← Blockspace: AI & Bitcoin
Blockspace: AI & Bitcoin
NEWS: $46m Heist Perp Gets Nabbed & Kraken Gets Fed Account
01:08:37
A Zoomer arrested for stealing $46M from the US Marshals, Kraken makes history with a Fed Master Account, and IREN builds to 150,000 GPUs.
Loading summary
Transcript137 lines
- [00:00]
A
So when I saw this screenshot, I couldn't believe that this was this kid because he's actually blonde. So this is a disguise. And he's got this like quasi soul patch here. I don't know if that's real or if that's like peach fuzz.
- [00:12]
B
It's more conspicuous than not doing anything. Like, I was in a restaurant. We're looking around like, who's probably the guy fleeing the FBI. Guy wearing bright red pants, sandals, and looks like. Looks like a Bond henchman.
- [00:26]
A
He's walking into these, like, five star resorts on the island of St. Martin. Where did he get all this money? Well, he stol. Stole it from the US Marshal Service. So John Dagati, also Dagita, also known as Lick, I guess that's one of his pseudonyms.
- [00:39]
B
Screen name, his telegram handle.
- [00:41]
A
Yeah. Stole 40 million plus from US government seizure addresses because his dad had a company called Command Services and Support, Inc. That was managing some of the funds from the U.S. marshal Service. Now, for those of you who don't know, U.S. marshal Service will often take control of cryptocurrencies that have been seized in criminal investigations or raids. And they contracted this company, Japanese John de Gida's father's company, to manage these coins. So then that raises the question, how did John get access to these? Is his dad in on this? Or did. Is his dad just really bad at managing coins? Apparently they paid this guy $27 million. The US government paid this guy allegedly $27 million to manage his money, and then his son ends up stealing $46 million worth of it. That's crazy, man. A zoomer who looks like he's about old enough to be raking your leaves. Just got arrested in the Caribbean for stealing $46 million from the U.S. marshal Services. That's our lead story today, followed by Kraken getting the first Fed master account for a cryptocurrency company. That's bullish, but it's not as bullish as it could be. Plus, we have iron expanding its GPU fleet to 150,000 and. And opening the largest at the market equity offering of a bitcoin miner yet. Those are our stories today. And for supplementary content, we've got the usual hash rate index update. And we have two banger guest segments from Chris Johansson of Ion Stream and Khan Firani of Luxor to talk first about the GPU rental market and then Khan to jump in to talk about what we saw in February during one of the most tumultuous periods of for bitcoin's hash rate yet. You're listening to Block Space Live and we are kicking it off.
- [02:37]
B
Hey, Charlie here. Guess what? We just announced our next bitcoin technical conference up next.
- [02:44]
A
That's right, y'.
- [02:45]
C
All.
- [02:45]
A
Op Next is back for 2026. We're running it back after a successful event at Strategies HQ in Tyson's Virginia last year. And this year we are bringing it to the the Big Apple. At the iconic Time center in Midtown
- [02:57]
B
Manhattan, we're hosting the big names and projects that you recognize like Robin Linus
- [03:02]
A
of BitVM, Nick Jonas of Blockstream, Antoine Ponceau of Chaincode Labs and Calle of Bitchat will also be present.
- [03:09]
B
And this isn't just for the devs. We have institutions talking with the developers. That's what UpNext is all about. We have Robert Michnick, head of digital assets for BlackRock in the building. We've got folks from mining pools, investor funds, bitcoin startups and other groups.
- [03:26]
A
With a ticket, of course, you'll get access to all the high signal programming and networking you could want. You'll also get coffee, catered lunch and access to the afterparty at Pub Key. If you want to go vip, you'll also get access to the speaker dinner following the event and an investor brunch on Friday.
- [03:43]
B
Tickets are capped at 300 and early bird tickets are already sold out. If you want to save yourself a spot, go to op next.dev that is op n e x t.dev use code podcast to save 20% off a GA ticket to the event.
- [03:59]
A
Ticket prices go up every few weeks, so don't wait. Y' all lock in that ticket today.
- [04:03]
B
We'll see you April 16th at the time center in New York City. This show is brought to you by CleanSpark. Make sure to go check them out later. And Colin, we we go live every Friday at noon now, so make sure you put that little thing on the Google Calendar and hit that notification button if you're on Twitter or YouTube so you get the push notification. Wow, Zoomer. Raking my leaves or stealing my money. Do we always lead? We'll get to that in a second. We'll get to that in a second.
- [04:35]
A
It's the probably the wildest story I've seen in a while, but we'll kick it off first with the the hash rate index update. And Charlie, you've already got it up there. So yeah, we'll go ahead and dive in and we're going to have some really good analysis and data from KH here in, in, in about 20, 30 minutes. Because February was probably it for me at least like the most wild month I can remember for bitcoin's hash rate for hash price and all of these different metrics and, and we're kind of coming out of this, you know, the depths of despair here. Although bitcoin just dropped back below 70K. So you know, whatever hopium anyone had for a renewed bull rally, there's at least a little bit of dimming there. But hash price is looking a little bit better for it. I mean it's still not Great. We're at $29.87 per PETA hash per day. That is up from an all time low around 27, which was set at the end of February. And the story as ever for this year is Bitcoin's hash rate is just not really going anywhere, man. We entered 2025 or we, sorry, we entered 2026. Excuse me. Having just come down off a all time high for Bitcoin's hash rate set in October when Bitcoin was ripping, we got just to under 1,200 exahashes. That's 1.2 zeta hash. And then as we see here, it just absolutely gets throttled at the end of January during winter storm Fern again more from that on Con in a little bit and then, and then it recovers here. But it's basically been, you know, if you zoom out and you wanted to flatten this out, Bitcoin's hash rate has basically been flat since September or October of last year.
- [06:17]
B
I mean the golden age of mining was, you know, led up till maybe post China 21, 22. And then it just got super competitive. There has been no consistent meta in mining for years now. It shifts and it morphs and it's turning into AI now.
- [06:35]
A
Yeah, and that's the biggest headwind against hash rate growth right now is all these big public miners are shelving their Asics, they're selling them and they're swapping them for GPUs. And also just the other big headwind obviously is the fact that there's, this is not profitable right now. And again, Khan will have some really good thoughts on this about what rigs are actually profitable currently. But, but it's, it's been brutal. So we'll, we'll shelve this for now because we'll be bringing in, have this data too much because we'll be bringing in a lot of the difficulty data and other things when Con comes on. But you know, I do want to
- [07:08]
B
point out that now we just had difficulty adjustment yesterday or in the last 24 hours. So we're just at the start. We're not even 10% of the way into this particular difficulty period. But we are targeting an upwards difficulty adjustment of 7.5%, which is pretty big in the overall context of difficulty adjustments. I would expect this to narrow as
- [07:30]
A
we get halfway along, potentially even go negative. I mean, it's basically just been this tug of war. We've had massive difficulty adjustments downward and also massive ones upward as a result of this winter storm disruption and Bitcoin's hash price being just crushed. Let's go ahead and hop on over to the first story. And Charlie, I'm going to share the screen here.
