it's a lot. It's easy to be an onlooker and see these deals and just think like you're kind of saying, Steve, this doesn't just magically happen, right? Like there's all this background work that has to be done. Whereas you know, you just read the press release and it's like, oh, there's another deal. But the scope and scale is just massive compared to bitcoin mining. Charlie, I'll toss it to you.

[Charlie]

Sorry, yeah, speaking of press releases, as Colin said, I'm gonna throw a curveball to you guys. Anthropic just filed for a confidential ipo. I don't know if you saw this, but I was wondering if I could get your guys takes on Anthropic, the makers of Claude going public. What does this mean for the industry? What are your takes?

[Steve]

I would say this is, this is great for the industry that I think what's even more significant is Anthropic is forecasting profitability, which really shows it kind of has the naysayers said there's no profit to be made. People are just going to be building all this infrastructure and not be able to generate enough revenue. We're seeing. Well, maybe that's not necessarily the case. Of course the amount of capital raising is significant, but as we look going forward, we have just kind of seen the beginning with clog code and the need, which essentially mirrors what the OpenAI moment we had a few years ago. I think we'll see more of these where all of a sudden work that was done in one way is going to be transformed. And it just says that there's probably a good bit of tailwind to this. What. What is happening in the market for a while. I mean everyone says this is a bubble or I shouldn't say everyone, but there's obviously talk and concern about that when we look at kind of where things are at and just it just still seems early. Obviously predictions are really hard, especially about the future, as Yogi Berra once said. And so don't worry. Just like trying to predict the price of bitcoin. Sheldon and I gave that up a long time ago. But it just feels that there's a lot going on here and a lot of just good stuff that's going to happen over the next few years with the market really constraining in terms of power availability and of course lead times on a lot of the equipment to build these data centers as well. So we just think this is. It's a good sign with the anthropic ipo and it's just kind of. I think there's still some tailwind here are.

[Colin]

The profitability angle is really, really key here I think too. You know, as you said, Sheldon or Steve. Excuse me. That's been one of the key criticisms against these, these LLMs is that there aren't enough customers yet and the cost of the compute is too high and anthropic proving that, you know, maybe not, but something to keep an eye on. And we'll also be looking out for open AIs running a public market here as well. Sheldon, Steve, thank you all so much for coming on. We'll no doubt have to get you all back on for an update as this site comes to be closer online. But best of luck and I hope you all have a good week.

[Steve]

Thanks for having us.

[Sheldon]

Have a pleasure.

[Steve]

Thank you, gentlemen.

[Francis Corvino]

Thank you.

[Charlie]

Thanks, guys. All right. I always love to talk to the people who are actually building stuff.

[Colin]

Yeah. And just to highlight, you know, DMG has been around for a long time. I believe they were one of I could be wrong about this, but they were one of the first publicly traded bitcoin miners as well. So they.

[Charlie]

15. Yeah, 10. 15 years. That's. That's like 5, 5, 10 years basically for this.

[Colin]

And also a reminder that before there was the great Chinese hash rate migration following the China mining ban, Canada was the foremost bitcoin mining mecca in North America. Roots go back deep.

[Charlie]

Some might say it still is. Anyway, all right, we're going to keep rolling. We're going to go to iron after this and then the Utah governor executive order. But before that, we have a word from our sponsor, Luxor.

[Colin]

