← Blockspace: AI & Bitcoin
Blockspace: AI & Bitcoin
SpaceX Acquires Cursor for $60B, Nvidia Closes $25B Debt Raise, Cipher Hires Former ERCOT Director
01:01:58
Loading summary
Transcript115 lines
- [00:00]
Colin
Foreign.
- [00:05]
Charlie
What's up, y'?
- [00:06]
Colin
All?
- [00:06]
Charlie
Welcome back to Blockspace Live. Coming at y' all fresh on another day of green. Charlie. Everything's ripping, including SpaceX. So if you didn't buy the IPO, congratulations. Your progeny will be poor. Your grandchildren will be working in the cobalt mines to make sure that we have enough resources to secure the AI future. We are going to be covering SpaceX and their cursor acquisition to start the show. Get some market reactions on where the stock is headed considering it continues to rip and it just flipped a few Mag 7 companies actually in total market cap. After that, we will look at Nvidia's oversubscribed $25 billion debt raise, even though they've got like 13 billion on their balance sheet. Why are they doing it? And what were the interest rates for that debt raise? We have that coming at y' all for the second story. Following that, we're going to be looking at who is shorting neoclouds right now. The contrarian trade that for some reason makes sense on paper, but we're going by vibes right now, y', all, so maybe it doesn't necessarily make sense given the current trajectory of the market. Also, data center costs. Chamath came out with a wild estimate for how much it costs to build a data center. We take qualm with that in our third section and then for our final section to close out the show, Cypher Digital signs on an ERCOT veteran to lead its large data center builds in terms of figuring out interconnection and what is becoming an increasingly saturated market block.
- [01:36]
CB Spears
Space goes live Monday through Friday every weekday at 1pm Eastern. Runs a little over an hour and we feature quick hits on AI markets, Bitcoin, data centers and emerging technology. If you like what you hear, this turns into a podcast. Wherever podcasts are found, search Blockspace in your podcast player of choice and subscribe to the feed. Drop us a review and if you're listening on YouTube, hit the notification bell. Get that push notification when we go live every day. This show is brought to you by CleanSpark. Nasdaq listed ticker CLSK. More on clean Spark later on in the show. And once again, we are talking about SpaceX. We're going to talk about SpaceX until this rocket collapses or explodes. And so far, no signs of stopping. This is wild. It's ripping. This morning I'm showing you here largest companies by market cap in the world. Obviously number one is Nvidia, then Alphabet and Google. But this morning SpaceX flipped both Amazon and Microsoft as SpaceX is up another 12% on the day overnight, ripped past $200 per share. So SpaceX is now kind of right in between the fourth and sixth most valuable company in the world. If it goes up even just a little bit more, it will only have Apple, Google and Nvidia.
- [03:16]
Charlie
It's up 60% from the IPO price.
- [03:18]
CB Spears
Yeah, I mean,
- [03:24]
Charlie
and we've got some takes on this, so I won't belabor it too much, but I do think this was my hesitation when not trying to get into the ipo. I knew that it was going to rip, but number one, I wasn't going to get probably enough allocation for it to matter anyway. And two, every part of me is saying this doesn't make any sense, but it doesn't matter, partly because the float is so low, there are so few shares trading right now and everyone wants to get in on a piece of the far flung future.
- [04:00]
CB Spears
We'll look at the SpaceX shares in circulation because that's the main narrative for bears right now. Bears realized they were wrong, that SpaceX doesn't make money. And rather, and I think they're more directionally right if they talk about shares outstanding. But let's look at this. Here's some takes for the day from Ryan Peterson says with today, that's yesterday, that's Monday. With today's 20% Space X Pop, Elon made more money today than Warren Buffett made in his entire career. If you are a boomer that melts your brain, that hurts.
- [04:39]
Charlie
I mean, the boomers are the value investor generation.
- [04:41]
Colin
Right.
- [04:42]
Charlie
And Warren Buffett famously faded a lot of tech stocks because he couldn't model why they were worth as much as they were. And I think that there's a similar generational divide with a lot of this stuff. I think.
- [04:57]
CB Spears
Yeah, here's another take. Not even a take. It's just Kalshee, which is really more of a news source now. Justin Elon Musk is now worth more than bitcoin. So if you're wondering why we don't talk about bitcoin on the show anymore, it's because we talk about Elon. Who's worth more than bitcoin?
- [05:14]
Charlie
Tap the sign. I mean, we'll just start framing this behind ourselves. You know, actually this would be a really funny thing to print out and like put on a wall. It's like motivation. Yeah, I don't know whether or not that should make me bearish or bullish. I think it makes me bullish because it just, you know, every Single time bitcoin would rip in a bull market, I would just think this is too crazy, it's going up too quickly. You know, once it crossed a trillion, people started thinking about diminishing returns, which will certainly happen. But you know, if you're bullish on bitcoin long term and still believe in the fundamentals that that should make you bullish. The fact that a single man is worth more than the entire market cap of decentralized digital gold.
- [05:54]
CB Spears
So yeah, and I, I think I'm
- [05:57]
Charlie
looking for, I'm just grasping, man, I'm trying to find something.
- [06:01]
CB Spears
We tell her, we make these myths together. It's got to be worth more. I feel it. But it's not, it's not on center stage right now. It's SpaceX. And so the really the material news on SpaceX is this morning because it's not just that it's ripping everyone's faces off, they're also buying, there's a buying spree. SpaceX just completed a $60 billion takeover of Cursor, the AI coding platform following its IPO. Now you may not remember, but apparently SpaceX had basically kind of like rights to buy or exercise this purchase. And this is for an all stock deal with cursor for 60 billion. This basically, very quickly, we already talked about this on the show for the past two weeks that SpaceX is an AI company who happens to have spent the past 25 years doing rockets. At least that's the narrative they're telling you with the XAI merger and now with the cursor acquisition. This puts SpaceX in a position where they have one of the top three coding platforms is kind of OpenAI anthropic and then Cursor, they have some of the best ability to stand up, compute both training and inference with their colossus and, and soon to be macro hard sites. And now they, they, they, they completed this deal. It's kind of amazing that in just six short months Elon has absolutely shown to be a pretty visionary and if in fact maybe leading contender in the AI race.
