Let’s talk about $WULF, and their announced Fluidstack/Google deal
Hosted by @Perry Lin · 2025-08-17 · Tags: WULF
TLDR
The discussion was strongly bullish on TeraWulf following its 200-megawatt Fluidstack agreement backed by Google, with participants arguing that the structure improves TeraWulf's credit profile and confirms Google as the likely end user. Perry Lin projected substantial valuation upside from the current contracts, a likely 160-megawatt option exercise, additional Cayuga leasing, and TeraWulf's ability to develop scarce power sites faster and with less dilution than competitors.
- TeraWulf has 262.5 net megawatts under contract, including 200 megawatts for Fluidstack, with Google providing a $1.8 billion backstop.
- Perry estimated a 95% probability that Fluidstack and Google exercise their option for another 160 net megawatts.
- Google received warrants for roughly 8% of TeraWulf, potentially rising to 14% or 15% if the expansion option is exercised.
- The speakers argued that Google's backstop and pledged warrants could materially reduce financing costs and support an investment-grade credit profile.
- Recognition-agreement provisions protecting Google's hardware and intellectual property were viewed as evidence that Google is the effective end tenant.
- Perry estimated that 422 megawatts could support about $700 million of revenue and roughly $500 million of EBITDA under his assumptions.
- Cayuga's 400 megawatts were described as a major prospective catalyst, with Perry predicting that one customer could lease the entire site within six months.
- TeraWulf's energy expertise, New York location, low-latency access to the Northeast, and ability to rehabilitate power sites were presented as key competitive advantages.
- Participants viewed Cipher, IREN, Bitfarms, Hut 8, and other mining-to-HPC companies as less certain, more dilutive, slower, or geographically disadvantaged.
- The discussion emphasized that projections about Waymo, future tenants, credit ratings, new sites, and valuation multiples remain speculative.
Speakers
- Perry Lin — Presented the central investment thesis, contract and capacity calculations, financing analysis, valuation scenarios, expansion expectations, and comparisons with other Bitcoin miners transitioning into AI and HPC data centers. He also answered questions about latency, power queues, site redevelopment, cooling, financing, and potential future tenants.
- Crypto Miami — Added legal and bankruptcy-document experience, discussed the recognition and warrant agreements, and emphasized Google's rights over hardware, intellectual property, and lease assumption. Also highlighted the extensive due diligence behind the transaction and argued that Cayuga may be an underappreciated catalyst.
- Ed — Identified prior clues that a major transaction was coming and contributed hyperscaler operations expertise on demand, latency, inference versus training workloads, GPU deployment cycles, retrofits, cooling, and carbon-offset practices.
- Andy — Asked how metropolitan proximity and latency affect site competitiveness, whether Bitcoin mining exposure discourages hyperscalers, and whether Perry truly expected Cayuga to be fully leased.
- Jamarco — Asked about Beowulf-related capacity, TeraWulf's ability to convert Bitcoin mining buildings to HPC, liquid-cooling design choices, and the investment prospects of Hut 8 and WhiteFiber.
- Jeffrey — Praised the analysis and asked about a proposed multigigawatt AI data center development near Amarillo, Texas, including its potential reliance on nuclear generation.
Notable quotes
- “And I would put odds of an option exercise at, like, 95%.” — Perry Lin
- “So I think there's a lot more room to run versus what they closed at on Friday.” — Perry Lin
- “Well, from TerraWill's perspective, this kind of changes the entire credit profile.” — Perry Lin
- “So really, this is kind of a win-win.” — Perry Lin
- “I think in this space, you can take Terrawolf off of the speculative list.” — Perry Lin
- “They basically just come and start making lease payments, and then Google's just basically assumed the lease and has the data center, right, transfer of ownership.” — Crypto Miami
- “Four months is actually incredible that they could put this together.” — Crypto Miami
- “And so I feel highly confident, and this is the alpha, that by the end of the year, all 400 megawatts of Cayuga will be leased.” — Perry Lin
- “I mean, it's almost insane, right?” — Ed
- “And if you're Terrell Wolf in Lake Mariner, there's no capacity in New York.” — Perry Lin
Transcript
Perry Lin: Hey, everybody. Thanks for joining. I'm going to get started shortly. Just want to say, you know, it's been a big week for Terra Wolf. Before I start, I'm going to kind of make some projections and use some rough numbers. So obviously, this is not financial advice. Nothing I say has been verified other than what's in the public domain. Please do your own due diligence, check my numbers, and again, not financial advice. So at the end, I will take questions and anybody else that wants to speak, raise your hand, more than happy to let you up. So let's give a kind of, I'm going to go over like a brief summary of what TerraWolf does and where things stand. They operate a facility in Lake Mariner, 750 megawatts of gross power, 250 megawatts of that dedicated to Bitcoin mining. And then they've recently added the Cayuga site, which adds another 400 megawatts. So we're looking at 1.15 gigawatts of gross power. Under contract, they have 62 and a half megawatts net at Lake Mariner dedicated to core 42, G42. Two and a half megawatts of that is operational at Wolfgen. 20 megawatts should be delivered this coming week. So we might get a press release or video in regards to that. And then 40 additional megawatts to be delivered in Q4 of this year. On Thursday, they signed a deal with FluidStack, Google backing them for 200 megawatts of net power, 250 megawatts gross, 40 megawatts of that being delivered in Q1 of '26, and the rest, the remaining 160 by the end of the year in 2026. So in total, that's 262 and 1/2 megawatts under contract today. FluidStack Google do have an option for an additional 160 megawatts net. It's a 30-day option. Now, the chances of that happening are almost certain. Paul Prager kind of went on CNBC and said as much. From my conversations, it seems like pretty much a done deal. And I would put odds of an option exercise at, like, 95%. So if you add in those megawatts of capacity, you're at 422 and a half megawatts net billable revenue by, let's say, January of-- You know what? It's probably going to be middle of 2027 before all of that is online, but I would say by January of 2027, definitely the 262 and a half. I think they're going to probably do, you know, maybe they can squeeze a little bit of capacity into the end of 2026 if they have to convert a Bitcoin mining building over. So let's call it 300 megawatts of billable revenue January 1, 2027. That's like, if you run some numbers, 422 megawatts at $1.65 million per is like $700 million in revenue. 85% gross margin on that brings it down to like $600 million. of net operating income, subtract, say, 100 million for SG&A, you're at 500 million of EBITDA. You apply a 20 multiple to that, that gets you to 10 billion, subtract the debt, maybe two, two and a half billion. So let's call it seven and a half billion, not factoring in any more excess capacity of what they still have to lease or any capacity that they do add. that's pretty much a double to what the current market cap is, so figure $18. So I think there's a lot more room to run versus what they closed at on Friday. So let's delve into this deal a little bit. Who is FluidStack? They're really good at building GPU clusters and delivering them in you know, super, super fast time frames, literally like GPU out-of-the-box, delivered and powered in, like, three days. You know, that's what they're known for. They work with some of the biggest hyperscalers out there. And, you know, I don't think it takes a rocket scientist to realize that the end tenant is Google. They're deploying this for Google. If you kind of delve into the AK that was filed, you'll notice two things. They filed four recognition agreements, and I guess a recognition agreement is kind of like a lease. Two for each building, two for CB3 and two for CB4. You know, the first one is pretty transparent. It lists FluidStack, TerraWolf, and Google. And it mentions a bunch of security guarantees for Google property on-site. So obviously, if Google is not going to be the tenant, why would there be any equipment there on-site? What's interesting are those two other recognition agreements that are not disclosed. You know, Terowolf mentioned that they existed and they were signed. but no further details as to who the counterparty is. Obviously, it's going to be related. My guess is, based on what Google's plans are, that it's Waymo, and that Lake Mariner is going to be used to power Waymo's operations in the Northeast. And I guess we will find that out eventually. But I'm speculating, but that's what I think. But based on the information that I've kind of delved into, that's what it leads me towards. So maybe not directly, maybe indirectly, but this is a hyperscaler deal. Now, the terms of what Google is providing, it's a $1.8 billion backstop in case FluidStack ever defaults. or is taken off the project for whatever reason. In addition, Google is getting 8% of the company in warrants. Now, I'm going to say when, not if, but when this option is exercised, that's going to go up to 14%, maybe 15% of TerraWolf. So quite a bit, Google's going to be a large equity partner, and you can do the math and what that's worth, let's call it half a billion dollars. So on the surface, it's like, wow, they're getting half a billion dollars for not doing anything. Well, from TerraWill's perspective, this kind of changes the entire credit profile. So what they're going to do is they're going to go to the credit rating agencies, And I think they're going to be rated at something around BBB, which is investment grade. If you look at corporate bonds of others that trade with the BBB rating, those bonds go for like five to five and a quarter percent yield versus if they were going to the project financing route previously, they were quoted SOFR plus four, SOFR plus five. So the difference is like 8.5% versus 5.5%, 6%. So if you take those two or three points of interest and you look at $10 billion of debt to build out Lake Mariner, the ROI, they're making back the grant they're giving, these warrants they're giving to Google in like three years. So you can say this is highly accretive, and these warrants won't be dumped on the market. That's the other thing. Google may end up with like 70 million shares, but these warrants are provided with consideration that they will be collateralized, pledged as collateral to the lenders of the build-out. So, in essence, these are going to be locked up until the loan is paid off. So, you know, Google is going to be a long-term partner of Kara Wolf. This is not, you know, a scheme where they're going to dump the shares like what Coreweed did with Applied, a completely different kind of deal. I'll also mention