Transcript 0:00 [upbeat music] What's up, y'all? Welcome back to Blockspace Live, brought to you by CleanSpark. We have a fat lineup today, Charlie. 0:12 We're gonna start with SpaceX's IPO being oversubscribed 3X because apparently, my friend, uh, the market has learned you don't bet against Elon Musk, even if the economics might not make sense. 0:27 After that, we have Jamie McCavitt of Corement on to talk about their new facility in Texas, and just to get his v- uh, f- you know, his takes on current Bitcoin mining economics and opportunity, and what we should expect over the next year as hash price craters towards all-time lows once again, and hash rate is following suit. 0:47 After that, we have notes on Keel and Hut 8 senior secured notes that they just announced the pricing of today. And then we have a story on Crusoe pulling out of a massive data center in Wyoming. 1:01 And then we will end with OpenAI planning a 10-gigawatt, $500 billion data center in Ohio. If you thought that we have reached peak hysteria for NIMBY data centerism, you haven't seen anything yet. 1:17 Blockspace goes live weekdays at 1:00 PM Eastern, featuring quick hits on data centers, data centers, data centers, and data centers. Also, some Bitcoin AI and emerging tech. 1:30 If you like what you hear, this turns into a podcast anywhere podcasts are found. If you are listening on the CoinDesk feed, we are leaving CoinDesk next week. 1:42 So head over to the Blockspace feed, search Blockspace in your podcast listener of choice and subscribe there. We'll see you over there starting next week. Also, check out our newsletter. 1:56 If you missed the live stream, you can get the highlights from the newsletter, newsletter.blockspacemedia.com. This show is brought to you by CleanSpark, NASDAQ-listed ticker CLSK. More on them later on the show. 2:10 Let's kick it off. Rather, let's launch it off because we're gonna talk about SpaceX, Colin. I mean, the IPO is this Friday, and boy, is it gonna be a heater. 2:24 The news this week, um, from Reuters is that SpaceX has drawn more than $250 billion of investor demand for what stands to be the largest ever IPO, dwarfing the $75 billion that the firm is seeking to raise. 2:42 This means the SpaceX IPO is oversubscribed three to four times the planned offering size. Holy smokes. Holy smokes indeed, dude. Yeah. And for context, the largest IPO in history was Saudi Aramco at $29.4 billion. 2:59 So- Which is basically- I agree... the entire economy of the leader of the OPEC countries. So- [laughs] Yeah... and then that, and then here we have SpaceX. Yeah. 3:09 And, and the $75 billion alone would be incredible compared, uh, for a new record-setting. If, if, if it's actually $250 billion, Charlie, 3:19 I mean, uh, d- I, I don't wanna, you know, I don't wanna be that guy, but just to be a contrarian, do you need another flashing signal that things are getting too heated? That to me- No. I mean, no. 3:31 I, you know, or you could just be wrong, Colin. So let's dive a little bit into the, the deal some more. So, um, just to review. 3:39 The, um, SpaceX plans to sell 555.6 million Class A shares, the meme number 555, at $135 each. It would raise $75 billion and value the business a little over $1.7, more like $1.8 trillion. And 3:57 30%, roughly, of those shares are allocated to retail investors. 4:02 You may have seen stories about, I think it's, like, Morgan Stanley or Fidelity lowering, like, the threshold for their retail clients to be able to buy in, and that's where we get this tweet that I've got pulled up from Luke Cannon, um, who tweets sardonically, "The cashier at Home Depot just asked if I wanted to round up to support the SpaceX IPO." 4:25 [laughs] I mean, that's kinda how it is. Like, this is very much, uh, this is, like, a Main Street, uh, IPO almost than it is a Wall Street IPO. 4:37 Um, I think people- I would be curious about that though because when you see that $250 billion, when I look at that, it's like, what, what institutions here are piling in at this point? 4:48 And, and which ones have been left out of the private raise and whatever? They've already, they've already piled in. That's the thing. This, th- SpaceX has been a private company for 25-plus years. Right. 4:57 But- Everybody who wants a piece, who's already one of these qualified invest- you know, institutional investors, has already gotten a piece. 5:03 But my, my point is saying that is $250 billion is not just retail though, right? That's a lot- Yeah... of money. Honestly. 5:08 So, like, there are institutions piling in now, and that is the one thing that I think I would love to see the actual allocation, how much of that is coming from, uh, retail, how much of that is coming from, uh, your brokerage accounts, like through Fidelity, where they've lowered their minimum to invest in an IPO to $2,000 from $500,000. 5:27 Um, pretty insane. But, uh, the, I think this will give people... There are two ways to read this, I think. 5:35 You either-- this is either gonna give people massive confidence in the IPO's immediate success after listing, or it's going to get people screaming for the exit because this thing is massively oversubscribed. 5:46 I think this indicates, given that we're 48 hours away, that the IPO is probably gonna go up after launch, and it'll be a very, very long weekend where you can't trade stonks, except if you go to Hyperliquid, which I've got pulled up here. 6:02 So you have the Hyperliquid pre-IPO SpaceX shares, which-Launched in mid-May at, like, 215 bucks, and it kinda was flat and went down the first week of June. 6:17 And so the, here's the rough math, is that a week ago these pre-IPO shares were trading about 40%-plus over, like, spot of the $135 per share. 6:30 Now they've dropped to $162, which means that they're only, like, 15, 16% over pre- uh, over IPO, uh, pricing. Which interestingly enough, Colin, 6:42 actually matches that of where the pre-IPO of, uh, uh, Anthropic currently sits. 6:50 So it's almost like if you wanna trade 24/7, and if you wanna get access to pre-IPO shares on one of these, like, you know, gray area market platforms like Hyperliquid, maybe the premium is 15% over spot. 7:06 So it's kind of interesting phenomenon we're seeing here. I, I agree with that, and I wonder if, uh, what we're seeing is almost, like, the same role that prediction markets play for political outcomes. 7:15 Like, th- this will be... I mean, nothing like Hyperliquid has existed for, uh, s- equities and for specifically the pre-IPO landscape, right? 7:24 And so with, with SpaceX and with Anthropic, and I would assume also eventually OpenAI, you'll get an interesting glimpse into investors' mindsets in terms of whether or not a company is overvalued or undervalued based on those premiums or discounts. 7:39 Maybe, or maybe that's just the premium to carry exposure to the stock on these liquid, you know, synthetic markets. Uh, do, do you see what I'm saying? It's like- Yeah... will, will this actually prove to be... 7:54 Like, these are testing grounds right now for where, where the market's actually going to value these companies, and I think it's going to be very telling once the IPO actually comes out as to whether or not these are legitimate proxies or actually good forecasts for where the stock is going. 8:10 Yeah. I wanna switch gears a bit on this SpaceX thing, and I wanna talk about the space part of the data center deal. 8:17 So as we know, and if you've been listening to our content for the past week, SpaceX is not a rocket company. 8:23 It's a rocket company that shoots data centers into space, where the majority of the revenue is going to be data center and AI HPC related. 8:34 And so, um, until about 48 hours ago when SpaceX put out their actual, like, more, like, specs on the satellite that we featured yesterday on, like, how they're gonna put this computer into space, 8:49 a lot of people were like, "Space data centers don't work," yada, yada, yada. 8:54 And I found this clip from Elon being interviewed on Dwarkesh Patel and, uh, drinking beer, a couple Guinness, with Dwarkesh, where Elon actually talks about one of the reasons he likes space, and it actually is the literally the most boring thing ever. 9:13 It's because it's easy to get permits to build a data center in space. In the same way that there's always sun in space, you don't have any n- neighbors in space. 