Who in the world is mining Bitcoin in Australia right now? I mean, I don't like a company going bankrupt, but whenever something like this happens, you can go into Pacer, find the court filings, and then you have all this juicy open source information about who is involved in these things. Some of the names that have surfaced from this. There's some really weird stuff in here. Like for instance, is the Chicago Blackhawks hockey team. This is real. For 1.26 million, at first blush, you might be like, 'Well, yeah, of course. Just, you know, check the boxes. You talk about you are the trade, bro. The trade.' That's not how it works. Now, this is very much a he said, she said story, riddled with drama.
What's up y'all? Welcome back to Blockspace Live presented by CleanSpark. For today's top story, Etched just raised $30 million at a $10.3 billion valuation to build AI computers in a way that no one else wants to. Big risk, big reward for our first story. Following that, we have Jay Patel of Ligos Finance on to talk about the recent shakeup at 21 Capital with Jack Mer's out and strike dropping from the proposed merger with Electron in 21. And then we also have Harry Sudok of CleanSpark on to talk about their Sandersville lease, what it's going to take to extend that tenant to Texas, and what else the company is eyeing with its AI expansion. To cap the show, we've got Gumshoe Charlie Spears on the prow uncovering the inside view of a new fluid stack data center in Oklahoma. Yeah, they said doom scrolling Facebook was going to ruin my brain. Well, actually, Mom, it's helping me with my job. So, Blockspace goes live at 100 p.m. Eastern every weekday with the exception of today because we're pushing it back for a guest. We got Harry Sudok. As Colin said, we feature the latest stories and hit on data centers, AI markets, and emerging tech. If you like what you hear, this turns into a podcast shortly thereafter. And our podcast and all of the Blockspace content, including a lot of written content you can find on sources such as Yahoo Finance, you can find those in full on our website at blockspace.dia. If you like this show, go to blockspace media because you'll love the rest of our content. This show is brought to you by CleanSpark, NASDAQ listed, ticker CLSK. More on CleanSpark later in the show. Colin, let's go back to AS6. Finally, a topic we could actually we used to know a lot about, but apparently AS6 can be used for a lot of things, including AI.
Yeah. Yeah, and we're having to learn from the ground up. But there's a lot of lessons from Bitcoin mining for this specific pivot or attempt to rewrite what makes sense in the terms of compute for AI. So the headline here from etched accelerating inference and the blog post here reads, quote, we've raised 300 million at a $10.3 billion valuation led by Sequoia Capital alongside Andre Horowits, Jane Street, Defusion, Argo, and SK Heinix. Interesting point there given the memory squeeze. We'll touch on that during the segment. We're grateful for their support on our journey to gigawatt scale. This is coming from Etched, which is a startup that was started in 2022 by two Harvard dropouts in classic Silicon Valley excellence fashion. And what they are building is, as Charlie mentioned, AS6 specifically for AI, but it differs a little bit from the ones that we've seen announced by Anthropic and OpenAI. We'll touch on that towards the end when we talk about risks for the business model. And one that could totally upend the current meta of relying on Nvidia for GPUs. So to set the or before we set some more context here, I just want to go over a little more of the history of the company. Um specifically, this is a series C round. The company has been busy over the last few years gobbling up funding wherever it can get it. And its prior raises included its first iterad. There was a $500 million raise between December 2025 and earlier this year, January 2026, $5 billion post money valuation led by Stripes, Teal, Positive Sum, and Ribbit. There was a June 30th raise um for roughly um I want to say roughly 280 million for 800 million cumulative at that point and this was led by Venture Tech Alliance which is a TSMC affiliated fund and so I find that interesting Charlie because it proves that TSMC is kind of backing a horse here with regards to a product line that could significantly benefit their business if it ends up taking off. So what is that business exactly? Edged builds a transformer only ASIC which is a chip with the transformer architecture physically hardwired into the silicon rather than executed as software on programmable hardware. So a transformer in this sense it's not the electrical infrastructure. It's referring specifically to the logic by which LLMs currently produce outputs and what the transformers allow them to do. Stripping it down to the most TLDDR explain like I'm five um explanation that I can is they allow the model to basically put every word in context with each other. So like if you're writing a sentence that said Terra Wolf is pivoting to AI in a move that will mean the infrastructure provider can provide compute other than Bitcoin mining within the context of a transformer. it would take that string of that would take that sentence and allow all the words to be processed in parallel with each other and in concert with each other so that the LLM knows that the that infrastructure provider refers to terrorolf in that instance. This is like basically the way that the LLM is allowed is enable to cognitively process all of the context within an input so that it can provide the appropriate output. The way that Nvidia's GPUs work currently is as Edged claims only about 30% of the entire computing processing power is actually used towards that end for the transformer logic while everything else is also baked into the other processes that make that GPU work because the GPUs can be used for multiple different things. So there is uh a decent amount of math and processing power in them that is geared towards functions that are not just specifically producing AI outputs. And this is where the correlary to Bitcoin mining comes in because GPUs were really good for mining Bitcoin before we realized that AS6 could do it much more quickly. And it's a similar dynamic here because what Etch is basically building is an ASIC that's only mandate is to perform that transformer logic. There's there's nothing else in it that you would have uh like with GPUs. There are no instructions for decoding. The chip doesn't the GPU chip doesn't know in advance what it'll be asked to do. So it needs circuitry to read instructions, right? To make sure that it can actually process the task. Um there's a scheduling and control flow circuitry to handle branching within the GPUs and again there's flexibility for other workloads but etch it the chip is uh just specifically for this transformer logic and etched claims that versus Nvidia's 30% for its GPUs it can produce 90% efficiency for the entire computing processing for these tasks. I've got a few more things, Charlie, but just want to toss it to you really quickly for second takes and color.
