WEBVTT
1
00:00:00.080 --> 00:00:10.960
On y'all. Welcome back to Blockspace Live, presented by CleanSpark. Charlie, we had earnings week the last two weeks, but we've got one more earnings to cover.
2
00:00:11.060 --> 00:00:19.080
That is Hive Digital's Q1 fiscal year 2027 earnings, which dropped at the end of last week after market close.
3
00:00:19.120 --> 00:00:24.000
They just had their earnings call this morning, so we, we will be unpacking those results for our first story.
4
00:00:24.120 --> 00:00:38.960
Following that, we've got Lagos Finance CEO Jay Patel on to talk about Jensen Huang in NVIDIA, specifically their new backstop of OpenAI, and just generally how the credit outlook is looking for the powered shell at AI CapEx bubble.
5
00:00:40.060 --> 00:00:50.460
Second for interviews today, we've got Dan Hillary of UTXO Management on to talk about digital credit and whether or not we think that stretch and strategy are in the clear.
6
00:00:50.570 --> 00:01:01.240
And then we will cap off the show with news that Blue Energy and GE Vernova are doing a little dual energy strategy for a Caruso campus in Texas.
7
00:01:01.300 --> 00:01:14.940
They are going to be deploying nat gas as they look ahead to deploying small modular reactors to power the roughly 1.5 gigawatt site. Blockspace goes live every weekday at 1:00 p.m. Eastern.
8
00:01:15.120 --> 00:01:25.360
We are compute's daily live show, featuring quick hits and stories on the latest in AI data center deals, markets, and emerging technology.
9
00:01:26.080 --> 00:01:36.560
If you like the live stream but you miss it, you can catch it anywhere podcasts are found later in the day when we turn it around and put it onto all platforms. Check out our RSS feed. Also,
10
00:01:37.780 --> 00:01:49.740
hit our website, blockspace.media. We have a newsletter that goes out every single day, as well as a bunch of other stories written and distributed in your favorite outlets, such as Yahoo Finance.
11
00:01:50.880 --> 00:02:04.660
This show is brought to you by CleanSpark, NASDAQ-listed ticker CLSK. More on CleanSpark later on in the show. Yes, we did earnings the past two weeks, Colin, but there are a few stragglers.
12
00:02:04.720 --> 00:02:17.040
The, uh, the first one today is Hive. I believe there's Iron later this month. Um, but there's a couple more of these that we will cover as they happen. The bulk of earnings, though, happened in the past two-week period.
13
00:02:17.180 --> 00:02:27.440
I would recommend anybody, um, who missed some of these, we had some very stellar CEO C-suite executive interviews, um, with pretty much everybody did their earnings. So make sure to go check those out.
14
00:02:27.500 --> 00:02:36.880
Um, check out the executive interviews playlist on our YouTube channel. And if you want to hear the earnings themselves, we've also got that on YouTube.
15
00:02:37.400 --> 00:02:48.540
So if there's anything you missed, you want the Q&A section, wanna see what management has to say about their strategies, go check that out. But we will go ahead and kick it off with Hive's latest earnings.
16
00:02:48.560 --> 00:03:05.720
I've got our coverage at Blockspace here. Hive Q1 2027 fiscal year revenue rises to 74%, rises 74% as GPU cloud ARR earnings reaches 110 million.
17
00:03:07.100 --> 00:03:19.160
Funnily enough, their cloud ARR is actually higher than that considering a new contract they signed, signed and, and, uh, announced during earnings. But at the end of Q1, that was the number.
18
00:03:19.170 --> 00:03:27.700
We're gonna go ahead and dive into the details here. There's a lot to unpack here, Charlie, and Hive, one of the more interesting case studies, because they're really kind of dividing their strategy.
19
00:03:27.720 --> 00:03:37.620
They're going cloud and colo, and they had a few questions about that on the earnings call. But before we get to the qualitative, let's focus on the quantitative really quickly.
20
00:03:38.180 --> 00:03:48.040
Revenue came in for the quarter ending in June 20- at 79.1 million. That was up 73% year over year and percent quarter over quarter.
21
00:03:48.049 --> 00:03:58.100
The biggest bump here was from mining, uh, revenue, 72.1 million coming in at 77% year over year and 7% quarter over quarter.
22
00:03:58.220 --> 00:04:15.740
HPC revenue, though, nearly doubling over the, uh, quarter at 7.1 million, up 47% year over year. Gross operating margin came in at 31%, and HPC margin at 44%.
23
00:04:15.780 --> 00:04:30.500
More on that later during the analyst Q&A. They had a GAAP net loss of 142.9 million. So what's behind the revenue increase? I've got one word for you, Charlie. Paraguay.