9:25 So I think, like, it, you know, um, the NIMBYism has to, has to evolve into not in my low Earth orbit. Uh, but here's, here's Elon talking about this. I think it's pretty hard to build solar panels. 9:37 You have to get, like, permits from, like, the permits for... Try getting the permits for that. So space is really a reg- [laughs] It's really a regulatory play. 9:45 It's, like, harder to b- harder to build on land than it is in space. It's, it's harder to scale, um, on the ground than it is to scale in space. Um, but, but also the, 9:55 the y- you, you're gonna get about five times the, um, effectiveness of solar panels in space versus the ground, and you don't need batteries. Um, I almost wore my other shirt which says, "It's always sunny in space," 10:09 which it is. [laughs] So, um, because you don't have a day and night cycle or, uh, seasonality, uh, clouds, uh, or a, or an atmosphere in space, uh, 'cause, uh, the atmosphere alone, um, 10:25 r- uh, results in about a 30% uh, lo- loss of energy. Um, so, uh, so you can, for a- any given, uh, solar panels can do abou- about five times more, uh, power in space than on the ground. 10:37 And you avoid the cost of having batteries to carry you through the night. So 10:43 this is really interesting because we will, we'll ta- we'll come back to this point that Elon made, and this was, like, last February that he made, which is, um... 10:52 And it, he actually seems unbelievably prescient here, because right now the top stories are about local and community pushback to data centers, and that is actually right now probably a bigger constraint and execution risk factor than, like, procuring power generation. 11:11 Um, maybe power generation is not as scarce as permitted land and permitted power. So you know, the, if, if you're an investor and you're seeing this change, then you might buy into Elon's, uh, 11:27 pie in the sky [chuckles] quite literally pitch that, uh, space is, uh, blue ocean figuratively. So- To me, it's just a matter of timelines, right? 11:39 A- as mo- well, as a lot of people have pointed out, Elon is very good at selling the future on an aggressively accelerated timeline that quite often his companies do not meet. So he's saying, what, like, 20... 11:52 What is he saying? 2028, 2030 when they'll start launching these things? It's, it's like- Yeah... he's basically signaling a few years out, right? 11:59 And if they do hit that, then do, to keep using corny space puns, the, the, the star is literally the limit in some cases for what SpaceX could do. The sky's the limit and the star is a lie. 12:11 [laughs] You're, you're metaphoring up your mixes, you know? [laughs] Um-So to me, it seems like a foregone conclusion that we will eventually do this with enough technological progress. 12:26 But there's a lot of healthy skepticism right now, I think rightfully so, over whether or not this is a to five-year out, 10-year out, or 20-year out endeavor. Yeah. 12:38 Um, I, you know, I, while we're kind of waiting for our guest, uh, Jamie McCavity from Corement to come on, um, I'll bring up some criticism to Elon's timeline that I've seen. 12:49 Um, I saw a tweet, dang it, I wish I had it ready to go, but it was something along the lines of this. 12:55 If we look back at Elon's big, like ambitious projects, the Hyperloop, various boring companies, um, robotic, entirely robotic like Tesla manufacturing facility, none of these really came to fruition. 13:09 A couple have gotten a little bit there. But, uh, it's, it's been a, uh, it's been a while since we've seen like one of these giant, big visions of Elon's, uh, like come to fruition. 13:28 I'll acknowledge that. I'm gonna sound a lot like an Elon bull. I'm gonna sound like a huge like Elon, um, uh, simp here, and I am, you know, to some degree. But, uh, consider that like I... 13:42 Like who's pay- Like, what's the economic incentive for to build a Hyperloop? What's the economic incentive to like build a giant like tunnel underneath LA or, or Las Vegas that, that Elon was planning on doing? Yeah. 13:55 You- Or Nashville boring through bedrock, which is what they're currently trying to do. Exactly. 14:00 Like, it makes a lot more sense to me that you have a client and you have infinite money pouring into space, and more so data centers. 14:08 So it's like you have abundant clients and capital waiting for you to go build that thing. Maybe this is ac- Maybe the, the problem was not really an engineering one, but rather a markets and demand one. 14:20 So, uh- You saying infinite money reminds me of the hats they were passing out at Bitcoin 2025 during the treasury company craze that said infinite money. And if you needed any top signal, then that was that. Yeah. 14:31 Anyway, side note. Yeah. But yeah, um, maybe there'll be something like that. We, we need a... 14:37 There, there's a abundance of merch and shirts, and if, if SpaceX can't fulfill the IPO launch demands, then I'm certain they have a bunch of memes they can put on shirts and- Yeah, they can sell those It's Always Sunny in Space shirts. 14:48 People will love that. Exactly. Yeah. All right. We're gonna come back down to Earth. 14:54 We've got in the wings, we've got Jamie McCavity of Corement waiting to come on, and we'll have him on here right after a word from our sponsor, CleanSpark. [gentle music] We are CleanSpark, 15:06 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. 15:23 We are setting the standard for what's next. Learn more about the intersection of energy and Bitcoin at cleanspark.com. All right. We've got our boy Jamie in the wings, gonna talk about Corement in Texas. 15:41 Let me bring him on up here. Jamie, welcome to the show. What's up, boys? Good to see you. Howdy. Good to see you too, man. Thank you for joining. Thank you for having me. So, um, you guys just announced a big deal. 15:59 I see it on LinkedIn. Uh, 58 megawatts, Jim Wells County in Texas. Am I correct that this thing is supposed to energize in August, is what the s- what you guys say? Tell me about this. Yeah. 16:14 Yeah, thereabouts August, uh, between August and Thanksgiving is what we're projecting, but it's, it's our first greenfield start to finish site. Our, um, our existing site in Fort Stockton was a, a brownfield, call it. 16:32 Uh, I don't know of any other one that's like it. We actually dec- We were part of a, 16:37 a transaction that saw the decommissioning of a wind farm, and then we repurposed the transmission infrastructure to, uh, to become a load for Bitcoin mining. So, uh, I'm kind of curious. 16:52 I, I, I don't know how much you can talk about this deal. I'd love to l- know a little bit more, 'cause you've, you kind of have, I would say, anchored the, the private world of Texas Bitcoin mining for many years. 17:04 Um, incredibly low cost of power, some incredible like, uh, execution feats. This is a different type of data center, different requirements. 17:15 Can you tell me about how you put this deal together and like what you look for when doing this? Yeah, I mean, our operating premise over the last four years, five years was 17:29 you have to be the lowest cost, you have to focus on electricity and commodity optimization, for Bitcoin mining that is. 17:36 And then you have to focus on infrastructure deployments, deploying them fast, deploying them at a low cost. Um, and we got very good at that as a... And w- and we executed historically very well in Bitcoin mining. 17:52 And as a, as a side part of our thesis, we said we're accumulating land and interconnections in these somewhat remote parts of Texas. 18:04 It is possible that one day demand for data centers could grow, that these locations right now, which don't carry the same valuation as premium tier one data center locations like Northern Virginia or Dallas, uh, or proximity to major metro area data centers, that one day this land portfolio could act like a call option on that. 18:25 And-I think we have seen that come true. The demand for data center capacity is now huge. 18:32 So while we, we are currently still mining Bitcoin, and that is a skill set that we have and, and we're very good at it, we're bringing this site out to the market to see what else is out there in terms of offers, um, where it's plausible that the, the best economic use of this land that we've been developing would be not to be a Bitcoin mining data center. 