This episode is brought to you by Luxor's commander bitcoin miner management software for enterprise operation. Commander gives you real time fleet monitoring, bulk remote commands across your entire fleet. And intelligent miner, that's an automated profitability engine that runs every five minutes and adjusts your power settings to live hash rate and energy markets. ERCOT back tests show 10% increased profitability with intelligent mining versus old fashioned binary mining. Commander Pro is $100 per megawatt or a 0.25% pool fee adder. Roughly half the cost of competition and you can try it out for 60 days. So if you want to get started, go to Luxor tech forward slash commander. All right, back in. Back in the AI news, we're. I'll get this one up here, Charlie. Yeah, yet another one. Another public miner putting bitcoin to the side. Maybe not totally, but moving into AI Iron closes a $3.65 billion GPU financing round to back its Microsoft AI Cloud contract. There are a lot of details in this one. A lot of financial games going on. And the, the TLDR for the deal is this financing comprises of $2.10 billion in a U. S in a private placement with a fixed rate plus at a fixed rate of 2.3 plus SOFR and then another 1.55 billion delayed draw term loan at the floating rate of SOFR plus 225. So iron is also hedging that second one, the delayed draw term loan, which is like a loan as I understand it, paid in installments rather than lump sum. But they're hedging that with an interest rate swap. So the way this works is iron goes to a bank, says that they're going to pay a fixed rate, a fixed interest rate to the bank that is higher than the terms of the current loan and this gives them protection on the downside. Now, if the floating rate for SOFR goes down, irn is obviously overpaying for this loan. But if it goes up, they're protected on the upside there so they don't get scalped by a surge in interest rates. They said the all in rate is roughly 6%. So you know, I mean if we like look back at ASIC financing deals because this is a GPU backed finance deal, so this is equipment financing. If we look back at ASIC financing when interest rates were low, much lower than they are now, Miners would be lucky to get in under 10%. And all of those loans blew up, funnily enough actually not to throw shade or anything. Iron actually had one with an IDIG back in 2022 that went belly up.

[Charlie]

And they survived to, they did survive it obviously.

[Colin]

Right. But you know, they took out this loan at like the market peak in 2021 when S19s were going for like $10,000 a rig. And the value of those S19s plummeted into the bear market and Nydig ended up I believe taking control of most of those rigs as Iron more or less defaulted on the loan there, but swimming in much different waters. Now this just goes to show, going back to what we've said on the show a number of times, these miners are in the big boy leagues now in terms of how they're rated for their credit and how they're seen. And on that note, Iron claims that this is the highest publicly rated investment grade GPU financing announced in the first GPU financing in the US private placement market. I fact checked that and it seems mostly true, maybe with a few caveats. Core we've had a GPU financing deal, I think that got downrated to like B a while ago. But this has received an A from Finch and a low A from dbrs. So overall rating agencies seem to like it. And they also. One point of clarity before I toss it to you Charlie, they mentioned in this, in this release that like they're all in rate for financing. This GPU route for, for Microsoft is like 3.31%. Now what they're basically saying there is this debt facilities or these debt facilities are for 3.65 billion. They also have prepayments from Microsoft that will make up most of the rest of their, their capex burden for these GPUs. And they're, they're saying that that's basically 0% financing. So when you see that 3.31 we'll say, you know, obviously the prepayments for Microsoft are great, but that also means that they're not going to see revenue from that or a certain amount of revenue will be lost from them because it's Prepaid. So the 3.31% is reflecting that is basically saying that's 0% financing and they still have a few million dollars worth of equipment to buy before they can get the full amount of GPUs that they need for this Microsoft build.

[Charlie]

Yeah. To me the most interesting part of this is maybe not that it's 3.3656 million, but rather just the financing side of it, because this is the story that bitcoin miners just could not. This is the nut. Bitcoin miners could not crack how to finance your rigs. You didn't even have the trilemma, you just couldn't get a good rate. It was really difficult to get financing from anyone legitimate and reputable. And it was hard to do it in a way where it was timely. And yet here we are, the AIHPC run. And lenders love this or lenders prefer it much more. And I think this, this particular deal, like this GPU collateral deal, if I'm understanding it right, the fact that it's like US private placement, gives it like a lot more legitimacy. And having Microsoft be kind of like Iron's like shadow Bank, I think. I mean, I'm assuming that's probably the reason why it got an A rating so.

[Colin]

Well, yeah, one of the largest tech companies in the world, hyperscaler scale, sometimes

[Charlie]

the biggest company in the world depending on how you measure it and pay on the time of day. So yeah, so like to me it's the fact that capital is just unlocked for this capital and financing is unlocked so the industry can really grow. I would, I really do wonder, Colin, what the bitcoin mining run of the past five to six years would have looked like had we been able to have like reliable banking, reliable financing and like reliable.