- [07:44]
Colin
Right?
- [07:45]
Charlie
And this is something that goes back to our commentary on SpaceX during this whole saga is ultimately, if you're looking at SpaceX only as a rockets company and an aerospace company right now, the valuation makes no sense. If you're looking at them as an AI play in near term, then it makes a little bit more sense. And now with the Cursor acquisition, they've got AI software services in the stack as well as their own LLM with Groq and they also have the infrastructure play. So I mean at this point they're kind of just a full stack AI company. Yeah, in a lot of regards.
- [08:23]
Colin
Right?
- [08:24]
CB Spears
Here's a take from Sudo. Sue, one of my favorite followers on Twitter says say hi to Kurksur. C U R X O R and look at what Elon actually just assembled. The best coding product. A lot of people believe Cursor is the best coding product with distribution straight into the hands of expert engineers. A frontier model Grok trained jointly with IT Colossus, which is this incredible supercomputer cluster. So all of those now sit under one corporate roof. I mean look at how the other companies stack up. OpenAI as Sudo says, GoT has the model in a billion users but doesn't have the editor of its own and it rents most of its compute from Microsoft. Anthropic has probably the best coding model and CLI or command line tool, but they don't have a flagship editor, they don't have compute of their own. They're very compute constrained. Microsoft owns the editor and the cloud, but editor rent the brain from the partner that it's quietly at war with. OpenAI, the most expensive arranged marriage in tech. And Google has sat on every single piece for years and still ships them like an unfinished science fair project. So Elon has all of those under one roof. Incredible value thesis. Frankly, I don't think I saw it coming a month ago when this IPO was announced. And as we followed it, it's become pretty clear that this is the play. The question is, Colin, is it worth two and a half trillion dollars?
- [09:54]
Charlie
I don't know. That is the question. But I mean if you look at even what they paid for Cursor, the multiples at this point, going back to the Warren Buffett jabs, they don't make sense if you're an old school investor. So Cursor brought in 770 million in 2025. So it's more than they paid more. It's an all stock deal for Cursor, but that's more than 60x its revenue. Its annualized revenue run rate is 4 billion and it was valued at like 29.3 billion earlier in 2026. So but at the, you know, at a certain point, so like yeah, these valuations are crazy but at this point the market, to me the biggest participants like SpaceX are basically pricing in the opportunity cost of not securing these things because you know, for some of these companies like SpaceX, no one's building what they're Building or if they are, it's not even in the same league in terms of how good it is like with cursor.
- [10:55]
Colin
Right.
- [10:56]
CB Spears
So yeah, and there's obviously the space moat. But we are in this interesting scenario where if you are kind of a student of projecting and seeing how companies can catch up to frontier models, we kind of have a set stage now for which companies have moats and are going to be players in the frontier AI space. And there are massive moats from compute to personnel to just path dependent like integration and user adoption. This is Elon basically buying. Not just buying, but assembling a seat at the table and maybe it might be the head of the table. I'm going to let you read this tweet from Bill Ackman, Colin, because this is a fun tweet and also doozy.
- [11:43]
Charlie
Yeah, it's also a little tautological. One of the things that makes Space X valuable so valuable is how valuable it is. I mean this is astounding analysis. I'm going to give him the benefit of the doubt here though. We'll keep going the cursor. Acquisition cost material less than dilution because of SpaceX's high valuation. SpaceX's ability to economically, strategically and technologically accretive acquisitions is an important component of its value. There's enormous value inherent to a company with a high value, particularly when it is controlled by an entrepreneur that the most talented people want to work and partner with. Value begets value, talent begets talent. Too many values notwithstanding. We get the point. And I do think that that is, that sums up the investment thesis here. And that's one of the things that people, I think, truly miss when critics of Elon say that he's not some genius like some engineering genius, like he's not building the rockets, like who cares?
- [12:40]
CB Spears
Who cares?
- [12:41]
Charlie
I mean you don't have.
- [12:42]
CB Spears
Not up there with a screwdriver screwing in the panels, but he, he absolutely is.
- [12:48]
Charlie
He puts the pieces in place. I mean.
- [12:50]
CB Spears
Yeah.
- [12:51]
Charlie
And at this point, you know, there really isn't. And you know, I don't want to like simp or anything because I don't. I'm pretty neutral on Elon. I think he built some amazing companies, you know. Yeah.
- [13:05]
CB Spears
I got a love hate relationship with them. I'm using, I'm, I'm using X, which is like the worst thing in the
- [13:12]
Charlie
entire world, but also the best thing in the entire world, but also. Yeah, sorry. It was better when it was owned by Dorsey. Sorry.
- [13:19]
CB Spears
Everyone agrees.
- [13:19]
Charlie
Yeah, it just was. But that being said, you Know, complicated guy, probably would be insufferable and you know, depending on his mood and what in your relationship with him. The multiple wives thing is weird.
- [13:34]
CB Spears
Seven ex wives or something.