18:56 So g- going back to that call option, Jamie, what makes it specifically po- potentially interesting for that? 19:02 Because it's close to Corpus Christi, because there is not a lot of competition in the area, uh, because there's still abundant power. Like, why would this site specifically be optimal if you decide to go that route? 19:15 Well, yeah. It's... I mean, this is way closer to a major city than our Fort Stockton site. Uh, it's, it's close to San Antonio and Corpus. 19:24 It's fifty-one miles from the coast, so I think that for major data centers, anything within fifty miles and in of the coast is kind of, uh, considered to be riskier. 19:36 But this was fifty-one, so we're just one mile outside of that, uh, that red line. Ri- risky because of, of hurricane, uh, threats or? Yeah. I- it's just the amount of scrutiny in, 19:48 in siting and designing AI data centers is a lot higher than it is in Bitcoin mining. Where at Bitcoin mining, it's just like, "Yeah, let's throw a few containers out there and see what happens." 19:57 Uh, you know, you've seen some Bitcoin mining sites, I'm sure, in your days that are... They aren't even built up to the standard of the electrical code. Um, 20:05 whereas with, with an AI data center, they're looking at a five hundred-year floodplain. 20:10 They're looking at what happens if a Cat five storm directly hits the data center and, y- you know, takes out all the roads, and they can't get in there to refuel diesel generators on day four a- after all your diesel's run out. 20:24 You know, they are really game planning for the, the worst possible implausible scenarios, uh, imaginable. So fifty, fifty miles in is, is where those scenarios start to become a little bit worse. 20:38 So Jamie, kind of hitting on this note of the, uh, tension and the tug of war between Bitcoin mining and AI right now that we're seeing, w- what opportunities do you see for miners right now? 20:48 Um, as, as someone, you know, who just signed this deal and is getting y'all's second data center up, where can miners actually still source power, and, and where... 20:57 And h- and what does that mean considering we're in a historically low hash price environment as well? Well, you know, it's funny. 21:04 I was listening to a Howard Marks podcast this morning, and he was-- He's kind of putting it very succinctly. 21:12 When, when everybody else is looking in one direction, uh, you know, that's the right time to look in the other direction. With Bitcoin mining, it's probably never been a better environment to deploy capital and operate. 21:25 We have... Difficulty is, is flat year over year w- in a, a reasonable bear market. I mean, a fifty percent bear market, I wouldn't call that like a, a scorched earth bear market. Uh, certainly we've all seen much worse. 21:39 But, you know, difficulty is flat over a year. That's incredible. That hasn't happened, uh, since twenty-twenty, and I think the last time that did happen was, was, uh, in the Chinese mining ban. 21:50 So I would call that a kind of a black swan event. And then prior to that, it was twenty eighteen, where there was a ninety-five percent drawdown in Bitcoin's price. 21:58 So great environment to get into mining, great environment to buy containers, to buy used ASICs, or to buy new ASICs because all the customers are focused on AI. And 22:12 I mean, they're focused on AI for a good reason, uh, because from a dollars per megawatt hour perspective, let's just break it down into how we think about Bitcoin mining profitability and revenue. 22:23 In Bitcoin mining, you have to build a data center, and then you have to operate it, and you're effectively making the operating spread between your input electricity cost and your revenue denominated in, uh, a megawatt hour or kilowatt hour equivalent. 22:37 So for people who scrape the bottom of the barrel, like CoreMint, in mining equipment, we're making, like, twenty, thirty bucks a megawatt hour. 22:45 If you buy shiny new kit, like a lot of the public companies do, maybe you're making fifty to a hundred dollars a megawatt hour. So if you extrapolate that out into a year, that will... 22:58 You still have to pay for all this equipment. And at a hundred dollars a megawatt hour, you're making approximately a million dollars per megawatt per year on a, uh, on a Bitcoin mining deployment. With AI data centers, 23:14 you could see just in the market right now, and this'll be... 23:17 We'll see where the valuations eventually flesh out to and how things trade once they're fully stabilized, and maybe we exit this euphoria period that we're currently in. 23:29 Bitcoin mining stocks that have made a transition from mining as their core business to AI, they trade closer to ten million dollars a megawatt just when they sign a lease. 23:39 And then they're able to finance those leases and the construction of those data centers using as eighty percent debt, sometimes ninety percent debt. Um, so the, the AI data center group, they really do pay top dollar. 23:54 And i- i- it's... If you have an on-grid site and you own land, there's not really a world where it makes sense, if you're trying to value maximize, where it, it still makes sense to pursue Bitcoin mining. Um, 24:10 so sad times, I would say, for the mining industry maybe, for people who are excited about the US having thirty to forty percent market share. But also, if you're an opportunistic Bitcoin miner, 24:22 it's a great time to be deploying hash rate.Uh, obviously there's dollars which are incentivized everyone to pivot to AI who's been Bitcoin mining. 24:34 Um, but we mostly talk to, like, the public companies, and those are the ones where the numbers are public. We can see the market react to the decisions and stock price and valuations. 24:43 We don't get a lot of insight to the private markets, a lot for good reason. Um, I'm curious about your insight to, like, what are the strengths or weaknesses of, uh, the private market in the AI... Bitcoin to AI pivot. 24:58 Are there barriers? Do private operators have an edge? I'm curious your insights here. Yeah. 25:04 I think private operators have an edge over public operators in, in every category, and it's the same in this, where you don't have your employees fixated on share price. 25:16 Y- you know, there's no mark to market if you have a setback or, um, 25:22 you know, let's say in this case, if there's a delayed energization of a project, uh, or some legal action or something like that, all that stuff isn't gonna affect your stock price 'cause it's not publicly traded. So 25:36 avoiding the, 25:38 all of the extra scrutiny, the administrative burden, the compliance, the regulatory requirements, the constant lawsuits, uh, and, and everything that comes along with being public is the main advantage of being private. 25:51 Th- the, the reason why people make the jump is 'cause in public markets, you can access a lot more capital, and you can access it at a much lower cost of capital. 26:00 I- in AI data centers, though, we are still seeing an enormous private sector push where very large private equity firms are deploying... I mean, I don't know a single private equity firm 26:13 that's a, like, a big player that is not doing something related to AI data centers. Crusoe, 26:20 historically one of the larger, more innovative private Bitcoin miners, has reached a ten billion dollar valuation as a private AI development company, AI data center development company. Uh, so it's... 26:34 If you can raise capital and stay private, and you're comfortable going down that path, then that's, it's, it certainly still works, but there are definitely strong incentives to going public. 