← Blockspace: AI & Bitcoin
Blockspace: AI & Bitcoin
IREN’s $3B Note, CME Compute Futures, Mike Alfred’s Stock Picks, Trump’s Q1 Bitcoin Equities
01:25:33
Loading summary
Transcript205 lines
- [00:00]
A
Foreign.
- [00:05]
B
What is up y'?
- [00:06]
C
All?
- [00:06]
B
Welcome back to Block Space Live. We are capping another week in AI and bitcoin with our final live stream here on Friday. We are your hosts, Colin Harper, Charlie Spears. We've got a packed lineup today. We are going to be leading with a story of iron raising the largest convertible note of any public miner to date. Go into our hash rate index update and then we've got a really interesting story on an anonymous Twitter user named Soup, I should say pseudonymous who recovered 5 bitcoin using AI. Yes, if you think that you have lost life changing amounts of money, maybe the LLM gods can save you. After that we will be having Luxor's Mike San Miguel on to talk about the GPU market. Then we will hop into some news about Trump disclosing some crypto holdings, specifically bitcoin equities for his family trust and and then news that Hut 8 is facing a stonewall with its Logan County Data center in illinois with a 90 day moratorium. And to cap off the show we have Kush Bavaria of Oron on to talk about how their index is being used for the latest and first I believe compute futures market on the cme.
- [01:15]
A
And don't forget we have a little guest named Mike Alfred also lined up.
- [01:20]
B
Thank you Charlie. We have Mike Alfred coming on as well to talk about his Alpine Fox hedge fund and which bitcoin miners turned AI companies he is most bullish on and also his thoughts on the current Bitcoin rally.
- [01:33]
A
Mike's going to give you the best stock picks that you should absolutely follow. Block Space goes live Monday, Wednesday, Friday Noon Eastern featuring quick hits on the latest in mining, AI, emerging tech and sometimes Bitcoin. Still, make sure to like and subscribe. Hit that bell to get the notification on YouTube. If you like what you hear, this turns into a podcast. Wherever you find podcasts, Spotify, Apple and hit our newsletter newsletter.blockspacemedia.com for those CoinDesk listeners. If you're listening on Coindesk through their channels right now, I would recommend you also head to a Blockspace channel to remain have show continuity going forwards. This show is brought to you by CleanSpark. Nasdaq listed ticker is CLSK MORON CleanSpark later on in the show. Let's kick it off Colin. When I got into this business I thought we were firmly out of the GPU era and we'd be talking about Asics, but we are only talking about GPUs. Every single story we're doing today is about GPUs just about.
- [02:45]
B
Yeah, I think that's right in some form or fashion. And before we get into that, and we'll actually cover, you know, the other side of the compute market that we usually cover with the hash rate index. Update here in a second. But just a quick market update. If you're wondering, are we so back or is it so over, guess what? Nobody knows. Everyone's confused. We have a salad bowl market opening up today on Friday. Pretty interesting considering it's called what's the salad Bowl?
- [03:11]
A
For those of us who are.
- [03:12]
B
I call it a salad bowl. When it's mixed returns like. So you've got got a lot of red Nvidia bleeding while Microsoft is pumping. You've got Apple gaining while Google is dumping on dumping. And what's funny about this is it coincides with Jerome Powell's last day as Fed Chair. So there's something maybe, you know, twistedly about that. I just wanted to throw that up there because seems like we really are reaching a point in the market where everyone's kind of groping for what valuations make sense. But we'll leave that where it is and go ahead and move on to our story of the day. Iron closing a massive $3 billion convertible note offering at 1%. These convertible notes are due 2033 and they have a 32.5% conversion price. So that 32.5% shakes out to $73.07 priced from, you know, roughly over $50 per share as of May 11th is the reference rate for the convertible note. And iron with this convertible, I think
- [04:22]
A
we may have lost Colin. I'll keep the show going while Colin figures it out. So, yeah, this is a massive.
- [04:36]
B
All right, sorry about that, y'.
- [04:37]
D
All.
- [04:37]
E
A little.
- [04:37]
A
Okay, you're welcome back. Welcome back.
- [04:40]
D
Thank you, Charlie.
- [04:43]
B
As we've seen with other convertible notes in the bitcoin mining ecosystem, they are entering a cap call transaction. That cap call allows them to hedge future dilution. It's basically a way to net out any dilution that would necessarily occur from the convertible note. That cap call transaction is capped at $110 per share. So anything between 73. Any conversion rate between 73 and $110 per share for iron will give them the opportunity to hedge future dilution. On this note, this is the largest convertible note for a public bitcoin miner to 2 billion from a year note in Q4 2025.
- [05:24]
A
So
- [05:27]
B
it's interesting to see fundraisers. Back in the news cycle, we kind of had a lull there For a little bit at the beginning of.
- [05:35]
A
Yeah, absolutely wild. I think we'll probably come back and talk about this with Mike Alfred too, he notable Iron Bull. So we'll hit that in a second.
- [05:50]
B
One last note before I jump to hash rate index here. Iron is going to need every ounce or every penny of fundraising they can get because unlike other bitcoin miners that are doing the infrastructure play where they're just building the data center, managing power, Iron is doing that and they are deploying the GPUs. We've got a really good report coming out next week on how to value megawatt covers the different capex scale for all these different AI builds. So if you're curious about the dollar per megawatt cost of a data center like Iron's building versus its peers, stay tuned for that on block space next week. All right, before we get soup on waiting in the wings here and Charlie, it looks like we are going to maybe get. We actually just got a difficulty adjustment on live during the show. So the current difficulty adjustment just. Yeah, a day of first it seems we had a 3% adjustment just now, negative 0.3% adjustment in the last difficulty epoch. And this comes as hash rate is gunning 4,000 exahashes 1 Zeta hash. Again, we are currently at 998 exahashes hash price sitting. Not necessarily pretty, but a lot better than all time lows at about $37 per PETA hash per day. So miners aren't starving. They're not exactly feasting right now either, though I would say.
- [07:09]
A
Yeah, I mean, not a lot to be said here. Again, we're just kind of sitting on our hands while bitcoin figures out if it's going to go up or not or if we're going to go down. And that's really all we could say about that. Do we want to talk? I think we were going to talk about the, the. The American royal family doting out shares of.
- [07:34]
B
Oh, that's. We have that later. So you can go ahead and get soup on.
- [07:38]
A
Okay, let's do this. Here we go. This is the interview I've been looking forward to. Pseudonymous guy, OG wallet. Let's bring on soup. Let's do this. Sue, welcome to the show.
- [07:52]
D
Hi. Welcome. How are you guys doing?
- [07:54]
B
Great, man. Let us be the. Maybe not the first to say, but congratulations on recovering a huge hoard of bitcoin. That's a really cool story.
- [08:06]
D
I never thought I would see the day.
- [08:08]
F
Yeah.
- [08:09]
A
Okay, so the tweet went viral. You were Main character for a couple days. Probably still main character. Just give me the, like, the TLDR on what happened and then we'll throw some more questions at you.
- [08:20]
D
Yeah, I mean, long story short, got a bunch of coins in college, went back home, you know, all my friends were like, dude, you are so dumb. I can't believe you bought these coins. Whatever. Ended up changing my password one time and, yeah, I just could never get back in. It was one of those blockchain.com, like legacy wallets. Kind of a forced hold. Forgot about it for like five years, then started to try to break into it, like, could not figure out bitcoin recover. Hired a couple of those, like, prominent hacking teams. None of them could do it. And then eventually just got really AI pilled in the last few months and just started dumping all my old documents, files, going through math notebooks, finding old seed phrases, anything, any context I could just feed into Claude. And then it was crazy.
- [09:12]
B
It.
- [09:12]
D
I ended up finding this old seed phrase, Claude, decrypted it, and then it was my old password. So I was like, oh, my gosh, that's like. That's not helpful at all.
- [09:21]
A
What? What? That password. By the way, can I. Can I invite you to.
- [09:25]
C
Yeah.
- [09:25]
A
Tell us what the old password was.
- [09:26]
D
Yeah, yeah, that was the LOL 420, the police with some symbols, which is hilarious. And just sums up like, who I was in college. And then I had one last ditch effort two days ago, woke up, dumped a bunch of. I found this other computer, just started dumping all the downloads, documents, etc. And it found this old wallet file. And I was like, great, like, old wild file. Oh, I have the old password. It doesn't matter. I changed it. And not thinking that the old password and the old wallet file works. And, you know, I had everything the whole time. I just didn't realize how to. How to piece it together.
- [10:07]
B
So.