- [07:49]
B
Zoomers still stealing your money. Get used to it.
- [07:52]
A
This is probably just the zoomers stealing your money, disguising themselves like Slavic drug dealers. I mean, so when I saw this screenshot, I couldn't believe that this was this kid because he's actually blonde. So this is a disguise. And he's got this, like, quasi soul patch here. I don't know if that's real or if that's like peach fuzz.
- [08:11]
B
It's more conspicuous than not doing anything. Like, if I were in a restaurant. We're looking around like, who's probably the guy fleeing the FBI guy wearing bright red pants, sandals, and looks like. It looks like a, you know, like a Bond henchman. Oh, my God.
- [08:26]
A
Yeah, serious. Honestly, it's like this guy is like fresh off the boat from freaking, like, Slovakia or something, you know? What, what is he. He's running in. He's walking into these, like, five star resorts on the island of St. Martin. Where did he get all this money? Well, he stole it from the US Marshal Service. So this is the FBI director, Cash Patel, tweeting, quote, last night, John Ducati, or, sorry, Dagita, a US Like a
- [08:51]
B
Italian mob boss now.
- [08:53]
A
Yeah, he does. Seriously, a US government contractor who allegedly stole more than 46 million in cryptocurrency from the US Marshal Services, was arrested on the island of St Martin by the French gender, Murray's Supreme. A premier elite tactical unit in a joint operation with FBI. Yes, you got to get the elite tactical unit out here so that this zoomer doesn't gun down any FBI officials. But, so this is a wild story because first of all, some background here. And this is all sourced from Zach XBT crypto, Twitter's resident sleuth. This man must never sleep or just have, you know, foresight and clairvoyance because if there is a scam that goes on or there's a hack, this dude's always on it. And he often traces the funds to the individuals who stole them. And his sleuthing here directly led to this arrest. And a few kind of key points before we kick this off. John d', Gotti, also Dagita, also known as Lick. I guess that's one of his pseudonyms.
- [09:58]
B
Screen name, his telegram handle.
- [10:01]
A
Yeah. Stole 40 million plus from US government Caesar addresses because his dad had a company called Command Services and Support, Inc. That was managing some of the funds from the U.S. marshal Service. Now, for those of you who don't know, the U.S. marshals Service will often take control of cryptocurrencies that have been seized in criminal investigations or raids. So, you know, bitcoin that ended up being seized from the Silk Road or from like the Bitfinex hack, for instance. That's one thing that people speculated here is that some of these funds were from the Bitfinex hack that ends up with the U.S. marshal Service. And they contracted this company, John de Gida's father's company, to manage these coins. So then that raises the question, how did John get access to these? Is his dad in on this? Or did. Is his dad just really bad at managing coins? Maybe a multi sig setup would be a good idea in the future for something like this and not letting your chronically online son run away with a bunch of treasures with $46 million in cash. The last thing I want to note, and I didn't verify this because I didn't have time to go on the website, but Penny ether in one of the reply threads to this, found the contract.
- [11:13]
C
The.
- [11:14]
A
Or found the details of the contract. Apparently they paid this guy $27 million. The US government paid this guy allegedly $27 million to manage this money, and then his son ends up stealing $46 million worth of it. That's crazy, man.
- [11:29]
C
How.
- [11:29]
A
How do you. I would do this for. For like a hundred thousand dollars, and I would make sure that my son, who I don't have, doesn't steal it, you know?
- [11:39]
B
Well, okay, I know, like, you know, when it comes to family stuff, this gets really complicated because, you know, how do you. How do you prevent your. Your son from installing a keylogger on. On your computer? Adam, I'm just spitballing different things that could happen here. So there are some highlights from this, which, beyond just the, The. The classic picture of him being led away in cuffs, like part of the way as you got it right here. Part of the way that Zach XBT traced like the funds to this guy, to this kid is because not only do you. It's not good enough just to steal $46 million from, you know, the government and by way of your father. You have to then flex it. And this is like the classic online crypto heist guy, he wants to go band for band, which is when people flex and demonstrate how much money they have.
- [12:33]
A
I'm a total boomer. I had to look that up. I had no idea what that meant.
- [12:37]
B
Maybe this more time. Spend more time on Roblox, Colin. Because this is what the kids are doing these days. And basically they show I have this much money. And then you have to reply, I have this much money. And so when you have $46 million in crypto assets, you can go very band for band. And in this case, someone recorded him doing this and showing his addresses. And then a lot. There's a lot more. Like he would. He would taunt Zach xbt, this investigator by sending him de minimis amounts of cryptocurrency to like dust attack as well. To taunt him.
- [13:12]
D
Yeah.
- [13:12]
B
What do you mean? Was that.
- [13:13]
A
Do you think that was him taunting him, or do you think that was him trying to be like, oh, if you investigate me, I'm going to send you dust so that now you're implicated in this and I can say that you were part of this, you stole some of this stuff.
- [13:24]
B
Yeah. You know, I don't think there's a whole lot of real tactical thought here, but yeah. So this story is. I expect a Coffeezilla episode imminently about this. You know, all the people who, like, only hear about the crypto scams, this is one of those stories, like, you know, that you're going to see it on Facebook in four weeks. Wild news. I don't know if there's anything. There's a bunch of details, but I don't know.
- [13:51]
A
Yeah, I would. I have all the wallets pulled up, but it's not worth. Worth going into. And. And I just one discrepancy. And I'm not really sure if this is showing that they just seized part of it, but, you know, Zach XPT originally said that the wallets that are tied to this kid have upwards of 90 million stolen, but the FBI is only reporting 46 million. That's the only discrepancy I found so far. And there might be more. Maybe Zach made an error at first, but it just goes to show, you know, going back to what you're Saying about the band for band thing is he was doing that in a private telegram, but then someone, this, this guy Dryton or some other anonymous crypto scammer streamed it, I guess, or just posted the recording on his Twitter. And so it's like, even if you have, you know, bad opsec, your OPSEC can be made even worse by morons posting about what you're moronically posting about. Even if it's a private group chat. Last thing I'll say, the United States government and military was able to go into Venezuela and capture a head of state. If you think you're going to be able to get away with stealing $46 million and you know, absconding away to a Caribbean island to live out your days in luxury, you're absolutely delusional. We'll leave that there though.
- [15:07]
B
But not if you dye your hair black and grow a grow tea. So maybe that can, that can foil the five eyes of intelligence. Okay, let's move on to the next one. We have some real news.