24
00:04:31.220 --> 00:04:44.800
Now, those of you who have been tuning in for a while will remember that Hive purchased a facility in Paraguay from Bitfarms, formerly Bitfarms, now Keel, uh, a little while ago.
25
00:04:44.860 --> 00:04:56.180
That facility accounts for about 18.9 exahashes of their total operational capacity out of 23.6 exahashes. Their hash rate has absolutely surged as a result of that.
26
00:04:56.260 --> 00:05:02.300
And they mentioned on the analyst call when they were asked, "Well, what are you gonna do with those assets in Paraguay?"
27
00:05:02.390 --> 00:05:12.040
And they kind of demurred on this point in, in with regards to how far they would go, um, in, you know, maybe deciding whether or not to turn that into an AI facility.
28
00:05:12.070 --> 00:05:21.420
They did say for the time being, though, that they would keep that as Bitcoin mining as a cash flow cow for the business. And it is the largest cash flowing arm right now.
29
00:05:21.520 --> 00:05:27.990
Uh, you know, two or three years from now, I doubt that will be the case given their AI deal flow. You would actually-- you'd hope not if you're a Hive investor.
30
00:05:28.520 --> 00:05:32.550
But they're going to keep that facility as pure play Bitcoin mining as it stands right now.
31
00:05:33.800 --> 00:05:44.030
As for their GPU line, the acceleration for that revenue largely came from a contract deployment with Bell Canada at their Merit facility.
32
00:05:44.880 --> 00:06:00.320
504 NVIDIA B200 GPUs were deployed, um, in May, and the-- so the June quarter did not include a, the, a full three-month contribution from that, and you would actually see a significant revenue bump, I would imagine, if that were included.
33
00:06:01.220 --> 00:06:13.192
Of the current clusters that cloud ha- or that, uh, um, that Hive has for cloud, uh, right now- They've got roughly 4,300 A series. Look at that, Charlie.
34
00:06:13.432 --> 00:06:24.212
The A100s and A series GPUs continue to hit the news cycle- That's been the theme this past week, is that everyone's just saying, "Yeah, we've got these, and they're still useful." It's been like the theme. Yeah.
35
00:06:24.232 --> 00:06:40.892
D- uh, you know, depreciation bears in shambles. They've got roughly 4,300 of those and another 340, uh, 344 H100s and 504 H200s in Montreal and Quebec. Um, about two or three megawatts estimated there.
36
00:06:41.412 --> 00:06:56.452
They have roughly 4,300 A series GPUs in a one megawatt, roughly one megawatt deployment in Stockholm, Sweden. Uh, the Bell AI Fabric in Manitoba, they have 204 B200s and a few others for roughly one megawatts.
37
00:06:58.192 --> 00:07:10.072
And in Asuncion, Paraguay, they've got a very small undisclosed, uh, deployment as well. Give me one second here, y'all. Google thought it would be a great time to boot me out of my prep sheet.
38
00:07:10.392 --> 00:07:15.412
[laughs] Gotta love it when you just randomly get kicked out of, out of the cloud. Okay.
39
00:07:15.792 --> 00:07:26.252
So let's move on quickly to the, the GAAP loss here because this is really im- this is, uh, quite important, and, uh, they, they got some questions about it during the Q&A.
40
00:07:27.232 --> 00:07:38.012
That loss is substantially less if you strip out a 804-- or, sorry, $84.65 million non-cash regulatory provision that they list as a liability.
41
00:07:38.022 --> 00:07:47.192
Specifically, they're in a legal battle right now with the Swiss government over that. Value added tax, the bane of many a business in Europe.
42
00:07:47.252 --> 00:07:57.202
There's 76.6 million in disputed input VAT, 1.5 million tax supplements on top of that, and 6.6 million in interest. This is a liability. They have not paid it yet.
43
00:07:57.272 --> 00:08:06.432
And in fact, fielding a question from KBW Steven Gladgola, they said they have no intention to pay it. They are going to ke- continue to fight this through the court system.
44
00:08:06.442 --> 00:08:17.912
They've already lost, I believe, uh, two cases with the Swedish Tax Agency on this. And the TLDR is that the Swedish tax authorities are saying that they didn't pay the right VAT for this.
45
00:08:18.452 --> 00:08:29.432
Hive, as I understand it, is saying, "We don't understand why these wouldn't be considered under the same VAT classifications as a data center." This is for the Bitcoin mining operations they have in Sweden.
46
00:08:30.141 --> 00:08:37.692
And they're basically telling the Swedish authorities, "Why would you classify this as anything else? Seems punitive to up that for no reason."