- [10:07]
A
Yeah, so I think, you know, a lot of people see the headline, like, guy uses AI to recover old wallet. I'm actually pretty interested because you gave a little bit of, like, insight into exactly what you were doing. You said you'd hired some of these hacking teams. I know these guys, you know, that you, like, give them all your info that you can. They take a, you know, a bounty or a pound of flesh, but they weren't able to do it. What exactly are you doing when you say using Claude? Like, what? Yeah, walk me through the process here.
- [10:39]
D
Yeah, I mean, it really just. I was just feeding it context, feeding it all of my old passwords. Asking it to like run bitcoin recover for me because I'm technical, like, but not a software engineer or anything. Asking it to like try different regex combinations, run it in the cloud, you know, save everything. We've tried. We essentially tried seven, I think, or 3.5 trillion passwords over like the last six months. And I, as you can tell, make really complicated passwords that like would never be brute forced. So yeah, essentially just feeding it a ton of context and then eventually it connected the dots.
- [11:22]
B
So soup, this must have been gnawing on you for years, right? Like, did you have previous recovery efforts and what was the like light bulb moment where you're like, yeah, I mean, quad for this.
- [11:32]
D
You know, it's like I pretty much put a thousand in. So. And then I stayed in the industry, like, you know, made whatever did all right to like didn't give up obviously and like lose faith in the industry and did want to shout out CB spirits because I was heavy early on ordinals. I had a few of the 500. That was life changing as well.
- [11:54]
A
Yeah.
- [11:56]
D
But yeah, kind of gave up for a long time and yeah, just decided this year, hey, I'm going to give this like a final effort. Going to go through all my old paperwork, documents, computers and just like get this thing out of my head.
- [12:12]
C
So.
- [12:12]
D
So yeah, sorry I went on a bit of a tangent there but yeah, just, just have gotten AI pilled and just figured that, hey, Claude can kind of probably help me figure this out.
- [12:22]
A
And this might not. This might be kind of a dumb question, but like when you say trying what 3.5 trillion passwords, you gotta like rent GPUs or something to do that. You gotta like rent compute to run all that.
- [12:35]
D
I spent $15 on vast AI credits to do that.
- [12:42]
A
So that was like the. Yeah, 15 bucks
- [12:46]
F
credits.
- [12:47]
D
And that was probably. Oh, sorry, you got. That was probably the thing that like prevented me from trying all this stuff before was like I was like, oh, I need to like buy a GPU or I need to like figure out, you know, how the cloud works. So it was nice to just kind of have, yeah, this, this little handy buddy piece it together for me.
- [13:07]
B
The reduction in the cost of compute is just insane to me. Like when you think about how much anything costs like in the 90s, you know, for your computer, it's crazy to me to think that you could run 3.5 trillion, you know, passwords with $15 in credits.
- [13:24]
D
Oh my gosh. Yeah, it was crazy. I forgot the like, it was running like Hundreds of thousands, like, per second. So it's. Yeah, that took a few days to try, which was. Which was quite ridiculous in itself.
- [13:37]
A
So I bet you everybody's asking, you've given a little bit of a hint. You know, you got this nice windfall. You know, are you. Are you gonna. Are you hodling it? Are you gonna reward. Are you gonna go on a spree? I don't know. What's your thought process here?
- [13:55]
D
You know, I would say I'm still bullish. Cypherpunk, believe in bitcoin. I'm slightly disillusioned with, like, the industry as a whole, and there's other cool sectors going on and, like, AI and defense and stuff. So I kind of want to, like, do that. I'm also at the age where, like, getting, like, some land sounds appealing. Stuff like that. I'm already a holder. Like, I'm already exposed. So, yeah, definitely not. Definitely not dumping or anything, but it feels like we're starting to run up again. So, you know, I think the best thing to do would be sit tight and. Yeah, just be. Be a little patient about it.
- [14:33]
A
I mean, I see we're not. It's no Lambos or we're buying land. I think that's like, the new land is the new Lambo. You could say.
- [14:41]
D
Yes. Yeah, we need some. We need some acreage out here.
- [14:45]
B
A better Lambo, some might say. I think I'd rather have that than, you know, a depreciating asset.
- [14:51]
D
Lambo over Lambo.
- [14:53]
B
So you mentioned, you know, you're just saying you're a little bit disillusioned with bitcoin right now. Can you unpack that a little bit? Like, what exactly? About, like, the current place and time that we are in. In the market and specifically the participants. That has you kind of being like, this isn't really what I signed up for.
- [15:08]
D
I would say more so, like, I feel like the crypto industry as a whole has, like, tainted bitcoin. But then on bitcoin itself, obviously, we have, like, the Michael Saylor owning, like, a huge chunk of the supply. And now all of a sudden, he says he wants to sell things like that. Obviously, there's the quantum fears coming around as well. I would say those are probably the main two things, which I don't think are. Yeah, obviously I see a future where bitcoin's going to hit hundreds of thousands and millions of stuff, and it still retains the cypherpunk mentality. But, yeah, I would just say those are some of the fears. Okay.
- [15:52]
A
Because. So last Question. You know, you mentioned you're interested in other stuff. You know, maybe you got some wind, you know, at your back to go pursue other things. I don't know. Tell me about what you're interested. What, what, what is most of your attention focused on these days now that you recovered the OG Bitcoin wallet?
- [16:09]
D
Yeah, yeah. I mean still in the industry doing like automation work, that kind of stuff. So I really just like, I enjoy using Claude in my free time, like building a ton of little scripts and bots. I love like chatting back and forth with it. I probably have a bit of like AI psychosis as they call it. So I would just say like getting more acclimated to the AI stuff. It's just, it's fun, you know.
- [16:37]
A
Soup. So happy for you. What an absolute fantastic story. You love to see it.
- [16:42]
B
Yeah, I actually have one, one or two more closing questions here. Charlie, you said you've been getting really AI pilled here. Soup. Like what are some of your favorite tools to like what, what is your, what does your stack look like?
- [16:54]
D
Yeah, I just have a bunch of little folders. Like I have one for health related stuff and I'll just feed it contacts or one of my doctor's appointments are and like stuff like that. I have one for finances. I have one bot that's like sending me land and houses in the area that I desire whenever they pop up. So just like little things that I do manually in my life, like you know, you can just kind of create these little folders and, and have it bounce back and give you some ideas and stuff.
- [17:24]
B
Are you tapping into like a public API on like Redfin or Zillow for the homeless things? Are you going into the like, what is it? The. There's a database for real estate brokers.
- [17:35]
D
Yeah. The thing is now it can just like do headless browser and like go right onto the website and bypass like the need for the API and stuff. So yeah, it's, it's, it's really neat. Kind of, kind of all the stuff you can do.
- [17:48]
B
Yeah, that's awesome. It's, it really is kind of this. Dare I like use a cliche? A brave new world of compute. I'm just amazed. Every single day I'll log on to X and someone used AI for, for an edge case that 10 years ago no one would have even thought about.
- [18:02]
E
Yeah,
- [18:07]
A
it's. You know, I was reflecting at the top of the show. Like I'm in a longtime bitcoiner. I thought my days of GPUs were way behind Me, now, I used to be like Bitcoin miners, ASICs, but now all we do is talk about GPUs. Kind of feels like a throwback now that GPUs are back, front and center.
- [18:28]
D
It's crazy. Yeah. I feel like we're in a really pivotal year for technology as a whole, and I think the crypto industry, too. So excited to. Excited to see where it takes us.
- [18:42]
A
Yeah, Kong, I think I don't want to waste any more Soup's time. He's got land to go buy.
- [18:50]
D
Thank you, guys. Also, shout out, shout out. Mike Alfred, excited to see him on here next.
- [18:56]
A
Yeah, we'll get. We'll. We'll have. We'll get him. We'll get him to give a take on the. The recovered bitcoin wallet. You know, he's always got a good take.
- [19:04]
D
All right, I'm ready for it.
- [19:06]
B
Cheers. Thank you. Appreciate it, man.
- [19:08]
D
Yes.
- [19:11]
B
One of the more interesting stories I've heard from AI recently, I mean, you always hear wallet recovery stories, right? And there's kind of this impossible outcome that people are chasing. And typically, like you were pointing out, Charlie, you would go to a kind of like, white hat and say, here's a bounty. I want this wallet back. You can have X percentage runs anywhere from like 10, 25, maybe 30 on the high end. Right. But it raises an interesting question that as these AI tools get much better, the cost for recovery should theoretically come down. Right. I mean, Soup just did it with what? With his Claude subscription and $15 in tokens.
- [19:56]
A
Think about it, because this is a pretty serious business model, and it can be very profitable because usually the recovery takes like a percentage of the wallet, like 30%, 25%. And if you're recovering wallets with 10, 20, hundred plus Bitcoin, you're talking one
- [20:15]
B
from like 2012 or something, right?