one other point that hasn't been I haven't seen mentioned anywhere else. I didn't think you know, this deal was going to happen at this kind of scale because Terawulf is guided all along that hyperscalers weren't willing to pay their yield that they're looking for. And there were multiple analyst reports in the last two quarters saying, Don't expect a hyperscaler deal from Terawulf. Expect a client similar to Core42 because the hyperscalers weren't willing to pay you know, the going rate that TerraWolf wanted. Well, something changed along those lines, but if you look at how this was financially engineered, you know, granting Google, you know, equity and upside to that equity, as TerraWolf stock appreciates, offsets the cost of capital that Google had to pay. So really, this is kind of a win-win. It gets the annual lease payments in a range where it's probably palatable to Google, and it benefits TerraWolf on the financing side to get a deal done. So major kudos to the team over there. I think Patrick, and Nazar, and Paul, and Kerry have done a tremendous job getting this to the finish line in such a timely manner. And you're also looking at a pretty quick build timeframe. I mean, they're delivering lots of capacity in like 12 to 18 months, something that you would not expect. So just something to keep in mind. When I hear of investors and shareholders talk about other hybrids, iron and cipher and galaxy and applied, in the same breath as Terawulf, the knock on Terawulf has always been, well, they don't have enough capacity, they don't have two gigawatts of power. And what this deal shows and what they've done shows is that it's not how much power you have, it's how quickly you can build it and pay for it. A mentor of mine early on always said, Listen to what these companies say, because they usually try to do what they say. And there've been multiple hints. Nazar hinted that the company's guided, let's say, 150 to 200 megawatts that they can realistically build. on an annual basis. But a lot of that is constrained by how much they can finance and pay for. And so nobody has better financials credit profile than Terawolf right now. And I think we will see that guidance go up over time, maybe not in '26 or '27, but definitely by '28. Yeah, I mean, I think they can build maybe three and finance three to 400 megawatts a year. So if you look at all these hybrids that have much, much worse credit profiles and they and they've got to sell a whole lot of equity or dilute, dilute UDF with ATMs, it doesn't matter that they got two gigawatts of power, they're not going to be able to build any of that. I think the first hybrid miner to lease out two gigawatts of power is going to be TerraWolf, and they don't have two gigawatts right now, but they have a pipeline of getting there. So the other tidbit that they dropped on the call, they've looked at 75 sites for additional capacity, and some of those are in, you know, serious negotiation. I think by the end of the year, we will see additional sites come on board. You know, what we have to realize is that these are energy people. Nobody else, none of the competitors are energy people. It's great that you can tout how, you know, data center prowess. That's not really what the hyperscalers need. They have that. You know, Google and Meta, Oracle, you know, have plenty of super, super smart data center people. and have partnered with a whole bunch of these companies like FluidStack. What they don't have is power expertise and being able to create value from that. And so if what we're hearing is true and there are no additional sites to be had or whatever sites that are out there are super expensive, then you've kind of, you know, for future growth, you kind of need to create that energy. And that's what these guys come from. That's their background. So, complete speculation. I don't know what sites they're looking at. But, you know, let me give you an example of something that may work. Paul, in his, I think, December calls mentioned that they were looking at Virginia, Maryland, and Montana. Now, Paul owns a site in Montana, the old Harden site, that did maybe about 100 megawatts of Bitcoin mining from MARA. Maybe they're going to expand upon that and bring in renewable generation there. That's a logical fit down the road. But Maryland and Virginia sounds really interesting. You might have heard in the news a site called Brandon Shores. It's outside of Baltimore. It's a coal-powered plant. I think it's two plants on that site. And it's owned by their old partner at Nautilus, Talent Energy. Now, they were going to shut down that plant this year. It's not profitable. It's highly pollutive. And it requires some environmental mitigation. I think they got an extension just because the grid needs that site to run. you know, but it's 1.3 gigawatts of power. You know, I think those are the type of projects that TerraWulf can take on that no one else can. You know, buy that cheap from Talon, who probably doesn't want it, and then spend the next three to four years mitigating and getting it ready to power data centers like they did at Lake Mariner, you know, a similar type situation. So, you know, if you look at things, You know, in the presser, they also talked about Cayuga. Now, Cayuga is 400 megawatts. I think they can get 150 megawatts of that, you know, powered by the end of 2026. I do know for a fact that there were multiple candidates in competition with FluidStat Google to take on that capacity at Lake Mariner. If you hear what Paul said in terms of demand, kind of a stark difference to what Tyler at Cipher mentioned in his last call. Both mentioned that demand was strong, but Tyler said it was hot and cold. It's picked up recently, and demand was strong, but he didn't say it was overwhelming. I think Paul pretty much said that demand was off the charts. So, the parties that lost out on the Lake Mariner sweepstakes, you know, part of the reason why Cayuga was dropped in the company so quickly is because that capacity is available. And, you know, my hunch is, based on conversations, that they will sign a third tenant for that capacity before the end of the year. probably for all 400 megawatts. And that basically leaves 250 megawatts at Lake Mariner left. All of that's being used for Bitcoin mining at this point. And there's probably a good chance we'll see an announcement that, you know, by 2028, all of that is going to be converted and Google or Core42 will take on that capacity. So really, by the end of the year, there's a chance. I'm not saying, you know, it's 100%, but there's a chance that all 1.15 gigawatts of capacity is spoken for at Terawulf. I think it's much more likely that, you know, 800 or 900 megawatts of capacity is spoken for by then, and they still mine until the halving, and then make a decision on that conversion. I also think it's pretty likely that we will hear of additional sites in addition to that 1.15 they have now. So if you look at that, what is the company worth that they have 900 megawatts under lease? Probably 4-5x, even if you discount future cash flows to the present day, of what they're trading at today. Now, that's based on a 20 multiple, which is what Equinix and Digital VLT and some of these other guys get, but they run legacy data centers. So, you know, the market, there is nobody doing this right now, and TerraWolf is the first. I guess if, you know, you can call them almost a pure play. Yeah, Applied and Galaxy have some Core Weave deals, but I just don't think it's the same. So, you know, what does a data center that does high-performance compute AI with a hyperscaler as a tenant, with two or three clients signed, what kind of multiple does that kind of company deserve? Is it 30x? If it's 30x, you're looking at a $50 or $60 stock price. So, When people ask me on X, What do you think of Cipher or Iron or these others, who's next? I think the best next play is TerraWolf. We've got to remember that before this deal was signed last year at the height of the rally, this was a $9 stock, and it's basically trading at that level. So there really hasn't been a premium attached to it. I bought some additional shares as soon as the announcement was made. I think I bought it when it was up 17% or 18%. I sold it at the close because it was just a 40% gain in one day, but I regret selling it because I still think this is a really, really good play. I think in this space, you can take Terrawolf off of the speculative list. I mean, if you're buying Iron or Cipher or HUD or Riot, it's basically a speculation. The risk profile is different. You're guessing as to what is going to happen. There's just a lot more certainty. And so TerraWolf is a safer investment. And if you ask me who is more likely to Forex from now until the end of the year, well, I kind of laid out what they could announce by the end of the year. and that could lead to a 4X. Does a 50-megawatt deal at Horizon One for Iron lead to 4X? Does Iron go from 18, 19 to 75 or 80 on one 50-megawatt Horizon deal? I don't think so. Now, I own some iron, but not for their AI and for the data centers. I own them because they're a world-class Bitcoin miner. And if Bitcoin goes to $250,000, absolutely they can 4x and 5x. But that's a different conversation. We're not comparing apples to apples anymore at that point. Cipher, I love Tyler and I love his transparency. And he's basically told us what they're doing, right? They're trying to sell or lease out, they're trying to lease out Barbara Lake. 300 megawatts to a hyperscaler type quality. But with their fortress financing, you know, they're giving up 60% of that equity. And then they're repurposing part of Black Pearl for either smaller enterprise customers, probably taking on chunks of 20 to 50 megawatts, or saving that for more Bitcoin mining. And it doesn't sound like any of this is going to happen to be delivered to an end user before 2027. So really, let's say Cypher signs a deal. I think it could double. I think it can go from 5 to 10 to 12, maybe. Do I think it's going to get to 20? Possible. But just if you just do basic multiple math, probably not. probably it's probably worth maybe $12 based on my back of the napkin calculations. So and you're taking a whole lot of risk. There's a lot of uncertainty. So really, if I would have put fresh money into this space and you wanted exposure to, you know, HPC AI data centers, I would tell my friends to buy Terra Wolf. It's not. dissimilar to Palantir. I bought Palantir, my first buy in Palantir was $17. You know, I told my parents to buy it at 40. I don't think anybody's upset at buying Palantir at 40. And it's kind of the same thing. If you didn't buy Terra Wolf at five, you can buy it at, you know, eight or nine today. And if it pulls back a little bit, you know, it's a screaming buy, you know, you buy at the high sevens, eights. I think you have a 4X in there eventually. I don't know how long. Maybe you got to wait a year or two, but it's definitely 30 plus in my estimation. So that's kind of my thesis and where I, you know, what I think of the kind of the space. So with that, if anybody wants to ask a question or speak up, I will raise your hand and I will let you in. Crypto Miami, you're up. Good.