47
00:08:38.452 --> 00:08:46.361
And Aydin Kilic, CEO, said that they have no plans to pay that unless they were forced to from a, from a new court order.
48
00:08:46.392 --> 00:08:59.182
They're moving it up to, I believe, the Supreme Administrative Court of Sweden, or they're m- uh, working on doing that right now. Um, moving on to a few other items before I get to some of the meat from the Q&A.
49
00:08:59.812 --> 00:09:11.492
Cash sits at 208 million on the balance sheet. This comes from 150 million in principal, 130 million in principal of convertible notes. The first issued in April, the second issued in June.
50
00:09:11.652 --> 00:09:18.762
Um, they had 109 point- 109.8 million proceeds from the April one and 124.9 million proceeds from the June.
51
00:09:19.322 --> 00:09:32.172
They also issued thirty million in at the market, uh, or, or they, they drew 130-- uh, they drew 30 million from an ATM f- throughout the quarter. And they have 265.9 million in total debt.
52
00:09:32.232 --> 00:09:42.792
Only 3.2 million of that is current, and 245 million of that comes from the converts. So unclear whether or not that will be converted to equity or paid down through debt.
53
00:09:44.322 --> 00:09:56.992
[lip smack] They have 174.5 million in purchase commitments sitting on the balance sheet right now, as well as a $66.3 million lease liability for their Merit facility [sniffs] in Canada.
54
00:09:58.012 --> 00:10:07.452
They also, big news from the earnings call, they announced a 350 million five-year deal with a, quote, "investment-grade enterprise customer," end quote.
55
00:10:07.952 --> 00:10:22.972
This brings ARR to 180 million, up from the 110 million from what was booked during the quarter. And they're expecting a deployment by end of year 2026 at the Merit facility with 2016 Blackwell GB300s.
56
00:10:23.692 --> 00:10:32.682
Now, the first question they got during the Q&A, Charlie, was on this site and how they specifically will finance the $185 million deployment. And Kilic was pretty straightforward.
57
00:10:33.152 --> 00:10:49.912
He said he expects a three-year payback, 20% cost down on GPUs to start, and they are going to equipment finance the rest, and those payments will start hitting h- or the, um, the interest and the terms for that will start hitting once the shipment is imminent.
58
00:10:50.672 --> 00:11:03.472
The customer for this, unnamed customer, also made a 10% upfront payment of 35 million. They're expecting a yield of 28 to 30% annual EBITDA/CapEx before financing costs on this deployment.
59
00:11:03.482 --> 00:11:12.772
This is at the Merit facility in Canada, the one that we just covered or talked about just a moment ago. Hive doesn't own this facility. They just run GPUs out of it.
60
00:11:13.712 --> 00:11:29.442
And that's important to keep into context with regards to their strategy of colo versus cloud computing. Mike Rondall of Northland asked directly, um, about, uh-- Or sorry, that's, uh, another, uh...
61
00:11:29.442 --> 00:11:38.712
Mike was asking another question. He was asking what's holding up their Big Boden campus in terms of signing a lease. Uh, Kilic said the lease is already being drafted and revised between parties.
62
00:11:38.752 --> 00:11:49.482
Hive is essentially fine-tuning commercial terms here. Management says it hopes to update investors sometime in September. [lip smack] Um, regarding colo versus cloud, to-- per my earlier point,
63
00:11:50.652 --> 00:11:57.482
th-this is kind of, uh, w-was one of the things that was brought up a few times in the earnings call, going back to the Merit facility.
64
00:11:57.552 --> 00:12:07.392
And a lot of o-other facilities that Hive has, like Big Boden, w-with regards to what they're going to do, whether or not outfitting it with cloud or run- doing the powered shell colocation route-
65
00:12:08.388 --> 00:12:19.468
Hive said it has roughly 400 megawatts of Canadian utility capacity in its development pipeline with own sites theoretically supporting more than 120,000 GPUs.
66
00:12:20.948 --> 00:12:31.988
And management basically said, "Look, we are going to do what makes the most sense depending on the site." It was a safe answer, but the fact of the matter is they have quite a bit of optionality here.
67
00:12:32.688 --> 00:12:41.208
So they've got their Bell.ai Fabric site in Winnipeg, Manitoba. Um, they've got enterprise customers for that, unnamed.
68
00:12:41.248 --> 00:12:53.408
They also have the Bell.ai Fabric site in Merritt, British Columbia, [lips smack] uh, with- assigned with Cohere via Bell.ai Fabric. Again, these are cloud deployments for Hive. They don't own the facilities.