- [20:18]
A
Yeah, you're talking massive, massive amounts. And this used to be kind of really an arcane, specialized practice. You basically had to run all the GPUs yourself, maybe, if you're really savvy. But now you can have people who are kind of technically inclined but not fully there. CLAUDE handholds them. They spend some Claude credits, plus a few bucks in compute, and now they're guessing their own passwords. Like, this is what. It goes both ways, both for the individual, like Soup, who's like, recovering his own wallet, or the hackers. So, I mean, it's about time you get on top of your stuff. Like, if you have that wallet, you might be racing against other people out there. To get your bitcoin recovered before they recover your bitcoin for you.
- [21:03]
B
So 100%. And one last note on this before we get Mike up here. He just entered the green room. When I hear him saying that he spent $15 to run 3.5 trillion passwords, the, you know, the math doesn't make sense to me in the sense of like, that seems incredibly cheap. And I keep wondering, I was talking about this with my dad the other day. When are AI is going to have their Uber or Airbnb moment? And what by that I mean anyone who used Airbnb and Uber in the early days was shocked at how cheap it was compared to the alternatives in hotels or taxis. And part of the reason for that is that to bootstrap, these platforms had discounted rates more or less or weren't taking as much off the top from the riders or the home providers on the platform. But that changed once the VC money dried up. So I'm just curious, like, you know, AI kind of almost has this, it has this almost chicken and egg problem where the revenues right now are not going to be enough to run, you know, you know, chat GPT, chatgpt or quad potentially. Chatgpt is thinking about using ads, right. To supplement that revenue. When are we going to have real price discovery on what it actually costs to use?
- [22:15]
A
They'll get the revenue from recovering old bitcoin wallets. That's, that's the, that's the near. That's the bootstrap.
- [22:21]
B
Yeah, or in OpenAI's case, they'll get it from TBPN sponsorship or something like that.
- [22:27]
A
Okay, we got Mike San Miguel from Luxor, but right before that, a word from our sponsor, CleanSpark.
- [22:32]
B
We are CleanSpark, America's Bitcoin miner. A publicly traded company with the largest operating hash rate powered entirely by self operated infrastructure across four states. This is our proof of work and we are setting the standard for what's next. Learn more about the intersection of energy and bitcoin@cleanspark.com
- [23:03]
A
all right, let's bring on up our boy, Mike San Miguel. Welcome back to the show, Mike.
- [23:10]
F
Thanks, thanks for having me.
- [23:12]
B
Always nice to have you, Mike. Appreciate you hopping on. Charlie and I were joking when this started that almost every single topic today covers GPUs in some way. And the most absurd one that I want to touch on first with you. I didn't even hear of this until we were prepping in our group chat for this. What is this pearl coin nonsense where you can apparently mine this Thing called Pearl coin by renting GPUs. Can you explain to me how this works? Because it's not like mining. Right. They're basically issuing a token if you rent GPU space. What is this thing?
- [23:49]
F
Yeah, there's small hardware requirement and that's mostly where my specialty lies. It does require running some algorithms on the actual GPU hosts itself. Not sure what that algorithm looks like. I know that's something other parts of my team are working on, but it's popped up, it's less than 10 days old and we've been getting hit up. I think I've been on two or three calls so far this week, just around it, basically. The tough part with the math here is that renting GPUs, the ones that can run at H100 and H200, are costly because they're in demand from normal AI users. There's a subset of GPUs called NVL that aren't really optimized for doing training at scale because they don't have like the back end networking features. It's a perfect fit because those don't have too much of a place in the marketplace.
- [24:46]
A
Yeah. Okay, so you're getting outside and we'll just. One more. This is the only time we're going to talk about like a crypto token. Okay, so who is hitting you up? Because it's not just like degens who want like a token. Are these like legit counterparties who are looking to actually figure out how to acquire this thing?
- [25:05]
B
Yeah, yeah.
- [25:06]
F
It's VCs, private financial firms, miners that are interested in getting in on this early. Nobody knows really where it's going. So the first two to three months are sort of where the demand is seeing some interest for 12 to 18 months. The difference with deploying these is that, you know, these come in a server, it's not a mining asic. So everything has to be engineered to work together and that has to be deployed in a way where you can actually manage the servers. The hardware requirements are very different because it's low ram, low storage, low CPU on the server, which is completely different from doing an AI install which requires multiple fiber networks that are very expensive. So it's a great way to use up equipment. The real question is whether or not this is going to be staying long term and people can make the investment in the hardware specifically for it.
- [26:03]
B
All right, so moving back into big boy topics, so to speak, getting away from the new tokens of the day. I've been hearing chatter about application specific integrated circuits, ASICs being rolled out or being R&D for AI workloads. You know, we're using GPUs currently. There is this future where we will use ASICS for those potentially. What's the latest on how far along this technology is? What is your understanding the feasibility of ASICS for AI as it stands today?
- [26:36]
F
No, they're totally feasible. They're just starting to hit the open market now. The hyperscalers have been doing versions of these for about a decade. So if you've used AI features on Google aws, you know any of the big players, they're already using it. The deals are already in place. Cerebra set a $10 billion deal with OpenAI. Earlier this year, Nvidia bought Groq for 20 billion. So these things are moving forward. What's different now is that as opposed to being bespoke chips for hyperscalers, we're starting to see new manufacturers that are producing these and as a result these are going to start hitting the NEO clouds. The feasibility is that when you look at a gpu, you can use it for inferencing, training and agentic. With these, they're purpose built for one or the other, they can't do both. But that's the advantage. You get a much higher output going to something like this.
- [27:41]
A
Yeah, and that's what I've talked to some people. So I try to imagine what this industry looks like over the next 10, 20 years with your GPUs. They're typically much more flexible with the type of compute. But if this industry matures, as I assume it will, billions are being poured in, you will have spent. What's the implication here? I think at the, at the manufacturer level, like, you know, are GPUs actually that great to own in 20 years?
- [28:12]
F
Yeah, that's a great question. So you know, what it's probably going to look like in the next, you know, two to five years, because these things take time to roll out, is a hybrid approach. And that's where, you know, a lot of the OEMs for these that I've spoken with and ODMs, that's where they're headed is, you know, maybe out of a deployment, 20% is going to be an inference ASIC. The supply chain is building out around it. A lot of these are optimized to make the supply chain more efficient. Where they've got SRAM that's printed onto the chip as opposed to using that HBM3 that everybody's trying to get their hands on. These Days, so it'll make things easier. In terms of the actual deployment, you still have a lot of networking, liquid cooling. All of the fun parts of doing AI at scale. The ODMs are starting to produce these and starting to figure out deals to get them into production. You're probably going to see them hitting in scale in 2027 is when they're going to start entering fleets for a lot of these NEO clouds.
- [29:20]
B
Interesting. So 2027 roughly for when we should see these rolled out with the NEO clouds and smaller players on the market. I'm curious Mike, what does this mean for ASIC allocation at top foundries like tsmc for the bitcoin mining ASIC manufacturers? This has to be kind of an alarm bell moment for them, right? Or am I wrong about that?
- [29:41]
F
Everybody's trying to get as much space in fabs as possible. So far, at least on the AI side, the hurdles I've seen are more around memory and storage than they are necessarily around GPUs and chips. I think overall the foundries are going to be doing more. I don't necessarily think this is going to be starting out a large enough drop in the bucket to make a lot of waves that will affect ASIC manufacturing.
- [30:10]
B
So basically the idea being like they're going to the foundries will scale accordingly with this new demand. I mean it makes sense, right? So perhaps all the other ASIC manufacturers won't really have any problem with the securing allocation for asics in the future.
- [30:25]
F
Not that I foresee, but I'm also not an odm, so there might be some part of knowledge that I'm missing there.
- [30:32]
B
How, how likely is it, do you think though that these ASIC manufacturers for bitcoin mining tried to move into AI? I mean, is that just it? Can they, is that like way beyond their capabilities? Do you think they could bring that in house to design tape outs for that? You know, because when you look at the industry at large, these ASIC manufacturers are kind of getting hit right now with a double whammy. Bitcoin price is down, hash price is down as a result. And their largest customers, public miners, are moving into AI services. And so the kind of massive orders that would give them a foundation for their annual revenue has kind of crumbled away. So do you think there's a future in which they try to switch into other forms of compute or is that off the table?
- [31:20]
F
I think it's possible. If you look at etched. Etched I believe has been an 18 month trajectory and they're one of the big Players in the AI ASIC market. So you can do this quickly. Cerebras I know has been around for about a decade. Tensordyne that's spinning up I think is maybe a year or two old. So you can move fast with this. What it's looked like historically is that Broadcom and Mellanox have done the majority of the actual designing for these, for the hyperscalers and you know, it's changing now is that the people that are designing that are not part of, you know, that traditional space. To my knowledge, Bitmain has actually developed some of these in house. I think they are for the Chinese market which, you know, there's a lot of US requirements around security. So that could be more of a limiting factor than actually doing the, you know, design and bringing it in. That being said, these are expensive things to design. They're not easy. It's a very different design from a mining ASIC.