- [13:36]
Charlie
Yeah, that's not. Lindy. That's bad. But, but I think it's indisputable at this point that he is one of the greatest, if not the greatest entrepreneur of his generation. He's built $2 trillion companies that do things that no other company can do. And that was, you know, the other. You know, when I look at Tesla as well, people were always talking about how overvalued Tesla is. The first time I was ever in a Tesla and saw what it can do, it made me realize is your exposure to that stock is an option on technology that no one else is building on the, on the continued curve of that technological advancement, because that's what, that's what he built. He builds things that no one else is building and probably standards or rather engineering feats that will become standard down the road, like self driving cars, you know, all that, all of that jazz. But this is, this is also good from.
- [14:34]
CB Spears
Yeah, pull up the. Yeah, here you go, Colin. Here's the Rittenhouse tweet.
- [14:37]
Charlie
Yeah. As disastrous as Xai's early investments have been, the NEO cloud pivot and Cursor acquisitions were pretty well executed. The deals with Anthropic and Google created a narrative that XAI can monetize GPU capacity at a premium solely because they can stand up data centers faster than anyone else. I think this is actually really salient point that's super important here. The willingness to sell to external customers also gives them a longer lease to invest in AI infrastructure. And stark contrast to Meta. For example, the combined 26 billion ARR from the Google and anthropic deals, the 75 billion roughly in total over three years, definitely juice the SpaceX valuation pre and post IPO. I think that's exactly right. I think without those deals, SpaceX wouldn't be trading where it is today. And it's not even about the revenue coming in. Although if you look at the revenue from those deals, the multiple, if you look at the arrival it's trading multiple is a little less nauseating if you factor that in. But to Rittenhouse's point here, it's the execution and the ability to pivot. Right. They couldn't actually use Colossus for their own model with Grok, because Grok, quite frankly, is not a frontier model. But you don't need to, because you can just lease that compute out to competitors who are now beholden to you for their own survival. In some ways, you know, I, I think that also the engineering feat of standing up Colossus in, what was it, like 120, 130 days.
- [16:05]
CB Spears
Yeah.
- [16:05]
Charlie
Which I almost think, like there's got to be some sort of fact check in there where it's like. Well, like technically there was certain things done before, but we'll give it the benefit of the doubt. That's absolutely insane.
- [16:16]
CB Spears
Yeah. And we cover the space like it. I, I think this is, it's funny because we come at this from like a data center perspective, which has actually shown itself to be a very useful way to view this IPO and how it plays out and Absolutely. The execution in the past, like two months with this acquisition and with the deals of Anthropic and Google on the backs on the heels of Colossus 1 and 2 being built at an insane speed really do showcase. Yeah.
- [16:53]
Charlie
Sorry to interrupt, Charlie, but can you imagine if Musk actually retained a controlling share and, or controlling interest in OpenAI? Like imagine if OpenAI and chat GPT were a part of this stack. You might think that it would already be the most valuable company in the world. Whether or not that's justified. And that makes me wonder what's next? Are they like they try to. They try to acquire OpenAI or anthology? I mean they're not going to do that because open. The OpenAI team hate Musk and vice versa. But truly the only thing that's missing here is a frontier model. And I don't know if they can catch up at this point. It probably also won't be able to. Like there's no way someone's going to sell a golden goose like that.
- [17:38]
CB Spears
Well, also, what constitutes a frontier model is an incredibly fluid dynamic situation. Like what is today's frontier model is three to six months from now. China's basically the same thing. China will sell you basically the same type of model. Maybe not quite as dialed in in half a year for a fraction of the cost. So does it really matter? And we can go round and round about this. I think the last thing before we move on is we've been talking about this on the show. Now I feel very validated because all of Twitter is obsessed with this now. Like they just weren't watching our show for the past two weeks, which is that it's not surprising that SpaceX has been ripping. There's only 4% of the shares are circulating. Look at this wall. Look at this absolute wall of waterfall Share unlocks over the next.
- [18:40]
Charlie
For people watching, can you break down the color codes? Because what is it purple is.
- [18:47]
CB Spears
We have a chart here from Tokenomist, which is great because it shows how it's built for crypto tokens and unlocks. And it shows basically the SpaceX token unlocked. So we have the public float of 4.2% of the 555 million shares went onto the market last week. So come early August, you have a lot of institutional VC and employees, shares start to get unlocked. Among the different categories of classes of different actors who've got SpaceX shares, you have the XAI merger. So people who have shares from the XAI merger, you have Alphabet or Google, which owns a considerable amount, employees of SpaceX, and then the institutional venture capital. And here's a chart which basically 15% come early August of those will come online and then every few weeks another 3 to 5% get added until early November Target, when we now have a cumulative of like 44% of all outstanding SpaceX shares. And basically by the end of this year, we'll have close to a little over 50% of all the shares outstanding. With the big one, which is Elon Musk's 42% of the company, which he retains, that all unlocks on day 366 after IPO. So basically looking at every few weeks for the rest of the year, giant tranches of shares will be available to sell. And I absolutely know that there are some engineers and some janitors who are sitting on 7 to 10 figure, maybe not 10, you know, 7 to 8 figure stock options which, which unlock. And congrats to them. But like they're gonna be selling. Yeah. And the market.
- [21:04]
Charlie
Yeah. And this here's a chart on Twitter that shows roughly how much will be unlocked per those tranches. And so, yeah, to your point, Charlie, by October 10th, you'll have like 53% unlocked. And then by the end of the year on December 9, you'll have roughly 60%
- [21:22]
Colin
there.
- [21:22]
Charlie
There are a few caveats here. Like after their Q2 earnings, if there is a. I think if it's. There's a stock price performance beat, I think it has to be somewhere in the ballpark of 30%. Then you have an additional unlock on top of that. But just generally speaking, we're going to see cascades of shares come onto the market throughout the year. And I'll be really interested to see how people play this because this is all public information. And you know, you would assume that most of these people would be paring down their positions. But some people might just degen and YOLO and continue to go long even though they already have exposure.