Crypto Miami: Excellent work. Excellent work. We did have a little, there was a little bit of a wolf iron space that just sort of popped up this morning. So we've had an interesting discussion already this morning. So this is adding some nice color to this discussion. And I also read the agreement, which I found fascinating. You know, I came into mining, I was not in the miners last cycle. I've been in Bitcoin for a while, but I didn't do miners last cycle. So I came in mining through the Celsius bankruptcy and through the core scientific bankruptcy, because that was related to Celsius. So I came in through reading legal documents in New York federal court and Texas court through both those bankruptcies. So going on to Edgar and finding the-- it was like reading lease agreements, reading warrants, core scientific warrants. It was very interesting, and it's all available for you all to see. If you go to SEC EDGAR and look up Terra Wolfe, you can basically find the agreements that Perry's talking about, which is the recognition agreement and the Warren agreement. It was absolutely fascinating reading, and you caught that early and tweeted about it, which I thought was fantastic. Again, it's just super duper interesting. When you read the Google backstop, it's very explicit that the Google backstop and the warrants are to create that credit facility for TerraWolf. And so it's very explicit that these warrants are going to be used to allocate for credit, to create that credit facility to benefit both parties. So that was great. But I think you highlighted something that was super duper interesting, which is a lot of that agreement is about Google's property on the site. And that's the kind of stuff like we just haven't seen before. Like no one's ever talked about that before in the sort of the core weave style agreements. But I mean, there's just a lot of language in there that Google itself has backstopped it. But number one, it maintains the right that in case FluidStack defaults on this agreement, that Google will basically just assume the lease. They basically just come and start making lease payments, and then Google's just basically assumed the lease and has the data center, right, transfer of ownership. Absolutely. Yeah, but a lot of a lot of that language is so fascinating because it's really about data protection. So protection of Google's hardware and IP within the data center. And you're kind of asking yourself, well, if FluidStack is signing the lease and Google has infrastructure that they need to protect, like Google, Google's going to have hardware, IP software that they need to like step in and protect in this data center like. If this really was just FluidStack being like an anti-cloud provider running Google's data, yeah, maybe you need to protect the IP, but they were sort of specific about hardware. And Google certainly seems interested in, no matter what happens to FluidStack, if FluidStack defaults, They have full access to the data center. They assume the lease and the hardware, the IP, everything that happened in there, like that's their properties, you know, they maintain those rights, which I think is fascinating.
Perry Lin: I'll also add that you're absolutely right with all of that. Companies like Google also don't assume leases just for the hell of it. unless they have an interest. Exactly. Because everything is planned with hyperscalers, right? They're just not going to be like, all right, if something happens, we'll take it on, we'll figure out what we're going to do with it. They know what they're going to do with it. They're on site. Patrick also shared with me the security of Lake Mariner is not going to be anything we've ever seen just because they demand it. We're not talking about a core weave site with Core Scientific where they got mall cops roaming the property. They're talking about 24/7 drones over Lake Mariner.
Crypto Miami: Exactly, exactly. Yeah, some other color I can just add, and this comes from being in a bankruptcy. And I only bring this up because having been a retail person involved in court and all this-- one of the things you learn in bankruptcy is about valuations, M&A, and legal, due diligence. Number one, companies are valued, because all of the companies are in bankruptcy court, and then, of course, These are the two miners that came out of bankruptcy that were completely revalued and came out as new entities. Like Celsius came out as Ionic Digital, and then, of course, of course, Scientific emerged from bankruptcy. So two years in a bankruptcy court reading documents, but the amount that goes into the due diligence of a deal like this, retail just doesn't understand. Like, oh, it's due diligence. Yeah, but you're talking investment banks, you're talking M&A. You're talking lawyers, M&A lawyers, legal, $2,000 an hour lawyers. Like you're talking blue chip, blue chip, blue chip legal firms. Like the amount of money that's spent to do a deal like this, just in kind of legal business fees. By the way, like the document also has the payment to the real estate to CBOE, I think $20 or $30 million payment. That's the broker. The amount of third-party expertise involved in a deal like this is insane. You could do this, but that's why it takes so long. Four months is actually incredible that they could put this together.
Perry Lin: Well, exactly. And they've dropped hints. I didn't pick up on it at the time, but when they had their last annual meeting, one of the things they voted on, they increased the share allocation to 600 million. That was to accommodate the Google equity warrants, because they're basically bumping up against that right now. So that was, what, two and a half months ago. So this was all in the works, probably the last six months. It's going to take that long. So all of these deals for everybody else is just going to take a while.
Crypto Miami: That's right.
Perry Lin: If Iron is talking about a 50-megawatt pilot, and I believe that they will get one, but these hyperscalers want to see that you can produce and build something. So let's say they build it on time. Yeah, on time. Let's say they complete it, and they like what they see. you know, and that happens at the end of the year, you probably won't see a deal until Q2 of next year if it takes six months to sign a real, you know, large-scale deal. So I think you have time for, and all of these others, they're just a little bit behind.
Crypto Miami: That's right. Terawolf is just a, you know, outside of Core, which is a completely different deal, right? Because Core had the existing infrastructure. Core Scientific had that data center available. Well, remember, Core Weave was an Ethereum miner within the Core Scientific family. So the original Core Weave were Ethereum miners that were hosted at Core Scientific in the last cycles. And they pivoted with those original ETH miners into AI. That was their genius. But they had that relationship with Core Scientific already. So, and then Core Scientific had that data center near Austin available in that original deal. So, in a way, we're watching the iterations of the deal flow, and you watch the original Core Weave deal, which was incredible at the time. but was perfect for this company emerging out of bankruptcy that was desperate for its shareholders. The shareholders of Core Scientific were a completely different animal than the ones that are in IRN and are in Wolf. I mean, those were folks.
Perry Lin: Absolutely. And before we were paying for everything, and these are much smaller sites, you're talking about 30 megawatt buildings at Core Scientific.
Crypto Miami: That's right. Completely different deal, and it was amazing for the time and a complete value add for Core Scientific shareholders that were basically reamed and destitute for years, suffering in bankruptcy court. And of course, Adam has done a great service to core scientific shareholders. Now, all these Nextdeals, the G42, which I think was really misunderstood by the market, I think part of that was retail just automatically thinking, oh, Nextdeal hyperscaler, but didn't really understand who G42 was. So when that news got announced, it was like, oh, not understanding. a sovereign wealth fund is and the amount of due diligence that would go into a sovereign wealth fund, like the colonoscopy that would be done by and the due diligence done by them, right? But what I'm talking now is about the colonoscopy and proctoscopy that's done by a hyperscaler. Like you've got Morgan Stanley on the investment banking side, but like I said, the legal, investment bankers, real estate, and M&A lawyers, like when you talk about the due diligence that's now been completed by number one, a sovereign wealth fund, and now you've got Google backstopping this deal with billions and taking equity stake, you're basically saying that Terra Wolf has been given a stamp of approval, like just the due diligence here is like outstanding, like ridiculous compared to what's been done before.
Perry Lin: Yeah. You know, G42 is, from what I've heard, is 100% supportive of this. They are FluidStack partners in Europe and Middle East for, like, their sovereign AI products. So they know each other very, very well. They're almost partners. And I wouldn't be surprised if, like, you know, Google is a G42 client at Lake Mariner. So the synergies are all there. This could all be part of one huge project.
Crypto Miami: Yeah, the interesting thing about FluidStack-- so when you read the agreement, and you see the Google backstop for FluidStack, but then FluidStack's a private entity. It is a UK-based-- I can't-- UK, so I don't know if it's Cambridge or Oxford back from 2017 as an entity, they've got into AI. They've got 100,000-plus GPUs. And so they're bigger than Nebius, right? Obviously smaller than Coreweave, but like bigger than Nebius, who like is relatively transparent on the cap table. But like when you go to look up FluidStack, it's super interesting that they're that big and it's a bit of a black box. Like on the website, you see who their partners are. What's interesting is that Google is not a partner, like Meta is a partner, Nvidia a partner, you don't see Google, you do see Mistral. which is like the European, like Google has a relationship with Mistral in Europe, which is a European AI startup. And so Google backs Mistral, and Mistral is a partner with FluidStack. But like, you don't like, it's really hard to figure out why is Google so invested in FluidStack. It's not perfectly clear, which again, I think you picked up on like, why is Google backing this deal? It's not that clear based on, like, stuff you can find, you know, on the internet.