Yeah. To summarize, these new chips can do what the GPUs do much cheaper, lower power, and at greater speed. The the kicker is is that they kind of have to be married with the model. So that you it it and maybe you're going to go down this road here, but um when models are changing so rapidly and we have such competition at the frontier and and we don't know like what even the the meta for building these language models will be in two to five years or at least I don't um it can make it difficult because the hardware and the software if we went to call a model software they have to be built kind of in tandem. So it is depend you know the the chip architecture is and how effective it is dependent on the software. There's also an interesting dimension here Colin on the memory side of the things. This is why I think SKH investing in this is really interesting because uh as we know we're in a multi-year memory crunch and this is why SKH Samsung and others Micron are ripping because we can't manufacture the memory. It's very difficult to scale up and everybody wants it. AS6 do not need onboard memory the same way that GPUs do. In fact, as you mentioned, the the GPU only being used like 30 40% for inference, the actual processing. The rest of that is actually largely used to kind of batch and keep track of the instructions in the memory. And this is one of the reasons why these the the new GPUs have so much memory uh so that they can hold all those instructions whereas an ASIC doesn't really reply. It doesn't apply as much here. The the instructions are embedded into the chip architecture itself. So um I can speculate as to why SKHIX might be investing in here. Probably a diversification play, but also uh you know they have tons of money makes sense to invest in chips. I'll throw it back to you.
Yeah, it makes sense to hedge, right? I mean, if you're if this competitor comes out and some people are hailing it as the an Nvidia killer, then maybe it would be best for you to have a piece of that so that you don't lose significant source of revenue once GPUs become obsolete if they do. And that, to your point, is the big question mark for all of this because Edged is specifically betting that the transformer architecture for how LLMs work currently will be sticky. And if that goes out the window, then their entire business model is in jeopardy because they're building their chip specifically for that. And this works because if the chip only ever runs transformers, you don't need instruction coding and the instructions are the same every time. So, it's fine and it's going to do that one task uh efficiently and better than the GPUs can. But if AI models converge on another form of processing this information and the logic behind producing the outputs, then etch a etch as doesn't work. And that that's kind of the last that's the note I wanted to leave on Charlie when looking at the competitors with regards to uh custom silicon and AI A6. um specifically you know we have Amazon's Tranium we have Google's TPU um they give up general purpose flexibility but they keep architectural flex flexibility and so they're not nearly as uh pigeonholed in the event of a change of of form fac or of form factor here for how for the transformers and for the logic behind the LLMs um whereas etched gives up everything for maximum efficiency on one workload. And also looking at OpenAI's Jalapeno and also uh Anthropics chip that they're reportedly designing with Samsung. Um they're just a little bit different. They're an inference optimized accelerator, not a fixed function transformer engine. So, um, even if the Frontier Labs, uh, so the front even, so even if the Frontier Labs building custom silicon around their own models aren't going as far as Etched is, uh, there's kind of a signal that Etch either has an edge that no one else dared to take or a signal that the people that are actually working on these models think that maybe the transformer only bet is too aggressive and they don't want to go that far because their A6 won't be nearly as constrained as etched in in the sense of a change. Lot of people taking shots at the king. The king being Jensen Huang and and Nvidia and CUDA. I mean, we have anthropic with AMD story yesterday. Anthropic with Samsung also building a custom chip. We've got OpenAI with Broadcom's Jalapeno and we have yeah Tranium. We have all sorts of other, you know, edge chips, but Nvidia still remains king for the foreseeable future. However, uh yeah, will the Kings uh will will who who will take a successful shot at the gate height? Who knows? We have Jay Patel in the backstage. We're going to talk about the latest going on, the latest drama in Bitcoin Treasury Land, but we'll bring him on up here after a word from our sponsor, CleanSpark. 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, and we are setting the standard for what's next. Learn more about the intersection of energy and Bitcoin at cleanspark.com.
