Power Analysis
Power Analysis is the premier destination for investors tracking the Bitcoin mining and AI infrastructure space.
Hosted by Anthony Power (@cazenove_uk) — one of the most respected voices in Bitcoin mining stock analysis — alongside Bryce McNallie (@McnallieM), Power Analysis delivers in-depth company interviews, earnings breakdowns, sector news, and market commentary — direct from the executives building the next generation of digital infrastructure.
Whether you're actively trading mining stocks, researching AI infrastructure plays, or just getting started — Power Analysis gives you the edge with timely, accurate coverage across the companies that matter most.
New episodes dropping regularly. Subscribe and never miss a move.
🐦 Follow Anthony on X: x.com/cazenove_uk
🐦 Follow Bryce on X: x.com/McnallieM
📊 Anthony's Patreon: patreon.com/cw/Cazenove_UK
📊 Bryce's Patreon: patreon.com/c/McNallieMoney
🌐 poweranalysis.io
For information and entertainment purposes only. Not financial advice.
Power Analysis
Data Center Demand Surges As AI Infrastructure Players Race To Deliver!
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
We track how the AI infrastructure trade shifts from headlines to execution, where megawatts, grid regions, and build timelines decide who gets paid. We also debate co-location vs bare metal, using real company updates to show how contracts, financing, and delivery convert into durable revenue.
• Bitcoin price action and why we zoom out on volatility
• Strategy preferred shares buyback, yields, and how the capital levers work
• Sector heat map pullback across AI infrastructure and mining names
• Data center demand growth toward 2035 and why PJM deserves more attention
• ERCOT capacity, gigawatt sites, and why “ready and waiting” power matters
• Keel’s deal expectations, rerate logic, and what to watch into earnings
• WhiteFiber NC1 delivery, retrofit speed to market, and contract economics per megawatt
• Core Scientific and CoreWeave buildout progress, revenue ramp math, and why who funds capex changes margins
• AI bubble question, plus the trade-off between co-location safety and GPU ownership upside
• IREN’s three-layer stack framing: infrastructure, compute, software and services
• Soluna’s transparency, renewable power pipeline, and the market signal from notable buyers
Let us know in the comment section below which of these companies is your favorite in the space or your top pick currently in the sector!
Anthonys Info:
Website:
https://www.powermininganalysis.com
X:
https://twitter.com/cazenove_uk
Patreon:
https://www.patreon.com/Cazenove_UK
Bryce's Info:
X:
Patreon:
https://www.patreon.com/McNallieMoney
Merch:
https://shop.mcnalliemoney.com/
Website:
Business Inquiries:
Email - powermininganalysis@gmail.com
Welcome And What’s Ahead
SPEAKER_02Hey guys, welcome or welcome back to the channel McNally Money, the official home of power analysis. A big episode to kick off the week for you. Updates out from Keel, Iron, Core Scientific, Mara, and White Fiber, just to name a few. We're going to be talking through the latest news and headlines in the AI infrastructure space. Before we do, take a second, smash the like button, guys. Big help to myself and the channel. Anthony absolutely loves it. If you're not already subscribed, McNally Money, feel free to join. And let us know in the comment section below which of these companies is your favorite in the space or your top pick currently in the sector. With that being said, let's get into today's video. All right, guys, Monday afternoon, and we've got a great one in store for you today. We're going to be talking about all things data center demand, capacity, deployment, timelines, capex, uh, giving some examples and what we're looking forward to in this round of upcoming earnings.
Bitcoin Check In And Saylor Moves
SPEAKER_02So it should be a good one, Anthony. Before we get into the AI side of the house, we did want to do some housekeeping on the Bitcoin side, just under 65,000 today, and some very interesting moves from Michael Saylor, the king of Bitcoin.
SPEAKER_00Yeah, um, not too crazy over the weekend. We saw it drop um down to sort of 63 and a half thousand level, but it's been showing a little bit of uh positivity since then, and like you say, uh hit 65,000 today. Hopefully, the rest of this week we can get back into that 66,000, 67,000 range and continue that upward trend that we've seen over the last few weeks. But um, as I say, Bitcoin uh you know uh does have its volatility. Uh, we're mindful of that, but we're also mindful that you have to zoom out a little bit and uh look at the bigger picture.
SPEAKER_02You definitely do now. Bigger picture, Michael Saylor. What are your thoughts on this, Anthony? I mentioned some interesting moves from Strategy. Uh, he's all over the place now. Buying Bitcoin, selling Bitcoin, buying shares, selling shares. Uh, what is the most recent update we've got from Strategy here?
SPEAKER_00Yeah, we've we've had an update today that they've um you know used some of their uh cash to repurchase uh 288,930 shares for about 25 million at an average price of $86.52. The share price has been depressed lately. So Strats are using that opportunity to buy back some shares at a cheaper price. Now, you know, this is just another, you know, uh leave they can pull to sort of like you know to keep it balanced. The idea was they would sell as many preference shares to raise capital when the preference share price was over $100 uh to buy Bitcoin, and you know, in recent um months they've had to sort of like um tweak that strategy a little bit. We've seen the Bitcoin price um you know rise recently, but the the pro the price of the preference shares has been lower than $100, so they can't sell any uh those shares at that price there. And you know, they've even had to try and raise the yield as well on the preference dividends well, uh, and that's gone up a couple of basis points as well to try and entice people to buy back in there. But this buying of 25 million is probably you know added maybe you know two dollars to the preference share price there. Uh, we can see that you know from the screen there currently $88.41 per share. It was $86 price to the view there, so um, you can see a bit of volume there, one just over a million shares there. Uh SATA playing a bit more closer to its target, the hundred dollars there. So both of them using very, very similar um uh strategies. And if you look at the yields on these two here, I mean, you know, I look at what I get from my bank at the moment, which is around about three to four percent. Um, and if you've got a business bank, it's probably even lower than that there. But the yield on these particular stocks at the moment, well over 13% for each of them there, and I know that you Bryce have got um some of the stretch um preference shares uh to attract that nice that nice yield uh that it is at the moment.
SPEAKER_02Yeah, my personal strategy is to use SATA. Uh stretch is preferred to act as basically a cash account. They pay a daily dividend, so your money sits there, you're earning daily interest. Strategy though, Anthony, you mentioned the preference shares of stretch uh trading actually below 90. Strategy moving up pretty nicely on the news today, getting closer to 100.