Perry Lin: I would guess that it's the services side of things. So you're almost kind of like the IBM of data centers, right? You're providing the services, and they're just really good at rapid, you know, cluster deployment and doing it really, really well. And if you're Google and you're running all of these data centers all across the country, you got to get them up and running. There just aren't enough Google employees to do that. And so, and it'd be interesting to see who was supplying the GPUs. I'm almost certain these are Google GPUs and not FluidStack GPUs that are going in.
Crypto Miami: Interesting, right. There you go.
Perry Lin: It's probably the Google allocation. And that's why the property is so important, right, because we're talking billions and billions of dollars in and GPUs that they want to safeguard. That's why it's in those agreements, if you read between the lines. And so FluidStack is probably going to make a lot of money, but as a service provider, managing that facility for Google, because Google probably is going to have 40, 50, 60 facilities all over the world and aren't that many Google employees, and they probably don't want to manage that in-house. That's my guess.
Crypto Miami: Yeah, it's interesting. And I would certainly say that if you listen to the morning earnings call, which was beyond bullish, like it was so bullish, and it took me so long to process that I'd like you-- I mean, whatever, I stacked some extra-- I mean, obviously, I've been a large Wolf holder since 2022. And if you-- Wolf is like the most transparent of the miners, like they brought-- this was so telegraphed. It was telegraphed in, number one, the announcement on Friday, which basically said in the headline, 200-megawatt deal, basically. It basically said that in the headline. But when they basically told you that on Wednesday that tomorrow, we're going to have our earnings call, and we're going to be attaching a supplemental sheet, and you're thinking, oh, my god, they're going to attach a supplemental deck. This has to be a big deal. They're not going to do that for a 50-megawatt deal. They're not going to do that for 50 megawatts from G42.
Perry Lin: Yeah. This is what I'll tell you about TerraWolf. There are a lot of things they can't say, but they like to drop hints to lead you in the direction of what it is they want to say, but they can't. And so because they're investor-friendly and transparent, and everything that we do, we say, other people say, other people write about is all good for the stock price. And they realize it, and they care about that. But there's just certain things that they can't say. But they drop hints all along the way. And sometimes you can pick up on it. Sometimes you can't.
Crypto Miami: So yeah, it's a bit of reading the tea leaves. So what I'll say is-- and I know you noticed. So the difference between the morning meeting and Paul Prager's appearance on CNBC at 2:30 that day was that Paul basically said on CNBC, I am 100% confident that Google will be taking the additional 170 megawatts, which was like, wow, OK. That's confidence. And you also see it in Carrie's tweets. Like they're saying we're going to be the largest data center, right? Or one of the largest. And so like the current megawatts don't do that. And the team TerraWolf does not play that way. You know what I mean? They don't like if they're going to issue that guidance, they mean it, right? Like this is not like a speculative bubble here. So it was interesting to see. between the earnings call and then that afternoon, they really ramped up in the guidance there, like quite a bit.
Perry Lin: Well, and it's developing. I think the reason why there's an option to begin with anyway, I mean, the announcement was Thursday morning. If you look at the filing, Google did not sign until Wednesday. And they thought this contract would have been ready for Friday's original call. So it got pushed back. So this this was like to the last second. And there probably just wasn't enough time to get this additional agreement done. Something happened in the morning or whatever. And that's why they're not, it's not going to be 30 days. I think we either hear an announcement this week or next week about this lease extension.
Crypto Miami: But it's. Well, and the other thing is like, which I also find fascinating, really, because Some of it got lost in the news. Like obviously, the headline and the algos were trading on the Google, right? Which even pre-market, like the algos just saw deal Google, and you start the 30% pre-market. Really, the Lake Cayuga thing, I think, is the sleeping giant here, because they basically pulling-- I was wondering if they were pulling Lake Cayuga forward earlier than they anticipated. And that all dropped on the same news. So you have to be a real wolf aficionado to get-- they're pulling that forward for a reason.
Perry Lin: Here's what I know about that. I mean, I think 150 megawatts was always in the cards. They guided that previously when they started the process. So there's power there. They've been working on that site for a while. But if you hear what Kerry said, what Patrick shared with me on his-- we had a little call the other day. They drop hints, basically, that they have-- and Paul said this, too, that demand is high. There are people they're talking to, and they don't want to lose them. There's a good salesman. If you've got a prospect on the hook, you want to close them. You don't want to let them go somewhere else. And so I feel highly confident, and this is the alpha, that by the end of the year, all 400 megawatts of Cayuga will be leased.
Crypto Miami: Yeah, yeah. That would certainly fit in with Kerry's narrative about, you know, that would put him in the largest, you know, that would tip them over. That would tip them over Core Scientific and Galaxy.
Perry Lin: Right? Exactly. Ed, I'm bringing you up, you got something to share?
Ed: Yeah, yeah, we're just gonna kind of comment on, 'cause you're talking about the hints, and there was kind of three main hints that... And I was definitely one of those pounding the table on wolves saying, look, something's about to happen. I don't know what, but it's about to happen. And there was 3 main things I was kind of looking at. One is that RSU award, right? They dropped, and that was a very, very evident clue, right? They dropped that RSU award, which is a much larger award than they've ever. got previously. I think it was still in cycle based on their performance cycles, but it was much larger, right? So to me, that was a hint saying, Hey, look, they just closed a big deal.
Perry Lin: Well, and they were very, very transparent about that with me. Patrick basically said that since he started working there from like the days of close to bankruptcy to getting this deal done, None of them had been paid, and they'd been working like 18-hour days. And so before this deal happened, they wanted to get paid. And so I don't blame them. I don't mind them getting paid if they're going to deliver for shareholders versus just like Ride Amar getting paid to exist.
Ed: Yeah, yep. And I think so the other thing... If you kind of go back to that previous earnings call, right? And I've read through those transcripts over and over and over. And there was one thing that Paul said multiple times in that last call. And it was like, once CB1 is done, you will start to see that interest in those new deals, right? And so when you look at the timing of all of it, right, like we knew that CB1 was energized this month. right? And it was getting delivered this month. And they said revenue was happening this month, right? So you knew that was kind of a--.
Perry Lin: Yeah, this week, I think. I think they turn it over, and the meter runs this week on CB1.
Ed: Exactly, exactly right. So you knew that was coming. So that was a big hint, in my opinion, that he's telling you that that's what you need to be looking for. Once this is done, other things are about to drop. And then if you really look at the share price, right, we were stuck in the five to 550 range for how long, right, a month, right, or greater, right? And it was so consistent, and if you look at how these deals were structured, if you look at the warrants, right, like those were all based on VWAP, right? And so it was tied to financing deals, right? They wanted that share price to be right there. So, that was the economic piece of their deal, right? Warrants, and if you look at that second milestone for financing, tied to Paul with the Beowulf seller. a sale. So there was a number of reasons to have that share price where it was, and it was so steady that it really just became even more evident that something is going on, right.
Perry Lin: Well, and that Beowulf deal, let's not overlook that, bringing all those employees in means that they can buy a site. that they pretty much told us, right? They have sites that they're evaluating that are under negotiation. They never said that they were under negotiation until this call. And they're doing this after they're bringing in the Beowulf 94 employees or so. Those are energy people. They know how to mitigate sites, to turn them around. It would not have made sense for Paul to own any more of these additional sites and then sell it to Terawolf, right? It's just easier for Terawolf to own them, mitigate them, and turn them around and reposition them into data center sites. So I think that aligns with, you know, there's going to be more capacity coming online soon.
Ed: Yeah, definitely agree. And I think one call out there too is, you know, there is very few sites And you may have already mentioned this, I kind of joined late, but with that low latency, right? And so, when you look at what hyperscalers or really anybody that's trying to run inference workloads are looking for, they are looking for sites close to densely populated metropolitan areas. Right, and I'll just to give you some background, like I work for a hyperscaler, right? I, I'm on the operations side of the house, so I don't always see what goes into site planning and that kind of thing, but what I do see is the demand, the scale at which which we're in speed at which we're trying to build. I mean, it's almost insane, right? Unlike anything in my past, you know, 15 years in the business I've seen. So, But there's a lot of complexities, right, in determining the right location. Sometimes you want to use it for, you want to be able to use it for inference, but you also want to be able to potentially use it for training, right? But if you want to use it for training, you have to have enough scale to use it for training. And so they're juggling so many different factors, right? And people wonder why these deals take so long, because There is just a lot to consider, right? Even on the GPU side of the house, right? The development cycle on GPUs right now is like six to nine months, right? You go to build a data center, and by the time you build it, the GPU that you want to deploy is obsolete. Maybe not obsolete, but it's old technology now, right? And so...
Perry Lin: Yeah, it makes total sense, and what we'll see in this market is if power really is that scarce and all of these sites are going to go, the best ones are going to go first. And we saw the best one just went. Yeah.