Alrighty, let's get Mr. Jay Patel on the stage. Jay, welcome to the show, sir. How you doing?
Yeah, fantastic. And um I know there's a bunch of other things going on in finance and treasuries and yada yada yada, but the hot news this week related to the type of things we talk about is Jack Valor stepping down from XXI Roman numeral 21 uh Capital Jay. What's going on here? Why did Jack step down? And what did you think of his apology video or I'm sorry, his announcement video?
Yeah. Um, so I guess his stated reason was basically there were differences between the board or majority ownership and him in terms of the vision that they saw for the company or at least the path that they saw to get to that end goal. Um, in which case I guess it makes sense like obviously he doesn't want to continue operating a company where he's a minority shareholder especially if he's bringing in strike. Um, it was interesting though because from day one there was obviously going to be this kind of contentious situation where he's the CEO of the public company 21. There was a plan to merge it with his private company Strike. Obviously, the public shareholders want to pay the least they can for Strike and Strike shareholders, Mullers included, wants 21 to pay the most they can. So, I don't know if there was a I don't know if the falling out was purely on like what was the valuation they were going to merge strike in at, but I imagine that could have been one of the big factors here. Um, and then I guess the next question is really like where does 21 go from now? Because obviously, you know, Tether put in a bunch of capital, they bought out Soft Bank. Um, I know that they're trying to push the mining business, but obviously Bitcoin mining has not been in favor by the markets in any way. So,
Yeah. Yeah. I mean, what do you think is the clearest path path forward for them? I mean, is is there an argument to be made that you just take Electron, you turn it into an AI play, and then you just have 21 Capital as kind of like the piggy bank? Assuming that the Treasury model actually has any, you know, desiraability with the with within public markets now.
Yeah. So I'm interestingly of the view and I think this is probably interesting in the you know the fact that there was like the the news about orange juice and you know the Lyn Alden back or was it Selenus you know Lyn Alden and the folks at Ego death running uh a fund you know basically acquiring cash flow businesses cash flow positive businesses and buying Bitcoin. I'm of the view that like these public companies, these treasury companies, I think the market's coming around to the realization that if you just have a pile of Bitcoin, like I'm not going to pay you a premium just to hold the Bitcoin for me, right? You can go the Micro Strategy Strive route and try to do the kind of financial alchemy and financial engineering and monetize the stack that way. Um, and whether you agree with what Strategy and Strive are doing or not, I think it's, you know, it's reasonable to expect that if we're back in a bull market, they will trade at a premium to NAV because they're able to do the things that are possible with a big stack of Bitcoin to increase Bitcoin per share. Um, I'm not as sold on this idea of like we have a bunch of Bitcoin and we're going to do AI and mining on the side to accumulate more Bitcoin. I'm almost thinking like, you know, if I believed in that kind of business model, which is cash flow positive, steady income, buy Bitcoin, you know, I would go buy like Proctor and Gamble stock or, you know, any of the tobacco companies or something like that and just hold the Bitcoin for myself. like um I I think that realistically the model which you know maybe MERS was pushing for is like you need to get a company that is actually able to use the Bitcoin in some creative fashion whether it's through deploying it as capital towards mining or financial products or whatever else to actually generate more BTC because people aren't going to pay you a premium to just hold on to the Bitcoin even if you have a cash flowing business on the other side. it maybe that's where some of that tension came from that you were intimating at earlier between what was going to be the actual right price for strike especially consider that you have I don't know uh exactly what strike's revenue profile is but you have the mining business on one side with electron then you have strike on the other side and so the question then becomes okay well which business is the one that's actually the most important for the cash flow of this company to make the treasury company actually more than just a glorified fund holding Bitcoin. You you Oh, go ahead, Charlie.
Well, I I know there's a question, but the one burning on my mind, and I I kind of vocalized this yesterday, is like, okay, what about the intangibles? Like, Jack was the frontman for this, and that's kind of what he was paid to be. And that's every good Treasury company play had to have a a front man, a hype man. However, uh that's also a fall guy basically. Um, to what extent do you think this is like a reputational thing or just like he's got to get out and he doesn't want to be tied to this thing from a reputational angle for the next six to 12 to multiple years?