- [15:18]
A
Yeah, this is, yeah, with the real news. We led with page six with the pulp, and now we have the front page for business. This really is, you know, one of the biggest news items for Bitcoin institutional adoption in recent memory or ever really. That's Kraken receives a Federal Reserve master account in cryptocurrency banking first. So these master accounts are essentially, they allow companies to have accounts with the Federal Reserve. They are typically reserved for institutional banks. Right, Institutional firms. And Kraken is the first ever crypto company to get one of these via one of its subsidiaries, Kraken Financial, which is a chartered bank in Wyoming. Now what this means for Kraken and what it doesn't mean, because this is important, this means that they now have access to the plumbing and financial infrastructure that the world's largest banks and America's largest banks like JP Morgan, et cetera, have access to. So they can now send payments through Fedwire. They no longer have to rely on an intermediary bank to settle those payments through the, the traditional financial system. So if we look at the Bitcoin ETF as being Bitcoin's big institutional adoption moment for institutions ability to purchase this asset through a wrapper like an etf. This is a watershed moment for crypto companies because now the first time we've had one let in to the secret circle of institutional plumbing within the US what it doesn't mean, though, if you saw this headline, you probably saw it labeled as a skinny Fed federal master account. And what that means is that Kraken actually can't earn interest on its deposits with the Fed like other banks do. So this is an important distinction because the way I kind of read this is like you can come to the party, but you have to come through the back door. And by the way, when you get your punch, you have to sit away from everyone else. You're not like quite, you're like almost at the cool kids table. You can look at the cool kids table, but you don't actually have a seat there. And this is maybe not correct to view it this way because obviously Kraken's not going to be depositing stablecoins with the Fed. But I couldn't help but look at this almost in dialogue with the Clarity act news. So the Clarity act is being stalled in Congress currently. This is a kind of foundational piece of legislation for bitcoin and crypto market structure. And one of the key provisions in it is this ability for companies that hold stable coins to produce yield for their clients. So it's like if you have stable coins that you hold like Coinbase or Kraken or whatever, then those stable coins can then produce yield and they can pass on that interest to the people who have deposited, deposited them. When those institutions, the banks don't want this because that kind of drains liquidity from their coffers.
- [18:13]
C
Right.
- [18:14]
A
And so when I see this Kraken being kept out of the actual, you know, yield bearing and interest bearing aspect of a Fed master account, the Clarity act just popped in my mind because they're not letting Kraken do this on, in the buttoned up way through the institutions. They also don't want crypto companies to provide interest on stablecoin deposits. So I don't know if they're related, but to me it's almost again like they don't want to let the crypto companies in too much. They don't want to give them too much power here because the banks see them as a threat to their way of doing business.
- [18:46]
B
Yeah. So a couple takes Jeff park, who's
- [18:51]
A
with Pro Cap, formerly Bitwise.
- [18:54]
B
Yeah, formerly Bitwise. He's with Anthony Poppy, on and out. He says, quote, people are not realizing yet, but this is by far the biggest win in crypto in, in far, is by far the biggest win the crypto industry ideological believers have, have ever had, I would argue, since the birth of Bitcoin. So he thinks this like, there's, like there's a bitcoin starts and then Kraken gets a Fed Master account, like as the two iconic points in bitcoin's adoption cycle will let him have that take. I think that would pro. I would probably narrow it to like that is this cycle's biggest milestone. Maybe. I don't really know how the Fed master accounts work. And there were some takes which are like, okay, well aren't we supposed to like disrupt the traditional finances? And I'd say, yeah, but banks have to figure out how to use this and they're locked out. And this basically is the. As you said, it's the back door with the party, but is a door into the park.
- [19:52]
A
Yeah, it is a door. I'm glad you brought up Jeff Parks tweet here. And this idea that this is kind of like Bitcoin's inception moment for the big time kind of reminds me of like the way US history is taught. It's like, oh, and then the settlers arrived at Jamestown in 1607 and then the American Revolution happened in 1776.
- [20:12]
B
Yeah, Bing Bang boom. It. Yeah. Then we just. They became a country.
- [20:16]
A
But, but you know, if, if Bitcoin does end up going on this adoption curve that everyone hopes it does, maybe this will be kind of looked at that. I mean I think there are, I think the ETFs still is. Will probably be the, like the watershed moment for most people in terms of. Because it's the big flashy one. This is kind of the boring procedural side of banking that most people don't understand and won't appreciate the complexity for. So the Bitcoin ETF will still be, I think like kind of the premier news event for.
- [20:42]
C
For.
- [20:43]
A
For that institutional adoption narrative. But congrats to Kraken. I think, you know, just one last take on this. We should expect Coinbase at least maybe Gemini to follow if. If this all pans out well for the industry.
- [20:57]
B
Yeah, I am kind of surprised it wasn't Coinbase because they've co. Coinbase and Anchorage have gotten, you know, have taken. Have tag teamed being the first and the premier like companies to get the various milestones of institutional option. Kraken gets to be the winner this time around. And so shout out Kraken. You gotta love this.
- [21:17]
A
You love Shout out one of the OGs. 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, four states. This is our proof of work and
- [21:39]
C
we are setting the standard for what's next.
- [21:42]
A
Learn more about the intersection of energy
- [21:45]
B
and bitcoin@cleanspark.com let's talk computers, specifically Neo clouds. We have our first guest of the show, Chris. Welcome to my show, Chris.
- [22:03]
C
Hey, pleasure to meet or well, pleasure to talk to you guys. I've met you guys a few times over at this point.
- [22:09]
A
So yeah, if you could give us a quick high level overview and just for more of a formal introduction. Chris Johannesson is the go to market lead at Ionstream. Can you just give our listeners a brief rundown of what Ionstream does?
- [22:24]
C
Yeah, so we specialize in bare metal server rentals across the GPU as a service landscape. So it's really a simple business. I always kind of like to equate it to car leasing. We buy equipment and then we lease it out. If you want it for a day, it's really expensive. If you want for a month, it's less. If you want it for a year, it's even less expensive. So it's really a simple business. At its basebone, a lot of the complicated jargon and pricing volatility comes in through the plumbing and that's kind of where you see this industry kind of forming a very similar path to bitcoin mining in terms of being a wild west at the moment at least compared to separate compute like different commoditized compute spaces like let's say storage or CPU on that side.
- [23:06]
B
And you come from the bitcoin mining world so you've got a long history, you know, plugging in ASICS over the years and now you're in the neocloud business.
- [23:17]
C
Yeah, so I was. It's been a long journey on that side. I was with Poolin previously. I directed business development for their US arm and I got to oversee their 100 megawatts of infrastructure in West Texas. I quit that probably about a year and a half ago at this point, went to impossible cloud, built a cryptocurrency token for as a filecoin competitor and then now we're over at Ionstream. So crypto is really bread and butter for me in the AI space is more so the new endeavor that I've been chasing over the past year or so.
- [23:47]
A
You said that this is mirroring the wild west days of cryptocurrency mining. And you made a point that other folks who have pivoted or expanded into AI from the bitcoin mining world, they've made this point, also made the point that the bitcoin miners somehow are like more well positioned than the traditional data center companies because those more traditional forms of compute like storage, everything's Kind of buttoned up with that at this point. I mean, we've had a couple decades to figure that out. In what ways does the current AI boom mirror what you saw in the bitcoin and cryptocurrency mining boom over the last few years?