26:48 Jamie, as a closing question, since we're talking about funding and the CapEx cycle in general, the biggest question everyone has right now obviously is how much farther does this thing have to run? Mm. 26:59 We talked to Mike Alfred a few weeks ago. 27:01 He says that he thinks over the next two years you're going to see the bull run continue because it's just going to follow the progression of this current CapEx cycle before these initial investments are through and the banks and investors get exhausted. 27:16 What, what are your thoughts on the current CapEx cycle and how much longer you actually think we have to run? 'Cause we, we were talking on the show yesterday, things already seem pretty hot. 27:26 I- if you looked at FactSet's earnings insights for S&P 500 companies, y- you know, it was like fifty plus percent, uh, or, uh, the technology sector had fifty plus percent earnings growth. Mm-hmm. 27:39 That just doesn't seem sustainable to me from, from where I'm standing, but I'm curious your thoughts. I think what you'll probably see happen is profits will shift and valuation and market cap will shift from, 27:55 from where it is today to... and it could shift materially to another place in the future. I don't think 28:03 that there's any putting the genie back in the bottle in terms of price discovery and market, product market fit discovery on the birth of a new commodity, which is the commoditized digital labor. 28:17 There's just so many applications of commoditized digital labor. It's getting better and better every day. They're break, they're, they're breaching into new categories. 28:29 How that is delivered to customers, what the model looks like, what moats are established, uh, open source, frontier labs, which hardware is used to deliver that, whether it's something more like a TPU or continues to be NVIDIA dominance, a custom ASICs. 28:47 There's, there's so much disruption that could happen within the category, but I don't think that the category is getting smaller. I think the category is getting bigger. 28:57 And then one thing that I am certain will happen at some point in the future is, like all commodities, this commodity will return to the marginal cost of production. 29:07 And right now it's in price discovery, and we're very far away from the marginal cost of production, and the market is subsidizing producers who are producing that commodity at a loss for user acquisition or market share gain and things like that. 29:21 But at some point, it will return to the marginal cost. It'll be just like Bitcoin mining. 29:27 It's gonna be about, you know, energy optimization, data center CapEx optimization, and you'll need to be a little bit responsive to power prices every once in a while. 29:38 I'm, I'm really confident it's gonna end up like that. It could be decades from now when that happens. Uh, but that, that's my two cents on sort of what it'll all look like. 29:48 S- so to recap just a little bit, and for- forgive me if I'm kinda putting words into your mouth, it's almost like the, uh, new paradigm argument in the sense that, like, it's hard to even say whether or not valuations are crazy right now because what is being built is unlike anything we've ever seen in the market. 30:03 'Cause I kinda come back to that with my dad sometimes. We'll talk about how CAPE ratios, P/E ratios, all of these economic indicators are approaching dot-com level, uh, the dot-com b- levels. Mm-hmm. And I always... 30:16 You kind of put it more eloquently than I have with him. I just kind of tell him, "Dad, we've never seen anything like this before. People don't know how to price this." 30:23 And if you're looking at the productivity gains and the possibility of, like you said, an entirely new commodities landscape, who, who's to say that those old valuations even make sense?Yeah. 30:35 I think that's one way to look at it. And look, I'm not gonna say we're in a bubble 'cause it's, uh, I, you know, I don't wanna be in a bubble. Uh, but it's 30:46 plausible to me that with the amount of leverage that's in the space and the amount of eye-watering valuation that we're seeing based on growth that has recently happened, there could be a correction or, or more likely a series of corrections that could also lead to de-leveraging, and people are going to get carried out, companies are gonna get carried out from, 31:10 uh, this industry for sure sometime within the next five years. I don't think that broadly the, the industry itself is overvalued. I think that that is... 31:24 It, like I said, it's a birth, a birth of a new commodity, how that commodity is delivered, reductions in the marginal cost to produce it, all of those things are gonna create disruption to the existing incumbents. 31:33 And there's gonna be some people who get carried out, for sure. But the trend is... 31:39 I think the trend is clear, and if you're disciplined and you stay focused on the core themes, which I think are power, power generation, utilities, 31:51 supplementary computing equipment, and you find good entry points and try to buy on pullbacks if you're sort of allocated to the sector, I think it's a, it's probably a great way to play it. 32:00 Um, however, buying the all-time high across the, the basket of names that represent the space has been a winning trade for two straight years, so, you know, just keep doing that if, if that's your thing and, uh, and good luck to you. 32:14 But, um, yeah. Before you go, before you go, I got one last question. This is coming in hot, sliding in my DMs from our, uh, from Blockspace founder, Will Boxley. Wow. 32:27 He wants to know what the cost of capital is for private AI builders. Because we've seen cost of capital go down. I think CoreSize, what, like in the single digits now. Uh, are you able to reveal any insight here? 32:43 Yeah, I mean, I think it depends. Um, it depends on who your counterparties are. It depends on what your pipeline looks like, what your track record is. Like if, if... 32:55 I mean, Corement as an unrated entity, our cost of equity and debt would be significantly lower if we were to have a l- a lease signed with someone like Amazon or, or any other hyperscaler, versus if we did a deal with, say, a CoreWeave or another Neocloud, where it just the, the cost of, of capital, of debt capital on those deals is higher, so then the equity capital becomes higher and you're looking at kind of a higher hurdle rate for those investors. 33:25 Um, 33:27 what I will say, Will, is that our cost of capital as a potentially an emerging AI company, we haven't even done anything in AI, but just as sort of potentially emerging AI company, is better than it ever was as a Bitcoin miner. 33:43 Not surprising to me. It was very expensive to be a Bitcoin miner [laughs]. Um- Yes. Oh, wow. I wanted to answer one question that you guys asked, though, about what Bitcoin mining is gonna look like, and I know I'm... 33:55 we're, we might be over time here. No, no, you're fine. That, that's a great closing question. Yeah. And just one thing I wanted to add to that, you know, you talked about hash rate. 34:02 I'm looking at the chart right here at hash rate index and, and we're about to test the yearly low, 34:08 and if I'm thinking about this, going back to you saying difficulty is flat, given that hash price is low and there's sentiment in the dumpster for Bitcoin's price, it's hard to see an environment where hash rate, 34:20 uh, grows this year, honestly. And so what, what is your outlook- For sure... for the next year or so? 34:26 Well, we're in four CP season in ERCOT, so you're gonna see the seasonality of lower mining hash rate just from curtailment in, in ERCOT. 34:37 Price being lower means more economic curtailment will happen if miners are being price sensitive. And the name of the trend here, boys, that we're gonna see is hash rate volatility. We are enter... 34:50 We have exited the era of high uptime, buy the newest ASICs, one year payback Bitcoin mining. That's the way it goes. We have exited that era. It is over. The new era is going to be where is the cheapest power? 