Crypto Miami: And people forgot about New York. Like, where is it? Like, lots of PGM Pennsylvania stuff, but no one's been talking about energy in New York, right? So that's always what made League Mariner like a unicorn that no one really... about the geographic diversity and the need of New York for this kind of power, right?
Perry Lin: Yeah, I want to bring up Andy because he's been waiting for it. Andy, you got something to say?
Andy: No, on that point, I guess my question was, and forgive me, I'm a little newer to the space. I'm still getting up to speed. But how important is it to be next to a metropolitan area versus someone like Iron, who has a big presence in Texas? How much of a differentiator is that?
Perry Lin: I think latency does matter no matter what anybody says. I mean, if you look at the data center maps for the hyperscalers, they're not all in Texas for a reason. And they all have gaps. Obviously, the areas that are hard to get power in, they want to be there. And so I think that matters.
Ed: Yeah, it's extremely important when you're running inference workloads, right? So for AI training, not that big of a deal, right? You look at Applied's deal with CoreWeave, you know, their site is in North Dakota, right? From an inference perspective, they can't meet the mark, right? But that site will be used for AI training. So, you know, like, yeah, it... It is important and it's driving those decisions greatly, right? And I think Deepseat came and changed a lot, initially on the training front and made everybody question, how big do they have to scale AI training? And so there was a lot of delays. And I think that's why it took so long for Applied to get a deal was because their site was just not suitable for inference, right? right? Maybe someday in the future when, populations grow out over there or something, it may become relevant. But right now it's, it, and they've told you specifically, this is a site for training. So yeah, it matters. If you look at like the Northeastern corridor, right? So right now, most of that is serviced by like the Northern Virginia, you know, data center central hub there, right? Largest, largest data center hub in the world. And there's some stuff in Pennsylvania, but New York, upstate New York, right there, where Lake Mariner is, is an absolute sweet spot that has really been underutilized. So yeah, I mean, all that to say, super important.
Perry Lin: I will say Pennsylvania is important. I just think-- but being able to know what you're doing also matters. And it's why I'm so bearish on bit farms. They've outsourced everything because they have no clue on how to run a data center, and they've admitted as much. They brought in worldwide technology, and they brought in T5, and they're so far behind the curve, they're still trying to lay out their site plan. I mean, we're talking two to three years away, and this is a company, unless Bitcoin takes off, loses money, they have going concern issues. And I'm just, you know, it doesn't matter how good your site is, you know, they should sell it. Someone should just buy bitfarms, because I don't think they can operate it. Got it. Yeah.
Andy: I guess I'm trying to contextualize, like, how much does this weaken IRAN or Cypher's negotiating leverage when it comes to these types of hyperscaler?
Perry Lin: I always thought that their negotiating leverage was weakened anyway, because they had to compete with each other. and Galaxy and everybody else in Texas and the natural gas providers like Diamondback who want to build sites. There may be scarcity in power across the country, but if you want Texas power, you have to compete with all these other providers of Texas power. And if you're Terrell Wolf in Lake Mariner, there's no capacity in New York. They're the only game in town.
Crypto Miami: To add some color to that exact point, Perry, It's about the queue. So this is going to be my next question. It's really-- everyone's used to hearing about the ERCOT queue for power, and it's a multi-year queue now if you don't have those megawatts locked in. So I was wondering if you had any alpha on-- so we know Lake Mariner is currently 500 megawatts, but it's up to 750, so they need to applied in New York power for the additional 250. And then with Lake Cayuga, it's 400 megawatts. I guess they have 150 or 170 available, then up to 400. Do we know anything about the queue for getting that additional in New York? I'm assuming it's nothing compared to Texas, because there's no one else applying for that kind of thing.
Perry Lin: It's a handshake deal from my understanding of how all this works. Just because they're in a zone, in the New York ISO zone, where there's excess capacity, so it's not hard to get. Because it's Niagara Falls, right? There's plenty of power, and it's too expensive to transport. There's not a lot of use case for it. But the grid needs to plan for the capacity. And so they don't-- it kind of works different than ERCOT. They don't ask for the power until they know when they want to use it. And it's usually like 12 months out. So they'll call up New York ISO and say, all right, we're ready. We need another 250 megawatts by, let's call it November of 2026. That's when they'll get it, because.
Crypto Miami: It just helps-- Right, the lines exist already, right?
Perry Lin: Yeah, so it helps everybody else out, yeah. The worst thing you want to do is you burn bridges and say, we want the capacity now and not use it, and it just sits there.
Ed: Yeah, I mean, I'm expecting that to come either this quarter or next quarter for that additional approval. I kind of thought that that would need to be in place before this extension is signed, to be honest, right?
Perry Lin: Well, you saw that the, I think it was last quarter, that they said that their additional 250 was approved. So, because they knew that some of it was going to come online this year. So I think they asked for the power for the last 250 just recently. And when they need it, it'll be there. That's just how they-- Paul is a utility. So he knows how New York works. He knows the people there. And it's more political than anything else, I think.
Andy: Yeah, I doubt it. Hey, Perry. Do you think the Bitcoin mining exposure that some of these other transitionary plays are turning off hyperscalers, the idea of backstopping them or investing equity in them?
Perry Lin: Possibly. I don't think they really care. I hate to say it, but there are a lot of clowns at these companies. I mean, do you really want to do business with Riot or Mara? If they don't have the financial chops to engineer a deal, you really need Wall Street guys running the show. Iron's probably a little behind because they're Australian, and no disrespect to Aussies, but I think it helps to get a financing deal across. have, like, 30 years experience in Manhattan working these deals in the back rooms. The banking relationships matter. And so them transitioning to US reporting helps. Bitfarms doing that helps. I think the Cypher guys are really, really good. And they did what they could, and they actually have a financing deal in place. I don't think that's by luck or chance.
Andy: Right. Did I hear you mention that you said that you think potentially, I don't want to misquote you, that you think potentially that Cayuga is going to be fully leased out to new customers by year end?
Perry Lin: I do. I think one customer will take all of Cayuga and we'll hear something in the next six months.
Andy: Wow. Okay.
Perry Lin: I think that's the reason Cayuga was dropped because they knew that they have client, they've hinted at it, they have clients on the hook. Patrick pretty much told me, one of them called, they saw the press release. He was like, WTF, why didn't we get that? And he said, well, we can offer you Cayuga. It's officially part of the company now.
Andy: What does that imply in terms of contracted megawatts? Is that like over 800, I think?
Perry Lin: Cayuga's 400. I think 150 is going to be available in 2026. So it's probably a 27, 28 story, but it's 400 megawatts.
Crypto Miami: Wow, okay.
Perry Lin: Jamarco, did you want something to say? Sorry, kind of left you out hanging there.
Jamarco: Yeah, I have a few questions. So first, what do you know about the other capacity that Beowulf has? So since they've merged with Terawulf, I've seen on Beowulf's website that they own four gigawatts of energy capacity. How much do you think of that can be retrofitted for HPC capacity to Terawulf?
Perry Lin: Okay, so this is kind of a tricky question. Beowulf, who they acquired, was a services organization. So they are asset light. They ran the operations for a lot of the sites Paul owned. Now, if you look at like, and so let's not confuse it, they don't actually own any sites. Paul Prager does through a lot of his operating companies. You'll see it in the documents, like Farewell never owned like Mariner, it's like Somerset Power, It's Harriot. It's all these other entities. So I think they've mentioned Montana, which is the old hardened Bighorn site that ran Bitcoin mining for Mara. I think that may be an option down the road, because they've mentioned that. I don't know of any additional sites. I know they run a lot of gas-powered sites. and whether or not that's suitable for what they're trying to do. Everybody wants to go green. That's one of Google's major initiatives. And maybe that was a driver in getting this deal done. And I don't know if that's the case in Texas. That's the other downside of Texas, right? You can talk about wind all you want, but all these guys pull from the grid, and it's like 99% natural gas. Like, gas trades at negative value. Like, they can't give it away. They have to pay you to take it. That's why power is so inexpensive in Texas. But it's not that green. And you can spin a story that you're using green energy, and I guess you can make that happen, but I don't know if all the hyperscalers believe that. But anyway, I'm going off on a tangent. I don't know if there are additional sites within Beowulf's umbrella that they could drop in other than Montana. But I do know that they are looking at sites, you know, and I think Maryland and Virginia were discussed at one point. And, you know, I think it's pretty awesome that if they can, you know, secure megawatts in the Mid-Atlantic, you know, that's something nobody else has.
Jamarco: But they should at least have the relationships and a know-how advantage, right? But just have an experience with a lot of power generation.
Perry Lin: Absolutely. I mean, that's kind of the secret sauce that nobody else has is they understand power, and they're willing to take on projects that no one else will take on. Everybody else just are real estate guys, right? They just want to take on real estate and build. Nobody wants to do the dirty work like mitigating environmental site. Like in Virginia, I think their clean energy initiative is like all the coal plants need to be decommissioned by the end of the year. So you're going to have a whole bunch of decommissioned coal plants, and like this team knows how to turn those into solar battery-operated green energy sites with existing infrastructure.