- [22:05]
CB Spears
And I've heard a lot of people say, I mean this is also entirely plausible that investors front run the unlocking of these shares. So we crash a bit before we will monitor the situation for you dear listener. So but let it be known that if this was news to you this day on Twitter because it suddenly became my entire timeline, we were talking about it for the past week and a half. So you should keep listening because we are just dripping with alpha on public markets here on the former bitcoin focused live stream. I think Colin, we should move on to more AI. So we're going to go to Nvidia next. They did a big raise. But before we do that, a word from our sponsor, CleanSpark.
- [22:51]
Charlie
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. We are setting the standard for what's next. Learn more about the intersection of energy and bitcoin@cleanspark.com. If Bitcoin's actually the best money and it's the thing that people should accumulate and it's the best risk adjusted asset, I lose zero sleep about whether or not that's going to happen. I just ask the question of when is literally matrix math that you're running
- [23:37]
CB Spears
on large pieces of data. The bitcoin miners can absorb that energy and in many ways this feels like
- [23:43]
Charlie
a second bite at the apple to build a new Internet. All right, Charlie. Nvidia. Nvidia. Nvidia. Nvidia.
- [24:01]
Colin
Nvidia.
- [24:01]
CB Spears
Nvidia.
- [24:02]
Charlie
Nvidia doesn't have enough money and it's trying to get a little bit more. Nvidia just closed a $25 billion raise. $25 billion debt raise that they actually upsized from 20 billion after they had $85 billion in interest going back to over subscriptions. We saw the same thing with the SpaceX IPO. Reportedly it was like 250 billion in interest on the $75 billion raise. And I don't know man, at this point you can't even call it frothy. I mean the market is just effervescent with enthusiasm and the capex cycle is not slowing down and the appetite is not dying out either. So here's their filing for the debt raise and they've issued a number of notes in multiple different tranches ranging from 2 to 330 years so some quick details here beyond what I've said so far. The longest dated bonds are priced about 65 bips over current treasury rate over the current 30 year rate. The raise was, like I said, 25 billion more than and they had roughly 3 to 4x interest. The bonds are expected to be rated double A, Moody's AA1 and S and P AA. So they're obviously investment grade. It's Nvidia. It's the most valuable company in the world, so why wouldn't they be? And if we look at the actual details of the notes themselves, the interest rates are pretty juicy compared to where things sit right now. The notes due in 2028, $3.5 billion worth, have a yield to maturity of 4.273. And then the notes due 2056, the longest notes, the 30 year ones, 3.5 billion of those. The notes are 3.5 billion in size and they have 5.628% yield to maturity. And I mean most people were looking at this and asking why is Nvidia doing this at all? They have like 13 billion in cash on their balance sheet. And as we'll cover in the next section, it's because data centers are really freaking expensive to stand up. And also as Jay Patel pointed out on our stream yesterday, there's an argument to be made that when the money's flowing, just take it, man. If you're confident in your position, you're confident that you're going to generate enough revenue to cover the principal and interest of those notes, if your credit rating is good, if your bonds are already trading well in the secondary market, you might as well take the money so that you have the cash and dry powder available to execute on it when you want to. Yeah, I mean, yeah, go ahead.
- [26:56]
CB Spears
When it rains, fill a bucket and store it for later. I don't know if that's what they're doing, but it's raining Money right now. IPOs are hot. I do think there is some concerns of all the money chasing the next IPO. And there's two next IPOs, there's Anthropic and OpenAI and there's probably other IPOs that will come like a freight train and blow us all away. But like you've got these two looming ones on the horizon and if you, and maybe it's you want to pick up your few billion dollars in front of the freight train. That's the analogy I was trying to communicate here.
- [27:35]
Charlie
Yeah, and I wonder too how they will divvy up this raise. How much of it will go to expanding R and D, expanding their own manufacturing capabilities, expanding their product lines. How much of it will go to actual infrastructure to house their gpu'? You know, we've seen recently that they are partnering with Anthropic for Anthropic's own ambitions to actually physically own computers and host them at powered shells. Nvidia backstopped or is reportedly going to backstop their search for those gigawatts. And as we know, they're involved in a number of build outs throughout the country. One of the chief criticisms actually of that kind of circular, that circular financial, you know, circuit that we've been seeing. So but this really, this story, Charlie, I do think transitions well into our next segment because again, people are asking why are they doing this? Because data centers are really expensive and people are still spending money on standing these things up. And in fact, forecasts for the data center demand and build out continue to inch upward. I'm going to get a note from our friend Matthew Siegel up to kick off this segment here, Charlie. He tweeted this today following a JP Morgan note where JP Morgan says that they're raising their CAPEX guidance from Matthew Siegel here, JP Morgan takes victory lap after strong year to date for high yield data center debt. Raises estimates by 13, raises power estimates by 13%, raises AI CapEx estimates by 8% through 2030. Now expects 138 gigawatts of capacity growth through 2030, up from 122 gigawatts in prior forecasts. It's just not slowing down every day. You think it can't possibly get any hotter or any frothier. It just continues. It continues to froth. It continues to heat up.
- [29:44]
CB Spears
The yeah, the market can stay rational longer than you can say liquid. We are from bitcoin. We are from the world of crypto. We know this. But I do ask is the market irrational? Is there something going on bigger that you don't know about? Well, if you do want to be short, there's you've got a, you've got a champion now and his name is Jim Chanos. We're going to talk about Jim here in a moment, but before we move on to that section, a word from our sponsor, Luxor.