- [32:31]
B
That's interesting. So, sorry. So Bitmain is already producing AI ASICs for the Chinese market to your understanding?
- [32:38]
F
Yeah, to my knowledge. I've seen a few articles on it that really hasn't hit where, you know, we play in because by and large the market that I serve is more US based and European based. But there are a few articles out there about them starting to make the shift back I think in 2017, 2018 was the first news that I saw about it.
- [32:59]
B
Wow.
- [33:01]
A
So it's, you know, every single news show like ours is always trying to like is now trying to like basically look at the stock market and try to make picks based upon like this crowd level intelligence. I mean like my cursory looking around, I don't see a lot of these AI man, AI ASIC ODMs. Like I don't see a lot of them. Like list on the US stock market. I mean where do I go to be long AI ASIC chip companies and manufacturers.
- [33:35]
F
Yeah, it's a great question. So the spot that people aren't looking at is really the people that actually manufacture all of these, these. So if you look at a lot of your OEMs, like your Dells, your Super Micros, your HPEs, they really don't actually manufacture everything. They're kind of like a US car manufacturer where they, they actually assemble parts, if not just order them directly from. There's these main manufacturing ODMs, they're sort of like TSMC but for actually putting the systems together. Those are Foxconn and Ventech, there's a few others and they do the actual manufacturing. They'll brand Them, whatever person ordered them. But these people are going to be doing the same manufacturing of Cerebras etched Tensor Dyne Tension Torrent is another one of them. It's probably a better bet to bet on the people that are actually making them than to bet on which out of these five is gonna be, you know, the big manufacturer there. Nvidia is also another one because Nvidia bought Grok for 20 billion. So they're including that. You know, I'd go for a portfolio of them. But the people that are actually making them, that's not going to change.
- [34:53]
A
Yeah, it's funny because all these, everyone's like, buy the picks and shovels and buy the. So you. And they're like, buy the, like the, the OEMs, but actually you want to buy the folks who are, you know, manufacture, smelting the iron ore to make the shovel head. In a way. It's kind of a funny inversion. Okay, so I guess another thing is a lot of folks we're talking to today, we're talking about GPU rentals, GPU compute futures. We're going to come back to with our guest later. What are your thoughts on GPU compute futures? The CME Group announced compute futures this week. I don't know. Open question here.
- [35:36]
F
Yeah, there's a lot of different people that are trying to do this. There's orn, there's Compute Exchange, There's a number of different people that have looked at it. It's tough to do because there's so many different factors that go into those contracts. I deal with them daily because you have to get an offtake contract typically to get financing to get the GPUs and the infrastructure to go from mining to AI. So I look at these a lot and there's a lot of variability. I don't necessarily think it's something that anyone has really made a good comprehensive product on yet.
- [36:16]
B
It seems to me the going back to what you were just saying about this being difficult, there are so many different forms of compute and so many different ways that you can use that computer that I struggle even to think about how you structure a futures product for compute when there are so many different applications for it. You don't have the same problem with hash rate. Right. Because it's like there's one function, but with these other things, you could do so many different things. You can do inference, you can do training, you can do a suite of different things even within the inference category. So I struggle to think how you even narrow that down and where the market starts. Right. Like what do they focus on first? Yeah.
- [36:56]
F
Within the actual reference architecture that all of this technology is based off of, there's a lot of variability not just between your GPUs, but between the storage technology, the network and then even just the network out to the site itself. It gets really complicated. I've seen a lot of people that are doing this come up with other products outside of just the futures as a way of generating extra revenue. I don't know if there's a good way to do this. Honestly, I haven't seen it done in a way that makes sense yet.
- [37:31]
A
Well, we have Kush, co founder of Orin Exchange on later on the show. So I'm going to take this question. We're going to hit it with it later. So, Mike, thank you so much for coming on the show. Very excited to talk with you. As this trend continues, continues over 26 and 27.
- [37:53]
F
It's exciting. It's going to change more than it stays the same.
- [37:57]
A
Cheers.
- [37:58]
B
Thank you, Mike.
- [38:02]
A
All right, and we have another guest, the Mike Alfred coming at you right after a word from our sponsor, Luxor.
- [38:17]
B
This episode of Block Space Live is brought to you by Luxor's Commander Bitcoin miner management software for enterprise operations. Commander gives you real time fleet monitoring, bulk remote commands across your fleet and intelligent miner, which is an automated profitability engine that adjusts your power settings to live hash rate and energy markets. Ercot backtest show 10% increased profitability with intelligent mining versus binary mining. Commander Pro is a hundred dollars per megawatt or a 25 basis point pool fee adder which is roughly half the price of competitors. And it's also free with a 60 day trial. So if you want to get started, try it for 60 days. Go to Luxor Tech forward slash Commander. All right, and now we got Mike and Mike, we got back to back. We've got the Mike Alfred, head of
- [39:06]
A
Too many mics on the show today. Bring them up, Mike.
- [39:11]
B
Welcome to the show, sir. Thank you for joining us.
- [39:13]
C
Hey guys, how you doing? By the way, I actually led the Luxor series a way back in 2021. A lot of people don't know that.
- [39:19]
B
Oh, wow. Early mover with that one and early mover with a few other things as well. We've got your disclosures for Alpine Fox up here. We really want to touch on some of your bull theses for some of the holdings you have here. I would love to start with two, two stocks that we cover a lot on the show. Cipher and Iron. Those are your two biggest stock positions as I understand it from your filing. Why did you choose these two as your front runners? What do you see in these companies that elevates them above their peers?
- [39:49]
C
Sure. Well look, this is critical to my original thesis. Way back in Q4 of 2022 when I started to look at taking these bigger positions in the open market, I was already on the board of Iron. So Irons, you kind of have to put that aside because when you make a big bet and you choose to join a board and participate in the corporate governance directly, obviously you're, you have a higher conviction but also you have more restrictions. So part, part of the reason why the Iron position has been historically smaller than ciphers because it completely limits my ability to trade. I can't use options, I can't hedge. Right. I can't, I can't reduce the size of the position or increase the size of the position very easily. In fact, you'll note that I haven't once sold a share since I acquired the Iron shares in the open market. So I just want to say that at the outset to lay that kind of foundation. But when I looked at the space four or five years ago, I basically talked to all of the top 20 or 30 management teams in the world, private and public. And my conclusion from that process was that Dan Roberts and Tyler Page were head and shoulders above every other executive in the sector. So I placed a big bet in Q1 of 2023. Looking at their site portfolios, I thought they both traded for about a quarter to a third of what I could liquidate the value. So the liquidation value, assuming the businesses didn't work at that time I thought I was buying like when I was buying cypher. I bought 7 million shares between 1 and $2. I thought I could sell all the sites for 4 or $5 at that time. Right. So to me it was a classic value investment with a margin of safety. I of course modeled that in the future there would be a desire to use that energy for either bitcoin, mining, AI, synthetic genomics, high frequency trading. Of course all of those categories are now competing for that Compute. And I placed a big bet three and a half years ago now on those names and thankfully they've outperformed the group. Right. I think Iron from the absolute bottom it was a dollar and 2 cents December 27th or 28th of 2022. And the high price in the regular market was $75. It went as high as 80 in the pre market way back in early November. Last year. So approximately as of right now. Like a 55x, right? Cypher was more like a 35 or 40x off the lows. It traded at 39 cents way back then. So you know, that was a good time to buy. And I think, look, the last couple months if you're on X, a lot of people were trying to scare other people out of stocks and bitcoin at the lows and that's kind of a story as old as time. I try to do the opposite of those folks.
- [42:17]
B
So you know, you mentioned you talked to a bunch of management teams from some of these companies. As I see it right now, there are like five formerly public bitcoin miners that are going hybrid into AI and eventually will be just totally AI companies. And the leaders to me are Iron Cypher, Wolf Cores and Hut. I'm curious, no shares of Wolf Cores or Hut in your portfolio for Alpine Fox. Why put those to the side versus the two front runners that you have?