[Colin]

I can tell you it would have been worse. Hash price would have gotten crushed. Because like, you know, my, my take on that is with the public miners, especially coming to the US you had, I would call it almost like artificial hash rate growth. And what I mean by that is a lot of these public miners, there's one in particular I'm thinking about and we won't just name them straight out, but they were able to just sell their equity into the open market and then use that to cover their operating costs and their capex. Regardless of the actual profitability of those plans at the time, these miners would just like the big ones would lose millions of dollars a quarter, you know, millions of dollars Every year. And their access to public markets and equity financing, really, I think, was one of the large reasons why we saw hash rate just go absolutely ballistic following the China mining ban. Obviously, hash price was really high after the ban because difficulty dropped. We had a hash price bull market amid the bitcoin bull market. But these guys were expanding aggressively in the bear market. And the reason they could do that is they could kind of just ignore the actual dollars and cents of what they were doing, because they could always tap the public market. Now, not always, but if you looked at what they, you know, their ATMs and things, they would sell hundreds of millions of dollars into them. So I think it would have been.

[Charlie]

Yeah. The other take I want to have here is because, like, the story this weekend was Nvidia. Nvidia, however you pronounce it from Jensen Huang's presentation over the weekend. There's a girl from Jensen Huang where He says that AI data centers could cost, quote, 80 billion to 100 billion per gigawatt in the future. So that would roughly map onto 60 to 100 million per megawatt. And if my vibe check on the industry right now is at all in the right ballpark, I think 10 to 20 million per megawatt is. Is like not too far off.

[Colin]

It depends on what you're measuring.

[Charlie]

It depends on what you're measuring.

[Colin]

Are you just doing the build? Because 10 to 15 is probably right. If you're doing the GPUs and the build, it's several. It's. It's much higher than that. It's probably somewhere between 30 and 50.

[Charlie]

Exactly. And this is just kind of how do you value power and land and assets? But basically, by whatever metric Jensen's referencing this, that would be, you know, a 10x from where we are currently, roughly. Which is significant. And I just think anyone in the game right now is incredibly well positioned, you know, so if. If you're iron or any of these other neoclouds or, you know, AI factories, if you will, can we.

[Colin]

Can we get. Can we. Can we, like, make a petition nomenclature. Can we just make a petition to just excise that term, like AI factory? I don't know why it bothers me so much. Yeah, but like, a factory assumes that you're building something. I get it. You're generating compute. But like, I don't know, you have all of these buzzwords like compute as a commodity, AI gigafactories.

[Charlie]

Like, see, Colin, this is why. This is why you'll never. This is why you won't raise you know, a trillion dollars on a $100 trillion net worth from just a single slideshow. Because you got a chop.

[Colin]

Also why I'm a poor talking head and not making money in pr because I refuse to use these terms.

[Charlie]

You just gotta go gig along on compute on Orin futures and ice. Okay, I think we probably need to wrap up this story and go on to. I don't know if there's a cry corner or just your data center hate corner.

[Colin]

No, it's actually a. It's actually a to reel myself in. You would expect this if we're going to have this massive shift, but we'll get to that.

[Charlie]

But first, but first a word from our sponsor. Lygos.

[Colin]

Hedge funds are getting liquidated. Is your Bitcoin safe? It's not just Bitcoin's price drying up. Big whales, lending desks, etc are reeling after the October 10, 2025 and February 5, 2026 liquidation events. Counterparty risk is rampant. So it's more important than ever to know who actually controls your Bitcoin. And with Lygos Finance, that is always you. Lygos Finance is Block Space's preferred Bitcoin lender. They use Bitcoin data smart contracts to make sure you're always in control of your Bitcoin and never relinquishing your keys. Don't become the next FTX or Celsius victim. Hold your keys. No wrapping, no bridging, no rehabothecation. Just competitive rates as low as 10% APR. Go to Lygos Finance to learn more.