SPEAKER_00Yeah, stretchies um uh well we come on to the shares in a minute that we come on a daily basis, but sort of booking the trend at the moment, up 7.28% as at the time of the podcast, $98.34. And you can see there on the one-day chart there been as you know just over $94 at the start of um at the start of the day there. So um, you know, a nice, a nice bump in, a bump in price.
SPEAKER_02It is unfortunately that bump in price doesn't uh transcribe into the AI infrastructure players.
AI Infrastructure Stocks Pull Back
SPEAKER_02We've got a few buck in the trend. DMG uh looks to be having an okay day, but a lot of the names we're gonna talk about actually today, the keels, the Wi-Fi, the uh irons, giving up some ground.
SPEAKER_00Yeah, DMG, Foo Foo, ABTC, sort of like you know, just hovering above there. DMG have been a decent day, 4.4%. There's no news out to suggest anything. We know they're Emily trying to wear that LOI into some sort of an agreement about 50 megawatts of compute power, and if you can get an agreement on that, there I'm sure that will certainly add value to the company because 50 megawatts, although it didn't sound a lot, it's a significant amount. And if you're looking at what the current uh contracts are providing in terms of revenue per megawatt, at about two million uh dollars plus there, that could easily bring in a hundred million dollars um of revenue per year uh just for that 50 megawatts there. Now think of the market capitalization the moment the market capitalization is in Canadian dollars, so you'd have to do the conversion be price, but the con Canadian value at the moment this company is just under 100 million dollars, it's at 96 uh million um and or just shy of 97 million in fact there. So um the the revenues from that one contract would bring in you know effectively nearly double the market capitalization um for the companies, which is uh you know pretty astounding, and you know, they're looking to even increase the the power at the site that they've got there at uh Christina Lake, and and and maybe that 50 megawatts becomes even bigger because the PUE is significantly lower that part of the world than it is in the in the Texas region that we talk about on a day-to-day basis. Canada doesn't have the heat problems that they have down in Texas, and therefore you've got that more natural cooling environment, certainly in the winter months. I mean, I wouldn't like to you know to to live up there without make making sure my my uh accommodation was fully heated during the winter months there. But for GPUs that you know are very, very hot when they're when they're on there, that cooling environment will provide a natural cooling uh for these GPUs, and therefore enabling you to spend less on ancillary power to keep the GPUs at a temperature uh for delivering you know maximum output. But uh they're having a good day there. But you can see from the other side of the coin here, we've got the likes of Core Scientific. We'll talk a bit more about them today. They're bucking the trend at the moment, they're down nearly 10% today. Uh Cypher as well, another big company. Well, you know, we've talked about them on recent podcasts, they've got significant uh contracts. Hutt as well, keel uh we'll talk about today. So some of the big hitters, some of the some of the companies that are expecting to uh deliver great things, uh you know, now and in the future, they're all having a challenging day there. So the majority in the in the red there, you can see from the day range, it's pretty much um since markets um open there. Most of these stocks are currently um trading at their day low. And when we look at the 52-week range, just to see per total performance, as I highlighted last week, since that 22nd of June, we've seen a big drop in the sort of like in the last four weeks of these stocks, um, and most of them were getting close to the 52-week high as at the 22nd of June. No longer that way there. There's probably a couple of stocks that are maybe halfway towards their 52-week high, uh, those being Cypher and HUT that we just talked about there. Uh, so you know, a challenging start to the week there, but we know that you know a piece of news can turn this whole thing around. We don't think uh this is uh you know a bubble at the moment. We'll talk a bit more about the you know AI and HPC, uh, whether it is a bubble or not. We don't think it is, we're using it on a daily basis. So uh, you know, um it's only going to get bigger and and obviously better, and as it's getting bigger and obviously better, these companies that are delivering HPC AI will then be able to charge uh a more reasonable figure for people to use it. And at the moment, um it's a bit like using that analogy. Remember when Uber go into a city and they have to compete with the current tax in the city? What Uber do is they charge really low rates for for people to use the apps. Once you're comfortable paying that amount of Uber and using those apps there, then what happens then is you know you'll start putting the competition out of business, and they'll have to do something really different differently. And uh, we've seen this um very, very often. I've just come across this this last weekend, actually, in my hometown where Uber had been there for about 12 months, and I was speaking to some of the the businesses there, and think there's only like two major taxi companies in the area now. There was probably a whole host of them, but they can't compete with Uber, and some of the prices I was paying for my trips this this last weekend were phenomenally cheap. It just didn't make sense to walk anywhere when you were paying that amount of uh price there. And I think the same is going to happen with HPC. People at the moment are paying, oh I'm paying 20 dollars or 20 pounds or 20 euros a month. That's not gonna be the same. You'll be paying for usage in the same way you pay for electricity and gas at the moment, and when that occurs there, those real um revenues that go through to these companies uh that are delivering the compute power um will then feed through to the companies that are delivering the co-location deals for these GPUs to sit in there. So I expect the future is bright for these companies here, but today we're having a bit of a red day, Bryce.
SPEAKER_02Yeah, they are, and we're gonna talk about those different layers of the AI cake extensively today. I wanted to get your thoughts on the heat map, Anthony. Anything standout here? You mentioned that pullback over the last month, pretty red across the three first columns. I know we talked market cap as well. Uh, keep these numbers in mind as we move into the next chart in terms of gigawatt data center growth capacity. Uh, but a tough day, tough week in the market here.