- [24:19]
C
Yeah, so it's like if you really, I like to equate this back to like, like 2020 in Bitcoin mining when China banned bitcoin mining and everyone kind of had this rush into the US and almost the mentality of like, if you build it, they'll come. Now the reason we say that in AI space is because you actually need enterprise demand. With bitcoin mining, if you build your data center, you can, you can mine. So it's very much the same mentality on that side in the, the beginning phases of these booms. And then it's like you kind of start to see as time goes on, people get weeded out, they have poor financial controls, operational controls, they might get rolled up into another larger enterprise for whatever reason or flat out just go bankrupt. And that's kind of what we're seeing in the GPU as a service space. It gets underpinned by most of the known operators, are the most professional in the room. So you won't see that with your land is Crusoe core weaves, but your random Neo cloud that no one's heard of, that has no experience operating data centers, you'll definitely see those issues.
- [25:21]
B
You mentioned that these AI firms, they are extremely compute hungry and they're starting to source from Neo clouds, which is because they're so desperate for GPUs. Why is that?
- [25:35]
C
Yeah, so I actually wouldn't say it's from a point of desperation, but it's actually more of an inflection pointing cost. Think about it this way, like if I were to spin up one B200 node, like and I have enough R and D funds to really cover it, I can AWS or Crusoe or core weave or any of these players are actually more attractive because I can set up an on demand instance, spin it up, spin it down for my R and D versus going to some of these Neo clouds. They more so expect production workloads so you'll see longer term contracts. A lot of them don't do on demand. It's like really the way those contracts work is you get a unit, you can trial it for 12 to 24 hours. If it works for you, you're signing the contract for a month, three twelve, two years, et cetera. So it really comes back to that inflection point in cost of compute. Think about it. It's if you're now running, let's say 32 servers of B2 hundreds, right, that the difference in cost will be, let's say it's $6 at AWS and 350 out of Neo Cloud. You're saving what, 80% on your cost of and at that 350 mark it's around like 22k a month in MRR at, at that point for a single server. So like with AWS you'd be spending closer to 3540. And then you multiply that by 32 servers and that's kind of where you start seeing that you can get large hordes of compute for significantly cheaper from NEO clouds versus like your enterprise providers and AI teams. See that? I think the only bottleneck that sits on the NEO cloud side is either access to supply, power or space. And it seems like there's a lot of folks in the industry that have two of the three, but not three of the three figures.
- [27:20]
A
We'll get to those capacity constraints in a second as it relates to some of these older generation models like the H100. But I wanted to zoom in on something you just said about these NEO clouds. If I heard correctly, the NEO clouds offer them more flexible terms where you don't actually have to come and say I'm going to buy this much compute for this contract for a long term contract. I'm curious what that means for dynamics of if you're running inference versus training because to me it almost seems like these AI companies like OpenAI are anthropic for the inference side of things. You'd almost need more flexibility, right, because you don't know how many people are going to be using your model at a given time and you don't know how much demand you're going to have from users outside of the models that you've run for your adoption. Does that make sense? I mean, does this fit into do these NEO clouds help with that problem or am I just making that up?
- [28:10]
C
So it's actually, it's actually inverse. So Neo clouds, they have much harder fixed term contracts where let's say you're hyperscalers and I don't know what you would label the step under that, but let's just call them like mid market hyperscalers because they're pretty close to it at this point and they're much more flexible with R and like let's say training and burst or buying compute and Burst say you need it for a very specific time frame. They're much more flexible on that side. And that's also another reason why you see a lot of the biggest foundation model labs go to them is because they actually have the scale to support their needs. Whereas if you take a step down and you look at a lot of inferencing as a service, whether it's like fireworks, AI at base 10 modal on that side, those guys, they have retail facing products like they're inferencing as a service where you can come in, spin up, spin down, let's say serverless inference on that side. But they also have a large book of consultant clients and clients that they're integrating workflows into on their backend. And that kind of like is what the direction of let's say Fortune 2000 companies are going is. They're going to the top inferencing labs and asking them like hey, how do I actually work this out or figure this out in my workflows? Because it just kind of rolls back to the whole question of like how do we use AI? And when you when multiply the amount of people under one system that just becomes a much more complicated question. So a lot of these inferencing as a service labs, they're the ones that are really chasing neoclouds because I think they're the ones that have the, they feel that inflection in price the most because I mean if you look at Google for example, I mean they have their whole TPU production under arm. Gemini's. Listen, don't, I can't quote or I can't speak for facts on this, but Gemini is absolutely subsidized by the search engine. With meta it's the same thing. Olama is absolutely subsidized by their social media platform. OpenAI is kind of the 800 pound gorilla in the room spending tons of money where people really can't pinpoint an end game to it. So that's kind of the way like the ecosystem is laid out at the moment. The biggest guys definitely want to work with the enterprises and the hyperscalers just from a raw scale perspective, but everybody kind of below that where they're working in between, let's say 128 to a thousand nodes. That's where neoclouds really come in. They specialize, they get better pricing and it's much more of a win win situation because all the, it's a bit more of a down to earth conversation when it comes to profit margins. Pricing and attacking how you structure these deals.
- [30:45]
B
You know, you mentioned it's a common model, might be trial a cluster, then buy a longer contract. Is that how clients are operating? I'm kind of curious about like how they, how they scope these deals.
- [31:01]
C
Yeah, I mean, so this one kind of rolls back to what I said initially where you have a lot of operators from different backgrounds. It's actually one of the things we specialize in at Ion Stream is that we cumulatively have like over 70 years of data center building experience under, under our hood. So like when it actually comes to building enterprise systems, that's what we specialize in. Now the issue is, is when you get like, let's say a bunch of Wall street guys or finance guys that come in and say like, oh, I want to buy 32 B2 hundreds and rent them out to the market, they're going to start looking at numbers on things. They're going to say like a dual 100 gigabit uplink, that's, that's a lot of money. Triple redundancy on our power, that's a lot of money. And then that's when it turns around and you start getting, let's say, lackluster clusters. And that's why the trial is there. Because what if I gave you a server and it had one 1 gigabit per second dedicated and you actually can't ingress or egress any data off of it. So that's why people trial and test, because the industry, it's very much like bitcoin mining on the side.
- [32:03]
A
Again.
- [32:03]
C
One of my biggest criticisms of the space at the time was there was a lack of standardization that made it hard to find finance builds. You're kind of seeing something similar but different because there's an expectation to be in an Uptime Institute certified data center, you're expected to have dual uplinks on your network, redundant power, and of course a cooling footprint that supports it. Because the other reality is imagine your cooling footprint is let's say 20% under where it's supposed to be. You're just going to be overheating machines all the time. They're going to be going down. It's no different to any other hardware class on that side.