[Charlie]

All right, our last story is Surprise. It's also about data centers. But this time it is that the Governor of Utah signed an executive order establishing a higher bar for data center development in Utah. This is Governor Cox who says Utahns deserve confidence that water resources, air quality, utility rates, wildlife will be protected. In Utah, environmental stewardship and economic opportunity go hand in hand. Now, for context, this is following a lot of public backlash against the stratos project. Kevin O' Leary's data center. A massive like 40,000 acre AIHPC data center campus planned in Box Elder County. If you've been on normie social media, you've probably heard about this.

[Colin]

This has become the totem for AI hate, you know. Yeah, like it's, it's kind of become. It reminds me a lot of when Terra Wolf was under scrutiny for one of its Bitcoin mines in New York. It became this like the coal plant one. Yeah, yeah, yeah. Well, like they were making this, you know, they were making these Claims in these lawsuits that Terra Wolf was like increasing the temperature, I believe of Lake Mariner or some other lake up there. Just this total moral panic. And this AI data center in Utah has become the punching bag for a lot of people who are against building these things for sure.

[Charlie]

So the. What does this executive order do? Well, I'll spoil it for you. Doesn't really do anything. It just says that there's a framework for executive agencies to follow specifically protect the Great Salt Lake. It's that big salty lake in Utah

[Colin]

and Water Resources specifically. Yeah. It suspends or it reinforces the suspension of any new water appropriations in the Great Salt Lake Basin. Which, you know, fair. Right. I mean Utah is one of the most beautiful states ever. If I lived there, I really wouldn't want data centers drawing from the ground. But also, I wonder, it's like I

[Charlie]

was, I was just in Arches national park and you know what I thought? You know what? I thought you could fit a big data center through some of those arches, you know, so.

[Colin]

But my question with that, and I would love someone to answer this, like can you even really use the Great Salt Lake? I mean, wouldn't you have to desalinate it to a certain point? The mineral content, or at least run it through like the mineral content in the Great Salt Lake is super high. You're right.

[Charlie]

Thinking about this too much. Let me get, let me keep going with the other stuff it does. It says it mitigates wildlife impacts. The framework executive agencies must follow should be that they should protect utility ratepayers, they should ensure energy generation, transmission and then here's an ambiguous one, lead on pro human AI development in Utah and then be transparent about it. Also, it's a guiding document. So the thing is, Colin, is that this, this, the governor received a lot of backlash because people say that this is not specific enough, it's toothless and it's too little too late assuming that something bad has already happened. But it's because like the way this data center got approved is not really like a rubber stamp that the government or the executive agencies of the government put their stamp on. Specifically the, the approval for this data center came through the Military Installation Development Authority shortened to mida, which is a quasi governmental entity and the county, the box elder county, like signed onto it. So this is not really like a state level issue, at least insofar as how the data center got approved. There's another details that the executive order does kind of update some language, like it's small language, like it changes what a Large data center is now considered to be.

[Colin]

This is a big change though, because it goes from 50,000 square feet to 10,000.

[Charlie]

Well, I actually have some insight to this. I don't know exactly what the language is for Utah. Oklahoma did almost the exact same thing. And it's, and it actually has less to do with the specific square footage and more so defining. Defining like what is what can be counted in the square footage. So Oklahoma had a similar reduction where the square footage encompassed the entire site, whether or not

[Colin]

just the site. Yeah, I mean that was like a big thing with the Kevin o' Leary's is that people thought it was going to be 40,000 acres when really it's just like that's the land that they had to acquire.

[Francis Corvino]

Yeah.

[Colin]

But for this one specifically, a large data center will now be considered 10,000 square feet rather than 50,000. Are you saying the use is big?

[Charlie]

But that's, I mean that actually narrows it a lot because some of the,

[Colin]

I mean it will encompass basically anything being built in the state at this point.

[Charlie]

Yeah, pretty much everything is a large data center now.

[Francis Corvino]

Yeah, yeah.