- [30:22]
Charlie
This episode 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, that's an automated profitability engine that runs every five minutes and tests your power settings against live Power markets and hash rate markets. Ercot back tests show 10 improved profitability with intelligent mining versus binary mining. Commander Pro is a hundred dollars per megawatt or a 25 basis point pool fee adder, roughly half the cost of competition and you can try it for free for 60 days. So if you'd like to get started, go to Luxor Tech Forward slash Commander to learn more. Okay, Chanos Mode.
- [31:05]
CB Spears
Charlie, what is Channels mode? We have, we have a clip from Jim Chanos here. Why, why should we explain it when he is perfectly good to give his pitch, Roll tape, turn our attention to
- [31:23]
Colin
the neoclouds like coreweave or Nebbyus. It's even worse because those are, when you boil it down, those are equipment leasing businesses, right? They buy a chip from Nvidia and then they turn around and lease the chip to a hyperscaler or to a pure AI software company like Anthropic or OpenAI. And there the bet is on depreciation and whether or not you can get the chips from the supplier. So we were talking a little bit earlier, it gets back to the point that we've made to clients that anybody that's just a middleman in this chain like the data center guys are either as REITs or equipment leasing companies should never trade at higher multiples than the company that controls their supply, Taiwan Semi and Nvidia AMD and so on, right? They're price takers in all sense of the word. They're middlemen. And so it's really, we're at the point in the bull market where all of these stocks have gone to premium valuations. And I suspect at some point in the near intermediate future we're going to start to get to the point in the market where people are going to begin to really analyze the business models and say, okay, what's special and what's a commodity?
- [32:51]
Charlie
I mean, I think that what he's saying is perfectly reasonable. Specifically the idea of these middlemen with the neo clouds trading at multiples higher than their actual principal supplier. The only thing I would say about that is, well, kind of obviously, yeah, because these companies are still in price discovery, they're high growth tech stocks. Nvidia is the most valuable company in the world, firmly established. Once you grow to a certain level, those multiples end up compressing. When you're in the growth stage, they are going to, you know, outpace completely your, your revenue or your earnings or in the space of SpaceX or negative earnings. But all that being said, I, I don't necessarily think that what he's saying long term is probably wrong. The only thing I'll push back on and, and maybe this is a mid curve take, but everyone talks about depreciation of the GPUs. I personally think that it's too early to say whether or not that's going to be destructive to some of these business models. Because if you look at the DI metrics dashboard here, obviously the Blackwells are the most valuable ones. Their rental prices are higher than the others and they've gone up more drastically over the last few months. But we've seen here that H2 hundreds and even H1 hundreds and A1 hundreds have all seen their rental prices increase, albeit marginally to some extent for this for the older generations over the last few months. But overall, to me it's not totally clear that you get the same,
- [34:35]
CB Spears
you
- [34:35]
Charlie
get the same depreciation death spiral like with ASICS that we had in bitcoin mining simply because the tiers of compute are so different. And yes, you might want a B200 to train a frontier model or for high top level inference, but an A100 might be just as good for a lower compute workload. My point in saying this is like there are so many actors in the market and there are so many different layers of compute at this point in terms of performance and affordability. Who's to say that there won't be certain use cases that can leverage those older models? And it might make more sense to do that kind of in the same way that we saw off grid oil and gas bitcoin mining operations defaulting to older generation rigs because the capex was worth the trade off for what it cost them in OPEX for lower efficiency.
- [35:31]
CB Spears
I mean look at this. I love DI metrics, but it doesn't go back to 2024. Here's orange dashboard H1 hundreds going down, down, down 160 something odd per for rental rates and now tripled in the past couple months. So we saw this happen with asics. We've seen that they are very much related to the commodity cycle of whatever they produce, in this case flops. And yeah, I think assuming that these are going to decline in value takes a lot of assumptions about just the overall trajectory of the business and the ability to produce and stand up compute.
- [36:19]
Charlie
Yeah, I think it's hard to say now in terms of like it could totally play out that way and for some people it will. Some people will be ahead of their skis and they might not structure their business properly. And so the DEPRECIATION will crush them, may crush them. But to me this also seems like a problem that we won't be sure that it won't materialize for sure until I think this first capex cycle really concludes and we actually have these data centers that are in the pipelines being built out and then that's kind of a make or break moment or at least a pressure test for the entire industry. Did we overbuild? Did we underbuild?
- [36:59]
Colin
Right.
- [36:59]
Charlie
And if you do have an overbuilding scenario, then yes, the depreciation will probably destroy some of these companies. But to me everyone's treating it like a foregone conclusion. They cite the three to five year useless usefulness cycle for some of these GPUs. I don't think anyone can fully know how that's going to shake out yet.
- [37:19]
CB Spears
I agree with this high level statement and analysis that yeah, If your industry is constrained by other businesses and downstream of that you probably shouldn't be trading at multiples of that. In this case Iron's trading, it looks like 79 price earnings. NVS is trading 103. No whatever between 80 and 150. That is pretty wild to me. The thing is Channels, it's just like I made this joke last week, investing is just like humor. It's all about timing. So Channels has a pretty good track record and he's one of those folks who's like public about it. He likes to be a publicly, publicly short stuff. I don't know if I'd say he's a good, I don't know if I say he's a contrarian because like I don't know if this is, is a contrarian take. There's a lot of people who this would resonate with. But if he, he likes to be the, the, the poster child, the guy out in front of this on the topic.