- [32:39]
A
So for a closing question here, Chris, we have this post from Matt Siegel of Vaneck here. I got it up here, Charlie, you got it. And this shows rental prices for H1 hundreds. So kind of a two part question with this. What's driving this? Is it the capacity constraints you were discussing earlier and 2 how is it that H1 hundreds are still viable for some of these new AI workloads, considering this is a model that was launched in 2022?
- [33:08]
C
Yeah. So it's absolutely driven by capacity constraints. I mean, the H100, it's still a great unit, it's a workhorse. And you also have to think like, when we're looking at model specs and we say Kimik 2.5 comes out with a trillion parameter context window. Yeah. That can be ran on a single B200 server. Right. And that's like the attractive side to that is you don't need to start interconnecting servers, but as time goes on, there's a lot of open source groups like sglang Vllm that actually optimize from the chip level and the kernel level to make these units communicate better. So it's very different from Bitcoin mining, where it's like you have your asic, you plug them into the wall and those are the specs you can really build. Let's say you can use InfiniBand to interconnect these, you can build external storage clusters, you could tag on external RAM to them if needed. So you can really make these things work, work the way you need them to. Now, obviously, the more you change away from OEM spec, there's concerns over what could break, what could not break. But a lot of this is actually happening from the software level proportionately. So that's where you're seeing these one hundreds just get bid back up, because that's really what's available. And like rolling back to the if you build it, they will come statement. The H1 hundreds were that phase of the market where everyone just bought out as many servers as possible. And they just have this mentality of, all right, I have the servers, they're there, AI companies will come and use them. So it's interesting because the H100 market got burned a long time ago because of the oversupply on them. And then you had the H2 hundreds drop and then the Blackwell chip series drop. But I mean, in the end of the day there's still three year old units. So as long as you kept them in good condition, you treated them well. They still should be working towards their nameplate capacity.
- [34:58]
B
So am I hearing that H1 hundreds are the S9 of the GPU world?
- [35:02]
C
Yeah, it's actually a good analogy in a sense, but I think you're going to see this with future models rolling forward. It's like you're going to see the H2 hundreds, they're going to have a long life. Same with the B2 hundreds. Like, even when Nvidia drops their next chip, Ruben, I still expect a lot of these to be useful, at least until the supply constraints lift themselves. And they eventually will. And that's just a manufacturing cycle for all of our ram, ssd, CPU suppliers on that side.
- [35:30]
A
Chris, thank you so much for hopping on, man. We'll have to get you and maybe someone else from the Ion Stream team as well on for a longer pod on the GPU market dynamics because obviously we're covering it a lot more. I find it super interesting because it is, like you said, kind of this wild west, a brave new world of everyone trying to build the car as they drive it. So thanks for shedding some light on it. And before you hop off, where can people find out more about Ion Stream and what y' all are doing?
- [35:56]
C
Yeah, so you guys can go check out our website at ionstream AI or find us on LinkedIn, Twitter. Although we're not very active on Twitter at the moment. That's planned to be changed. And you can always just reach out directly to me if you ever have any sales needs of it. I'm available on LinkedIn under Christopher Johannesson. I was available on Twitter this week until I got a random suspension out of nowhere.
- [36:16]
B
You're gpuing too close to the sun, dude.
- [36:19]
C
Yeah, me and my three followers, right?
- [36:20]
A
Yeah, it's like getting delisted from the NASDAQ for being under a dollar.
- [36:24]
C
Yeah, exactly. Like I saw the suspension and I was just like, okay, it's fine. I guess.
- [36:29]
B
Maybe later. Chris, thanks so much for coming on.
- [36:33]
C
All right, have a good one, guys. Really appreciate having me.
- [36:36]
B
All right, from GPUs, we got to go back to our roots. We got to talk hash rate call.
- [36:40]
A
Gotta go back to the asics.
- [36:42]
B
We gotta go back from a.
- [36:44]
A
We went from a capacity constrained industry with no difficulty adjustment to also a capacity constrained industry because no one's running ASICS anymore, but with brutal economics.
- [36:54]
B
And with that, we got Khan from Luxor back on the show. Welcome back, Khan.
- [37:00]
D
Thank you so much. Colin and Charlie, great to be back on the pod.
- [37:04]
A
Yeah, man, thank you for joining us. So y' all just published your. I'll get it up here, Charlie. Your February look back series. And for anyone who is interested in data around bitcoin mining, first of all, you know, I mean, they sponsor the show, but we love hash rate index for all of our bitcoin mining data. And these look back series are packed with insights from each month for changes to hash rate difficulty and mining economics. Also, Kahn leads research and content at Luxor and he puts out fabulous reports on a lot of the bitcoin mining network. They just published their year end report a few weeks ago. Recommend you check that out as well. But Khan, February was for me the most tumultuous month I've seen in bitcoin mining. Probably since somewhere around the China mining ban or perhaps the halving in 2024. When you were compiling this look back article for February, what stood out to you as the most surprising find for the month?
- [38:07]
D
That's a good question. There was a lot that happened in February. It was a wild story both in terms of bitcoin price and network difficulty. But what stood out to me the most was looking back on prior bitcoin price cycles and sobering up in terms of where we might actually head for future bitcoin price moves. We've seen that prior peak to trough cycles have had ranges anywhere from 75 to 85% in terms of drawdowns. And this has some interesting implications for where bitcoin price might go in the near future and where hash price might land as well. So we'll get into a few of those details towards the end. But I found that to be the most surprising and personally sobering observation throughout this month.
- [38:54]
A
I love these charts. This was, this was what really stood out to me as well. Looking at this for the reading this over for the first time and I'll have this beautiful one where it's current bitcoin cycles, price and drawdowns November 2022, a drop to the present, updated March 3 and right now from the October peak we're down about 50%. When you all publish this and you also have this really good bar chart here that compares this to prior cycles and this one currently is it the third worst at least in the modern era. I don't think y' all included data for anything creased 2017 for because that would produce too much noise. But you know, within the last almost decade, this is the third worst drawdown in bitcoin's history.