[Colin]

I mean anything that, that falls into that category will, will be considered that. You know, going back to what you're saying, Charlie, about the kind of like the push and pull here of the state level government and local government, one of the things that I find interesting about this, there was a lot of pearl clutching and kvetching in the comments of anti data center people saying this doesn't go far enough. Some of them are probably bots, who knows. But you can't keep anyone happy apparently. But I do wonder, this is basically telling executive agencies, here are your mandates, go follow them. And to me this, this reflects a kind of, I think disease in the federal bureaucracy where we have moved so much of the legislative process to the executive branch via agencies which we don't vote for, which we don't approve and which can be expanded almost limitlessly unless there is some sort of, you know, court case that goes up to the Supreme Court to challenge their authority. So I, I do wonder, like you this, on the surface this doesn't do very much. But with the right interpretation and a strict interpretation against data centers, they could probably do a lot with this until they run afoul with the courts. And that's another question I have, you know, will some of these mandates kind of butt up against each other for exist. For example, there's one about protecting ratepayers by making sure data centers, you know, don't cause a spike in electricity prices One of the biggest concerns, actually in the newsletter last week, there was a, there was a Gallup poll where the two biggest concerns on a data center impact for localities were energy prices and resource use with water. So that they were neck and neck. And so the water and energy problem are like the two biggest concerns in the minds of the average American when it comes to one of these things. But I wonder, okay, well, now you have to bring your own power. I believe they're doing that at the Utah data center. They're going to bring their own generators or turbines. Excuse me for nat Gas, but does that run up against the mandate for air quality and wildlife? Like, could you interpret that as, oh, well, if you're burning that gas on site, that's going to impact the air quality of the surrounding area, so you can't do that. Do you see what I'm saying? I feel like there's. It's so broad that this could be interpreted as strictly or as loosely as the specific federal agency and the zeitgeist of that agency at the time wants to.

[Charlie]

So America, this is what happens when you let the balance of powers, the checks and balances between the three branches of government get too heavy in favor of the executive branch. Because now the agencies are all appointed by one guy sometimes who you elect into office. So, you know, this is, this is a pretty difficult problem. I gotta switch gears and show you this tweet from more Perfect Union announcing that Bernie Sanders will introduce a bill to have the public take a 50% ownership stake in the country's biggest AI companies. And the American AI Sovereign Wealth Fund act would have the government tax AI companies, take 50% of the stock and put it under control. This both seeks to mitigate the supposedly runaway asymmetric wealth generation that's happening as well as try to take control of AI from an existential perspective. Is it risky? So a lot of people, you know, it's funny, is on the block space, internal slack. We all had different takes on this

[Colin]

because get this communism out of.

[Charlie]

Yeah, I think a lot of people might, might find it might be surprised for me to say that I don't think this is a crazy idea. Before you crucify me, I think it's crazy to do the specific things that Bernie suggests here in that 50% public stake and public control of that. But this idea of figuring out how to produce some kind of sovereign wealth fund from the AI boom is directionally in this, something that I've talked about on this stream before, in that there needs to be a better way to generate public wealth from this industry is right now you pretty much have to be a shareholder of these companies and you can't even buy the public stocks of these companies. So do you want to trade your way out of the permanent underclass? No, the average person can't possibly do that. Look at how Norway has structured their natural gas and oil and gas extraction and use that to build a trillion dollar sovereign wealth fund. So directionally I think this might make sense.

[Colin]

I would.

[Charlie]

Yeah. Of it. The specific details of it in classic Bernie fashion are insane.

[Colin]

Well, because he hasn't thought it through. You know, like all love to burn.

[Charlie]

Oh, he's thought it, he's thought it through. This is. Yeah, this is platform aligned.

[Colin]

I mean here's the thing. There's difference between thinking through the idea and like for public relations reasons and actually thinking through the ramifications because like 50 stake is insane. Right? Like that's just, that's nationalizing.

[Charlie]

It's like the wealth tax in California.