69
00:12:53.868 --> 00:13:07.348
They also have this new deal that with an unnamed tenant at the Merritt facility. Bigbot in Sweden is 32 megawatts utility, 25 megawatts critical IT. Um, they have a non-binding LOI, not definitive.
70
00:13:07.388 --> 00:13:09.988
Like we said, we might get an update there in September.
71
00:13:09.998 --> 00:13:23.348
But they also have a number of these other sites, like their Grand Falls, New Brunswick site, which is an a- a Bitcoin mining campus that they actually own, and they are thinking about converting it into a tier-three-plus AI center for 70 megawatts gross, 50 megawatts critical IT.
72
00:13:23.888 --> 00:13:32.608
I think, I don't know for sure, Charlie, but if you were going to see, um, if you were going to see any sort of retrofit where they do colocation,
73
00:13:33.668 --> 00:13:40.828
it's probably going to be in places like that Grand Falls, New Brunswick site. That's the route they're going with Bigbot in Sweden.
74
00:13:40.908 --> 00:13:49.408
That is the template, I think, that we will see with Bigbot and, and with other sites that they own outright as well, that they will probably go the colocation route.
75
00:13:49.808 --> 00:14:01.128
The other site in their portfolio that they are currently thinking about standing up a pretty big operation at is what they call the Greater Toronto Area AI Gigafactory.
76
00:14:01.608 --> 00:14:14.568
It's 320 megawatts of utility, 240 megawatts of critical IT, um, and it could h- house 100,000 GPUs at full build-out. There's currently no customer for that site, [lips smack]
77
00:14:14.588 --> 00:14:24.868
and they were asked questions about whether or not that will be colo or cloud. They did not explicitly say anything during the call about that.
78
00:14:25.278 --> 00:14:40.828
There will be three phases to this, 80 megawatts of critical IT each, and they said this could include a government colocation contract, hyperscaler colocation contract, multiple separate colo customers, maybe even some Hive-owned GPU cloud capacity.
79
00:14:40.848 --> 00:14:52.028
They're targeting integerization at the end of 2027 with GPUs online early 2028. Um, you know, it, it bears mentioning there that for a site that large,
80
00:14:53.128 --> 00:14:58.208
you would imagine that they're probably moving more towards the colo route. That will be very expensive.
81
00:14:58.748 --> 00:15:11.368
And they were asked about this directly of like how do they weigh colo versus, uh, versus cloud in terms of, uh, return on invested capital, and,
82
00:15:12.808 --> 00:15:21.188
uh, they, they really didn't, uh, th- uh, they really did not give an answer as to what exactly would happen at each individual site.
83
00:15:21.668 --> 00:15:33.828
But Killick did give some numbers that kind of illustrate how much you could expect from either, uh, either deployment with regards to how much the, how much revenue they could earn.
84
00:15:34.228 --> 00:15:41.088
For colo, roughly 1.8 million per megawatt per year. GPU cloud, 14 million per megawatt per year.
85
00:15:41.208 --> 00:15:54.348
The illustrative revenue multiple for colo is 10 X, for GPU cloud 3.5 X, with an implied, um, enterprise value density of $18 million per megawatt for colo and 49 to 50 million megawatts for GPU cloud.
86
00:15:54.848 --> 00:16:02.728
The takeaway here, obviously GPU cloud, as they were asked by analysts, this is the more lucrative one. Fact of the matter is, though, incredibly CapEx intensive.
87
00:16:03.148 --> 00:16:16.488
I would imagine that Hive for the larger megawatt sites is going to go for a more colo-centered strategy because there's less risk involved. There's less reward, but there's less risk, and it's easier to finance, so.
88
00:16:18.368 --> 00:16:33.258
So, uh, Hive up pretty big today. Actually, I would say it's the first, um, stock, Colin, the first equity that we've covered, where it's, uh, really soared on its earnings the first day. We saw CoreWeave.
89
00:16:33.268 --> 00:16:47.288
We saw some of the Neo clouds do it, but, uh, they floundered a bit, um, you know, during the intraday trading, whereas Hive, uh, earnings dropped Friday. Uh, uh, call was this morning. Now up 20 per- percent.
90
00:16:47.668 --> 00:16:59.008
Up 50% of the year. I'm looking at DI metrics here, dimetrics.ai. Link in the, in video description. Colin, can I say something without everybody getting mad? You can try. [laughs] Okay.
91
00:16:59.068 --> 00:17:12.528
Okay, we spent the whole time just covering their, their AI HPC deals. Um, but if we look at Hive's real, like, revenue-generating business today, it's really overweight their Paraguay Bitcoin mining operation.