- [42:44]
C
Well, I've, I've owned all of those stocks on and off over the last three years, right. And at one point I had a, a fairly large Ter Wolf position. At one point CleanSpark was in my top three or four. Right. Or much earlier in the cycle. But I think, you know, I'm a professional invest investor, right? So like I already am heavily exposed to the sector. It really doesn't make sense to have your third, fourth or fifth best idea in a category. Like I need to express other ideas idiosyncratically that are going to behave differently. Like what's the point of having 17 data center developers, right? Like if you look at my 13F, you'll see a bunch of other completely different names that have nothing to do with the sector. In fact, coming into this year, if you look at the previous 13F, I had quite a large exposure to consumer staples. Like I was heavy into sodas and alcohol, right. I was heavy into the railroad industry just as some examples, right. And so I'm looking for where the best leverage is, the best asymmetry at any given point in time and along a curve that changes. I think those teams are wonderful. I would say Terawulf is right up there. Terra Wolf I think is a fast. It's in the group of the top three. Now the question is there's a separation of the pack between the folks that are doing cloud and the folks that are doing colocation. So Cypher and Terra Wolf I think are the closest peers because they've really decided to go all in in Colocation. I Think they're both quite good at it, although we need to see how well they execute. And then on the other side you got Core Weave, Nebius, Iron Crusoe and others that are really trying to go more the verticalized cloud stack. But again, even within that group, there's quite a few differences. Some of the companies don't really own all their infrastructure and some of them are really trying to go fully vertically integrated. So what I think is happening, the paradigm shift that I really think is important to call out is we've had this period over the last two years, basically since Core Weave signed the original deal with Core Scientific, where there's been this rush to see who can sign the best deal. And I remember being told two years ago in spaces that no one would ever sign a better deal than Core Weave deal with Core Scientific. And I thought that was complete lunacy at the time. And I called people out for that. One of the reasons why some people don't like me is because I can't help myself when I see something being said that's that dumb. But what's happened over time is we've seen a lot of other deals and the pricing has just gone up. So actually like now, the pricing is accelerating to the upside, the demand's actually accelerating to the upside, to the folks who have the most capacity in the future may see better monetization opportunities along the curve. And then separately from that, just because you announced a $20 billion or $10 billion deal, that doesn't mean you'll actually be able to deliver that compute successfully or that you'll keep that customer across a 10 or 15 year period, that that customer will be as profitable as you model originally.
- [45:15]
E
Right.
- [45:15]
C
If power prices go up or if the relationship with the grid changes or you have to shift out chips to a, to a more modern generation chip and that requires a change out in the cooling or some of the other infrastructure. So I just think a lot of investors are reading headlines and not focused on who's actually going to be operationally excellent. If you look back to 2023, my argument is that Iron and Cipher were better than Marathon and Riot, that they were better than Hut, that they were better than pretty much any of the legacy bitcoin miners. And I thought that was true because I looked at really closely at their operational effectiveness and talked to the teams and really studied them and that ended up being true. And it turns out that that created the most shareholder value. I think the same thing is going to happen in the AI space where it's gonna be a lot of noise at the beginning. It's sort of like a group beta experience right now where everybody trades together and everybody kind of goes up when one company signs a deal and everyone goes down when there's a negative news item like Deep Seek or something. And then over two or three years you'll see real dispersion between the folks who can actually deliver and the folks that can't. And that's not clear yet. Right. So like I think it's way too early to draw conclusions. I see very like less, like less sophisticated, more novice investors drawing conclusions by very recent price action. And they made the same mistake in 2023 by saying oh, because Marathon was up as much as Cypher that year, that marathon was just as good of a company as Cypher. I was very confident that was wrong and I'm very confident there's still mispricings in the market even right now.
- [46:35]
A
You mentioned the AI trade broadly and these within the broader. AI is carrying the stock market right now. This is a subsector, the data center subsector. How long do you think the AI trade has left? Is this a super cycle of the AI trade? Are we still talking about this three years from now or at some point do we see clear leaders pull from back and differentiate?
- [47:00]
C
Well, people ask the same question in 2002, 2003 about the Internet. And it turned out it had 20 plus years to run and it was actually the wrong question. Right. Because basically ubiquitous information became ubiquitous. Right? Like you can use the Internet instantly anywhere in the world to transfer information that didn't exist in large scale in 1995. And if you did access it was through a very sort of dated way, like a dial up modem where you block the phone connection. Your kids couldn't call their friends because they were trying to use AOL Instant messenger, right. So a lot of people called the top of the Internet in 2002 and 2005 and 2010 and 2015. But these megatrends, when something's just better, like ubiquitous information is better than a fax machine. Right. And AI is better than less intelligent intelligence.
- [47:50]
B
Right.
- [47:50]
C
If we have ubiquitous intelligence, I think that's a 50, 100 year, 200 year type of trend. And for certainly the Capex cycle is signaling at least two to three more years of pretty parabolic growth. So I don't see a topping or even a flattening of that demand till 2028 or 2029. The markets can certainly front run that. And some like sandisk could sell Off a year in front of that. Like Sandisk could go down 80% or something, right? Micron could go down, GE Vernova could go down a lot. Anybody who's serving the AI value chain can certainly go down. But those will probably be buying opportunities for approximately 20 to 30 years if it mirrors the Internet. So basically anytime people are dumb enough to be distracted by short term macro or technical factors and you have a longer term view, you close your eyes and you look out five years and you know we're going to be higher or 10 years we're going to be higher. Then you basically want to buy every single artificial external driven dip, right? So there will be interest rates will go up or wash will mess up and try to shrink the balance sheet too fast or we'll have another geopolitical entanglement like Iran. People get scared. The Algos go nuts. They try to liquidate all the stops and trigger a bunch of people who bought too many call options like they do every Friday, like today, which is just a kick in the face risk off day, which literally will be retraced probably in two days as it always is. But people will be scared by that. And so I try to focus on 10, 20, 30 years out and then I make sure my philosophy, my investment philosophy in the short term mirrors that. Right? So instead of getting sucked into will we top in AI in the next week or month or whatever, like I certainly think with the IPOs coming, like you're not going to see any sort of top of SpaceX and OpenAI and Anthropic still in front of us. Like that's just not the way markets work. When you have big inventory of basically the best private companies in history and they haven't even gone out yet. There's going to be a building of sort of fervor until that happens. We may top around that time in the mid to short term, we may top three months after that, we may top a month before whatever. Right? But I think when we look back in three to five years it won't matter whatever the next dip is, you should probably buy. That is my opinion.
- [49:48]
B
So Mike, you mentioned that there are some companies in the bitcoin mining space that are pivoting to AI that you think are overvalued. Kind of a two part question here. What are some of the companies that you think are attracting more market share than they should right now? And are there any companies that haven't signed deals yet but are actively in talks that you're excited about?
- [50:09]
C
Look, I think the Best ideas are the best ideas. Like if I, if I thought there were better ideas, then I would change the positions that I have, right? So like I still think, I think people are underestimating because, because Cypher has gone from 39 cents to 25. They think it's over. And I think it's going to go from 25 to 100 or 25 to 200. And I think iron, you know, this is not financial advice and this is, this has been my view from dollar a share. Like way back in December 2022. If you listen to my spaces, I got yelled at by kid analysts because I said I thought iron could be a 200 stock. I still think it could be a $200 stock. So I want to own the stuff where I think you have the most certainty about execution. All that said, like I do think CleanSpark is going to sign deals. Marathon's going to sign deals. Riot's going to sign deals. Hut 8 is going to sign more deals. Terra Wolf's going to sign a deal. Probably in the very short term, right? Like that's what's being sort of signaled in the market. Applied Digital is going to sign more deals. The demand is just so high that a lot of people can win on the deal signing. The question is just how much of that gets translated to real economic net profits for shareholders. And that's where I think there's going to be something lost in translation along the way. In terms of like who I'm a little bit bearish on. I've said this publicly a few times. Like I think Core we've is. Is out over its skis a bit. Like I'm not sure that they really have what's required to be a world class company at scale over time. We'll see. Certainly they'll probably be motivated if they hear these types of comments from people like that's why I don't spend a lot of time like being too bearish. I think like in the short term everybody's stock can go up, right. I just, if I needed to use a hedge, which I'm not going to hedge right now because there's too much upside left in the sector. But when I do need to hedge, right, like when I want to sort of net out some of my long exposure, I would probably start with a company like Core Weave.
- [51:53]
B
Mike, last lightning question because we got to get you out of here in a minute. Bitcoin's rallying thoughts on the rally. And do you have an end of your price target?