Yeah. I mean, I think there's definitely some aspect of that. like most of these treasury companies there is the front m you know there's there's the the podcaster you know Twitter personality you know whatever whoever it is that's kind of I don't want to say shilling the stock but telling the story of what it can be and why you should buy the stock um I think in Jack's position it's kind of tough though because he doesn't have control of that company right and so maybe his his perspective was really hey if I'm attaching my personal brand to and I'm not going to have control long term of the decisions we make or direction we go in. You know, maybe that's a the risk he didn't want to take. On the flip side, though, there's kind of like the double-edged sword of like you already kind of attached to your personal brand, right? like he had been pitching the fact that 21 was going to do something that no other treasury company had in that it was going to build profitable Bitcoin businesses and you know really kind of follow through on the promise that a lot of treasury companies had made and obviously you know most of the media for 21 was just jack to his followers and his audience and so I'm sure there's a decent amount of backlash from folks who bought into 21 believing that they were going to own a piece of strike at some point and you know it doesn't seem like that's gonna happen.
So wanted to switch gears to another unfortunate story in the realm of Bitcoin treasury companies and we've talked about it before but we need a kind of an update from you and that's micro strategy or sorry strategy old habits everyone die hard strategy is down 5.5% today which doesn't surprise me that much considering the entire market's actually getting routed right now. Nvidia is down 2%. Tesla down 14.3% on the day. I mean I know it's no longer the golden boy of Elon Musk companies, but still that's an incredible bloodletting. And the thing that surprises me about it is or not, it's not totally surprising, but Bitcoin's down about 2% today, but strategy is suffering much worse. Strategy also down like 9% this month, while Bitcoin's barely up 1%. And Stretch is still teetering. It it it almost broke 90 earlier in the week, but since then it has fallen off like a rock and it's back to where it was basically a month ago. Actually, a little bit down. All that to say, Jay, we've seen strategy sell Bitcoin. We've also seen them sell common stock and then not deploy it and just add it to the cash reserve. Has that done anything to ensure the market to what they see as existential risk for stretch and for strategy? Or is the market signaling that hey that's that ain't doing it chief?
I I I think you know as we talked about last time I'm still in the same camp that I was before which is you know at some point you have enough dividend reserves that getting more dividend reserves is not the thing that's going to get stretched back to $100. And I think that, you know, the market is basically screaming like no matter how much you continue to dilute common shareholders and just sit on more cash, like that's not going to be the thing that moves the needle because in in reality like you know, a couple hundred million against all of the potential converts and you know or you know the debt that could be um you know due in the next two years. Um, I think that's the bigger thing, right? Is like if you look at the difference between Stretch and Seda, Strive hasn't had to deal with as much pressure from the fact that they have this debt overhang. Um, and, you know, in an up market, you know, there is a sense that, oh, this debt will get equitized. There's not actually, you know, an outflow of cash. And so, maybe, you know, we had the RA rally um earlier in the week and folks thought, hey, we're going to head back up towards 100K and we don't need to worry about this debt. But the longer that the market thinks that we're not going to be above 100, 120k, wherever you need to really clear that debt, um I think that's going to be the bigger issue. Um and obviously, you know, strategy has a tough job to navigate this because the worst thing that they could do is use the cash to pay back the pay down the debt or repurchase it because they tried that once and obviously that's what kind of kicked off this whole whole situation. So, I I do think that they're kind of in between a rock and a hard place. And maybe it's just survive until Bitcoin price rallies.
I mean, that's that's pretty bleak, man, when you think about the kind of ultimatum that they're having here in in a way, or rather a dilemma in the sense that there doesn't really seem to be a good option. I wanted to flag one tweet really quickly because you retweeted a response to this uh saying that strategy I believe it was from yeah Josh man saying strategy seems to have a short attention span and that specifically was underlining this disclosure from one of their filings quote our current intention which is subject to change and our sole and absolute discretion is to adjust the monthly regular dividend rate in such a manner as we believe we'll maintain stretches stock trading price at or close to its stated amount of 100 per share. And that was in a a quote tweet with a reply to this that said stretch is the right idea but without a real buyback program it will continue to be volatile. So my next question for you Jay is with that in mind is the only thing that strategy can do to actually address this problem with stretch to buy back stretch at this point and then building on your prior response does that just further endanger strategy because then they have less cash to actually address these converts if the puts are exercised in a year or two?
Yeah, I think look, you could probably get stretched back up to 100 or near it by doing a buyback program, but I I think that that probably puts a ton of pressure on the common because what you're saying is we're going to dilute the common just to keep the preps towards 100 even though we can't issue any more preps for liquidity. Um, and then you got to think about like, okay, when these converts come due, if you're not gonna be able to tap the press, you know, there's only there's only one one party left to screw over, and that's the common again. So, I I don't I don't see them doing that. Um, I think the short intention span is across kind of all parts of the capital stack, though, right? You know, before they said they weren't going to issue common below 2xmnav and then 1.6. Now they're issuing it at or around one. Um, and then clearly, you know, I don't think that they're going to take extreme moves to get STRC back towards 100 right now if that puts the rest of the company in danger. Um, but I think that, you know, that's the risk with like the kind of company strategy is, you know, all of their followers, everyone reads into every word Sailor says, everything that they say on an earnings call. And I don't know if it's them not being careful enough about what they said or you know their view evolving but obviously you know people kind of cling on to um the the guidance in the past even if it's changed.