35:08 How can I get that? And the answer is the cheapest power is not available around the clock. The cheapest power happens when it happens, when there's low demand, high solar, stranded gas, you know, stranded wind. 35:22 It happens intermittently and it's not always available, and Bitcoin mining will continue to feed on that power like the dumpster diving commodity industry that it is. 35:33 And what, what I think that will look like is block time volatility, hash rate volatility. 35:39 And we're, we have now entered the new era of Bitcoin mining, which means that you might need to pay more fees to get into a block if you send a transaction at the wrong time of, of the evening. 35:51 With the amount of solar that's excess stranded solar that exists in the United States, if there's a lot of Bitcoin mining that's co-located with solar or nearby to solar, you could see the Bitcoin network confirm transactions most reliably during banking hours, which would be somewhat ironic for the debank money. 36:10 Uh, and I think that trend could continue where, where there is a coincidence of solar co-locations with Bitcoin mining, that's when you see lots of quick confirmations and, you know, power will effectively be free. 36:23 If you get into the period where it's an overnight or the sun is over, call it the Pacific, before it rises in, 36:31 in, uh, in Asia, presuming that there's still some mining in Asia, you could see really slow confirmations and fees could get very large during that period. 36:39 So I think-Hash rate volatility, mempool volatility, fee volatility, that's the new era. We've just entered it. 36:47 Like I say, um, the best traders of hash rate futures and perhaps fee rates are going to be meteorologists and weather wonks. So once again, Oklahoma meteorologist armchair enthusiasts on top. 37:01 Jamie, thank you so much for coming- [laughs]... on the show. Appreciate your time. Thank you, boys. Yeah. Thanks for having me. Charlie wanted to be a meteorologist when he grew up. Yeah. 37:12 Oh, man, you're super doxing me. I did. 37:14 There was a career day in, uh, preschool, and my mom has a picture of me, like, with, uh, like, a rain gauge pretending to be a meteorologist, so it was also the longest word I could say at the time, so... 37:25 Also, one- one takeaway from that, that somewhat jokingly, but, and somewhat seriously, a PE getting interested in AI should terrify everyone. 37:35 If you think your models are degraded right now, just wait until PE firms come, and then they strip everything out except the nuts and bolts. You thought the nursing homes were bad, but what about the models, everyone? 37:48 I'm partly joking, but we can- Nah, with the new Meet Those model, get rid of all your, all your analysts and your quants. 37:55 My quant is, uh, a data center in Memphis, Tennessee that Elon owns, is trying to sell to the public markets, to unsuspecting retail investors. Okay. We are gonna keep going. We've got a couple data center stories. 38:10 We've got Keel. We have s- Hut8. But before we go to Keel, let's hear a word from our sponsor, Luxur. 38:19 [firework launches] This episode is brought to you by Luxur's Commander, Bitcoin miner management software for enterprise operations. 38:30 Commander gives you real-time fleet monitoring, bulk remote commands across your fleet, and Intelligent Miner. 38:35 That's an automated profitability engine that runs every five minutes and tests your power settings against live energy markets and hash rate markets. 38:44 ERCOT backtests show 10% improved profitability with intelligent mining versus binary mining. 38:49 Commander is $100 per megawatt or a 25 bip pull fee adder, which is roughly half the price of competition, and you can try it for 60 days for free. So if you wanna get started, go to luxur.tech/commander to learn more. 39:06 All right, Charlie, let's get back into data centers, data centers, data centers. Shout out to Jamie for a fun little quasi-Bitcoin mining data center segment there. 39:15 Yeah, that was the most Bitcoin we'll do on the show today. Yeah. And I, I do think that what he said makes sense. You know, buy... What is the saying? It's like buy when the streets are running red, you know? 39:27 Like buy when there's blood in the streets. And right now, CapEx for Bitcoin mining, I would guess, is probably the lowest it's been in a long time. Like- Yeah. Is there such thing as negative CapEx? 39:39 Because people are- [laughs] We'll pay. We'll pay you to take the ASICs so that we- Leave-... don't have to pay for the inventory. Yeah. 39:45 Anyway, we'll, uh, we'll get on with our next story, and that's a note on Keel closing a $458 million convertible note deal for its data center build-outs. 39:56 So deals for 458 million gross, uh, for a 1.25% convertible senior note due 2031. The deal was upsized twice. 40:06 Initially, it was launched for 350 million on June 4th, priced at 400 million the same day, and then there were initial exercise options, uh, to buy 58 million additionally, bringing the total value of the note to 458 million. 40:21 But the net cash injection for Keel will land around 400 million after closing cost and a capped call consideration. 40:30 Keel went into a capped call to protect against dilution on this for roughly 41.7 million, and that capped call goes up to $11.86 per share, which is... 40:43 was 100% premium over the June 4th stock price that that, uh, capped call was priced on. So there will be no dilution with that capped call or with the converts unless Ke- uh, Keel's stock goes above that price. 40:58 The conversion price for this note is $7.41 per share, 25% premium over that June 4th close. [lips smack] And like I said, it will be due 2032. So a few interesting parts about this. 41:15 Keel really kind of beat its chest in the press release for this, saying, quote, "The existing liquidity expected on their balance sheet is expected to be sufficient to develop Panther Creek, Sharon, and its Moses Lake s- uh, sites through leasing." 41:28 So to recap, Panther Creek and Sharon are in Pennsylvania. It's Keel's coal-fired plants that they purchased and data centers that they purchased from [lips smack] from Stronghold Digital. 41:39 And Moses Lake is in Washington. And they said, quote, "The opportunistic capital raise is expected to improve flexibility to make value-added investments across their current developments." 41:50 So they're kind of-- They're, they're phrasing this as an offensive rather than, like, a necessity-driven capital raise, basically saying, "We're just trying to pad the balance sheet for what comes our way for when we need it." 42:00 Charlie, any thoughts before I throw a little teaser for what might be going on at Moses Lake? I don't have a ton of thoughts. Ben Gagnon, uh, executing- [laughs]... incredibly well, raised a bunch of money. Great, 42:14 uh, great cash deal. I'm super impressed. I mean, this is a kid who went and built Bitcoin mines in China in the 2010s, so a little bit of lore there for those of you who don't know. 42:28 Yeah, and cool to see in the way that he's ri- risen through the ranks of Bit- formerly Bitfarms, now Keel. 42:34 Now, the question for this, Charlie, is w- you know, they're saying they have cash on hand for their current data center build-outs that they're going to move into AI. 42:42 As we've noted h- on this show recently, Keel has been surging this year-Leopold, uh, Aschenbrenner, um, I messed up his last name. No, that's good. It's good enough. 42:54 Anyway, Leopold Situational Awareness ha- is, has taken a stake. 42:58 They're bullish Keel, and we've raised the question: What do the insiders or, and not insiders in Keel, but insiders with the knowledge of what's going on and investors know about Keel that, that we don't? 43:11 Thank you for pulling up the chart there. Yeah, it's, it's been, it's been doing super well. It started really moving up in Q4 of last year, and it's really taken off this year. 43:21 It's about specifically after the rebrand, you know, and then they, uh, after the rebrand, they're now domiciled in- The people like-... to US market... the people like, yeah, the people like boats instead of bits. 43:31 I mean, the, uh... Look, everybody's been trying to sell their Bitcoin to buy a boat, so name your company after a boat. 43:40 Yeah, and I think the biggest thing here, too, is now that they're on the Nasdaq, there's just much more capital for them to- Yeah... acc- I mean, or rather there, uh, there's much greater investor access. 