Jamarco: Yeah, the other thing was, you're mentioning that there's 250 megawatts of power left at Lake Mariner that's being used up with Bitcoin mining right now. So with that, what would they have to do with that? Are they able to just plug in GPUs in those buildings, or would they have to retrofit it, or would they have to knock the whole thing down and rebuild it?
Perry Lin: I think it's definitely a different building. I mean, they might be able to keep some of it, like the slab or whatever, but I think they're basically teardowns. And I think they're going to-- they don't need to make a decision on those 250. I think they're going to take out one building because they're about-- if you look at the guidance, they're probably 20, 30 megawatts short of CB4-- I'm sorry, CB5. So and you see their Bitcoin guidance go down from 12.2 x hash to like 10 x hash. But I think it's a knockdown. And I don't think it's going to take that long to put up a building in that spot. But it's probably 2028, 2029.
Jamarco: So they have done what Iron does and make those dual-use buildings, where they can just-- unplug the miners and plug in the GPUs, would that have been impossible? Was that something they should have done?
Perry Lin: I think the lesson here is that it doesn't matter what they want to provide the customer. It's what the customer wants. And CB4 is a 200-megawatt building. So, you know, like, imagine Iron going to Google and be like, We got these, you know, 50-megawatt you know, drop in buildings that you can take on. And Google's like, no, we want one big 200 megawatt building. You know, I mean, that's basically what they asked for, right? Because Terawill pretty much said that we've got 50 megawatt blueprints for, you know, that we did for Core 42. Can we just drop in a whole bunch of those? If you look at the site plan from the last investor presentation, you'll see like seven or eight buildings. But that's not what Google wants, right? Or FluidStack want, they want CB, CB4 and CB5, which are going to be two separate 200 megawatt, 200 actually 250 megawatt buildings. No, I'm sorry, 200 megawatt buildings. There's 200 and 160 megawatt net. So, you know, that's four times the size. So when you're trying to lease capacity as a co-location entity, I mean, it's not what you want to build, it's what your client wants.
Crypto Miami: Yeah, I think the dual use you're thinking about is the air-cooled with Iron, right? So they have air-cooled GPUs next to Bitcoin, ASICs.
Ed: That's right.
Crypto Miami: And so your question, why should they have built for AI ahead of time? Well, no, because what we're talking about is liquid-cooled. So no one was-- you're seeing the first liquid-cooled data centers come up now, right? So yeah, no one was building liquid-cool ahead of time. That would be insane because the CapEx.
Jamarco: Yeah, that's true. But there's still air-cooled Blackwells that exist.
Ed: Yeah, those are the B200s.
Perry Lin: Yeah, but again, that's depending if the client wants to run those.
Ed: Yeah, right. I mean, I think that was for the CSP business, right? So I think they had more choice there. But when it comes to, yeah. wolf, I think, and if somebody's asking for a GB 200 liquid cooled, I mean, that's just what they're going to have to build a spec to. Now, in regards to retrofit, like you can, right, they may be suitable for retrofit, really, and I base it on some experience, right, because a lot of the data centers we're building now, they weren't, they weren't, you know, they were designed two years ago, three years ago, right? I mean, and so, but they still need to be able to deploy GB 200s in these racks, rack spaces, right? And so to do that, you have to get creative and retrofit. And so you can bring in liquid cooling relatively easy. You know, you'll still have to have like that, the discharge, you know, heat rejection, right? You can still bring in liquid cooling relatively easy. So I think those could be retrofitted if they decide to, to convert that 225 megawatts to, HPC and get rid of Bitcoin mining altogether. I don't, I'm not saying for certain, but they may be able to just, you'd also have to change the bus ways and, be able to provide that rack density, whatever, whatever's required. But, that even that's changing so fast, right? I mean, the VR, I think it's the VR200s from Nvidia, I mean, those go all the way up to 1 megawatt per rack, right? So you just have to have that visibility of what you're gonna need and when you're gonna be able to bring it online and what you'll need to do to retrofit, but it can be done.
Perry Lin: Speaking of that, though. You've heard that from Core Scientific and others, right? like the later core scientific buildings that were basically teardowns. They're like, it's going to cost more to retrofit than it is just to build a new one. And you're seeing Cypher building these hybrid Bitcoin or HPC buildings at Black Pearl. But they just announced that, right? Like what they built before that doesn't seem like it's easily retrofittable. That's why they announced that they're building these hybrid type, you know, buildings in the future. So I think the base case you got to go in is, you know, maybe there's some savings there, but not a whole hell of a lot.
Jamarco: But speaking of the liquid cool, what was with Terra Wolf saying that they're actually going to use a closed loop build without using any of the water from the lake? Why were they saying, why were they going with that when they have the lake right there and the permits to use the water? And that's supposed to be an advantage of those lakes.
Perry Lin: Again, you build to spec to what the client wants. So, you know, I don't think they have a whole lot of input in that.
Ed: And it could be some environmental reasons around that too. I don't know if anybody pushed back on that. But with a closed loop, you're not using any water really. Once you fill it, it's done. So there's some trade-offs between the two. Yeah, it's more expensive. Right, more expensive and from an efficiency standpoint too. I think I think, water cooled might be a little bit more efficient, but then, closed loop, you're not, you're not using any water. So, who knows? Somebody, somebody, some of the local jurisdiction could have pushed back and said, we don't want you using all that water.
Perry Lin: Yeah, it is more expensive. They did raise the CapEx guidance from, it was like 6 to 8 million for a core 42 to 8 to 10 million for.
Ed: But the main reason for that was they're going to bring in people from out of the area to supplement the, their existing construction crew.
Perry Lin: No, that's true. There's a lot of it is labor and tariffs and things are more expensive.
Ed: And to another point, any time like some of the some of the big hyperscalers when they bring on capacity, right? And I know like when you look at their, their carbon and climate goals, and how much of an advantage that is. Right. Just to kind of quantify that a bit, when when you bring on if, for example, if I bring on 100 megawatt data center, I'm I'm also offsetting that by, you know, 100 megawatts of solar. Right. So they remain carbon neutral somewhere. Right. So it's not necessarily tied directly to the data center. They just they just they do a solar project somewhere else that makes them carbon neutral. So yeah, it's an advantage to build, where you're using almost carbon neutral energy, but it's not a showstopper because they got ways to, they'll supplement it if they need to. It'll come at a bigger cost, of course, but that's kind of how that works typically.
Perry Lin: It's like buying, it's like saying you're carbon neutral by buying energy credits or carbon credits.
Ed: Yeah.
Jamarco: Yeah, yeah, but the last question I had is, Harry, what do you think about Hutt and white fiber? So for Hutt, they supposedly have a lot of power in Louisiana and Meta supposedly having a big... Building in Louisiana that I still think that's with what and meta that and they also mentioned on their their earnings call that the reason why that's been delayed is because they're trying to expand that that project to one gigawatt or or or one plus gigawatt. So so you do you think they're they're a good investment right now since they're they're under the radar, but they they might be. signing a deal with a big hyperscaler soon.
Perry Lin: Yeah, I think there's some catalysts there with Hutt. One, they have the Trump brothers, you know, on their side. They have Kotu on their side. We're really, really smart guys. I think they can get a deal across the finish line probably sooner rather than later. But, you know, they gave away a lot of the You got to look at the deal terms with HUT and KOTU and the financing. It's like, how much of that project do they actually keep? That's one concern. And then Crypto Miami knows Asher very well, and he's not a huge fan of that management team. So there's that. In terms of white fiber, it's not at the scale of any type of company that I'm interested in. It's just too small. They're building tiny. They're not a real player in the data center space, where I think you'll see multiples on your return. It's the same reason no one else is investing in it now, and the stock hasn't moved that much. It's just they're not taken seriously, good or bad. I'm not so sure this whole strategy of spinning them out and having the parent focus on Ethereum Treasury is, it's just a, it's an even more muddled story that I have no interest in. Gotcha. Jeffrey? Yep, I've added you. Do you have something to say? Jeffrey, you're on mute.
Jeffrey: Sorry. Hi, yeah, hey, thank you. Love all the information here. I've been calling you for a while, Perry. And this morning, Crypto, what's your name, Crypto Miami? I thought you had a great discussion this morning. I was super bummed that thing wasn't recorded because I missed part of it and I wanted to go back and listen to the stuff that you were talking about, the genesis of the, kind of Paul Prager's role in Beowulf and the leasing of different things and all that stuff. I thought that was just incredibly insightful and a lot of time spent figuring all that out, which, you know, I'm a big investor, but I haven't spent that type of appropriate due diligence looking at stuff. But I was came across this weekend talking to people and I haven't been paying attention to the news, but I don't know if you guys have any comments on the stuff that's supposedly going on in Amarillo, Texas. The with Rick Perry, some investment thing and supposedly trying to build like a 20 gigawatt data center. Do you do you have more specifics about that or understanding about that you can shed light on?
Perry Lin: I don't. The only thing I know about Rick Perry's been doing recently is like trying to lobby for like, you know, psychedelic mushrooms.