- [38:34]
Charlie
So yeah, and one other thing that I think might, may bear highlighting
- [38:41]
Colin
and
- [38:41]
Charlie
maybe this isn't necessarily a fair comparison because the CapEx constraints and the CapEx requirements are so much different. But when you look at the trajectory of bitcoin mining hosted bitcoin mining plays did well enough up until a certain point until hash price was incredibly compressed and then you had the largest public miner at the time, Mara, having to completely eschew their asset light model which you know the selling point with that is like their time to enter. Their time to enter Jason was, was, was shorter. The, the capex for their builds was smaller because they only had to buy the A6 but their operating Costs were much higher as a result of that and you basically had towards the end of the golden age of bitcoin mining within the public markets in the US almost none of the big players were hosting equipment at a colocation facility. And I do wonder if that model will eventually play out for these data centers as well where maybe through some combination of JV or just full pure vertical integration, the powered shell and the NEO cloud model, or rather the powered shell hosting model will kind of give way to a vertically integrated NEO cloud model like Iron. But as your next clip would indicate, maybe that's not possible just because the actual capex requirements are so insane and the specialization is so insane between managing the power building data center and actually running the GPUs that we might not get that here.
- [40:21]
CB Spears
Is Chamath not going to try his last name to explain his view on current datacenter?
- [40:29]
Charlie
Capex bot now costs $100 billion guys okay so there's a huge capital moat that's a problem here as well. Sachs which even if we wanted to go and endorse and breathe life into the open source model community, where the hell am I going to come up with $100 billion now when I started this project it was like 4 or 5 billion and it's increased by 20x. So what now costs $100 billion guys okay, so $100 billion, that's almost as much as retail investors lost on Chamath Spocks. And maybe he can come up with 100 billion if he, if he launches a few more, you know, get gets on the board of a few more Spocks. So you know, he's not, I don't think he's right about this. I don't know where he's getting the hundred billion dollar figure. I did some, I looked around, he doesn't cite anything. I don't, I'm not really sure where that's coming from. That being said, you know it is quite a lot. So.
- [41:29]
CB Spears
So I did confirm Colin that Jensen Huang did kind of also echo this. Jensen1 did throw out similar numbers in the Nvidia presentation recently.
- [41:41]
Charlie
Is that 100 billion forward looking because it says as racked as racks Densify.
- [41:46]
CB Spears
Yeah, I think, yeah, I think Jensen said it'll cost 100 billion in the future.
- [41:52]
Charlie
Oh yeah, probably will with the rate of inflation. So sure, Jensen saying that and perhaps we'll get there. I do think though that currently the research that we have right now and also looking at actual expenses from some of these companies doesn't really add up. So this is from Shift Lab by Orenia. They estimate that it costs $60 billion per gigawatt to build a fully integrated data center. The bulk of that is hardware and IT equipment and then the rest of it is infrastructure. That includes labor, materials. This also includes having gas generation on site. So, you know, if we're looking at all of those, the sum of those parts, 60 billion, still a lot makes a lot of sense to me. I would also say that according to some of my research, Nvidia's Q2 2026 call or fiscal year Q2, Nvidia's own figure, which Bernstein notes in its estimate, reflects 50 to 60 billion. So I'd imagine that 100 billion is probably projected out into the future from Jensen. But Chamath is quoting it as though it's happening right now. Now some people will quibble, but like, there's a big difference between 50 and 60 and 100 billion. The fact is though, going back to why Nvidia would be raising money right
- [43:22]
Colin
now,
- [43:25]
Charlie
that is case in point right there. 25 billion wouldn't even cover a gigawatt data center under current pricing.
- [43:35]
CB Spears
So yeah, I got one last tweet on the, on just the data center story here relevant to this, which is maybe it's not the capital that's scarce, maybe it's not the, you know, institutional competence, maybe it's not the amount of, you know, switch gears that are scarce. Maybe it's where you can build the data center. Here's a tweet from Seanu Matthew which echoes a lot of what we've been talking about here, which is according to Bernstein, there is a 325 gigawatt pipeline for data centers of which 63 gigawatts are under construction, 34 of those are now defined as stranded due to pushback. What does pushback mean here? It's basically, it's the anti data center movement in a nutshell. Now there's, it's a broad term here, but a lot of data centers are being postponed or subject to moratorium or having challenges actually breaking ground because people don't want them in their backyards or anywhere really. And we've talked about why this is, this is not the time for that. This is just time to look at this crazy number. Whether or not it's like fully accurate, I think it's directionally right, which is how many data centers are getting delayed or postponed or potentially canceled.
- [45:07]
Charlie
34 gigawatts. Crazy. I mean, one gigawatt can power San Francisco or Denver. So that's Just that, just to give context. So 34, you're talking about 34 major metropolitan cities in the U.S. you know,
- [45:20]
CB Spears
I would also say, like if I were to apply some bitcoin miner historical subject, subjective evaluation here is there's a lot of bitcoin mining that was never feasible in the first place, that for whatever reason they say we're canceling the project. And I wouldn't be surprised if we've got some of that here where it was really ambitious. They were already over their skis and because of that they selected sites which would never have worked anyway because you're never going to. You really shouldn't build one of these things in the middle of a neighborhood. So maybe due to that ambitious strategy that could be counted in this 34 megawatts to find as stranded, but regardless, it's not an immaterial amount is what I'm saying. There's no way this is directionally correct and that there is an immaterial amount of data centers stranded or paused. So I think this is just besides the AI trade, the pushback story is the biggest story of the year. And it's just so clear to me that that is just going to define the economic landscape of AI and data centers in the west for the time being 100%.
- [46:27]
Charlie
I mean, this is the new climate hysteria in a lot of regards.
- [46:30]
CB Spears
You say it's hysteria. I take the, I take the different approach. It's not hysteria. It's a, it's, it's a, it's a movement against. Which is different.
- [46:39]
Charlie
But I think it is. What is hysteria like literally being hysterical reactions is something that people think, think they understand because they're being told by trusted experts that this is how it works when they don't actually understand how any of it works. And their entire model for how it works is actually wrong.