43:49 But I wanted to get up this from a, uh, this finding from a pleb on Twitter. This is one- Now, now, this is, yeah. So if you remember, for context, 44:02 the Keel deal with AWS was sleuthed out from the job posting. Well, we don't know... Uh, here's what I'll say is behind Keel- We don't know if- We don't know that there's a deal yet... 44:12 it, it, w- this sleuthing happened, and speculation occurred that Keel may have AWS as the tenant. That's- Right... the line. 44:21 A- and the reason for this is, per this note, there is an AWS job post to manage a data center in Moses Lake, Washington, and the listing says the facility is, quote, "still under development." 44:32 Another note, Wayne Duso is one of the board of directors for Keel, and he spent, I believe, like 12 or 15 years at AWS, so there's some- something of a connection here. But here's the job posting. 44:44 Job details in, uh, Washington, Moses Lake, operations IT and support engineering. 44:50 You'll join a diverse team of software, hardware, and networking engineers, supply chain l- uh, specialists, security experts, operation managers, and other vital roles. Blah, blah, blah, blah, blah. So 45:01 that, that's been my biggest question with this Keel run-up is it, to me, it, it reminds me of when Hut 8 was running up before they find, signed their first deal. We've seen this with a number of the other stocks. 45:10 When, when there's a deal in the works, there's enough people who know outside of the company who are going to try to capitalize on that, or they're reading the tea leaves. They're looking. 45:20 They're sleuthing for things like this. So I'd be curious to see if we're actually going to see something come to fruition here. 45:26 And the other question I have, too, with Moses Lake, what I find really interesting about that is on the surface, you'd think that the Pennsylvania sites would be much more interesting because the power is on site and is fully owned by Keel. 45:40 Uh, but I do wonder if the coal-fired power plants still carry a little bit of ESG risk. 45:47 I don't really have much to base that on other than vibes and the fact that, you know, coal has been demonized for the last few decades. We've been replacing coal with nat gas on the grid for cleaner b- generation. 45:59 I wonder if a hyperscaler or a Mag 7 company m- may not still feel comfortable with tapping that power. I don't know- I could be totally off base about that, but that's just a question that I have... 46:11 look, Colin, you're living in 2021, back in the before times, f- you know, back when people still cared about those things, um, but- I think they still do with the NIMBYism, man, right? 46:23 I mean, we've seen people freak out about the Utah data center. We've seen disinformation about water use for these data centers. Could you imagine if one of the Big Five ended up leasing a coal-fired data center- I-... 46:40 and people in the normies found out about it? They would be picketing that, that [laughs] you know? They would be picketing out in the streets. 46:47 I mean, insofar as G, governance, in the ESG relates to, like, civic unrest or civic, you know, rejection, I don't know, man, because I really do struggle to see that this water criticism of data centers as having a long-term sustainable legs. 47:04 I, I, this, just because it is so misguided and so misleading, I think the real criticism is gonna be energy prices, capital extraction, and, uh, just how ugly data centers are. 47:19 You just can't get around the fact that we're building a giant, you know, altar to Moloch coated in, in, uh, you know, beige or gray, and it, and it is just this giant wall of, of concrete, you know? 47:35 You can't- And I could imagine, like, driving to, um, Scion National Park or something in Utah, and then- Yeah... out of your way, you just see this giant gray dirt- You know what would go great here is a data center. 47:46 [laughs] So we need to be, we need to build them like cathedrals. Bring, bring back stone masonry. 47:52 Uh, I no longer- Asher Grenut of Hut 8 has a great comments on this in an interview we, we did with him recently, by the way. 47:57 He thinks that's actually, to your point, Charlie, one of the things that they need to do better at, is actually making them aesthetically pleasing. 48:03 One of the things the Industrial Revolution actually did kinda cool is the buildings looked really cool. 48:09 You go to these giant factories and power plants, and as much as you can imagine them spewing, you know, coal and, and stuff into the sky, into the residential areas around, they look pretty cool even 150 years later. 48:24 So data center builders, spend that extra money. Make your data centers look a little bit more like a satisfactory, uh, factory tour video than a, uh, you know, suburban, uh, big-box store. It's my, my request here. 48:40 More, less like an IKEA. Yeah, yeah. [laughs] Actually, IKEA's are a step up, too. Okay. We, we gotta go to the next one. Um-Speaking of Asher Knuth, Colin, uh, I think we're gonna, uh, roll over to the HUT8 news. 48:55 But before that, a word from our sponsor, Lagos. [static noise] Bitcoin is back at 60K and hedge funds are getting liquidated again. Is your Bitcoin safe? 49:12 With Bitcoin's price drying up and whales and lending desks going under, it is more important than ever to know who controls your Bitcoin, and with Lagos, that is always yourself. 49:23 They are our preferred Bitcoin-backed lender here at Blockspace because they use Bitcoin-native smart contracts to make sure you always hold the keys to your kingdom. 49:31 No rehypothecation, no bridging, just pure self-custodied Bitcoin collateralized loans with smart contracts, and competitive rates as low as 10% APR. So if you'd like to learn more, go to lagos.finance. 49:45 All right, Charlie, we will h- hop on over here to HUT8. An- a, uh, another senior secured note. We've had a number of these this week, and this is a continuation of a thread we're starting to see where 50:02 these Bitcoin miners turned AI companies are really starting to see- be seen by the market as legitimate credit-worthy operators. As I'll note here, and we noted with Cipher, they're... 50:13 The notes that they're issuing are no longer considered junk bonds. They're in a tier above that. 50:18 Now, they're not quite in the A tier yet, but they're in the triple B tier, as with this one, and their interest rates are reflecting that. 50:25 Their cost of capital, going back to our segment with Jamie, is significantly lower than when they were Bitcoin miners and if they were private. So a few notes on this before we get into it, Charlie. 50:39 This close is for a $4.25 billion offering at 6.129% for senior secured notes secured by HUT8's Beacon Point Data Center LLC. 50:50 They're due 2042, and the proceeds are going to fund development of a 352-megawatt AI data center in Uvalde County, Texas. 50:59 The deal is HUT8's second investment-grade construction bond and is n- non-recourse to the parent company. 51:04 And what that means is that if Beacon Point fails, the bondholders can seize the project and the equity pledge for the project from Beacon Point LLC, but they can't actually reach into HUT8 itself. 51:16 So HUT8 has gated itself, the parent company, from any sort of recourse from the lenders should Beacon Point fail. I think that's important. 51:24 Something important to note, um, this is similar project-level financing that we saw with Cipher for its Stingray and Black Pearl LLCs. 51:32 Another really interesting point about this deal is that interest starts accruing in November, but principal amateurization doesn't begin until May 2030. 51:43 So about a 3.5% interest-only runway that lines up with the construction and lease ramp-up before the loan starts paying down. 51:52 So what this means is that HUT will only pay interest payments for three and a half years, and once the site is online and cash flowing, then they will start paying down principal. 