SPEAKER_00Yeah, it has been a tough week. We have five or six companies in that sort of $10 billion bracket, not too far far long ago. And now we can see just iron and up and iron are getting closer and closer to sort of like 50% of where they were at their high. In fact, the share price $35.65 is less than 50% of where it was at $76 there. Market cap uh $12 billion. We'll see that change when they come out of their earnings because they will have issued a significant amount of shares from that $6 billion ATM they've had over the past sort of three or four months, and we'll know exactly to the date um when their earnings come out of how many shares they've issued there. So likely see that market capitalization will change dramatically on their earnings date there. HUT, um, also fortunately, still remaining as a double digit in terms of billion dollar market capitalization, but then Wolf Cyber Cypher Riot, who've been you know over $10 billion, you can see the pullback there. 31% down, Wolf is on just a one-month change. Uh, you can also see BitDear down 30%, Keels down 31%. These are some of the big names that are you know delivering at the moment. The year change, year to date change and the one-year change uh looks okay. But look at the year-to-date change now, just for iron, down six percent year-to-date change. Long time since we've seen that as a red number there. Um, also bit dear down one percent uh from a year-to-date change there, uh, bit digital down and ABTZ down uh largely because of the fact that they are effectively uh you know treasury management companies in terms of that their business revolves around holding Bitcoin or Ethereum um on the balance sheet there. So uh the challenge is ahead there. One year change is still showing an average of 79% up there, but that was showing you know significantly three uh figures in terms of percentages. Uh Hutt's still showing over 403% up. Uh Keel now only showing 213% up. Remember, that was four or five, six hundred percent up about four weeks ago. So you can see the drop, the impact of that drop, um, and the timing uh which has affected it. So Bitcoin actually having a better period over the last three timelines there, only one percent down today, one percent down of the last five days, and actually um in quite reverse to the companies with eight percent up um over the month there. But when you start talking about year today and one year change, there you see that the difference then 27 down year to date and 45% down on the one-year change. So, uh and you can see that correlation with the 45% down with ABTC and Bit Digital. We don't put the Ethereum price on here, but it's pretty much correlated to Bitcoin generally, and we'll you know, we if we looked at those periods as well, it would be double digits plus in a similar way to what we're seeing for Bitcoin there. So Ethereum and and and and Bitcoin affecting Bit Digital and ABTC from that year, year today and and one year change. Um, as I say, a lot more red than we saw last week. We were getting to the stage where two of the columns were green, but um, the move today um and the move on Friday, not helpful to us at all, Bryce.
SPEAKER_02No, it's not. Maybe we throw Ethereum on here for future uh updates. It would be nice to compare. Now, let's get into the AI side of the house, Anthony.
US Data Center Demand To 2035
SPEAKER_02We've got a big lineup today. We wanted to set the table with this slide here, talking about data center demand. So just megawatts or in this case gigawatts allocated to data centers from now until 2035. Some interesting takeaways from this chart. Not only the amount, the scale, the rate of increase here, Anthony, but you made an interesting observation about the ratio PJM, ERCOT, MISO compared to some of these other markets. Uh, wanted to start there.
SPEAKER_00Yeah, I mean, you can clearly see when we get to 2035, and bear in mind this is probably the largest uh number we've seen in terms of uh demand in the US for compute power there. 194 gigawatts. I mean, I think when we had Dan Roberts on the podcast uh last year, uh we were talking about four gigawatts at the moment rising. We know that figure is probably today closer to between eight and 11 gigawatts, maybe nine gigawatts of power, um, you know, in the in the in ERCOT itself. And you can look at 2025, it's around about just shy of 50 gigawatts there, and where we're gonna be in 10 years' time from that, that graph is really um steeply curving upwards towards that 200 uh gigawatt mark that it's took uh you know 15 years to go from a couple of gigawatts all the way to 50, and then we're gonna use the next nine years to effectively grow three times more than what we've already achieved. So you can see the demands here, um, the supplies coming through because we can see that there. But you're quite right to mention we talk a lot about ERCOT, and when we do a look at like the comparison of ERCOT, we mentioned PGM because we know that Kiel are delivering sites in PGM, but we don't give it enough credibility. If you look at 2035 there, they're gonna have probably uh 70 gigawatts or 75 gigawatts of the whole 194. That's a significant amount, and we should have been maybe sort of highlighting this, but we have on the podcast said where keel are lining up some of their sites, like Panther Creek. One thing that Ben said when we asked him about the area and you know, was it uh was it was it the right uh area to to do this? He said he was surrounded by hyperscalers. You can see from there that PGM uh has got a massive amount of uh you know data center power at the moment and going to grow significantly over these next nine years.
SPEAKER_02Yeah, good good commentary there. Now, also wanted to get your thoughts on ERCOT. You and I did a bit of digging. There's about, say, 10 gigawatts of capacity already allocated to data centers now. If you project this out to 2035, I think you and I figured maybe in the 25 to 30 range, we know a lot of the companies that have gigawatt sites in ERCOT coming online within those years. It seems the companies we talk about are taking up a pretty big piece of that pie.
SPEAKER_00Absolutely. And just just you know, just off the top of my head, all we have to do is, I mean, you know, Sweetwater itself is a two uh gigawatt site of which 300 megawatts has already been energized, even though they haven't got a client for the site there. Sweetwater one and two totals two gigawatts there. That's already got the power approved for that. And Dan Roberts has clearly said when that power is needed, it'll be switched on as required. Um, that's two gigawatts of that of that junk there, and then you look at the likes of uh Corsi Carnot, which is a one gigawatt site. You can also say Rockdale and at Rockdale, you haven't just got right at Rockdale, you've got BitDeer at um at Rockdale next door as well. Both of those companies heading towards maybe 700 megawatts each, there's another 1.4 gigawatts. Um, if you think about where the power is, it's about the companies that we talk about regularly. And you know, if the hyperscalers are out there looking for power, then they don't really need to look too much further than some of the companies that we talk about regularly. They've got the power already, they're not having to wait in that queue in the in the ERCOC queue there to to await approval for their power. These companies are effectively ready and waiting for you know the right client to pay the right uh value for the for the for the for the for the assets that they have, and most of these companies now are actually starting to to build out their sites, right? Even though it hasn't got a client, it's already got a design to drive 756 megawatts of compute power at their Corsicana site, um, which is phenomenal because they've changed the design and effectively got an extra 50 megawatts of compute power. We just talked about DMG having an LOI for 50 megawatts, and that could bring in a hundred million dollars a year. Well, right, have managed to find effectively another hundred million dollars benefit by just tweaking the design that they had. Um, let's hope that you know uh in the not too distant future, i.e., hopefully roundabout earnings, they start to give some update on where their potential clients are because I know a lot of people out there uh now, and we're including that group there, we want to see an announcement from Riot. They've had some good announcements at Rockdale with AMD, but the flagship site, which we've been to visit twice in the last uh couple of years, that's the one that uh people are looking at there. That's the one that we've been you know focusing on in terms of location, in terms of fibre, ticks every box, it's got its own lake there, so they've got that water for cooling, etc. etc. And and that's the one we're looking at there. But there are many other sites. Sweetwater is a great site there, and we've been to some of the other sites there in Texas, like Granbury, like like uh Dorothy One, Dorothy Two, and Soon to Be Dorothy 3. All these sites have power, and uh when you're looking at you know where can the power come from and how it's going to grow between now and 2035, these are the sites that are gonna be included in that uh in that box for Urcot there by by no means.
SPEAKER_02Oh, for sure. Yeah, you think about Mara, Saluna, Cypher's got a big presence there, HUD 8 has property in Texas as well. Uh now, speaking of deals and anticipation of deals, an interesting one for you here.