- [52:02]
C
I mean, I Don't really do price targets generally. I joke around on Twitter and I make calls like Babe Ruth, but those are largely for entertainment purposes. But I do think the longer term. Yeah, a little bit of that. Like you point out, people get so angry about this stuff, they don't realize that a lot of it's just a shtick that you do and that people enjoy. A lot of people who follow me know that like a lot of it is for entertainment. But I still think 300,315 area is possible before the end of the cycle. I think what we just went through is a mid cycle blip, not like any sort of end of true end of cycle. Yes, we drew down over 50%, but I think Bitcoin can draw down more than 50% and not really be at the end or beginning of a cycle. So my argument would be we're still moving off the 16 level from Q1 of 23, like January. And I think the high price for sort of the next year or two could be 300 to 500. And then in 2033-35, I'm looking for like a million potentially before that. Right. You know how bitcoin is. It's very reflexive, it's very impulsive and it hasn't behaved that way in so long, people forget. But when it goes back into a more impulsive, reflexive cycle again, it can go up quite quickly. And so I think it's kind of silly to try to predict it with any, with any exactness. But I think it's more important to just constantly be adding it or just holding. And so in the fund I express some of that. I do have bitcoin in the fund, by the way, but it's not considered a 13F asset by the SEC. So like actual spot bitcoin doesn't show up there. And then I have some call options which I use just to create a little extra convexity. Those are usually between two and call it six months out and I keep kind of rolling them forward.
- [53:38]
F
Right.
- [53:38]
C
I'm going to catch one of these next up moves. Thankfully, I've made tons of money in cipher options, so that's covered all my continued losses and IBIT calls. But that's the nature of the game. Right. It's probabilistic. If I keep betting and eventually I'm right, maybe that covers all the previous bad decisions. So I think bitcoin is just something you hold. Personally, I'm just going to hold it forever. So everything else I do is just to support my bitcoin habit.
- [54:06]
B
Mike Alfred, thank you so much for joining us, Mike. Appreciate the comments and best of luck with the rest of the year.
- [54:11]
C
Thank you, guys. Have a good one.
- [54:14]
B
Making me bullish, man. I hope he's right. Like you said though, you can't, you can't. You know, it's all for fun, but I do think that there's something to be said about like bitcoin surprises when people are expecting one thing versus the other. And everyone is so bearish.
- [54:29]
A
Everyone's so bearish and neutered.
- [54:31]
B
On the timeline right now we do seem kind of primed for people to get their faces ripped off.
- [54:36]
D
Maybe.
- [54:37]
A
Yeah. I mean, look, this is a GPU show now. That's so it's perfect. It's perfect for bitcoin to do a ripper upwards. Colin, let's talk about the Royal family.
- [54:49]
B
Yes, let's talk about the Royal Family. So we covered this story, this broke yesterday at the end of the day about Trump's holdings at the end of Q1. This is his disclosure to the US Office of Government Ethics. But Trump has increased his exposure to bitcoin and crypto, but through more, how should I say, buttoned up means than the Trump coin and the World Liberty Financial of yesteryear. So according to this disclosure, Trump holds
- [55:26]
A
Mara,
- [55:29]
B
he holds Robin Hood, he holds Sofi, and he also holds strategy as well. And there's one other here missing, Coinbase. Of course. How could we forget Coinbase?
- [55:42]
A
Does the Trump family have the inverse Mike Alfred portfolio?
- [55:49]
B
That's a good thing to point out. Man, that's hilarious. Because actually that's kind of true because in that 13F we didn't cover this because I don't really want to talk about bitcoin treasury companies. But Alpine Fox holds exposure to strive. Right?
- [56:02]
A
Yeah.
- [56:03]
B
So which is interesting, they just released their daily paid out preferred this past week. Again, we're not really covering treasury companies right now. But to get back to the Trumps, these are held in a trust controlled by the President's children. Just as a reminder, Trump is a family man.
- [56:21]
A
Family man.
- [56:22]
B
He's setting up trust for children across three baby mamas. So if you're yelling at him for doing this for personal gain, gotta do it for the kids. But anyway, just to break down some of the details here, the largest of any of the crypto stocks dated February 10th was some positions taken out in Coinbase between a hundred thousand and two hundred fifty thousand. There are two purchases of Mara in the filing valued at between fifteen thousand and fifty thousand.
- [56:52]
E
And,
- [56:54]
B
and then there's also the strategy holding and in aggregate, in Q1, there were 2,000 transactions with roughly, for roughly 220 million to 5 to 750 million. This includes Nvidia, Apple, Microsoft, Amazon, Meta, and Oracle, with those trades ranging between 5 million to 25 million individually. And just for more context, all of this buying Q1 in the run up to the Clarity act, advancing through the Banking Committee in the Senate, which just happened yesterday in a 15 to 9 vote. So thank you, Mr. President. Please pump our bags. That's all I have to say about that. You know, I actually know I have one more thing to say. If he's going to be trading anything, let it be these equities. I don't want any more of this Trump coin, Melania coin, World Liberty Financial chicanery. You know, there's something really disgusting about that. But this, you know, you could say the President shouldn't be buying equities. He shouldn't be buying stocks. Well, maybe that's.
- [58:03]
A
Now he's aligned. Now he's aligned with the American, American economy. Finally, at long last, so aligned with the American economy.
- [58:11]
B
Right as he comes off of a trade deal trip to China. Never, never forget that, everybody.
- [58:18]
A
I don't know. You see the stuff he was saying about China this morning. He was singing Xi's praises today, being very happy that the. How do I put this politely? The Chinese are buying up American land. He loves that. He thinks it's a great idea.
- [58:34]
B
GGPing is going to make America great again. You know what it kind of reminds me of, like this kind of diplomatic dance of saying nice things about a country when, like in Back Doors, you're probably cursing their leader. And this goes for the US Or China. It reminds me of the movie Mean Girls with Lindsay Lohan and Rachel McAdams. There's this scene where they're all calling each other and they're jumping between. They're merging calls and jumping between calls and they forget that they're on a merged call or one character forgets that she's talking to the person that she was trying to hang up from and then she was bad mouthing that person. Like, it's just so catty, man, and it's so duplicitous. You know, statecraft really is a kind of like Janice face business.
- [59:14]
A
So, yeah, I mean, I love, yeah, I do love the image. Somebody make one of those AI edits of like, Trump and Xi and all the, the, you know, American and Chinese oligarch titans all hopping between each other and like a Mean Girls edit. That's a freebie. There's some free, there's some free question
- [59:33]
B
is, who is Rachel McAdams and who is Lindsay Lohan? I think that probably Trump is Rachel McAdams and then, and then Z is Lindsay Lohan. Like the up and comer with China. Or maybe it's already flipped. I don't know.
- [59:47]
A
I. I don't know. I think Lindsay had a rough few years there. So.
- [59:51]
B
She is back, though, dude. She's back Hallmark movies now she's doing the Hallmark Christmas movies.
- [59:55]
A
Oh.
- [59:56]
B
Anyway, we won't, we won't touch on that.
- [59:58]
A
But yeah, I'm so sorry. I'm so sorry that you had to listen to us talking about Mean Girls
- [60:01]
B
on the show about AI but now you'll have to listen to one more story about AI that will throw up here. And then we will wrap up with our interview with Kush. Logan county panel backs 90 day data center moratorium hut 8 says delays could sink project. Now, we've covered this story a little bit on the POD in the past few months, I believe maybe just at the turn of the year or before the turn of the year. But basically Hut 8 wants to build this 500 megawatt facility in central Illinois in Logan county. And they are getting stonewalled by this local zoning committee. So most recently, the zoning committee voted three to two on Wednesday to recommend a 90 day moratorium on the data center construction. This follows a recommendation for a 60 day moratorium early. And hut 8 is on the clock here. This is why this is notable. Hut 8's Greg Irwin warned the committee that the moratorium would likely kill the project because the company's agreement with utility provider Ameren would expire. So, you know, you can file this just under the continual PR nightmare that the AI Capex boom has been for some of these businesses. We can cite other examples if we want to, in terms of all of these different, you know, these different rural communities or even urban communities, the people or the citizens in them kind of uniting against this, this, you know, data center. Creepy for sometimes I don't want to totally dismiss concerns. I'm sure sometimes there may be some legitimate concerns. But let's just look at some of the headline numbers here for what this would bring to the county. Now, Admittedly, this is HUD 8 reporting this. This is their, this is their numbers. I imagine it's probably close to accurate, though. Hut 8 said the project would create about 200 permanent jobs and more than 1500 temporary construction jobs and generate $65 million a year in property tax revenue for the local government. 65 million for a rural county in the middle of Illinois is a huge windfall. So for me, I just kind of get confused sometimes. Again, as my grandma used to say, there are three sides to every story. So I know that there's a truth somewhere in the middle that I'm missing from the detractors and the proponents, but I just don't really understand, you know, and I'm sure the two council members who voted for the zoning are probably banging their heads against the wall. 65 million is a huge number for a place like this, but still there was a protest of like 100 people recently against the data center. So again, just another headline to show that, you know, this whole data center boom is really kind of turning into a more vicious version of the anti bitcoin mining stuff that we used to see of like, you know, the kind of Greenpeace marketing pushes and astroturfing campaigns to demonize bitcoin mining. You know, there were stories in Granbury, Texas of locals really, really kind of beside themselves over the noise from one of Mara's data centers there. And so this to me seems an extension of that fight, but on a much larger scale, right? Yeah.