- [39:40]
D
Yeah, exactly. And as you point that out, Colin, I'll. I'll get into it. To talk about bitcoin price, I think we can sum up February's story with two main points. The first is around bitcoin price and the second is around network difficulty. When it comes to bitcoin price, if you scroll up a little bit, we'll see a bitcoin price and network difficult difficulty. Charts right here. So as you pointed out with those bar charts, we've seen the decline in February extend what is now the deepest drawdown in the current cycle. Back In October of 2025, we were at 126,000. In late February of 2026, we touched 63,000. So that's around a 50% drop from the peak that we saw back in October. Bitcoin opens at 78,000, declined by just under 24% throughout the month and closed at around 65,000. We averaged $69,000. And obviously this dragged down dollar denominated hash price with it as well. So February marked a milestone that miners wanted to avoid. But unfortunately we're back here again. We've seen new all time lows in hash price. Hash price began the month at $34.91, ended at 28:57, touching a new daily all time low of $27.89. And we averaged at $32.31, down 18% month over month. So that's the story with bitcoin price. Obviously mining economics from a revenue side are down bad. Now, if you scroll down towards the network difficulty section, this is where it gets very interesting. February saw a historically rare trend in difficulty, what we like to call a difficulty whipsaw effect. So what happens early in the month? On the 7th of February, we saw a significant difficulty drop, just over 11%. And this was on the back of two consecutive difficulty declines throughout January as well. Now this provided temporary relief for miners. Bitcoin denominated hash price went up throughout this difficulty epoch, spanning from the 7th to the 19th of February. And this difficulty decline stood out as the seventh largest decline we've seen since 2016, which is what we consider to be the modern steady state. Asic era since 2013 to 2015 saw a lot of step changes in hash rate and difficulty. As we transition from GPUs into FPGA into ASICS. We look at this time period from 2016 onwards and we see that February 7th decline ranked as the 7th largest decline in difficulty. So that's the first interesting point. But then what's even more interesting is the resolution and the rebound, which is we saw later in the month. So just 12 days later, on the 19th of February, we saw a difficulty adjustment, positive upward difficulty adjustment of close to 15%, which brought network difficulty back up from 1:25.86 trillion to 1:44.4 trillion. And this ranks as the 12th largest positive adjustment since 2016. Together, combined, these two consecutive swings total a 26 percentage point change in network difficulty, which is the fourth largest back to back opposite direction swing we've seen since 2016. And it's essentially the most extreme difficulty whipsaw we've seen since the 2021 mining ban in China. All other previous extreme swings in difficulty were were related to the resolution of the mining ban which took place throughout the summer of 2021.
- [43:49]
A
I really love this table as well for showing how severe that whipsaw is. And going back to the kind of the, you know, analogy or the analogous situation here with the China mining ban, you have an inverse of the mining economics at the time in terms of the fact that with the mining ban we had this crazy reduction in Bitcoin's hash rate and difficulty fell accordingly. And then it was so profitable to mine because of that that people were revving up S9s and older units. Now though, even though we had this massive difficulty reduction, it is still so brutal to be out there mining. You know, Bitcoin's hash price hit an all time low over the month of February multiple times. So for my next question I have for you, what does this mean for the average ASIC that's operating on the network today? Like what generations from Yalls research are actually even still profitable at these levels?
- [44:45]
D
That's a really good question. And to find that answer we look at our quarterly mining economics projections model which we provide every quarter, we refresh it every quarter and it gives us 18 month interval projections for hash rate, hash price and network difficulty. This is exclusively available to Gold Tier Hash Rate Index premium members and it gives us a lot of insights around the hash rate supply curve that exists around the globe which accounts for all machines that have been manufactured and that haven't failed, and also the active hash rate which is economically viable as well. So I'm going to take you down towards the end of this article where we look at concluding thoughts and Colin, I think there's two questions that are relevant here. The first is at current hash price levels, do we expect more hash rate to come offline? And I think we have a pretty interesting insight here. Well, the first point that our projections model shows us is most legacy hardware is already offline. It's not as if they they were hashing throughout the recent period and now turned marginal offline. And most legacy hardware has actually been offline already. Now the rigs at risk of imminent shutdown stand at an efficiency range of around 25 to 29 joules per terahash. So this puts us into the class of around the S19 series and equivalent machines. At the moment what we find is that breakeven efficiency at current hash price levels, assuming an average power cost of 5 cents is around 24.5 joules per terahash. So any rig that boasts an efficiency above 24.5 to 25 Joules is essentially underwater at this stage. So this means that from a practical perspective we're looking at machines around the S90K Pro and an equivalent. So that's one thing. Legacy hardware has already been offline. 25 to 29 Joules is currently underwater. But then we also have newer generation machines, let's say spanning from 14 to 17 joules per terahash, which still boasts profitable mining economics. So there is an incentive to still deploy these newer generation machines. So these two opposing forces are interacting with each other. On the one hand we have an economic incentive to continue deploying newer generation machines, let's say the S21 series and equivalent. But then on the other side we have marginal machines being curtailed and essentially forced out of the network at the same time. So the interaction between these two forces will determine net growth in difficulty and hash rate.
- [47:36]
B
You know one thing that I noticed from your machine breakdown by efficiency and break even power cost is this the choice of power cost at $0.05 per kilowatt hour, you want to roll the clock back two, three years that a kind of assumed median price might be around like 6 or 7 cents. I don't know if you have any insight on like just the, the trend of lower power costs and why $0.05 is like the break even. Yeah, so I'm, I'm, I'm, I'm kind of curious you if you have any thoughts or insight on that.
- [48:06]
D
Yeah, so the power cost we get, this is an estimation that we provide for average industrial power costs that bitcoin miners generally face. As we know, mining operations are highly sensitive to changes in power prices. Even a marginal change from 5 to 5.5 or to 6 cent per kilowatt hour is going to have a significant impact on mining profitability. What stands out to us is that this is forcing the network to get leaner over time. So we have legacy hardware or let's say current generation machines ranging in the 24 to 29 jewels per terahash efficiency ranges, they're, they've now been made marginal so they need to move out of the network. But at the same time the machines that are coming in already boast profitability under most power price scenarios. So even if your power price is a bit over average, your efficiency, your fleet wide efficiency makes up for it and still provides positive margin which allows these machines to be continuously deployed well. So we have more hash rate coming in from these new efficient machines that are able to withstand the current mining economics whereas legacy machines are being forced out. So depending on your power price, if you have more efficient machines spanning from 14 to 17 joules per terahash, you can still make sense and deploy these for your operations.
- [49:29]
A
Give us a ballpark or roughly estimate where y' all think hash rate is going this year. Because when I look at all of this I, I would be shocked if we even crest the all time high from, from October. Maybe not shocked, but I just don't see much room for growth here.
- [49:48]
D
That's a good question. And we have two different sources that we can touch on. The first is the projections model which anticipates slower but still positive and modest growth throughout the rest of the year. And the main reason is because we expect to see a continued deployment of newer generation machines coming in, although this will be somewhat muted from legacy marginal machines exiting the network. In our projections we provide five different scenarios ranging from super bowl all the way to a stress case with base bullish and bearish scenarios in between. And I took an average throughout all of these scenarios and we see that the model anticipates a year end network hash rate of around 1,5 zeta hash with a yearly average hash price of around $31 per pet hash per second per day. So that's what our projections model is telling us. And then on the other side in terms of nearer term expectations for the next six months out, our forward market is also pricing in further hash price compression. So the forward market is also expecting positive but somewhat modest growth for hash rate and difficulty through to August 2026.
- [51:08]
B
1.5 zeta hash. I mean that seems, I, I'll say gut check to me that seems like a reasonable projection and that's only from where we are currently right now. That's only like a what, 30, 40% increase, maybe a little bit less than that. What are we at 1 1.1 zeta hash right now? Barely.