Keel Deal Timing And Rerate Math
SPEAKER_02Kiel has now come out with their Q2 results. I just got an email from the team. We're gonna be having Ben on to follow up, uh, ask some questions there, which will be great. That's Monday, August 10th. So following uh next week, which is already a pretty busy earnings week. But what we're watching for here, Anthony, is these deals and this re-rate. And we wanted to bring up this slide that we talked with Ben about last time. Interesting now, hindsight being 2020, to look back at the numbers they used on this slide and essentially the re-rate or re-rate potential that Keel felt they could get as they sign these deals.
SPEAKER_00Yeah, this was an interesting slide. I don't think any other company brought out a slide showing this type of explanation. Uh, we do know that many of the companies, and keel would be included, you go back to 22nd of June, I can tell you now the share price of Keel on that day was $7. The reason being is I did sell a portion of my keel, you know, believing that was a good price to sell out there. And since then, as the share price has come down, I've added some of those shares back to my portfolio. So took a little bit of a profit during that period, and it never hurts to take a little bit of profit. But as the share price at the moment is around about $4, it's clearly from that chart there in the position where no deal has been uh announced yet or no lease execution. Um, you know, with with Ben now explaining there's three deals by the end of the year, and when we look at um some. Of the updates uh from some of the analyst on Wall Street, and our good friends um at uh Van Eck have actually put together a you know a table of when they believe some of these deals will be announced. They've suggested by the end of summer that two deals will be announced by Keel, and they do have uh these three uh sites that they're looking at. They've got Panther Creek there, that's a gross capacity of 350 uh megawatts. They've also got Sharon, 110 megawatts, and the site that's you know most likely to be announced very early is the 18 megawatt site, Moses Lake, which is in Washington. Now, interestingly enough, Panther Creek and Sharon are in that PGM energy market which we've just talked about there, and another site that they are developing, which is still going through all its permitting and application, etc. like that, is the biggest site that uh Keel potentially have. That's the Scrubgrass uh site there. That has the potential for up to 1.3 gigawatts of power. Remember, their total um pipeline of power is 2.2 gigawatts, so the vast majority of their power is actually in PGM, and we don't give that PGM as much uh notice as we do to ERCOT because predominantly most of the companies that we talk about do have sites in ERCOT. But this is just highlighting how you know where PGM, where the where are the companies there? Well, keel are doing their fair share, but you can bet your bottom dollar that the likes of Meta, the likes of Microsoft, Google, um, Anthropic, and some of the other big names out there are also operating in that PGM area because power is the bottleneck and they'll go to every um uh effectively uh power run um you know organization to seek uh the best opportunities going forward. So um I'm hopeful that when when Ben comes out with earnings, we maybe get something towards an announcement or at least a timeline of an expectation because now we're down to literally five months to announce three deals, and you know that the the the clock will be ticking and the expectation will be there to deliver that by the end of the year. But um, I'm hoping we get some sort of an update with earnings to say, you know, this is where we are, and um, you know, even if it's the stage of like LOIs have been signed uh in readiness to go through and finalize a full agreement, that'd be really helpful. It doesn't look like too many of these deals are priced in with the share price at the moment, has come back down to four dollars from seven dollars just in the last four weeks, so that's nearly a drop of you know 40-45 percent in the last four weeks. Um, and you know, like with many of the many of the other shares out there, having that significant pullback over that period. Let's hope some positive news comes out with a with an update and see that share price get back to where it is, and as the previous chart suggests, heading towards the $9 when more deals are announced. So there's plenty of scope there.
SPEAKER_02And I'll give you guys a hot tip. You mentioned some timely sells we were making on Keel. We've also been making some timely buys uh to get back into the position, and we document all that on Patreon, uh individually in real time, and on the channel membership for both of us on a weekly basis. So we had a few questions about where do you go for what? There's the answer to
WhiteFiber Delivers NC1 Fast
SPEAKER_02that. Uh now, talking about delivery here, Anthony, another company we've been very keen on here, White Fiber. They spun this out of Bit Digital. Bit Digital is the strategic asset company or parent company. White Fiber is the AI uh infrastructure component of that. It appears they've delivered NC1 or very close to delivering NC1. We've got a nice little flyover video here talking in past tense, uh, very similar to what we just saw in terms of the electrical completion for iron at Childress, but great to see the companies we're covering again, not only signing the deals, having access to the power, but now actually delivering these facilities.
SPEAKER_00Yeah, we've we've been covering uh Bit Digital and its sister company Wi-Fi, you know, for the last three years, and you can see how they've grown. You know, they they weren't the first to announce HPC, but what they were, they were the first to announce a sizable deal, even going back to you know the end of 2024 when they signed that first 50 megawatt uh deal uh to provide HPC uh for a for a large client there, and you can see from that you know from that period onwards there they've been signing clients, and I think when we looked last time, it may be 23, 24, 25 clients at the moment. This N scale deal is one of the biggest that they've done there. They've they've literally took um you know what was probably like a factory building and converted it um into you know a tier three building for N scale there. Now the IT load 40 megawatts the term, 10 years. All these contracts are 10 years and uh and and plus, and there are probably you know a number of extensions to this this contract. What what gets my eyes is the total contract value of this of this particular deal $865 million, and that's more than two million dollars per megawatt. That puts it in as one of the as one of the top in terms of revenue earners by megawatt by any of the companies that we talk about on a daily basis. There, now the great thing is they've executed this really, really quickly. I think it's took them, I think, you know, less than seven months to to do. Uh, the retrofit uh enables that fast speed to market and lower costs, they haven't started from the ground upwards and found the land, you know, they've they've took a building, and this is a model that they're going to continue to use. Now, the great thing from Nscales uh position is they get priority notification for future site capacity. And we do know white fiber are looking to at least get another 40 megawatts out of this site. I believe there's a total at the moment, 99 megawatts of power there, and uh you know that should be able to deliver about 80 megawatts of compute power. That's certainly um in their sort of vision to deliver that there, and um you know you can see from the from the slide potential expansions towards a two times initial deployment by the end of 2027, and the project financing expects to support remain deployment, and that's what happens with these deals. You sign a deal, and then the next thing you know, within two or three weeks, you announce how you're going to pay for it. You know, generally, a lot of the time, the companies have raised an element of the capital, and then they're able to go to the big banks or the big institutions and and and and use some sort of leverage to use that capital raise to then deliver the rest of the of the financing via you know a convertible uh note or uh you know uh you know an interest paying paying bond uh for the period of the of the contract there. So there's lots of different ways uh they can do it, and because of the types of clients that these companies have, um, you know, is really dependent on then what attractiveness you get in terms of interest rates, and it's all right saying, you know, yes, we're getting you know two 2.1, 2.3 million dollars per megawatt if you're if you're not able to attract the right client that will give you a low cost of capital when delivering that, because you know, you