52:02 And at first glance, that you might be scratching your head and thinking, "Well, that seems like a worse deal in the sense that they're s- going to pay more interest over the lifetime of the loan than they would if they were paying immediately." 52:14 But, I mean, this is a $4.2-25 billion loan. 52:19 So unless they actually have the cash flow coming in from the site, they would have to find some other means of raising that capital to pay down the loan, um, w- pay down the principal immediately. 52:30 So they're trading off the fact that they're gonna have more interest to pay over the lifetime of the loan, but they will not have to cover the bulk of it with the principal payments immediately, and they won't have to start paying that until the data center comes online. 52:43 So another kind of important note. And last point, Charlie, uh, the... There's pa-- this note is for par issuance, and this goes back to the credit rating here. 52:52 The principal is 100% of raised amount, um, which signals clean investment-grade demand, and, like, there's no book... There's no, uh, original dis- origination discount to clear the book for this one. 53:04 And this bond is triple B-rated, which is two notches above junk. HUT8's other inv- investment-grade bond for its River Bend campus was triple B minus. 53:16 So they're even m- they're moving up the ladder in terms of investment-grade, uh, bond ratings here. Yeah. Investment grade, uh, gets better. 53:24 I think the market is just figuring out how to price these and getting a little more confident. Um, I think this is pretty exciting. I've pulled up a picture of, uh, the HUT8 render. Um, very, uh, very good rendering. 53:41 Looks pretty. Uh, some shrubbery and external ornamentation. Pretty sleek looking. Maybe le- not art nouveau, but art techno, we might say. [laughs] Yeah. Um, at what point do 53:57 these, uh, miners, AI HPC factories just become banks themselves? We saw this with cars, car manufacturers. See this with, like, airline and airline, uh, manufacturers. 54:12 Uh, just the scale of capital involved at, you know... Eventually, these folks are gonna build banks internally. Look at GM. So I wouldn't be surprised if we see, uh, 54:24 long-term that the financial banking and computation businesses coalesce. So I think we've got a couple more stories, Colin. Yep. We've got two more. I don't have anything else on HUT8, so we can go- Yeah... 54:40 ahead and dive into Crusoe. Yeah, Crusoe, this one's a doozy. Not even... Well, not really a doozy. Here's the headline. 54:47 Crusoe's Project Jade up in, I believe it's in Wyoming-Uh, planned to be 1.8 gigawatts, potentially scaling to 10 gigawatts data center campus, one of the largest in the US. Crusoe announces... 55:03 Or the project has announced that it's paused, and then this morning, I believe, Crusoe announces that it's pulled out of the project entirely, or at least that hit the wire. Why did it pause? 55:15 At the request of the customer. So the project itself will continue, but this time it's Black Hills Corp, the utility providing power, who's going to, who's gonna continue working on this with the customer. 55:29 So why did Crusoe pull out? That's the big question. What does this mean for the project? That's also the question. Let me pull up the tweets. 55:40 So here's the tweet from Negligible Capital, quote, "Crusoe was working on Blackstone Inc.-backed energy company Tallgrass to develop a 1.8 gigawatt campus in Cheyenne, Wyoming, for an undisclosed tenant." 55:54 So the customer is likely a hyperscaler hitting the pause button. Um, this poster, uh, says it may be attributed to the tech dump. But wait. 56:07 Rittenhouse Research, the third time we've referenced them on the show here this [laughs] week, uh, has this great take. 56:14 Um, quote, "If I had to guess, the Crusoe data center pause in Wyoming is much more likely to be attributable to the growing community backlash than a lack of demand for AI infrastructure. 56:24 Wall Street Journal had a story yesterday on how the Laramie County Board of Commissioners pulled the approval for a proposed 5,600-bed man camp at the last minute, despite the planning commission having previously voted unanimously to advance the project." 56:42 So again, what's ha- what's, what's the ha- what's hanging these up is permitting. 56:48 'Cause when you build one of these data centers, you basically have to truck in hundreds, now thousands of people to go live in basically Winnebagos or, you know, modern, uh, quick fab, like, trailer long houses, and they're just these giant camps. 57:05 So, um, Rittenhouse then speculates that, uh, on the possible tenant, who is undisclosed, Rittenhouse says, "Meta and Microsoft already have sites in the area, 57:16 which suggests Crusoe's Project Jade campus will be leased either to Amazon or Google." 57:22 And so Rittenhouse pushes back against the speculation that it's the tech pullback because both Amazon and ra- and, and Google are raising money, quote, "hand over fist to invest in AI infrastructure." 57:33 Recall Google's $85 billion of equity last week. And so he says that, "The idea that either decided yesterday that they actually don't want or need more data center capacity is absurd. 57:45 It's far more reasonable, then, to assume that the pause allows Crusoe and the tenant to address the backlash and work with local officials on a path forward." 57:53 Now, this tweet was before, I think, Crusoe backed out of the project entirely, but I wouldn't be surprised if, uh, if this is a major factor. Um, what are your thoughts? It seems plausible to me. Yeah. 58:09 I think it's a good reminder, though, that when you see one of these headlines, there's almost always more to the story than what's being portrayed. 'Cause who knows if Crusoe's actually totally pulling out. Yeah. Right? 58:21 They, they could just be waiting and seeing. I mean, if, if they are pulling out, then that sort of situation is dire. 58:26 But that if this is a zoning issue, then that means that they just need to press pause and wait for things to shake out. It makes a lot of sense to me, though. I, I, I... 58:35 That it would maybe be more related to zoning than it would be related to, like, the market pullback. I mean, if that's the case, then you would s- expect to see a lot of these being put on pause, right? Because... 58:46 And we haven't even [laughs] seen that aggressive of a pullback, to tell you the truth. Yeah. I mean, I'm just, like, spitballing here, but I'm looking at this, and Crusoe's, uh, been an ambitious company. 58:56 They come from the world of Bitcoin mining off grid and flare mitigation, if you recall, from many, many years ago. But, um, this is from Bloomberg. Crusoe touts five gigawatts of data centers. 59:09 Um, and it says, "Crusoe's total project pipeline, including contracts and sites in discussion with tenants, is more than 40 gigawatts." Colin, you know, when I see... Look, 59:21 I can imagine companies like OpenAI with a 10 gigawatt site in the pipeline. To hear that it, that Crusoe, which is a relatively young startup, 59:32 uh, has more than 40 gigawatts, I wonder if it might just be biting off more than they can chew. But again, what do I know? I mean, that is an insane... That's, that's absolutely insane. Yeah. 59:45 That is so much freaking power. I, I have a hard time even with 10 gigawatts, man. I mean, that's- Yeah. Right? I mean, the, what, what, the biggest data centers in the US are... What is it? 59:55 Is it Colossus right now, and that's just one gigawatt? Is that, is that correct? Oh, y- uh, I think so. Colossus is two 300 gigawatt sites, I think so. 1:00:05 Well, yeah, I mean, it's planned for one gigawatt though, i, is, my, my point. And, um, I mean- We- And 10, 10 multi- uh, uh, 10X on that is crazy. Yeah. 1:00:16 We, yeah- I mean, we've talked about this a lot between, like, the billions that get thrown around and the trillions for CapEx and yada, all this stuff, and all of the gigawatts. 1:00:25 The numbers stop meaning anything until you actually ground them in what that looks like. And, and this, you can see this with money too. 1:00:32 Like, um, okay, uh, I as a Bitcoiner have been trained to accept, like, surprising amounts of money being thrown into something, and I keep joking about, like, normalizing the $300 billion slugs we threw in after the great financial crisis, which now seem trivial. 