- [51:27]
D
So 1.05 around.
- [51:29]
B
Yeah, yeah. And so we'll be covering this a lot I think. Yeah, it's a, it's time to get your end of your hash rate projections in because otherwise they don't count because it'll be too far along in the development cycle. So Khan, thank you so much for hopping on the block. Space Pod live show. We will count catch you again soon.
- [51:54]
D
Looking Forward to it. Thanks so much, guys.
- [51:55]
B
Adios.
- [51:57]
A
If y' all like that, go to hashrate index.com go to their blog, you can find this and other Look Back series and you can also go to the research page if you're interested in some of the premium research or in some of the free research as well. Wealth of information with the data over there at Luxor. I always appreciate hearing what they have to say about Mark Market Dynamics.
- [52:17]
B
Okay, we're the last. The, our last story again. Another data center.
- [52:22]
A
It's.
- [52:22]
B
This is just the, the data center Compute Week. Because this, we're talking iron energy. So all of you iron iguanas out there, whatever animal. What, what's their animal?
- [52:36]
A
I told him iron emus. I don't know.
- [52:39]
D
Emu.
- [52:40]
A
You know something, they should think Australian. I don't think they have iguanas in Australia, do they?
- [52:48]
B
I don't. I've never been. Okay, so IRON is expanding their GPU fleet to 150,000 GPUs as well as opening a $6 billion equity offering. I actually just like the numbers stop making sense at this level.
- [53:07]
A
This is. Or they start making a lot of sense.
- [53:11]
B
Yeah, a lot of dollars.
- [53:13]
A
Yeah, yeah, yeah. A lot of dollars and cents. So two, just two things here.150,000 GPUs is massive for a company that was running pilot clusters two or three years ago. So just to back up a little bit, IRON was one of the earlier entrants in the AI expansions for miners. They've decided to go a completely different route than a lot of the other miners. They're saying we're going to build the infrastructure and run the GPUs ourselves. Whereas most Bitcoin miners are just going the PowerShell approach and saying we don't want to manage clusters, we don't want to have to worry about networking. We are just going to build the data center and then we will court companies to come and host their infrastructure at our sites. So that's Core Scientific pioneered this model with Core Weave. You see Tarawulf doing it, you see Cypher doing it and all of these different. These companies have different contracts, whether it be Fluidstack, Amazon, all these different companies using their services. But IRON out here just stacking GPUs and they just. So they just purchased 50,000 Nvidia B300. Brings it up, brings them up to 150,000. I don't know who is going to be using these GPUs because Aaron hasn't made an announcement for a partner or a client for this compute. Yet, as far as I understand, except for Microsoft, some of these will be going to the Childress Texas facility which Microsoft will be renting capacity from. Some of them will be going to its McKinsey British Columbia facility. So when I say I don't know who's going to be using these, I don't know who's going to be using the entire rollout because it seems like it'll be split between those two sites. And as far as I know from my research, only Microsoft is listed as one of Iron's premier partners. That is in a. That what that deal was really kind of going back to like things no longer making sense. That deal was eye popping when it came out. So. So for that deal with Microsoft, it is a $9.7 billion agreement and I believe the contract terms are like 10 years or so. Also includes a provision with that deal that iron would purchase $5.8 billion worth of GPU equipment from Dell. Now, going back to what you were saying Charlie, about this kind of stop making sense. You know, Iron's advertising this, saying that this will this new GPU purchase and their fleet will give them 3.7 billion annualized recurring revenue by the end of 2026, when they expect to have all of these built, when they expect to have all of these deployed. JP Morgan, though, to your point, Charlie is asking where's all of the money going to come for this CapEx? Now, you know, Iron beat its chest in this press release saying that it has over the last nine months secured $9.3 billion in funding. This is from a mix of prepayments from Microsoft, potentially other customers. Again, they haven't really advertised any other customers for these clusters, convertible notes, GPU leasing and GPU financing. So they've kind of cobbled together this massive fundraising. You know, they've cobbled together this liquidity for the CapEx for these builds. But JP Morgan estimates that the 200 megawatt deployment for Microsoft will cost at least 3 billion and then they will need another 2 billion for a 40K GPU cluster in Canada. Now they have, I assume that's what we just saw with this 50,000 cluster or this 50,000 GPU purchase. Right, But I quote here from the, from the J.P. morgan report. The company expects to leverage these and other capital sources to finance approximately 3.5 billion of additional CapEx for these orders expected in H2 2026, including GPU servers, storage, networking, labor and ancillary equipment. Now, as part of this fundraising, and this was Part of the announcement, these kind of happened concurrently with each other. Iron also opened 6 billion. A $6 billion at the market offering. Now, an at the market offering is when a company goes and they say we are going to basically have a rolling offering for our stock for any one who wants to buy it on the open market. These are usually investment banks and other institutions purchasing these shares, hedge funds from the company directly through these programs. $6 billion is absolutely massive for this caliber of company. A bitcoin miner has never had an ATM this large. Like, I mean, I remember when we were on the show like last year, I was talking with Matt Kimmel. You know, we were kind of talking about the first time we saw like a billion dollar ATM that was a huge deal. Because usually the bitcoin miners, they like $100 million. Like, you know, anything within the cent a million range for a bitcoin miner. A few years ago was a massive at the market offering, and now iron is doing 6 billion. They have a market cap of roughly 13 billion. And so this does raise questions about how much pressure is this going to put on the stock price. Also, if they're doing the at the market offering, what does that say for private credit for these things? Is, are. Is credit starting to kind of dry up for some of these builds? You know, can IRON not go out and get a convertible note deal done or a secured senior note deal done? They're having to lean on an at the market offering to help finance some of these builds. I would love to have, you know, someone like a Matthew Siegel or someone else who is enmeshed in the world of credit to explain what this might say, where we're at in the cycle in terms of funding these things. So going back to what you said at the beginning, Charlie, and then I'll shut up at some point, and I'm not saying this is happening now. At some point, the spend doesn't make sense. You know, we, we see all of these numbers of how much money these miners are going to spend on these build outs. And at some point there's going to be something that breaks where they actually can't get the funding and maybe they don't get the payback to actually recoup their investment on these things. I'm not saying that's happening with iron, but we are getting the point where it's like, how many more billion dollar headlines can we really take here?
- [59:37]
B
Y' all counterpoint. Everyone doubted them when they switched hard and leaned deep into the GPU neocloud model before anybody Else and now the market's rewarding them. The iron CEOs, you know, C suite has profited individually a whole lot. Those boys have taken a lot of good packages home and put those funds to personal use. So maybe it'll work. I don't know. Like it seems like every other week it's, it's. We, the, the market and myself oscillate between, oh, the clock is ticking and it's about to be done too. We've just barely even started the, the true insanity. So I don't think it's clear if their deal like they're big deals with Microsoft. Microsoft's probably pretty good for their money. So I would say that's a pretty strong high conf. I don't know what other moving parts we don't, we have to adjust for.