sign a contract and then it's all about the costs incurred within that contract, and that's what they have to work out. And if you remember when Patrick came on the channel recently, he told us you know about this latest $19 billion contract for their 401 megawatt uh site. Uh, you know, they were able to you know race on via an ATM, and he was saying, you know, that's probably about five basis points, and then they were able to go to the market because this was a uh a very valued uh client there with you know that would open doors and get financing around about the six percent. So again, 11% weight average cost of capital for delivering that there, um, it it's you know that will then you know set you at what sort of margins you're going to uh you're going to achieve. Because the rest of the cost in terms of you know delivery, they can probably have a reasonable expectation of what they're going to be. They've done enough of these uh types of contracts now to know what their build costs are, what their operating costs are. The key points for me is you know, revenue in terms of contract value and financing costs when you've got it before you've actually managed to get those costs. Those are two elements that are you know that are not altogether in your control. I mean, yes, you'll negotiate as high as you want, but if you've got a hyperscaler the other side, and we've seen that with in terms of applied digital or with hyperscalers, when I looked at the value of those contracts there, that the revenue suggested about 1.6, 1.65 million dollars per megawatt. That sounds quite low, but it's when you go and arrange the financing associated with that, that's when you're able to sort of like you know, uh earn a lower cost of financing, giving you that margin back. So, you know, it's not just look at one figure, you've got to look at the whole whole contract, the whole pricing of these. And as we start seeing more and more of these contracts being delivered and handed over, we're gonna get a better idea uh of planning for the future from the out from the analysis part of it. You know, end of the day, these companies will have that experience of negotiating, but we'll see more and more information. We'll be able to sort of like make some sensible suggestions about what future contracts look with regards to their pipeline.
SPEAKER_02Yeah, and NScale has been growing a lot in their own regard. We would expect white fiber grow alongside NScale there as they pursue uh going public. I know they've done a lot of capital raises, and then you talk about the retrofit strategy there. Uh, initial rejo knee-jerk reaction is yeah, great, you have the building, the frame, but think about the permitting, the environmental, the water, the power infrastructure. There's a lot of benefits in terms of the timeline there by retrofitting these buildings.
Core Scientific Execution And Revenue Ramp
SPEAKER_02Now, one of the companies we've seen probably one of the earliest uh large-scale contracts alongside Bit Digital White Fiber was Core Scientific. We actually went into one of their first uh decks covering this. So I think it was Q2 2024, talking about their timeline and in terms of delivery, Anthony, it's almost spooky how accurate they were in terms of bringing these sites on for Core. We've I'll let you talk about what we expect going into earnings, but this is really where the rubber meets the road uh to your last point.
SPEAKER_00Yeah, and they they were the sort of like I think they set the the ground rules of what was going to be provided in terms of transparency. We were fortunate to be invited to Denton when they announced their first 200 megawatt deal with Coreweave, and we were there, sat with about 40 of the Wall Street analysts in that room there, uh, listening to Adam Sullivan, the CEO, talk us through that deal there and also talk about the potential for further deals going forward. And at that point in time, that was phenomenal. They were the like the first big deal named out there, and then it literally was like every few weeks later for the next you know few months, they started to announce extensions to that deal, and what we see now is a delivery plan target of 590 megawatts, which originally uh you know, by the end of 2026, so set the marker out there. Um, it looks like now when we look at where we are at the moment, um, they're very much on track with everything. Dalton is the is the will be the last of the um uh phases to be delivered, um, and that's on track to be completed for early 2027, and that's very much a similar time frame when I might with a core scientific team at the AIME Summit in London a couple of months ago. Um, they were telling me that you know by the end of June, by the end of July, there, hopefully between 400 and 450 megawatts of that total contract compute power will have been handed over to Core Weave with the remaining by the end of the year or very, very early 2027. So I still think they have a time frame there. One other thing they did also say, as well as as much as you've seen press releases about delays from from core scientific perspective, they were very, very much on target throughout this contract there and haven't had any penalties at all uh pace to them. Now, when you talk about earnings price, what are we gonna see? Well, you know, they're they're building up to that 400-450. So if we said they're at 400 megawatts by the end of June, then that means to me that they're probably two-thirds of the contract completed. And if you look at that $850 million per year in terms of um average annualized co-location revenue, uh, two-thirds of that is very, very much close to approximately about six hundred million dollars. So if you know looking at that from a daily basis, that'll be bringing in close to two million dollars a day of revenue from July the first onwards. If we hear that figure, 400 million by the end of June or even 450 by the end of July, you're certainly looking two million or just higher than two million per day of revenue. That from a monthly basis, 50 to 60 million dollars from a quarterly basis. That's getting on for you know a couple of hundred million dollars as well. So you can see how it's going to start increasing. But if people think that they're going to be have received 450 megawatts worth of revenue um in this last quarter, then they won't have done. They started to receive revenues in the first quarter, they should be advancing quite nicely, and you'll see that sort of escalator all the way through. Now remember, uh, when we look at that $850 million of revenue, and it was the first big deal, and we were very pleased with it. One thing that wasn't as transparent to everybody at the time, and it's come out gradually there is is when you look at the the the cost of capital for core scientific delivering that 590 megawatts in terms of those sites that we've highlighted to get them to tier three. Well, the vast majority of that funding actually is coming from Core Weave, they are paying for the vast majority of refurbishment. I believe Core Scientific have to make good on the last 70 megawatts of that 590 megawatt deal. You can see how much that is of a savings. When people go, Oh, we've got uh you know this other client's you know getting two million dollars per megawatt, that other client has to then go out and borrow the funding at a significant cost to deliver that there. Core Scientific haven't got that cost in their in their uh in their income statement at all. Uh core we will be paying for that, and the other great thing is at the end of the contract, the site is handed over to Core Scientific. So, you know, yes, it will have been depreciated down, and you know, obviously it'll be maintained all the way through, but that becomes an asset on Core Scientific's balance sheet, and so another tick in the box there. So when you look at the deal there, don't just look at the one figure, there's a lot more going on than people probably realize, and you can see from the slide there this 12-year contract is going to provide over $10 billion of revenue potential, it's going to achieve anticipated 80 to 85% profit margins. There's no there's very little um interest costs in there because they're not having to borrow significant amounts like their peers, and they've also got the ability for two five-year options taken into a total of 20-22 years on this contract itself. So um, I I think this was one of the best contracts out there. There are some good contracts coming out now, and as we've seen, you know, with the fact that you know that there's a real demand for power, um, you've seen some of the pricing go up, but no doubt about it. You know, this is still a great deal for Core Scientific. And bear in mind, Bryce, I think they've got a few more megawatts in hand to issue uh once this site's completed, or as this site's completed, with potentially other clients.