- [46:56]
CB Spears
Yeah.
- [46:57]
Charlie
One thing, one. Before we get into our next segment though, one thing I wanted to highlight on that. I wonder how much of those projections from Bernstein include phantom load, which is something we'll get into in the next segment in terms of. So like for ERCOT interconnection requests, because URCOT is a free market for energy, you have multi. You typically have multiple large loads going to multiple different utilities to request the same amount of load and it all gets added together. So number no doubt inflated, but still hundreds of gigawatts.
- [47:28]
Colin
Right.
- [47:29]
CB Spears
The other thing, I think your point there, Colin, is probably dead on. I think this type of assessment deserves a look from the perspective you just gave. Where are those true gigawatts in the pipeline. Yeah.
- [47:41]
Charlie
And the other thing that, the other bottleneck here that I don't think most people are appreciating more than power, more than land, more than pushing through the data center pushback. Can you find an actual contractor or construction company to build this thing for you? I mean, how many of these companies, how many of these construction companies that are, that are, you know, that have a history of building massive data centers or just large industrial sites? How many of them are there out there who can actually do this? How many of them are starting to pivot into this? We've seen the bitcoin mining to AI boom. Perhaps there is another shift going on in the construction sector in the US where a lot of contractors are looking around saying, hey, we figure out how to build these things, man, we're going to be making money hand over foot. I don't know if that's come up in earnings call as much yet. I would actually love to do some research on this because that to me seems like one of the bigger problems that could surface from this is there might just not be enough labor to even get these off the ground.
- [48:47]
CB Spears
Yep, I think that's the, I think that's the story we've. Speaking of stories, we got one last story. Talk about Cypher and Urcott. Before we move on to that and wrap it up, we have a word from our sponsor, Lygus.
- [49:07]
Charlie
Michael Saylor is selling Bitcoin and that means everyone's freaking out and losing your money. But not if you use Lygos for Bitcoin back lending. That's right. Bitcoin's price is drying up. Counterparty risk is rampant as ever as lending desks and big whales and hedge funds go under. So it's more important than ever to know who controls your Bitcoin. If you are using Bitcoin backed lending and with Lygos, that is always yourself. Lygos is our preferred lender here at Block Space. They use Bitcoin native smart contracts to make sure that you always hold the keys to your coins and you're not relinquishing them to a counterparty. No rehypothecation, no bridging, no wrapping. Just competitive rates as low as 8.5% to 10% APR. Go to Lygos Finance to learn more. All right, Charlie. Speaking of large loads, Cipher Digital. Cipher Digital has announced that they have appointed Bill Blevins, formerly a director, a grid director at ercot, to their board. And why are they doing this? We're going to get into that in a second. It's, I mean it's obvious. Is there power constraints? But we're going to get into some of the numbers here in a second. But yes. Bill Blevins was formerly an ERCOT director of grid coordination and he brings 30 years of experience in power systems at ERCOT. This is really interesting. He was one of the principal members responsible for grid operations during the transition to the nodal market structure in 2011. For those of you who don't know how ERCOT works, you know, it has these, these nodes, these specific points in the grid, which is like a substation, generator, interconnector, a load zone where all of the prices for these nodes are calculated independent of the independently from the rest of the grid based on local supply and demand. There are roughly 4,000 nodes across the ERCOT network. This is one of the reasons why it has become the premier location for bitcoin mining and data centers. Obviously they have a lot of generation out there, but part of the thing that makes it so desirable is that you can actually pick and choose what node you want to be on depending on supply and demand. So he was pivotal in getting that off the ground and now he's coming over to Cypher to a grid focused leadership role, I would assume to help the company hunt for power. That's the obvious, that's the obvious takeaway from this. And they're going to need every ounce of power they can get. Because if we look at what ERCOT is projecting for large load interconnection requests, it's getting pretty top heavy. I'm going to move this Charlie, to this view because we can't actually see the chart. So this is large load interconnection request. As of March 26, 2026 through 2023, they are projecting 410 gigawatts of interconnection, or rather there is 410 gigawatts of large load interconnection requests. Currently in the Que, 87% of these are data centers. Now, as we just covered on the last segment, some of this is what's called phantom or ghost load, where it's one, it's the same, it's the same company seeking an interconnection request from multiple different utilities in ERCOT's grid. But even so, let's just look at where it was at the start of 2026. It was at 233 gigawatts. So we've already seen like a 75% increase there in terms of interconnection requests in ERCOT. ERCOT projects that only 2.168 gigawatts, or, sorry, only 2.681. Roll it back, Jamie. Only 2.168 gigawatts were approved for energization in the trailing 12 months leading up to the report. And ERCOT itself assumes only 49% of the claimed data center load will actually materialize. So everyone wants power. Texas has some of it. But this is not like the wild west of bitcoin mining or the post China mining ban Bitcoin mining landscape where there was competition for this load, this generation. But not nowhere close to what we are seeing now. I, I mean these are just gargantuan numbers, Charlie.
- [53:44]
CB Spears
Yeah, I think word of the week, honestly probably word of the summer is going to be ghost load. It's because it's, it plays a lot into like projections. I think savvy investors will be able to look at this and identify what is a feasible, you know, load, added load scenario. But I think for the Facebook, for the retail traders and the Facebook data center folks, these kind, these types of nuance will be lost upon them. I gotta say, shout out to our boys over at Cypher, they just keep, they just keep grabbing people that we like. Add it, add it to the C suite.
- [54:30]
Charlie
Add it to the C suite. And I would. Maybe with this new hire they can get some new company swag. Like, you know, you'll see ski bums with the like big dumps on their Subarus. We should get them bumper stickers that say large loads.