Jay Lios Finance thank you so much for your time today. Uh I hope that our conversations about Bitcoin treasuries get more abulant in the future as opposed to doing feel like doing like a postmortem. So thank you for your time today. All right, cheers.
Love Jay. We have Harry Sudek in the audience. We are going to bring him on up here in just a moment. Talk Clean Spark right after a word from our sponsor, Luxer. This episode is brought to you by Luxer's Commander, Bitcoin miner management software for enterprise operations. Luxurious Commander gives you real-time fleet monitoring, bulk remote commands across your fleet, and intelligent minor. That's an automated profitability engine that runs every 5 minutes and tests your fleet's power settings against live energy and hash rate markets. Oo back test show 10% improved profitability with intelligent mining versus binary mining. Commander Pro is $100 per megawatt or a 25 basis point pull fee adder, but you can also try it for free for 60 days. So, if you'd like to learn more, go to luxur.tech/commander to get started. All right, we've got chief business officer of CleanSpark, Harry Sudok, in the wings. We're going to be talking about Sandersville and what comes beyond it. Let's get Harry on the stage. Harry, welcome to the show, man. Thank you for joining.
Good to see you guys. Well,
Harry, prior life, we might be grilling you on ASIC orders and maybe even ask you to weigh in on strategy, but CleanSpark came out with a banner AI deal earlier. Um, and or a month or so ago, a few, sorry, last week.
Last week. I lose track of the timeline, man. I mean, there's so much news flying around. Can you give us a brief breakdown of the Sandersville lease and specifically um how y'all got to this point? Like when was the groundwork laid because this seems I mean y'all announced your pivot later than some of the other Bitcoin miners and this deal seemed to come out pretty quick considering um the the pivot timeline. So give us a little background on this.
Awesome. I'm I'm happy to um really proud that our first AI and HPC data center project is going to be in Sandersville. Um it's been a flagship mining location for us for close to four years. And so I think that there's just a lot of a lot of poetic continuation for our business having that be the first one. Um you know, let's talk high level about what the what the transaction specifics look like. It's a 250 megawatt campus. We're going to be turning that into 175 megawatts of critical IT compute. The headline number on the deal is 6.6 billion. What's really important is that it's a true triple net transaction. So there's a lot of you know complexity out there in the market like what do you know does triple net really mean triple net? Like how do those margins and cash flows break down? This is true triple net which means that we expect you know near 100% NOI margin on that 6.6 billion. It's a 20-year transaction. So I think only one other one of those is in the market and I believe we're the only one to be interacting directly with a high investment grade counterparty as the other side of the tenant relationship. Let's rewind the clock. Talk back to um middle of last year. you know, we were not the first Bitcoin mining company powered land acquirer to say, 'Let's go build HPC data centers.' Um, you know, our friends at at a number of different shops made that transition sooner than we did. And that was deliberate because we wanted to look at two key indicators in the market to validate the thesis. The first was, is there going to be a durable demand profile for tokens that's going to grow over time? I think everybody knows the answer. The demand for for AI tokens and intelligence tokens is is ridiculous. Um and then the second is are the hyperscalers and the largest counterparties going to be willing to do business with our sector. Um part of that is a a asset quality perspective like we don't own acreage in Northern Virginia or in you know downtown Chicago or or you know right in central Dallas. like those are traditionally the hottest data center markets. And so are we going to see the largest technology companies take more of a geographically diverse demand viewpoint for their rack space? And both of those were proven true. We aggressively migrated the business to be ready for this capability and this type of transaction. And the list of customers isn't that long. And so we spent a ton of time engaging across all of them and and ultimately landed at this transaction for Sandersville.
Um Harry, I'm curious. One point with regards to the announcement is that the tenant itself was not named. This is a trend that's becoming increasingly common with a lot of the Bitcoin miners in the cohort. HUD 8, for instance, has has demurred from talking about its own tenant at Beacon Point. What's the rationale behind that specifically?
You know, look, I think it's from a from a shareholder and an open market perspective, like it's not good for the news day and everybody wants to know exactly who it is. Um, but on the other hand, it's also just a huge positive because the the close to the vest approach that these tenants are asking for is a signal of how aggressive demand really is. Um even if you know you saw the the progression with Cipher as an example they came out with their um with their 70 megawatt lease with they didn't they didn't even say which site it was at in the original announcement and then ultimately through additional filings especially around the financing we got a lot more detail about that project um but I think that the viewpoint on the other side of the table more broadly you know not not just for for our particular circumstance but across the sector is that even 60 or 90 days of anonymity is hugely beneficial in the data center strategies that these companies are rolling out. And so if you're in a market where 60 or 90 days is is high value, then it means that demand is extreme and supply is constrained and and I think we've seen a very similar behavior in the market.