- [63:16]
A
I mean, bitcoin walked and stumbled so that AI hyperscalers could run in full face plant. If we're talking about like, you know, public perception, this is one of those things that the beauty of the mechanisms of capitalism and supply chain doesn't really have a good solution for because you can move mountains to get semiconductors in the right place and to build out supply chains and figure out how to move those giant transformers and substation parts through the Panama Canal because you've got money and capital flooding. But what you can't do, you can't just make people like you. And so this is one of those things where like, you know, I don't know, do you have to up the size of the check? Do you have to figure out comms issue? Like this is an industry wide plague. Now your conservative voting counties are now protesting the new.
- [64:17]
B
And that's really the interesting part about this too. Like it, this is really a cross aisle thing with certain things people, right? I mean, maybe the hundred people who are out there were like, you know, they want a hundred. You know, it might be the hundred
- [64:30]
A
Democrats, it might be the 100 Democrats in the county, but it's a. That's Illinois, baby. That's cornland.
- [64:35]
B
Yeah, but you know, you'll listen, you'll listen to These county meetings. And it's a guy walking up who like, I'm pretty positive, didn't vote for Biden or Kamala, you know, kind of worried about this thing and it really is kind of becoming a cross aisle problem. And also it's really starting to pick up steam and social media in terms of people like spotting when areas might be scouted for a data center. You dropped this in the channel. Can you unpack this?
- [65:00]
A
Yeah, so I didn't. I have not done a lot of research on what this is, but this is a tweet showing a helicopter dragging what appears to be a giant like metal loop of some kind. And the tweet says if you see a helicopter towing one of these over your neighborhood, bad news, your town is getting a data center. They're running airborne electromagnetic surveys to map groundwater in the area.
- [65:26]
B
So like some form of like lidar or something like that. Right. To ground penetrating radar to see where the watersheds are within a given area. Yeah, right. Because that is one thing that is true. These data centers require a massive amount of water. Now there's some quibbling over a closed loop versus not, but I've seen a few of these tweets actually of people taking photos or videos of this exact setup and being like, what the hell is happening here?
- [65:52]
D
Yeah.
- [65:53]
A
And this goes along with, I would say a mixed and probably ill informed public understanding of the, the amount of water and context that it's used by AI data centers. But nonetheless, these are like imagine you're on Facebook and you see this and you're like anti AI data center, you know, Facebook group and you, you know, you're retired, you voted, you know, you voted Democrat the past three times and you're angry and this is a very scary looking picture and they're coming to take your water. So like that is a, you know, I, I think this has, this gets 10 times bigger before it turns into something else. I mean we'll, we'll have to see where it goes.
- [66:38]
B
It's going to end with, you know, you know, roves of rabid anti data center folks breaking into data centers and pulling GPUs out of the racks and then destroying them office space style in the lots outside of the data center, you know, curve stomping them and hitting them with baseball bats. I, I do. To your point, Charlie, this is only going to get worse before it gets better. When you look at the capex spend for these things and I think that what you're probably going to end up having is you're probably going to end up having a real bifurcation between states that allow these data centers to be built or have fewer regulations and restrictions on them and those that, that are, you know, just like, nope, we're going to try to fully restrict them within the laws that we have on the books, you know, without stepping on anyone's constitutional rights. And you could almost argue that we're already seeing that. I mean, Texas is emerging as one of the premier data center markets in the US Hard to say whether or not they'll catch up to Virginia. In terms of percentage of percentage of the grid used for data centers, I believe Virginia is insane. It's like 25%. But in terms of probably gross megawatts used for data centers, they have to already be, if not ahead of Virginia, close. So definitely a story to keep in mind because yeah, I don't really see going back to your original thing, Charlie, of like, is this a PR problem? Is this a comms problem? Honestly, dude, I think this is a cultural problem. Like there is a large. No, I'm being serious. There's a large segment of this country that legitimately thinks degrowth is the, is the, is the best thing we need to do. Like I've had conversations with well meaning progressives who say we need to use less energy. And you can try to explain to them that there's like no civilization on earth that has quality of life improvements if they reduce their energy consumption, but then they'll just, you know, claim climate change is existential or something like that. Right. So like I legitimately think we're seeing like with many other issues in the US social, political, etc, there's a completely divergent view of reality and what priorities should be. And I don't know how you solve for that.
- [68:51]
A
Honestly, there absolutely is, I think, and I go back to a point I made on the show earlier this week, which was, which is, okay, so like look at this. I mean hude, look at this Hut 8 deal. It generate, you know, an estimated 65 million a year in property tax revenue to the county. That's great. That's huge. That's significant. That county apparently is not making much tax revenue right now as it is. This would be huge. But what we see is while these data centers do often generate significant revenue to the townships and counties that they're in, they don't typically, like, they'll bring some jobs but they don't typically enrich like people who have been there. And then I contrast this with oil and gas. Yeah, oil and gas. You have both and you have revenue in the form of like gross production tax or excise of some kind to the state often or the, you know, and so you have an extraction tax which goes to the public revenue, the public offers. But then you have the increase in value and through the royalties of minerals and property. That revenue goes to individuals and property owners and families. And you have many stories of a family who had a bunch of land in West Texas, oil came in, drilled a bunch, maybe built a pipeline, and now they're collecting near generational wealth royalty checks. I think this is a big tough nut to crack for the data center industry because what speaks louder than like a public outcry against, you know, energy use and degrowth money. So if people are getting rich, people will feel more amenable to these data centers, I think. Let's wrap this topic up and we'll go to our last guest Kush here right after a word from our sponsor, Lygos Foreign. You're muted.
- [70:59]
B
There we go. Hedge funds are getting liquidated. Is your Bitcoin safe? It's not just Bitcoin's price drying up. Lenders, OTC desk, big whales are reeling after the notorious 10, 10, 2000 or 25 liquidation events. Counterparty risk is rampant. So it's more important than ever to know who actually controls your Bitcoin. And with Lygos, that is always you. If you're working with another lending provider, do yourself a favor and switch to Lygos. They are our preferred lending provider at Blockspace. And that's because they use Bitcoin native smart contracts to make sure that you always self custody your Bitcoin that collateralizes your loan. With Lygos, you always know where your Bitcoin is. Hold your own keys. No wrapping, no bridging, no rehypothecation. Get competitive rates for as low as 10% APR and go to Lygos Finance to learn more.
- [71:46]
A
All right, let's wrap up the day and talk about compute and compute futures. Let's bring up Kush. Kush, welcome to the show.
- [71:57]
E
Hey guys, how are you? It's exciting.
- [71:59]
A
Yeah, fantastic. We're a bitcoin show now turned now we talk about GPUs. So that's just how everything we're a
- [72:07]
B
bitcoin show that identifies as an AI show. This is what I'm gonna start saying.
- [72:11]
E
It's, it's you guys Corweave, Nevius, Crusoe, iron terrible. Every other player.
- [72:17]
B
So 100. 100.
- [72:21]
A
Yeah, that's the story.
- [72:23]
B
Yeah, I. Sorry, Joe, I just have To I have to, I want to take. To kick this one off because I just have a burning question. So really interested in this news of the CME's futures and them using yalls index for the H1 hundreds. I'm just curious just to kind of get to set the stage for this. We had Luxor's Mike San Miguel on and I kind of asked this question, so I'll also ask it to you. When I think about a compute future for GPUs and for AI workloads, I kind of my head starts spinning because there are so many different forms of compute and there's so many different applications and the only analog I have for this are hash rate futures with bitcoin mining where there's only one form of compute. So it's easy to standardize. So when structuring a product like this or like your index, for instance, for the H1 hundreds, how do you narrow down what you're going to focus on so that you can give a more or less accurate picture of what the cost of compute truly is?
- [73:17]
E
Yeah, so I think the easiest way to standardize it is we think of it very similar to how oil works. So if you look at oil markets, right, every barrel, the WTI crude for example, I'm just going to take that example. There's like Brent crude, Saudi crude, there's different ones. Let's just take a look at WTI in Cushing, Oklahoma. There's an exchange that happens between oil. So oil from all parts of the Permian Basin in Texas come there and they get transacted and that's what the benchmark of WTI crude is. Now if you drill oil in Odessa or drill oil in like El Paso, obviously they're going to be different, like sulfur quantities, different gradients. There's going to have different variations in that, yet they all go through the same benchmark and they're all classified as the same type of oil, even though they're very different inherently. And I think the same thing is true about GPUs where the way you kind of do this is you have a minimum grade or above sort of specification where it's like if the quality of the GPUs meet a minimum grade, which we have a specification on our index methodology for, then that GPU can be considered a valid H100 to be priced into the contract. It's the same way financial contracts for other sort of commodities work, right? You take a look at corn or cattle, for example, cattle futures. And like you probably know, all cows are different from each other, not the same cow. And yet it's a very liquid financial market that exists for that.