1:00:49 Right. So, like, I'm used to money accelerating. Um, but th- this is, like, actual physical infrastructure, and 10 gigawatts is not something you can print. 1:00:59 It's something that takes years.Um, I think we should go into the OpenAI 10 gigawatt- I think we should. I just wanna cap this with just a quick factoid. The average gigawatt production in ERCOT is, like, 60 gigawatts. 1:01:13 Just, just to contextualize where 10 and 40 falls within the scope of gigawatt or, or energy production in the US. All right, let's wrap it up with OpenAI's 10 gigawatt site planned in Ohio. 1:01:27 I'll try to be quick with this, but it, it's mostly, with this story, there's a lot that we still don't know because this is not, this hasn't been announced by OpenAI. 1:01:34 This is based on reporting from The Information, and I believe also Reuters confirmed some parts of the reporting, but not all of it. 1:01:42 OpenAI is eyeing a 10 gigawatt facility in Ohio, expecting it could cost half a trillion dollars. Now, that's- Just half a trilly... just half a trilly. I mean, that was Stargate, right? 1:01:53 It was the, this is basically Stargate 2.0, or maybe this is Stargate's [chuckles] rebranding at, at 500 billion. But, I mean, that's an insane CapEx burden, and the proposed 10 gigawatt data center is 1:02:08 for a s- for a campus on federal land in southern Ohio in partnership with NVIDIA. Uh, NVIDIA's potentially backing the project, [clicks tongue] 1:02:17 which is an important piece of information which we'll get to here in a second. 1:02:22 Now, nothing, again, I wanna reiterate, nothing has been signed yet, and this hasn't been announced by either of the counterparties or any of the counterparties in this reporting. It's also unclear who will own the GPUs. 1:02:36 The article said that OpenAI will control them, but I don't know if that means that they have full ownership, or they will just be utilizing and networking them, or if NVIDIA will own them. 1:02:45 And I think that's an important critical detail. Honestly, by 2020, by 2028, Bernie Sanders is gonna own a GPU. [laughs] So a few more details for this. Total expected capacity, 10 gigawatts. Phase one, much more modest. 1:03:00 A m- a modest 800 megawatts. A modest 800 megawatt. Expected in 2028, cost estimate 500 billion, lease term 20 years. 1:03:09 The developer, this is also interesting, SB Energy, a SoftBank-backed energy infrastructure company that specifically deals with transmission and energy builds. 1:03:18 It's unclear whether that means that they will actually be the powered shell operator for this site, or if they're just managing energy. 1:03:26 There are a lot of question marks on what the ownership for this would look like, and assuming that OpenAI or NVIDIA announces it sometime later this year into the future, we'll know a little bit more. It's, it's... 1:03:39 Oh, sorry, go ahead, Charlie. You've got something. Oh, I will say that whenever you see numbers which are so high that it shocks even, like, people like us, there's probably SoftBank involved. 1:03:49 Masayoshi Son can't, like, not write an insane check. [laughs] Uh, you know, so if it's one of these dizzying amounts of money, it's gotta be him, so. A few more housekeeping items. 1:04:04 It's Portsmouth site, Pike County, Ohio, and what's interesting about that is it's- Where, where is that even? I gotta find this on a map. I mean- Well, it's a... Yeah, if you could pull it up, please. 1:04:15 It's a former uranium enrichment facility that produced weapons-grade material during the Cold War before ceasing operations in 2001. 1:04:23 Now, the Department of Energy announced a public-private partnership with SoftBank and A- and AEP Ohio in March 2026 to redevelop that land, and that's part of why we're seeing it on federal land currently. 1:04:36 And this includes 10 gigawatts of new power generation and a $4.2 billion transmission upgrade, and that's where SB Energy, I would assume, comes in at least for the transmission upgrade. 1:04:46 Again, we don't know if they will actually run the powered shell. Uh, the power mix is 2 point, well, 9.2 gigawatts from natural gas, partially fu- are funded by a $33.3 billion 1:04:59 sh- uh, Japanese capital injection that is tied to the US-Japan Strategic Trade and Investment Agreement. So it's there, right on the Kentucky, Ohio border, 1:05:11 um, Portsmouth, Ohio, right here on the Ohio River even, if you will. Um- I mean, this goes back to what a lot of people were saying, including, um, I believe Andrew Burchwell, which y'all had on while I was out. Mm-hmm. 1:05:25 Ohio and the PJM market is a sleeper for a lot of the data center activity, as long as, I, I would, I would say, as long as, you know, red tape and bureaucracy stays out of the way. 1:05:37 It's where the backbone of the American industrial economy during the 20th century happened, uh, so- Rust Belt revival, baby... Rust Belt revival, baby. Let's go. La- last note on this. 1:05:51 According to the reporting, NVIDIA is backstopping this build, so they will supply the hardware and potentially provide financial guarantees for OpenAI and SB Energy's financing. 1:06:03 This plays into a criticism we've seen with the circular nature of the financing and the AI CapEx boom, where NVIDIA is guaranteeing a financial backstop for, it's the ch- the chip vendor basically underwriting the credit of its own customer's landlord, and I think this gets people a little uneasy in terms of the concentration risk of some of these builds, and the fact that, you know, you almost have to, in some... 1:06:28 Y- you know, the, the piper has to get paid, or you have to bend the knee to a, a king maker like NVIDIA to get anything done, but that's just where we're at in, in the current AI landscape. 1:06:39 All right, Charlie, that does it for me. Yeah. I don't really have anything else to add to this one. I'll, uh, I'll, I have, I got a, I got a video I'm gonna play as our, as we go out. 1:06:47 Um, this is a clip from It's Always Sunny in Philadelphia that I feel may, uh, may be a fun reference as we talk about the circular nature of money. Roll tape. That's right. How much fresh cash did we make? Fresh cash. 1:07:01 Yes. Uh, well, zero. Zero if you're talking about US currency. People didn't really seem interested in spending any of that. That's okay. 1:07:08 So, uh, when they run out of booze, they'll come back in, and they'll have to buy more Paddy's dollars. Right, uh- Keeping it moving... that is assuming, of course, that they will come back here and drink. They will. 1:07:17 They will because we'll redistribute these to the chanties, thus ensuring them coming back in, keeping the money moving. Well, no, but if we just redistribute these, then people will continue to drink for free. Okay. 1:07:28 How does this work, Mac? The money keeps moving in a circle. But we don't have any money. All we have is this. How does this work, dude? I don't know. I thought you- I thought you... What? 1:07:40 I thought you were on top of this. You're the one that came up with the plan. I ca-... Did I come up with this plan? Last night, dude. Oh. With the DMV power card and the- Oh, I blacked out. I blacked out that night. 1:07:48 Sh- dude, I'm following your lead. Oh, Jesus. Ah, shit. Okay. Okay, we have no money and no inventory. There's still something we can do. That's still a business somehow. 1:07:58 How does- Anyway, it's a shame that it never won an Emmy. [laughs] You know, they have a great bit in the show where every year they get passed up for the Best Pub Award. Yeah. Yeah. 1:08:07 And it's, it's, it's, it's an inside joke about how they've, they w- they never won an Emmy. Yeah. All right, thanks, y'all, for listening. We do this every weekday, 1:00 PM Eastern. Catch us live. Follow the podcast. 1:08:21 We're leaving Coindesk. Go search Blockspace in your podcast listener of choice. Subscribe there. This ends on Coindesk next week. Otherwise, newsletter.blockspace.media. I'm Charlie. I'm Colin. 1:08:37 And we are Blockspace. [upbeat music]