SPEAKER_02Yeah, I was gonna say part of this transaction or proposed transaction with Core Weave, we actually found out about their other portions of their portfolio. We've seen that number now grow up to four and a half gigawatts of gross utility power. So almost in line with a BitDear and Iron, someone of that sort. Uh, but it's interesting, we've been following Core Scientific literally since bankruptcy. And now hearing that they're one of the first companies to be pulling in literally hundreds of millions of dollars per quarter. And the nice thing now, Anthony, is they can start recycling that money back into the business, start that flywheel a lot of these companies have talked about, rather than always having to go to raise money externally. So we're pretty excited for that. We're also excited to see if they announce any of these other sites. Now, you just mentioned AI demand. This
Bubble Debate And Competing Business Models
SPEAKER_02has been probably one of the biggest questions we've toyed with over the past few years. Is this a bubble? Is it similar to the internet bubble? Is this real demand? Is it circular revenue? There's been so many different kinds of debates or conversations. Uh, Fred Thiel actually sat down with Natalie Burnell for an interview and was asked specifically about some of the aspirations with MARA, their shift to AI, uh, obviously their Bitcoin mining strategy long-term. So I'll play you guys the clip here and then we can get your thoughts.
SPEAKER_03Today, power is is the thing, which is why ourselves and many of our peers have pivoted because you get a lot more money per electron if you're doing it for AI than for personal. Find a tenant. Right. Because you can't get something permitted that you don't know how you're going to build, and that what you're going to build is very dependent on the tenant, unless you're going to be a um a bare metal shop uh like an iron where you're essentially um building a data center, putting GPUs in it, you're just going to sell that GPU capacity to somebody else, rent it out.
SPEAKER_02So a couple of interesting comments there, Anthony. The first talking about the pivot to AI, the value of the electrons, and then comparing or essentially calling out one of his peers, iron, and their bare metal strategy.
SPEAKER_00Yeah, and and and you know, Fred seems to be along the lines of they're going to remain a Bitcoin mining company, but they obviously want to get into this space with the rest of their peers because you know when you see the level of these contracts and the and the and the unpredictability of the Bitcoin price, um, you know, it makes sense to to have some predictable revenues and margins coming into the company when you have the amount of power that uh that Mara have. Remember, they're one they're still one of the biggest Bitcoin miners out there. I think there's only a bit deer that can um that can uh provide a larger hash rate out there. They have a bigger hash rate than CleanSpar. We'll probably find out on earnings what that hash rate looks like at the moment. But they're not shying away from mining, they've got a quite a sizeable uh power pipeline, and you can see Fred's definitely keen on utilising some of that power. Think of the fact that they've signed up with uh Starwood, who've produced you know six or seven gigawatts of data centers, and uh they've made a number of uh you know recent acquisitions in terms of you know companies that they've purchased, especially in my neck of the woods, uh, that already starting to deliver HPC on that behalf as well. The key thing with Fred though is you know he's not moving away from Bitcoin mining and he wants to try and you know use these two strategies in tandem. I've seen him speak um at the AIM summit. Um, effectively, you know, where you know where there is downside in you know in energy requirement for HPC, then he believes Bitcoin mining can provide that load balance to data centers delivering HPC. If they're not using power 100% of the time, um, you know, 100 of all the time 24, 7365, then why not have miners there with the ability to switch on as you have that um as you have that power become available rather than just you know not utilising at all? And generally these companies have power in cheap areas. I mean, one thing that's great about MAR is they've they've got their own behind their meter power, they have a wind farm, and I'm I'm I'm sure you know uh as you know the challenges with Bitcoin, you know, with energy prices, that's maybe an opportunity for them to continue mining uh Bitcoin, not having to go out and buy all the latest machines because if you're using your own source of power and it's extremely cheap, then you're gonna get a margin out of some of the less efficient mining machines, and that's potentially what what Mara could do there. But I do know it seems now that their focus has gone away from being one of the very few remaining pure players. We at one time it was Clean Spark and it was and it was Mara, and you could probably put BitFoo Foo and BitDeer into that equation as well. But we've seen from BitDear, they've been announcing HPC for a considerable amount of time, and Clean Spark have also pivoted, and and one of the last ones obviously Mara has now pivoted, uh, just like it's uh uh one of its biggest peers from the from the from 2021 riot platforms did uh you know literally a couple of years ago, because these transparent contracts that are coming out now look a little bit too good to be uh not be involved in, and so you know, Fred, although he's still five years in the Bitcoin space, wants to maintain that and he still believes the place for Bitcoin. He also believes that there is definitely a space for power uh for high performance computes, and um, you know, he's very very much in favor of having. Of that power uh dedicated to that strategy alongside his Bitcoin mining strategy.
SPEAKER_02He is, yeah. So some interesting remarks there, good discussion. Now, Mike Alfred uh also making some comments, interesting timing here, and essentially uh kicking back on a few topics. So saying AI infrastructure is not a bubble, this isn't an overinvested sector. Actually, owning the GPUs is an even bigger bet on the AI infrastructure space, and specifically talking about that multi-layer strategy. Now, we heard Fred Thiel say, hey, Iron is going after bare metal, but we know with the acquisition of Marantis, and we've actually got a clip from Kent Draper here from the Rays conference as well. Uh, they're going after the full stack. So I wanted to get your thoughts on Mike Alfred, the AI bubble, and this full vertical integration strategy, how that maybe compares to Fred's direction.