Yeah. And that makes sense to me. We asked Asher Gnut the same question we had him on recently and he said more or less what you just laid out that the tenants are asking for this partly because it chips away at a competitive edge they have when they're negotiating for other deals. So um makes total sense to me. I just it's very notable compared to the early days of these AI pivots where a tenant was in like every release and then now it's becoming much more commonplace that tenants aren't announced at all. So Charlie, you looked like you were about to jump in there. You know, Harry, you mentioned financing and uh this again, it seems to me that the that is kind of where the meta has shifted in that creative financing deals. Uh can you speak to like the the broader trends in financing these sites overall and any insight you can give us on specifically the Sandersville build and how uh we could think about that from a financing standpoint?
Yeah, I mean I think look the the the first thing to know is that the capital intensity for these projects is, you know, orders of magnitude more than a traditional Bitcoin mining build where you, you know, you were looking at maybe up to a half a million bucks a megawatt for infrastructure and then a couple million bucks for servers at the peaks. Um, this is a totally different ballgame. It's 10 to 12 million a megawatt for infrastructure alone. And then when you layer the chips on that, you know, the chips aren't part of our obligation for this project, but the value of those is going to be, you know, three plusx the infrastructure price tag. And so you're looking at, you know, billions and billions of dollars of assets sitting on on these data center campuses. So the financing is interesting because while the capital intensity has gone up, the creditworthiness of the underwriting has also gotten a lot stronger. The way that these leases work, especially under a triple net structure, is that you're getting the best guarantee for cash flows um of anybody in the market. You know, the the high investment grade counterparty that we're working with on this, you know, they fall into a short list of folks who have the true best of the best kind of credit ratings that are out there. And so it means that while traditionally I think Bitcoin mining companies used a lot of ATM products to finance growth then there was a big wave of convertibles to finance you know depreciating assets. Basically what we're seeing in the HPC data center landscape and it's going to be a playbook that we're going to look to mirror. These are comments that um that Gary made Gary Vearelli our CFO and president who talks about this more eloquently than I do. Um he you know you know he loves it because it gives us the opportunity to use project level debt financing oftentimes for our sector that's looked like the high yield market but they're also examples of you know more investment grade kind of construction real estate credit products. But what's important is that these these financing sit at the project level. They don't sit at the parent level and they're collateralized against the project, the power contract, the hard assets that sit there as well as the lease value and economics in the data center that the money ultimately gets used to fund. And so the collateral package is incredibly high quality. The counterparty who sits on the other side of the lease in our case is incredibly high quality. And that lets us unlock this lower cost of capital even in the
event that we're going down this more capital intense road. It's very, very accretive to shareholders. It's protective of share count and dilution over time and it lets you scale the business without needing to scale the equity base.
So, if I'm hearing that correctly, I'm not trying to put words in your mouth, but looking at potentially some project level financing, something secured, not putting the whole parent company at risk in something that is unsecured further up the stack.
And that's representative where the sector is, right? Like we have the benefit in our view of this second mover advantage. You know, you called us late to the party, but we said fashionably late. And so ultimately other people have done an incredible job building their businesses down this road. And so what we have the opportunity to do is, you know, we had this opportunity in the lease negotiation process is to kind of pick off the menu of the things that we thought were the most attractive and accretive. And then we're going to have the same opportunity on the financing side to say, hey, there's 20 deals in the market. We loved these features. We didn't love those features. And we're going to be able to pick and choose a little bit to arrive at what ultimately we think is most accretive from our viewpoint.
I'd like to ask you something for kind of a blunt take and a blunt question for the sector as a whole. I'm curious if you think that the true thing separating the serious operators at this point from the Bitcoin miner pivots to AI is this financing piece where unless you can actually get investment grade credit for a project, I struggle to see how you're going to be able to procure enough to actually pay for some of these builds. And I won't name names specifically, but there have been a few names in the usual cohort where they have pretty grand plans and they do have tenants that they've signed up, but they're not approaching project level financing or they don't have investment grade backing. And so we're seeing things like equity issuance. We're seeing things like converts. The debt stack is not maturing to the extent that I think that you would hope to see seeing a Bitcoin miner going to an AI pivot. All that's kind of a long-winded way of asking to in your mind is the investment grade credit piece kind of a make or break for whether or not these companies can really move into this industry at scale.