Jeffrey: Seriously, there's like there's like something down. There's something really like supposedly big in that space, and someone I know and one of the big four people spoke sideways about it, saying that something big's happening in Amarillo, and he can't say specifics, but it was, and then there's a search where you can pull up something on... Texas unveils plan for world's largest AI data center.
Perry Lin: Yeah, they're trying to I'm reading now. It's like 11 gigawatts on 5800 acres. I mean, if anything like that is even possible, we're talking like years and years and lots and lots of money and power. So they've got to build nuclear there. I don't know. I'm not so sure that that's relevant today.
Jeffrey: Right. So it's too much of a long time to horizon for that type of project.
Perry Lin: Too many times we've heard projects like that never take off, you know, get off the ground. So.
Crypto Miami: Yeah, I see four nuclear reactors in the article. So that's 10 years, right? That's a 10 year build up. Unless it's Oklo, unless we're doing an SMR, but that's Oklo.
Jeffrey: Yeah, I mean, Oklo has that stuff that they can spin out a lot faster and supposedly safer, smaller. And then they can, you know, kind of a piggy bank them together, Jimmy rig them together to create.
Crypto Miami: Yeah, that's yours. That's yours. That's yours off, right? Yeah. Okay.
Perry Lin: Yeah. So that's going to be a while.
Crypto Miami: That's NRC. That's NRC. Yeah. You're dealing with nuclear regulators, different, different beasts. Yeah. Now you're talking federal regulation bonanza. So.
Perry Lin: Yeah, I mean, what's more interesting are... Yeah, yeah. What interests me in Texas are these like natural gas producers, you know, like Diamondback, who have, who are trying to find a spot for the stranded natural gas. And they're gonna just build behind the meter sites. And I've always thought that it made sense for somebody that had infrastructure prowess to partner with these providers. And you know, they're looking for partners, right? Because they don't want to run a data center. They just want to offload their natural gas that they can't give away. And they've got plenty of land. And so I think that's the next iteration on the horizon are these behind the meter type arrangements?
Jeffrey: Interesting. OK. All right, well, thanks. That was my question.
Perry Lin: No, no problem.
Crypto Miami: Harry, do you have any speculation on Brandon Scholz, Brandon Scholz coal plant in Baltimore? I mean, I'm looking at a 1.3 gigawatt coal plant that was supposed to be decommissioned this summer. It's been extended to 2029, run by Talon Energy. And if you know Wolf History, you know that it's PGM Interconnect. Paul has.
Perry Lin: That's why I mentioned it.
Crypto Miami: PGM Interconnect run by Talon Energy and Paul Prager's political influence in Maryland is epic. So I mean, obviously this-- Yeah.
Perry Lin: I'm just-- that's why I mentioned it. And it got extended to stay open until 2029, and they're paying Talon a whole bunch of money to keep it open. And I think there's political pushback on it as recently as this month. And they did-- there was a plan for battery-powered storage or something to reinvent that site. and the grid operators weren't for it. I mean, it's a perfect type of situation where, you know, Paul buys it or TerraWell buys it. They don't even have to convert all 1.3 gigawatts of it, right? They can make a deal where they, you know, they allocate a certain portion of that to maintain the grid and just and be able to offload or or, you know, provide some grid flexibility. So, you know, these are the type of things that Terra Wolf can do that, like, no one else can. You know, I mean, that's where their, you know, where their future growth is going to come from. But, you know, I use that as one example because, you know, all of the players lined up, right? Tree? Thank you. Do you have something to say? You can unmute yourself. No? OK. Anybody else before we shut this down?
Ed: Yeah, just-- I don't know, Perry, if you've looked at the convertible debt that Wolfe holds. But I do have some expectations. At least it makes sense in my mind that coming up, probably pretty soon, that those debt holders would convert their notes to shares. I think if it's above 11, right, is the early trigger conversion.
Perry Lin: I think the interest is really low on that, that there's no urgency to worry. And I think the trigger-- and the upside, I think it has to get over like $18 before there's a huge impact. I mean, $18 can come pretty quickly, don't get me wrong. No, I think the.
Ed: Cap call covered them to like $12.80, if I'm not mistaken. But yeah, so anyway, I think that could dictate the near-term stock price to some extent, especially if they do convert right, that means they deleverage a bit. And if they deleverage knowing that they're going to have to bring on a lot of additional financing, it may make sense that the debt holders convert above 11 and then they exercise that, or Wolf exercise that cap call. So we may get a little volatility around that range. But yeah, it could be interesting.
Perry Lin: I think that's a good point. We're definitely going to see a series of financing you know, events coming up just because they're going to run out of money for the build. You know, they've got like 90 million on the books and they need a lot more of that. And I think they want to take some money out of CB1 that they front loaded. And, you know, with, you know, with kind of the Google news, you know, their LTC is going to probably go down to 90%. And Google will backstop it with some collateral. So I think we will see a series of different financing things to drop. And then who knows how that plays out? I mean, I expect volatility in the stock anyway. And there's just a whole bunch of events that could happen between now and the end of the year. you know, I would stay long, you know. But not as a trade, just, you know, it's an investment, and you should have a long time horizon. If you do, it shouldn't matter what it does in the short term. If you're in it for a trade, you know, good luck to you. This is the wrong space for that. I don't have any alpha for you.
Jamarco: What's your opinion on Coreweave right now? They're The stock is down a lot from its highs. And their share unlocks just happened. So supposedly, all those locked up shares, they're available to sell. Maybe they've already dumped them. But do you think it's a good valuation right now? I think it's under 50 billion. And they have, they said on their latest earning calls that they have, they're up to 30 billion in signed bookings. So that seems just on the surface, that seems like a low valuation, right? But do you buy into some of the rhetoric that people, Mike Alfred and other people that say that since they're asset light, they're not solid and they're going to fall apart since they don't own their infrastructure? And what's going to happen with the... the core scientific deals are going to fall through, are going to restructure or whatever. What's your opinion on it?
Perry Lin: I think I'm a core scientific shareholder, have been for a while, just not in love with what's going on right now. The first thing is that there were two large cell blocks of tranches of insider positions. I think they unloaded maybe a couple of billion of stock, like Thursday or Friday of last week. I think the going price was like $97. So I think we've hit a floor in the short term. I mean, they got the discount, and those shares should be on the market. Whether or not we'll have more shorts come in, hard to say. You can say that it's down a lot from the ties, but you can also say it's up a lot from the IPO, which is not that long ago. Fundamentally, I like Core Weave. I don't buy into the Mike Alfred rhetoric. I mean, they have a huge debt load, but all of it is secured on contract. And because of that, there's really no risk. I mean, it doesn't matter how far down the value chain black walls go. It's contracted, right? It's already these companies, when they sign a deal, are taking on X years of lease payments, and it's paid for already. As long as they continue to pay, and we're talking the Microsofts and OpenAIs of the world, as long as they pay, it's profitable to CoreWeave. So I don't think there's any risk in the short term. The risk in the long term is that the music stops and then the business model no longer works, right? But I think we're years away from that. It's not dissimilar to like Chesapeake Energy, like when they were just building gas, natural gas everywhere and doing generation, right? Like you could make that flywheel work for many, many years until it doesn't work and then you go bankrupt. But we are a long way from that. And their order book is strong. So they're just-- I'm not worried about the debt on the current order book. I'm worried that the order book eventually stops, dries up, and then you really don't have a business. But that's not something in the short term. In terms of the core scientific situation, I think it's 50/50 right now. It's a game of chicken. I think if they raise their price, they can almost guarantee that a deal will go through, or if they at least backstop it. It was signed-- it was basically a $9 billion deal for Coors. If they guarantee that on a ratio of their stock at $9 billion, almost certain a deal goes through. at its current price. And I don't know. I mean, from what I've read, it's like 50-50, because there aren't huge blocks of institutional investors that are going to vote this down. Those investors don't exist. There are only like a handful that own more than 1% of the company. So I just don't know if there are enough votes there to vote it down, but it's going to be close. And so if it's going to be close, it's kind of like a game of chicken. Does CoreWeave really want to make that risk? you know, take that risk or they're just going to up the offer a little bit and just know that they have a short thing and even at 9 billion, they're getting a bargain. Now, I think eventually that's what they're going to do. But, you know, we probably we're probably six weeks to two months away from a shareholder vote. So it's probably going to languish in this in this trading range for a while.
Jamarco: I also heard you say that. I saw you say that. Either way, Adam Sullivan's definitely got to go after.
Perry Lin: Yeah, I think he needs to go. I mean, so I don't know if I'm going to be a core shareholder after this deal, after this all plays out. Just because it's kind of a dumpster fire, then what, right? So let's say they vote the deal down. Well, then what? Nothing has happened in the company. They're repositioning it for Coreweave, they haven't added new sites, they haven't added a new customer, and Adam's really lost some credibility. So, you know, something needs to shake out. I don't know what exactly.