[Colin]

Like but, but this is for. But this is not like a tax. This is 50 of the like of the outstanding shares of the company. Like.

[Charlie]

Oh, I totally agree. But yeah, are these other. Is the public and elect board members.

[Colin]

Right, right. And then so but who actually controls the voting shares? Like probably the government. And this is concentrated at the federal level. Even though these data centers are scattered across the country. Like this like Texas succeed, SEC succeeds. If this goes through probably. I mean like this is wealth extraction to the federal level under the guise, and this is my opinion, under the guise of being owned by the public. But what does that even mean? You're going to give everyone like a fraction of iron stock, like a Dow

[Charlie]

call and everybody gets to vote. It's like, you know, we're going to do it on the blockchain, that kind of thing. I don't know.

[Colin]

I'm not, I'm making this up.

[Charlie]

But I know you are idealistic. It's this idealistic. Like everybody gets to participate in, you know, the decision making process.

[Colin]

But just a few more notes on that. Like that's my. And obviously like you said, he hasn't fleshed this out at all. But like so what? The federal government takes a 50% stake in all these companies? Like they just took a 10% stake in Intel. That makes sense for national security reasons. This is just a. Because we can. And then I guess the idea then is that money those shares are then somehow capitalized on and that's rolled into a wealth fund that we can use to, to benefit The American people. In theory that sounds great, but we're a nation of 50 confederated states. And so who gets to decide where the money comes from? The majority of the revenue for these things probably comes from Texas. Right. In terms of the shares that they're, they're, they're clawing from these companies. Do, do we then invest in Texas infrastructure? No, it goes to all these other states, depending on who's controlling the government at that time. You could see a scenario in which this gets really messy where the Republicans or the Democrats just take this wealth fund and then decide to distribute the spoils amongst their constituents and amongst their counterpoint.

[Charlie]

Counterpoint. Alaskan sovereign wealth fund with oil.

[Colin]

Right, but that's local. That's, that's a local thing.

[Francis Corvino]

Right.

[Charlie]

So you know, well, what I'm saying is directly there, there is a nut that has to be cracked here or else we're going to actually create the permanent underclass scenario.

[Colin]

I totally agree with that in the sense that I think if you're going to do this two things, I would rather it be somewhat voluntary. Now that's like a libertarian idealistic BS take because like that's not going to happen. But I mean, if you're one of these AI companies, everyone hates you because they think you're killing babies with the diesel exhaust from your non existent generators on site and poisoning the water supply. Like you should just go ahead and bootstrap something like that. Get an industry coalition together, get the funds together and maybe distribute it, you know, locally how you see fit. The other thing is, I do think that this doesn't make sense on a federal level. Again, these data centers are not accretive to the federal level. And that's part of the point. That's part of why I think the political class hates it. It's something that they can't totally.

[Charlie]

I don't want Washington telling, you know me what my data center company can and can't do.

[Colin]

Exactly. So doing something like the Alaska Wealth Fund for like oil and gas, the local level, state by state, I think makes a lot of sense as long as it is done in a cooperative manner and not an actively antagonistic manner with the companies. Because make no mistake, this is just antagonism. This is populist economic gobbledygoop that like no one's really thought through. And I don't disagree with you, Charlie, like ideally there would be some sort of fund to make sure that the wealth created by these data centers does flow back to the community more than just through property taxes. But keep that communism out of my data center is my final take.

[Charlie]

Seize the means of compute. That's absolutely going to be a line sometime soon. Okay, yeah, enough about this. Thanks for sticking with it to the end of the stream. Colin and I. You can find Colin I on social media. We are doing this every weekday now, so make sure you tune in at 1pm Eastern, runs a little over an hour block space live, covering AI Compute, some bitcoin still and other emerging tech stories, as well as markets. If you're listening on Coindesk, make sure to subscribe to block space feeds. We are leaving Coindesk very soon. Only on block space feeds can you find a search block space wherever you stream, anything and you can find us. Otherwise, thank you so much for listening. I'm Charlie. I'm Colin and we will see you tomorrow.