- [54:45]
CB Spears
Oh man, I come from the.
- [54:49]
Charlie
That might not, you know that we're
- [54:51]
CB Spears
playing with fire here on the show. I will say in oil and gas, you know, you. Maybe I'll save this for a. Okay. You don't, you don't call it cement. You call it cement. Yeah, that's like, that's, that's industry vernacular.
- [55:05]
Charlie
Because why? Because. Why not why? One last thing here, Charlie, or go ahead and share what you were.
- [55:12]
CB Spears
I gotta show the cipher digital chart here. Been doing great this year. Year to date, up a lot. I forget, what is it? Year to date up almost double the past five years. Chart looks great. Caught a bid this time last year and has just not looked back. So every time they hire one of our buddies, the stock goes up. So bullish.
- [55:41]
Charlie
He's calling it right now. Not financial advice.
- [55:43]
CB Spears
Not financial advice.
- [55:45]
Charlie
I wanted to cap off this segment, Charlie, by talking about some of the changes that are coming to ERCOT as well. Because the game is changing for how large loads actually interconnect.
- [55:52]
Colin
With the grid.
- [55:53]
Charlie
We had a show on this back in, I believe, first part of the year. And you know, some of this is still in the works. Some of these are proposed changes are being debated on, but some of them have actually been cemented, we should say. And a lot of these changes are coming after Senate Bill 6 was, Senate Bill 6 was passed in the. The Texas House. It establishes new requirements for interconnection and operation of large loads projects. So those are over 75 megawatts of demand. And it directs really two things that are really important for these large loads. One, it's going to reform the 4CP transmission cost methodology. That's the part that's really still largely in the works. 4cp is the stands for 4 coincident peak. It is the period from, I believe, what is it, Charlie? It's June, July, August, September in Texas where they have the most stress on the grid because it's really freaking hot in Texas, y'.
- [56:56]
CB Spears
All.
- [56:57]
Charlie
And you got to have that AC ripping as, as our friends from Europe are realizing. It's great. Just, just install it. It's a great thing.
- [57:06]
CB Spears
Yeah, it's a great thing.
- [57:08]
Charlie
The World cup, hopefully going to lead to world peace in the AC discourse.
- [57:12]
CB Spears
We're going to save lives by hosting the World cup in the US this summer.
- [57:16]
Charlie
Yeah, that's a crazy statistic. More people die of heat stroke in Europe every year than from gunshot wounds in the U.S. anyway, anyway, going back to what we're supposed to be talking about. So these reforms would look at changing for cp. They might extend it. We had Haley Thompson Deluxe drawn recently saying they might even roll it out for a coincident peak during the winter months when the solar and wind assets that Texas is now laden with aren't producing as much. And it also creates new large load interconnection and curtailment frameworks. So the biggest, the biggest change here for what's being considered is that they are considering requiring large loads to pay a portion of system upgrades in addition to their direct interconnection. So I believe now large loads are mandated to pay 100% of their direct interconnection, meaning if they're requesting like 200 megawatts, they're going to have to pay for the transmission lines, the substations, the transformers to deliver that load or that capacity to their site. But what's also being proposed, and this is somewhat controversial, is that they're also, they might have to invest in indirect infrastructure, which would mean expanding interconnection costs to include system upgrades in order to, you know, make sure the, the, the grid writ large is upgraded for all of its users. That is is highly contested. I would imagine the direct paying for direct interconnection makes a lot of sense to me. I could see a case for the highway provision where they have to do indirect costs as well. But I would imagine that they're going to kind of take that kicking and screaming. Also when you think about this, for the direct interconnection alone, the cost of building in Texas has gone up as a result of this. You're no longer having to think about just your capex for your site. You also have to think about it for getting the energy to your site itself. The other thing is that new large load, new large loads are required to be curtailable. So there's two parts to this. Whereas prior you could, you could opt into these curtailment or demand response programs, now they are going to say large loads will be required to curtail at a moment's notice and large loads at CO located sites won't be paid for curtailing either. And by co located here we mean sites that are on power generating assets. So those large loads won't be able to get remuneration for their curtailment or credits going forward. And that, you know, I haven't dug into the details. It somewhat makes sense if you're just thinking about it from ERCOT's perspective of look like we need all the generation that we can keep get during peak especially considering all these data centers are coming here. One last thing to note on the required curtailment this is they're, they're mandating that these data centers have gas turbine backups behind the meter generation as well to make sure that they can be curtailable.
- [60:25]
Colin
Right.
- [60:25]
Charlie
If you're going to say oh we can't curtail otherwise we have to shut down our data center and we'll lose out on revenue. You're going to have to have that behind the meter generation to make sure that happens. And one of the other changes that they made recently is they're actually allowing these data centers to operate before their interconnect with the behind the meter. Because the data center companies and operators said look, timing to market is everything. Right now we're going to put these gas turbines on site. Are you okay if we rev those up and then start drawing power before we're even connected to the grid? So that's already happening. We've seen stories about this, right? Like the jet engines being repurposed for turbines on data centers and whatnot.
- [61:02]
CB Spears
So we had a comment to wrap up the stream on YouTube. Quote Is this an AI generated video stream? Wally I the answer is no. This is the real deal. You've got Charlie, CB Spears, Colin As I, Lee Hodling and this is Block Space Live. We go live every weekday at 1pm Eastern featuring quick hits in AI, data centers, Bitcoin, emerging tech and markets. If you like what you hear, you'll love it as a podcast goes live everywhere podcasts are posted. Search Block Space Leave us a review. 5 stars preferably, but we'll take any stars you're handing out. This show is brought to you by Clean Spark NASDAQ listed ticker clsk. Thanks for tuning in. See you tomorrow.