- [74:39]
A
You mentioned Cushing, Oklahoma. I'm an hour outside of Cushing. I could go there and check the national strategic oil reserves if I needed to. You bring up oil in the context of the CME Group announcing they're going to do futures. You retweet a really interesting write up from I believe Caleb Shaq at Variant Fund talking about the challenge of bootstrapping one of these futures markets. And one of the challenges you have is the compute class is not fragmented enough right now. A lot of the compute kind of exists in a handful of companies control that may make a futures market difficult to build. What are your thoughts on this? Am I accurately characterizing the landscape? And like does this, does this have to. Do we have to have more diversification of like who. Who manufactures and provides compute in order for a futures market to develop? Yeah.
- [75:43]
E
So I don't know if you guys have Dylan before semi analysis, but you probably know that there's around like I think, I don't know what his number is like 2, 200, 250 Neo clouds now and there's all sorts of people getting the Neo cloud space. So in our perspective the neo cloud space is becoming more and more liquid where there's more participants. For example, the funniest one recently is Allbirds. The shoe company is now a NEO cloud. So that was pretty funny to see. And I think there's going to be a lot more of that happening given the scale of infrastructure that's needed to support all the AI demand. I think there needs to be tons of more compute launched into the world and you're going to start to see more and more fragmentation as companies that control electricity and these like regions of real estate are like hey, like I could play in this game too instead of just having the like top players play. I mean obviously I think that's still going to be a large market, but you start going down and there's going to be a lot more fragmentation. And it's very similar to how early oil markets fall. Right. You start off with a few oil miners and then it totally becomes more and more liquid and fragmented as time goes on.
- [76:45]
A
So everyone I know wants to be long compute right now. Who is hedging with these futures? Who's short compute?
- [76:53]
E
Yeah, yeah. So if you think about it, look at, see who shorts oil. Right? So it's the same in oil markets. The biggest like shorters of the natural shorters oils are the miners themselves. And it's very similar to like farmer short corn because they want to basically lock in a price that they can sell that corn or sell that oil. And the same example, like who is the farmer or the oil rig? It's the data centers. The data centers need a lock in a price that they could sell. Their future computer, they could do these through bilateral contracts, but they also want to make exposure if they're selling on demand through these various contracts. And so they get less exposure to the price. It's also the lenders to the data centers. The lenders are heavily exposed to compute prices. And so if you're lending like billions of dollars to Stargate, for example, and the price is like, if you're buying on demand, those prices just drop like you are. What are you going to do? And you've lost like millions of dollars on that actual trade. So I think it's a lot of the data centers of natural shorts and at the same time, right, it's a liquid market. So if you take any futures market, there's always at some price, there will be an equal amount of long and shorts. This let the market determine what that price is.
- [78:00]
A
So one thing I'm, I've looked, you know, I see the markets that or provides you, you know, the H100 index, there's other indices as well. What do you think is the best representation of the blended value of all compute? How do I even estimate that?
- [78:20]
E
Yeah, I think the easiest we could probably do some arbitrary combination of H100B200B3 hundreds or H2 hundreds and blend it all together, do some weighted average by estimating how many there are. Honestly, it's probably a good idea. We'll throw that up after later this weekend, so we'll see how that goes. But yeah, we could probably blend together an index with that. Just make it clear how we're doing the percentage weighting.
- [78:46]
A
So one deal, which I thought was really interesting this week, and I'm going to ask you about the Anthropic XAI deal this past week. Anthropic's been compute constrained. People are complaining that Claude's not performing as well, but they just signed a big deal with XAI or for the Colossus 1 data center, which kind of surprised me because I Aren't they competitors? What's your take on this deal this past week?
- [79:16]
E
It's funny to see. I think a lot of it comes down to just like the Elon Sam beef that's going on. And you see like Elon starting to play with. It's like, the enemy of your enemy is my friend. Sort of dynamic that's going on. You look at, like, OpenAI. OpenAI probably is the largest compute sort of capacity that they've confirmed. Sam has done an incredible job of gobbling up every single compute provider in the world and making them an OpenAI. And I think Anthropic has done a poor job so far, honestly, about how they get compute. And so this is like a huge deal for Anthropic now that they get access to the same similar amount of compute that OpenAI is getting now, and they can certainly push that frontier and get more access.
- [79:58]
B
So you think Elon's doing this to own Sam Altman?
- [80:04]
E
Look, my opinions, there's different reasons. I mean, economically too, it makes sense, right? XAI had underutilization. I think it was reported on these GPUs. And so economically, it also makes sense. But if you just think about it, if this truly is a race to AGI and whoever gets there first, what do you think, like, Elon is, like, who do you think he wants to win in the race? Just look at the incentives on each side.
- [80:29]
A
We had. I mean, we had. I don't know if you saw the story. Some guy cracked his bitcoin wallet with. With Claude earlier this week. We had him on, and he. He says he's gonna name his son Dario. So I think the. The lines are being drawn. Okay, So I read that this week, Larry Fink, BlackRock said that he's extremely bullish on the emergence of COMPUTE futures. And you talk about this a lot. You tweet about this a lot. How big is the compute futures market in 10 to 20 years? Is it the biggest market on earth One time?
- [81:14]
E
I mean, our opinion, obviously the business is built off of that thesis, but all, I mean, the easiest way to explain is like, today you look at oil is in every single good that we use. The mics that you're using, the covers on them are built from oil. The clothes that you're wearing are built from oil products. Every single tool that we use today has some variation of oil within it. And that's come across the span of 150, 200 years where we slowly started integrating oil into everything that we do, everything that we use as a society. And I think it's very similar to how compute works right? Today, every single enterprise might not be powered by all of the GPUs in the world, but in the future, they will be the economic input for the 21st, 22nd century. Even is going to be compute, where most of the money that's to be made and economic progress is a result of being powered by compute, whether that's GPUs or any sort of other formula take. I think the high performance compute is going to be the strongest power.
- [82:10]
A
For a former bitcoin show, we're still a bitcoin show. For a former bitcoin show we talk a lot about how the international economics and systems work and the petrodollar and my family, I'm near Cushing, Oklahoma, family tied to oil and gas. I see this trend, I see this personally, this trend that now oil and gas is almost going to be like the bridge to the compute industry. I'm curious, opine a bit for me on the transition of oil and gas being the large geopolitical driver to maybe compute in the era of nation states and compute.
- [82:56]
E
Yeah, I think the transition is pretty natural, right? The easiest example, and it's not a great one, but if you look at the war in Iran today, they bomb what oil fields and AWS data centers, those are like the economically strategic targets now for a regime that's looking to take down its enemy. And if you think about it that way, that means that in strategic importance, a data center is worth just as much as an oil field and they're both nationally strategic. And you would have said that 10 years ago. 10 years ago everyone were like, okay, data center, whatever, we can just go build another one. There's infinite amount of capacity, etc. Etc. But today, even in geopolitics, we look at how the world's moving. China exports of H200 H20 are a very contested topic within the US government. Is it safe for us to give foreign nations access to our high performance compute? Where is this compute going to be located? Is the US going to have enough power to power this and lead a revolution? There's so many questions that arise from just this, like how does this look? I think it becomes, and you just start looking at it like, okay, this is a very strategic sort of commodity where you're trading geopolitical resources based off of this one chip or this one resource with countries and trade deals are being negotiated off of literally a computer chip that if you ask this probably like 10 years ago, yeah, all the gamers use it to play League of Legends or whatever other game that they use. And now it's literally like a tool that they use for bargaining.
- [84:22]
A
Yeah, well, I'm one year sober from League of Legends and now I can use that. Those compute cycles to maybe not mind Ethereum anymore, but now I can do some local cycles. Kush, thank you so much for coming on the show. Fantastic. I'd love to have you back because this is going to be a big story over the next year. Thank you so much.
- [84:42]
E
Sounds good. Thank you guys. Appreciate it.
- [84:47]
A
That wraps up our livestream for the day. Make sure to like and subscribe. Go to Blockspace Media for all of your AI, hpc, emerging tech, and sometimes Bitcoin news. And if you made it to the end and you're listening on CoinDesk feeds, make sure to now follow us on our proprietary Blockspace feeds. You can search Blockspace on any podcast provider and hit up our YouTube channel. Otherwise, thank you so much. I'm Charlie. This is Colin. That's Colin. I will see you next week.
- [85:26]
B
Have a good weekend, everyone.