SPEAKER_00Yeah, I mean, he's looking at obviously as a non-executive director of IRAN, he's he's believing that the AI application, you know, will prevail. You know, when when each contendant, you know, owns the power, the LAN, the cooling, and the data center capacity. That's something that IRAN are doing, they're fully integrated in this system here. They will sell the actual compute power to a client. Whereas the vast majority of companies that we talk about on a day day-to-day basis, including you know, Core Scientific, HUT, Cypher, Digital, TerraWolf, and many others, are using uh you know the co-location model there. Um, you know, they believe that uh, and we've had Patrick Fleury on the channel uh exactly giving his view, and Ben Ganyon also giving his view on the fact that the unknown for them is is that you know the life of these machines and whether they can be utilized, you know, 24-7 and bring the necessary revenues in to pay not just for the machines but to provide a margin. They feel more confident delivering the sites for these big operators to bring their own machines, manage their own machines, actually arrange a uh a power deal with the power provider, and it's effectively then you're just paying for a facility charge, then but these facility charge uh bringing significant margins. Now, time will tell who has the right uh strategy. Um, I quite like the iron strategy. I'm I I will say that I've sort of many podcasts I am in favour of the co-location because when I look at risk and I'm a risk practitioner, you know, I want to make sure that I can manage risk, and at the moment, um, you know, with GPUs, uh, you know, we are seeing you know some of the older machines starting to last longer, but we don't have enough data to say if they're gonna last six, seven, eight, nine years. Um, most of these machines are bought over a five-year period, so they're gonna be depreciated over that period of time there. So if they are to achieve six, seven, eight, maybe even more years, then you're starting to bring in some significant profits because all the depreciation has been um has been removed, and so we we just don't know yet. But we're hearing from the you know the A100s and the H100s that they're still working well and they're delivering the level of uh revenues um you know higher than probably anticipated for a machine of that of that age. So there is some optimism there, and uh, you know, there's also the likes of uh Wi-Fi who are dabbling into this area here, they've got a number of clients. Hived Digital are also looking at you know, they also provide this service on a smaller scale, and Iron has been providing this service on a small scale for the last couple of years. They've got this major contract with Microsoft, which is due to start imminently, as Horizon One is in the process of being finalized and handed over, and so we'll start seeing maybe in a couple of earnings time uh the level of sort of like not just the revenues um but also any costs associated with delivery of those contracts because that's also the unknown. What are the what are the contract deliverables? What's the payment mechanism like if you're delivering GPUs for a service? If you've got an expected uptime for these GPUs or or performance of these GPUs, what happens when it's not quite at that level? Will you have to pay or be penalized uh with a payment against the contract payment there? And so that again is the uncertainty. Uh, there's less risk with a co-location, but there's probably more upside, and I think that's where Mike's coming from. You know, um, if you choose to own them, you have a good reason why it's potentially more lucrative, but also potentially more risky. And I think he's that's the point I'm trying to make. And he's made that point at the conference, and so we are aligned on that. Um, one thing we all say, we've been to the children's site, and um, the team there are phenomenal, and so you know they've achieved every target that they've set over the last couple of years in terms of building that phenomenal uh Bitcoin mining hash rate up to 50x a hash, and then all of a sudden the switch occurred then to start, you know, all focused then on delivering HPC and delivering a CSP model as opposed to a co-location, which many of his other um colleagues and peers have decided to go down.
SPEAKER_02Yeah, and I will say if you vote with your wallet, obviously Mike Alfred owning a lot of shares in Iron as a director, but he's actually been purchasing a lot of shares in the colo players as well. So think of Saluna, we'll talk about that in a second. Uh, cypher digital. So again, diversifying within his own portfolio between the two strategies. Now, we just talked about that vertical integration, the various different layers of the cake. You talk about the LAN, the GPUs, the cooling. Uh, interesting couple clips here from Kent Draper at the RAISE conference. So we know Iron was the keynote sponsor for this entire conference. They didn't stream it live. We got our hands on a couple of these clips. I'll play them one at a time here, Anthony, but some very interesting dialogue.
SPEAKER_04How much of the next five years gets decided at the foundational layer, which means energy data centers, before a single GPU is being wrecked?
SPEAKER_01Yeah, well, as was mentioned, uh, it is a fundamental bottleneck in the industry today. So I don't think it's any secret in this industry that demand is far exceeding supply today. And just to touch on your previous question, I think one of the biggest misunderstandings in this industry is still how long it takes to bring infrastructure online in the real world relative to this exponential demand cycle that you see in the digital world. And with every breakthrough in technology, whether it's increased efficiency of GPUs, whether it's lowering the overall cost and efficiency of compute, it actually only increases the demand.
SPEAKER_02Now we've heard Dan Roberts say a few times the physical world can't keep pace with the digital growth. Essentially, what we're hearing Kent say here, talking about the strategy IRIN's put together and the fact that those initial layers, the infrastructure, the land, the data centers, that's what everything else is built on.
SPEAKER_00Absolutely. Um and they've gone out there and and they've sourced, you know, the power, they've sourced some of the biggest sites available to them, and they've got these sites now in a position where power's already been energized. Sweetwater 102, probably the largest site that we actually talk about on the chart, it's a two-gigawatt site, already energized 300 megawatts of that. There, we've been to the Childress site, it's like a small town. The amount of movement on that site was phenomenal. We spent the morning there back in November as we did our five-day uh tour of Texas looking at five different sites. Childress was the first site we went to, um, and there was about a thousand workers on that site there. But vehicles moving in, and we were told, you know, by the security there that you know on a Monday morning you'll have a queue of vehicles all the way down the highway uh waiting to access the site. And as we were there, they all move in unison around the site there, it's all clearly uh planned and everything like that, and it's just a just a great facility. That's a 750 megawatt site, it still operates quite a sizable Bitcoin hashrate there, but you know that they've dedicated 300 megawatts already to um uh HPC with that massive contract with Microsoft. Um, and you can see what they've been doing since those times there. They've purchased uh more sites um in the US and they've even stretched across now to Europe and uh you know this this this the site in South Australia where they want to deliver an 800 megawatt compute site in their own neck of the woods uh for the two for the two co-founders and two co-CEOs.
SPEAKER_02Yeah, and really seems they're going after the CSP model, but not necessarily with a partner. They're actually going out there selling the GPUs. We heard Fred say the bare metal, but we know now with Mirantis, the software stack, there's a lot more to it. I'll play you the second clip now as well.
SPEAKER_01We refer to as the three layers of AI compute. So you've got your physical infrastructure layer, this is the data centers, access to power, grid connections, substations, your second layer is the compute layer, GPUs, back-end networking, storage, and the third layer is the software and services that unlocks the first two layers. But the first two layers are really the key foundation, and without those, your your ceiling is capped.
SPEAKER_02So again, we hear that three-layer cake talking about the physical infrastructure. That's really what we've been focused on primarily on the channel up until this point. Now they're moving into the compute and then specifically calling out that software as unlocking the full stack or full value uh opportunity.