Jamarco: Yeah, I agree with you to me that this deal pretty much exposes him that he's doing it for his own benefit. I don't know if he has some RSUs that get accelerated or something, but I don't see how you signed this deal. Yeah. With no backdopping or having it tied to the core we share, which is what, 400% or something that it definitely was going to tank, which it did right after the deal signed.
Perry Lin: So, well, it's telling, right? Because the core scientific stock has performed really, really well these last few days as CoralWeave has continued to go down. So it's almost as if the further down CoralWeave goes, the less likely a deal will go through, and the market is reacting to that as a good thing. So it's interesting. I mean, the other news in the space is Zack being ousted at CleanSpark or resigning at CleanSpark, which doesn't make a whole lot of sense to me. only because if it was a performance thing and they wanted a change in management, then elevating Matt does nothing for me or anybody else. They were like Batman and Robin, right? So you're just replacing one with the other. Like, what difference does it make?
Jamarco: Yeah, but performance about what, though? It's not... I was off the CleanSpark train when they came out and said that they're not going to get into HPC. But I don't see how they would be in HPC anyway. Their sites don't really seem to be suitable for HPC.
Perry Lin: Well, I think the board and shareholders were frustrated that the stock price wasn't better. And, you know...
Jamarco: Yeah, but I don't know what they could have done about that, unless they were going to get into HPC. they were going to under-- because let's be honest, Bitcoin miner really isn't a great business. I mean, unless you're really, really, really good at it, you're pretty much going to lose money because of depreciation on the mine. Unless you're making your own miners like-- Like BitDeer. BitDeer. But yeah, you're just going to pay all that premium to the BitMain, and you're not going to make.
Perry Lin: Any money. No, I mean, yeah, I mean, well, CleanSpark is not profitable yet, right? We thought maybe this last quarter, they can turn a real profit, but it still wasn't there. So it's tough being a Bitcoin miner.
Crypto Miami: You mean the cost of power matters?
Jamarco: Uh-oh.
Perry Lin: Well.
Jamarco: Yeah, I mean, it's useful if you're using it to bootstrap a bunch of megawatts like Terawulf did with it and Iron did with it. But just building a bunch of buildings and plugging in those miners from Bitmain and it's just, yeah, I'm not surprised that there's-.
Perry Lin: It's not the cost of power. It's the, you know, it's the cost of the miners, right? 'Cause you're depreciating it. I mean, you can be positive cash flow, but it's still not profitable overall. That's the real problem.
Ed: Yeah, I think we'll see a lot of them exit the space. And I think that's what's needed to allow some of the others to be profitable.
Perry Lin: Well, you're going to have them, when is the next halving? It's not that far away, like two years, right?
Ed: Yeah, yeah.
Perry Lin: So the flywheel starts again, they're going to have to figure out if they're going to buy miners all over.
Ed: Right.
Perry Lin: And that's an issue, because you can only dilute so much. I'm curious what these companies with huge stacks of Bitcoin are going to do.
Jamarco: What do you think about the small nuclear reactor companies like Oaklo and that-- are they a potential disruptor to everything that these companies got going on with their power, the grid-connected power advantage? Is one of these small bigger apps going to just come along? Because they've signed some big deal. I mean, I don't know what the-- It's.
Perry Lin: A game changer, right? The stock is like $70 or something. I mean, it keeps on going up, and they're signing customers. So I think it's disruptive. I don't even know if it's disruptive because they have no competitors. So I just think that there's going to be greater nuclear demand everywhere. It's just how long is it going to take for them to build it, and it's going to take a while.
Ed: I think the other thing is the cost, too, right? They're incredibly expensive, and you're only talking for 200 or 300 megawatts, right? So you're in the billions just off just to get 300 megawatts. You know, from a cost standpoint, it doesn't seem too economical at this point.
Perry Lin: Like, well, cost for who? It depends, right? Depends who's fronting the bill for it. Yeah. But I think there are enough power constrained area, you know, geographies where, you know, the rate base can absorb it.
Jamarco: Yeah, I mean, I would assume the.
Ed: Hyperscalers, right, that's going to bother them.
Perry Lin: I mean, the hyperscales can afford it, definitely can afford it.
Jamarco: Some of the deals that they signed are just crazy deals. Like, Switch signed like a 30-gigawatt deal with Opal or something. I don't know what the details of that deal is, but just, I mean, it sounds like they would want to go with them if that technology exists.
Perry Lin: No, I think you're right. I mean, it's definitely interesting technology. We've got to see how it plays out once we see the end product. But it's definitely interesting.
Ed: Yeah, I mean, I think it'll probably get to a point here soon where they simply just don't have a choice, right? I mean, the demand is-- And it's so great. It's multiplying so quickly. You're just not going to have a choice. You're going to have to do whatever you can to bring capacity online. I mean, but a lot of times you say, oh, well, yeah, hyperscalers got it, right? But you got to realize how much money they are spending too, right? And if you remember Microsoft, they were under pressure for a long time in their stock because because they were spending so much and investors like, it's too much. They weren't rewarding the stock. And then they finally started getting the returns and then everybody's applausing them for doing it, right? So there's a little bit of back and forth and juggling of that balance sheet and how much cash on hand that they need to be able to spend. And investors will give you credit for spending, right, or cheer you on for spending it. And yeah, sometimes when you have to build out gigawatts of data center infrastructure, on top of that, all the GPUs, on top of that, you're depreciating, so you're constantly swapping out. Like, the cost is just freaking incredible, right? And so, yeah, I mean, there's just a lot of concessions. that need to be made. And if you look at like what Mark Zuckerberg did with the tents, like that just kind of goes to show you one, how important speed is, and two, like how they need to be cost conscious of doing it, right? You can't, you can't just throw a, you know, and that's why people say, oh, why, why aren't every all these getting, you know, bought out? Well, yeah, if you, if you pay 10, 10 billion to buy it out, now you got to put another 10 billion in to build out the infrastructure and another, another 5 billion for GPUs, you know, or, you know, just, just rough numbers. But, but yeah, the cost is, the cost is incredible.
Perry Lin: And well, I mean, the Oclo story is a power story. It's a national security issue. It is not about any one hyperscaler. It's much bigger, right? And if you hear what BlackRock and what they're saying and what the hyperscalers are saying and what Stargate is all about, it's collaborative. Everybody has to do their part to build up the grid, build up the transmission lines, and then to create enough, to have enough generation, a lot of it needs to be the government. and the private sector, everybody working together. And so it's not about any one entity.
Ed: Right, right.
Perry Lin: It's just going to, you know, and it's going to take a long time and it's going to take a long time. Yeah.
Jamarco: Yeah.
Perry Lin: With that, unless anybody else has something to say, I'm going to shut this down. We've been going for a while.
Jamarco: Well, I guess one last point is what I've been trying to figure out is with these with With Oracle and maybe to a laser extent, Core Weave, they are basically asset-light AI hyperscalers. So on Oracle's last earnings call, Larry Ellison was talking about all this stuff about how much demand and capacity that they're gonna need. I'm trying to figure out who's gonna service that because it You know, there's some companies like Iron and maybe even TerraWolf for now that can get a lot of capacity online. But is it real though? Like, is Oracle really going to be able to sign contracts for tens of gigawatts and do it all asset light and do deals with the miners that have a bunch of capacity? Like, do you think all that stuff is real with those asset light? hyperscales like Oracle and CoreWeave.
Perry Lin: I mean, Oracle's got deep pockets. I think they're part of Stargate, right? So they, I mean, how asset-light are they going to be? I mean, I'm sure they're building out their own data centers, but, you know, it's never enough. That's why you see CoreWeave selling so much capacity to Microsoft and Google and everybody else, right? So they can always go that route and go to the neo-clouds and get capacity. But Oracle's still a bit player in the cloud. I mean, I think they're fourth or fifth. I mean, not that it's fourth or fifth, doesn't matter, but they're still trying to grow and scale up. They're not Microsoft, Amazon, or Google yet.
Jamarco: Well, what Oracle was saying was that the reason why they're going to end up being the biggest cloud, AI cloud, is because it's going to be all about data. And since they're originally, they have the Oracle database, which is the biggest database in the corporate world, that all these companies that are gonna end up, you know, having their AI strategies, they're all gonna have to do it probably with Oracle because they have the data already. And that's why they're so confident that they're gonna grow so much and become the biggest. So that's why I'm trying to figure out what's going to happen with Oracle and are they going to really be asset-light and are the miners going to benefit?
Perry Lin: Well, I mean, the miners don't exist in a vacuum. There are a lot of players trying to build data centers and capacity outside of the miners. Let's not forget that. You have the Brookfields of the world still out there trying to do deals and related companies and everybody else, right? But it goes to a broader question of, and I think this is true, there's not enough, everybody's said it, there's not enough power to go around, right? And no one has come up with a solution yet as to what that means. And I don't think Oracle's in any different situation than anybody else. If there's not enough power to go around, you're just not going to have enough power. All right, with that, I'm going to end this. We've been going almost two hours. Appreciate everybody being on. It is going to be recorded, so whoever missed anything can go back. Thanks, guys.