It's a hard question because I think like number one, equity financing and convertible financing like those aren't dirty words. Those are great tools that get used sometimes incredibly effectively, sometimes less effectively is what we've seen, you know, in our market, but even more broadly than that. So, I think that the key is are you marrying the right source of capital to the right business activity? In our view for the data center build piece of things, project level financing, debt financing, it just creates the best return profile, you know, for our business as we scale in this kind of way. But I think that those other types of capital formation opportunities like have a time and a place depending on your business and depending on your growth story because at the end of the day when you look at our portfolio and our asset mix, you know, we've got Sandersville now off the table and leased. We've got 885 megawatts of power in Texas that's under LOI with the same counterparty. We've got some other sites that have some AI applicability, but when you look at our, you know, if we've got a 2.1 gigawatt portfolio and we've got 1.15 gigawatts either leased or under LOI and we've got a segment of those remaining megawatts that are still going to be used for mining or maybe they're not big enough or in the right markets for an AI use case, we want to be hunting land and power and we spend a lot of time investing in our pipeline. And so when we think about site power and growth acquisition, we're not thinking about project level debt to do that. We've got a balance sheet we're able to use to do that. We've got Bitcoin back collateralized revolvers that we can use to do that. We've got all the different types of market activities. So to pair it, Gary again, you know, he would say we have optionality which I totally agree with. And what's important, the way that we think about it is that can we put incredibly high quality counterparts at our projects because that unlocks the debt component for us. And then can we continue to rinse and repeat the powered land acquisition thesis that has power that has gotten us to where we are today and I think we've been tremendously successful at. I think we've added a gigawatt since just this most recent fiscal year. So, we have a growth engine there, but really running a business that's as capital intensive as the HPC data center business is. It's about understanding all of your different funding levers and then all of your different business application layers and marrying the right sources to the right uses up and down the stack. So, that's a long way of saying basically there's a time and a place for lots of different things. Internally, we've got a very clear-eyed view about the type of tenant quality that's important to us because of what it does for the financing and what it does from a confidence in the longevity of the cash flows. Because the other thing that I think is not talked about maybe enough is that it's not just about getting these projects financed. It's also about being able to have sufficient confidence that they are going to pay the lease bill every one of those years all the way out for two decades from now. And so when we went through our, I don't know I wouldn't call it speed dating because it's not quite that quick but ultimately, you know, when we went through the exercise of match finding for our portfolio having a viewpoint on the financing was top of mind but just underneath that was making sure that the counterparty was one that's going to thrive for decades into the brave new world and be able to fulfill the entire duration of the lease term.
Matchmaking for data centers. Triple net is the new 6'5, six figures. You mentioned the 800 plus megawatts in Texas. That actually leads well into one of my final questions. We got a few more and then we'll get you out of here. It's kind of a two-part here. KBW Steven Gladola argued that the Texas LOI exclusivity was more significant than the Sandersville site itself. And I think his reading on that is most people expected the Sandersville announcement. It was if you were reading through the tea leaves, you could see it coming. But the Texas expansion could be massive in the sense that it's, you know, triple what the Sandersville lease would be on a gross megawatt basis. First question, do you agree with that? Second question, what specifically needs to be done and what milestones need to be met in order to get that LOI to be actually executed on?
I'll never say anything is bigger than the name at the bottom of the paper on a definitive lease agreement just because the amount of work that it takes our internal team to go through that exercise, the rigor, the technical expertise, all of that, the work product that they brought to bear to get that over the line is just unbelievable. And so on behalf of their hard work, I'll never say anything's bigger than the definitive agreement. I think from a scalability perspective of our portfolio, I agree. You know, I think that it was important to us to demonstrate that this is, you know, CleanSpark is not a project-based company. It's a scalable platform that we're building on. And when we think about our core competency, it's about the ability to acquire and mature powered land assets into HPC ready campuses. And then it's about the financial and technical acumen around that to convert those campuses into commercialized assets and then built assets and then cash flowing assets. And so, you know, I think that when we went through the process, getting zero to one is the hardest part in anything, right? Any business, that first dollar of revenue is the hardest. That first megawatt of leasing is the hardest, you know, always. And then as you get into a more repeatable cycle, you're able to refine your process and enhance, you know, what you're able to achieve, whether that's value or speed or quality, you know, all the different metrics that we're going to be assessing our portfolio on a regular basis. And so getting into that repeatability position was critically important to us because we want to move quickly and take full advantage of the second mover advantage that we think we have. So I think that, you know, and listen, Stephen is incredibly sharp and I think from the market's viewpoint, I think that he's probably right. I think we came with a bigger, you know, a bigger push out of the gate than would have potentially been anticipated. But the other thing that's really important to us is that we want to be able to give the market a tremendous amount of confidence that when we go out and we secure an additional powered land asset, we're going to be moving through the commercialization process rapidly because speed is king in all things. And so we want to move with discipline and deliberate approach, but also on a very, very aggressive and accelerated time frame.
Charlie, did you have something? Because I have a closer unless you've got something else.