- [60:36]
A
So yeah, that's the thing is these deals are, there's so much complexity in them just, just to give an idea for the way that they can be structured. For instance, one of the reasons why Cipher was able to get some pretty cheap debt for its Amazon deal is Amazon had a, like there was a clause in their contract of Amazon where they can't back out of the deal. And not every bitcoin miner has that. When they're signing these AI deals, if something goes wrong, their client, you know, their Partner, these big Mag 7 companies could back out. And so we've also seen, you know, the backstops that Google is doing on the fluid stack deals with terra Wolf and Hut 8 and others. One fact checking thing that I want to say. The Microsoft deal with iron is five years, not 10. It is $9.7 billion over that contractual period and there are no iguanas native to Australia.
- [61:27]
B
So we glad we got that cleared up. We're very, yeah, we, we need to make sure we get our facts straight on this on Blockster's live. Especially the important ones such as, yeah,
- [61:37]
A
emus are native to Australia though. So I'm gonna go ahead and call it the Iron Emus as, as the, as the complement to the Mara pigs. Speaking of Mara, for our cry corner this week, Charlie, what are bitcoin miners doing with their bitcoin?
- [61:53]
B
Selling it. And honestly, about dang time. Yeah.
- [61:59]
A
What other commodities, you know, business, they just like hoard the thing that they, that they produce.
- [62:05]
B
So you say that and I actually put that, I, I said this in another chat today or this this week, which is like no other commodities production business does this like bitcoin miners. And someone provided me with a material counter, counter example of a gold miner who did spot purchases of gold. So I'm wrong. But I will say like I don't think this is definitely not the case in oil and gas, which is kind of what my mind oriented, which makes
- [62:29]
A
sense in terms of like I don't know what the, I don't know what the production cost of gold is. I assume when it's ripping, obviously like anything, it's much higher or it's much lower. The margins fatter. Excuse me for the miners, but oil is a thin margin business and bitcoin mining is quite quickly evolving into a very razor thin margin business. If not quickly evolving, it kind of always has been. But to highlight this tweet that we have from Bitcoin News, I have the Transcript from Mara's 2025 earnings call and they say quote here now historically we held the bitcoin we produced as a long term investment in the second half of 2025. We began selling Bitcoin to fund operations in 2026. We expect to continue to monetize bitcoin opportunistically to enhance our financial flexibility and to provide liquidity or to fund capital projects and other initiatives that we believe enhance long term shareholder value. All of the kind of PR buzzwords. But here's the thing is this is a trend that is picking up now. So we have so far and I'm not going to pull up the articles just because it's kind of redundant, but I will pull up bitcoin treasuries here. So far this year Bit Deer has sold a thou, roughly a thousand bitcoin
- [63:38]
B
that it had plans to sell the rest of their bitcoin.
- [63:41]
A
Oh no, they've already, they've sold all of it. Well they, to your point, they're going to sell all of their mind production going forward.
- [63:46]
B
Yeah.
- [63:46]
A
Which is something that other miners like. Iron's always done that and Ter Wolf has also always done that. Ter Wolf had like a few, they had like little bit of bitcoin on the balance sheet. Not very much so. But Bit Deer has already sold everything and as Charlie said is going to continue to sell their mine production. Cipher said they intend to sell throughout the year. They have roughly 1500 bitcoin kango in January sold 4451 bitcoin to pay off a loan, restructure some of its debt. And if we look@bitcointreasuries.net which is a great resource for keeping track of all of these bitcoin boards, there's a lot of room for These miners to sell, man. Merrill holds 53,822. Bitcoin riot holds 18,005 bitcoin. Ryan hasn't said that there. Hasn't really made a big splash about selling. Maybe if I look in their Transcript from the 2025 call, I would see, you know, like, basically same similar language. Amerigave. I would imagine they're probably thinking about it. In fact, I think they have actually started selling some of their mine production. A lot of these miners in 2024, from 2024 on, started actually selling mined production because a lot of them got burned in 2021 for not selling at the top. Some of them actually, like bought on the way down, which was really funny. Let's see. Hut 8's got 13,696 Bitcoin and clean sparks got 13,300. On and on and on.
- [65:13]
B
1.8. Bit Foo Foo 1.7. Kanan 1.7. Yeah, cipher still has some. According to bitcoin treasuries, I mean, bitcoin
- [65:24]
A
miners collectively hold roughly. I haven't done the math, but roughly or just above like 100,000 bitcoins. So a lot of that's coming to market over the next few years, I would imagine. Right. Because these companies used to. These were the first bitcoin strategy companies in the sense that before microstrategy was cool, or before strategy was cool, these bitcoin miners were trading at premiums during bull markets because they held bitcoin. And it was one of the only ways that institutional investors could get exposure to the asset.
- [65:54]
B
Right.
- [65:55]
A
And if you actually look at correlations of these bitcoin miners to bitcoin, that, that, that beta that they have with bitcoin starts to break down after, I would say around 2024, 2023, because you had strategy and then you had the ETFs. And so there was no need to trade these things as proxies to bitcoin anymore. And so the benefit for them holding this bitcoin is totally gone. No one's buying them based on their bitcoin coin treasuries anymore. So it makes total sense that they would deploy that capital for more productive means.
- [66:29]
B
Yeah, you know, I'll observe that there's a lot of folks, and this is a, for some reason, popular view that persists year to year, which says miners selling their produced bitcoin is a main driver of bitcoin price behavior and they usually relate it to the halving. And I just think that is wrong. Those of us in the industry think that's kind of a red herring generally. However, a hundred thousand new bitcoin coming onto the market of just actual straight sellers, that does seem more material to me. Okay, I think that probably wraps it up. Thanks for sticking with us. If you like this show, make sure to leave a review on your podcast player of choice. And if you're watching live and you haven't already, hit that button bell or notification button, go ahead and do that. Because we'll be doing this every Friday. And if you're lucky and you slide us a 20 over the Internet, maybe we'll do it more than one Friday a week, but you'll have to check in later. Thank you all so much. See you next week. Hey, this is Charlie and Colin from Blockspace Media, and you're listening to the Blockspace Podcast Podcast, a show about emerging tech in Bitcoin, AI, energy and markets.
- [67:50]
A
We publish two interviews weekly with CEOs, investors, analysts, and anyone else of consequence within these spaces. Plus, we have a weekly news roundup for all the important stories you might have missed from that week. The show is perfect for retail and institutional investors, analysts, and really anyone who wants to keep their finger on the pulse of the stories that are moving bitcoin and energy and data markets.
- [68:14]
B
So if you've stumbled across us, make sure to search block space wherever you get your podcasts. And on YouTube, hit the subscribe button. Give us a rating. We produce bonus podcasts and other content on our main feed, so you don't want to miss that.
- [68:27]
A
And if you have any feedback or comments to give us or shows that you would like to see and topics you would like us to cover, hit us up at. Hello, Block Space Media.