SPEAKER_00Yeah, and as they start delivering more and more of these, we're hopeful that uh what we have seen in recent periods is that is the increase in revenues per uh uh GPU kilowatt hour, and and and you know, the deal with Microsoft, I think we came out, I think $2.91 was the average for the five years, but we're already seeing uh those rates go higher than four, five, six dollars. Um, so potentially, you know, you've got your first deal in place there, you're maybe in a better negotiation position. Now, as we see this massive demand requirement, and and the fact that you know uh you know, it took I think Facebook 10 years to get to to 100 million uh Facebook users, it took Chat GPT about two months to get to the same number. This is um this is here to stay, and it's getting bigger and bigger. And I've just been to a to a to a family party at the weekend where I met you know 56 members of the family, and I was asking people, you know, are you using you know AI? And the majority of them were saying yeah, and they're really heavily involved in it, and and a lot of them were saying, Yeah, we pay this um you know this monthly fee. Well, as I said before, that's all right now, that's the lost leader, the monthly fee. Once uh the company's then determined they've got sufficient um customers, then we'll see the usage come in then and people will be paying for what they want to do on AI and HBC because they'll it'll they'll be they'll have been drawn into it. It's a it's a day-to-day requirement, especially for small businesses as well, to try and you know to deliver some efficiencies. Um, you know, it's it's you know, I I I sort of like rely on certain aspects of of AI on a regular basis. I know you do, Bryce. It's got to be the pretty much the norm. I remember I was against buying the iPhone when the iPhone came out. I thought it was a gimmick, and I was happy with my sort of knockier phone that just did texts and calls because that's all I wanted to do. But my son, who was probably about 18 at the time, he dragged me down to the uh to the to the Apple shop, and um I was shown the I think it was the iPhone 3, and um I haven't been without an iPhone since. So, you know, I got hooked on that, you know, from the very early days uh of the of the iPhone 3. I missed out the first two versions, but um, you know, I think more and more people will be certainly um you know tied into HPC to give them that ability to do things you know in a more efficient way um than they do at the moment.
SPEAKER_02I'm pretty sure I'm 16 for 16 or 17 for 17, whatever they're at now. Uh and I just get a chuckle as you're saying about the Microsoft contract opening doors, like on the financing side, you got to think in terms of brand recognition and marketing too. If you're a smaller company and you're thinking about who you want to run your GPUs with, you look at Iron, you go into the meeting and they say, hey, our main client is Microsoft. That's obviously very validating for a brand as well. Uh now we wanted to finish with this one. I
Soluna Signals And SpaceX Texas Close
SPEAKER_02just mentioned uh Mike Alfred, John Belazaire as well from Soluna. They were actually speaking at the Energy Investor Forum down in Texas last week. We mentioned that Mike Alfred voting with his wallet again is buying some of these colo plays. He's been accumulating Saluna shares. You can see a photo of the two gentlemen together there. But this is interesting again, Anthony, as we see more companies, differentiated strategies. So Soluna Renewable, they've got behind the meter, wind farm, still coming to the same conclusion that these megawatts need to be dedicated to AI. That's where the money's flowing.
SPEAKER_00Absolutely. Uh, Mike was very fortunate to meet up with John at the Bitcoin conference when we were there in Vegas uh back in back in May, and um, you know, obviously got to hear more about the company and and within a within a you know, I think within a day or so had highlighted the fact he took a sizable uh position in Saluna. And since that period, I think he's made at least two additions to that sizeable position there. So um, you know, we'll try and find out uh the latest uh 13F filing for Mike's company, and uh we'll maybe show that on a on a on a future podcast just to show you what he's buying in at the moment and how those positions are doing, because it also tells you from a cost perspective and a and a profit perspective. We know he's probably his biggest um uh uh share in his portfolio is currently Cypher Digital, and followed by Iron. And I think as you might mention earlier, those two positions there probably amount to about 80%. But Saluna, um, certainly one that he's watching, and I know from my social media uh that I I follow as well, that Saluna's name is getting mentioned a lot lot more now. We talk about you know the likes of Iron and Keel having sort of like a big, big following out there. I can tell you now that Saluna's building that following, and it's one of the smaller companies out there. Mike's obviously doing his bit, but it had a big following before Mike got involved this year. Uh, a lot of people um optimistic. And I think one of the reasons why it has that that following is the fact that John is extremely transparent, you know. Think of it on a monthly basis, he doesn't ask me anything podcast, uh, you know, where he's able to sort of go through and answer some of the retail shareholders' questions or concerns that they have. And we've had him on the podcast numerous times over the last few years. He's never ducked any of our questions, um, and we certainly don't um we don't give him a softball. We tend to sort of like, you know, we'll put in a few challenging questions, but the thing is, Saluna um probably highlight more things than all the other miners, you know, they do put out a lot of updates, they are growing, they do have this great table that shows them their pipeline of power and all their projects and the expected delivery of those projects and what those projects are going to be used for. So they are looking at sort of the hosting model for Bitcoin mining and they're looking at HPC AI, and you know, in recent months, that table has sort of like you know pivoted more and more to HPC AI when you've got the likes of Dorothy 3 and with Cati 2, both of those sites potentially more than 300 megawatts of HPC AI. Now, we're not going to be delivering 300 megawatts in one go. I think he's already in uh you know alludes to the fact that they have this agreement, this LOI agreement for 100 megawatts at Cati 2, so they'll be working hard to get that finalized there. But again, for the size of the company, these are extremely positive times. And I think this is what Mike's you know, his understanding of the space, having been a director of IRIN since the IPO days back in 2022, he's got a lot of knowledge of a lot of these companies. He speaks with many of the uh CEOs, uh, you know, and we we also have the opportunity to speak to some of these, but I know that Mike's friends with quite a few of the COs. He has a a lot of understanding in the space there, and he can probably see from what John's explaining the positivities in this company here, and that's why he's got a sizable, sizable uh holding there. So look forward to uh more updates on that. There, great to see John and Mike getting together uh over a glass of wine at the big conference there, and hopefully we'll catch up with with both of them at future conferences in not too distant future.
SPEAKER_02I'm sure is a pretty nice bottle of wine as well. Now, just a carrot on top or cherry on top here for you guys. Uh, SpaceX also planning more expansion in Texas, in Urcot, announcing plans to build or construct data centers in the state. So bringing things full circle, talking today all about deals, delivery, demand. Hopefully, you guys enjoyed. Let us know your thoughts in the comment section below, specifically as we get into earnings next week. Start to see some of this AI revenue flow in. Anthony, great commentary as always. We'll see you back here tomorrow.