I got one more kind of curveball for him and I'll let you close. Okay so Harry I don't know if you saw OpenAI announced they're doing 3.2 gigawatts in Georgia and CleanSpark has I believe over half a gigawatt in Georgia. It's kind of like a stronghold of your mining operations and they're doing it off Georgia Power and I'm a little bit curious, can you give me any insight to what's going on with Georgia Power? Do you think that utility can scale to service 3.2 gigawatts and like what needs to happen for OpenAI to actually realize that scale of a site in Georgia?
Look, I mean, I think that from a utility system perspective, Georgia functions a little bit differently in that it doesn't necessarily run all of its own gen while they do distribution, transmission, and distribution. They can also use the open market to fill their wires on a more kind of liquid grid-to-grid basis. And so the way the RFP process works and the bidding process works to secure long-term capacity doesn't necessarily mean they have to have that gen on their system, but it does mean that they need to have the import capabilities cross systems to be able to achieve that kind of scale. So look, I think it's possible. I think it's achievable. I think the tougher part is going to be around the actual delivery of the power into the data center over what kind of time scale. But this is, you know, we saw this in Bitcoin mining many years ago which is that when you bring a demand profile that the market has never seen before to an existing behavior but the economies of scale get unlocked for the first time the levels of innovation that are able to be achieved on a rapid basis are incredible. You know, I think about, listen, I'm a left curve guy. So, when we look at like the PSUs, the power supply units that the ASICs ran against, if you go to the airport, you know, all the outlets that sit on the terminal seating areas, the PSU that runs those outlets is basically the same size and wattage, etc. that a Bitcoin miner is running against. If you buy them in the airport, they're like $600 when you're building the airport because there's all this red tape and you got to do it and bid it out and competitive and there's state involvement and whatever. Those don't cost $600 when you put them on a Bitcoin mining rig anymore. They cost, you know, $170, maybe $70 and keep scaling down. But that's just because if you're going to build a whole airport, maybe you're going to buy a thousand of them. If you're going to build a Bitcoin mine, you might be buying a hundred thousand of them. And so the supply chain got brought into focus with a wave of demand that was for a component that already existed, but it had never existed at that type of concentrated scale. And so I think we're seeing a similar behavior with the way that AI and HPC data centers are working is that data centers have been in demand for 30 years, but they've never been in demand at a gigawatt or 3.2 gigawatts at a single location and able to realize full economic value in that way. And so what about the supply chain? What about the construction process? Can we reimagine because we've got an order for a million units, not a thousand units? Those types of dynamics, I think, haven't been broadly understood or digested yet. But I think it's where we're headed. I think that, you know, the growth and proliferation of this industry is going to be relentless.
Last question, Harry, in terms of charting expansion. Sorry, not going to give you all a chance to rest on your laurels as I know y'all wouldn't anyway. So, Texas obviously on the docket if that LOI gets executed. Where else is CleanSpark looking at expansion? Would you look in Mississippi or Georgia or Tennessee for your current sites? I know some of those are smaller. I believe all of them are smaller than the Sandersville site. I'm wondering if there are opportunities there or if you'll look to Greenfield similarly to what y'all are doing in Texas going forward.
Yes. And we're looking at the existing portfolio. We've got a double-digit gigawatt evaluation and growth portfolio that we look at. From a pipeline perspective, we're pretty conservative about what we put in our, this is ours and we are going to point your focus to it. We think it's an incredible asset and we have certainty. Once we sign a power contract with certainty of delivery, that's when we talk about it as part of that expanding 2.1 gigawatts of contracted power. But there's a huge piece of pipeline that sits out beyond that that frankly we don't talk about often because we want to give the market sure things. And it means that we're saying no to more than we say yes to by a wide margin because not every project is mature enough or living on the timeline it needs to or in the right jurisdiction where we're going to have the type of community tailwinds that we've enjoyed in Sandersville. So we're looking all over the US. We've seen some interesting stuff outside of the US, but we're focused here to begin with because we think there's just so much fertile, you know, room to grow. But I think that it's, you know, it's a lot of the same kind of story around here. There's valuable growth in Georgia still. There's valuable growth in MYSO. There's growth, you know, to be had in Urkott once some of their permit, you know, their approval process shakes loose a little bit later this summer. There's behind the meter opportunity. There, you know, all this wide range of electrons moving into their highest value form. That story is in the early innings right now and so we're just excited to look at new projects and kind of put them through our internal process and add to that 2.1 as quickly as possible.
Well, Harry, thank you so much for joining, man. We'll have to get y'all back on later in the year once that LOI is signed. Not using if, using when. I know you can't, but I'm gonna put...
I would never dream of coming out, but I look forward to coming back.
Harry, thank you so much, man. Have a great week.
Awesome. Thanks, gentlemen.