92
00:17:12.568 --> 00:17:27.688
Now, they do use language such as looking forward to Boden, looking forward to these AI HPC deals. But, um, I kinda have a bit of an interesting insight here, which is, uh, so their, their Paraguay site in...
93
00:17:28.168 --> 00:17:42.348
How do you say the name? It's, uh- Iguazu... Igu- Iguazu. Iguazu. Um, yeah, they're the largest client of the Paraguayan, uh, power monopoly elec- uh, you know, grid monopoly down there, ANDE.
94
00:17:42.788 --> 00:17:48.448
I won't, you know, I won't explain what that... It's just the abbreviated version of the utility down there.
95
00:17:48.548 --> 00:18:03.688
Um, and, uh, their power contract goes up for renewal the end of 2027, and they're adding, I believe, another 100 megawatts this quarter of Bitcoin mining. Um, they have a ton of negotiating power with Paraguay.
96
00:18:05.080 --> 00:18:18.970
Uh, because of this. End of 2027 is very, very far away if we consider the, the speed and scale of build-outs as well as just the, the churn of markets. Where will the AI trade be in 2027?
97
00:18:19.400 --> 00:18:34.540
Where will the Bitcoin mining risk-reward trade be in 2027? We're looking at an interesting scenario where all of the North American miners are as quickly as they possibly can converting their Bitcoin mining sites to AI.
98
00:18:34.620 --> 00:18:47.440
If Hive is in a position where they're, they have added and are the largest client of one of the, like, lowest cost energy, uh, grids in the world in Paraguay, and they have 400 megawatts of Bitcoin mining,
99
00:18:48.540 --> 00:19:02.180
there could be a scenario if Bitcoin turns, everyone else in the world is locked into long-term duration HPC contracts, who's got hash rates online? Hive. So this is, uh, this is me trying to...
100
00:19:02.720 --> 00:19:08.500
I'm grasping for straws for bullish Bitcoin mining narrative. And this is what I can put out, so yeah. Yeah.
101
00:19:08.580 --> 00:19:17.940
And just going back to the Paraguay site, and I'll cap it after a few more comments with regards to the margins for Bitcoin mining versus HPC.
102
00:19:18.640 --> 00:19:25.780
If they could deploy this for, like, a powered shell colo for AI, I think they would've announced that already and tried to- I think so too, yeah.
103
00:19:26.120 --> 00:19:34.740
There, there has to be something either with investor risk of being in that region. Perhaps it's just not well-suited for this. You know, it's, it's, it's in the middle of the Amazon.
104
00:19:34.860 --> 00:19:47.180
How many of your clients, how many clients near Paraguay need inference? Because look, I mean, the closest major city is San Juan, I believe. Um, yeah, so. Uh, yeah, potentially. I'm not, I don't know the geography.
105
00:19:47.280 --> 00:19:56.620
I would've considered it, would've thought that it was the capital, but I could be wrong. Um- Yeah, but I mean, do you need 400 megawatts of, of inference, Hive needs in- But, you know- Yeah. Right. Right.
106
00:19:56.780 --> 00:20:02.300
I mean, that's a good... You know, but you... There are other things we could use it for, potentially, like training, right? Yeah. I just, I wonder how much...
107
00:20:02.320 --> 00:20:16.520
We saw this with the reason why Keel got out of the region is that, you know, quite frankly, most North American investors don't really want what they perceive as the political risk or other risk of being in, in Central or South America.
108
00:20:17.160 --> 00:20:22.719
They, they don't know how to price it properly. They don't understand the governments, they don't understand the local cultures.
109
00:20:22.800 --> 00:20:31.160
So I think that if they could, they would've done something more with this already, and the margins are kind of proving that out. I'll just leave on this point.
110
00:20:31.200 --> 00:20:44.860
Chris Brendler of Rosenblatt asked about future margin expectations. Hive didn't give any future expectations, um, but they did mention that, you know... So right now it's 44% on the HPC business line.
111
00:20:45.420 --> 00:21:02.560
There's no target, but Killick said that their A Series fleet of GPUs is running out for 40 cents per GPU hour right now, while the 844 H100s and H200s they have a- are driving 20 million of the current 35 million ARR for the contracts that are already active.
112
00:21:02.589 --> 00:21:06.570
That's not contracted. Um, potential ARR, that's what is running today.
113
00:21:07.320 --> 00:21:15.710
So when the Blackwells hit, you could imagine that the margins would potentially be even juicier, and the fact that they're getting that much out of these older models right now really tells you something.
114
00:21:16.280 --> 00:21:26.080
I will say spot price for A100 80 gigabyte models are, is, like, close to $1, so 40 cents would be lower than that, but that's, like, for spot- Yeah... um, and not for contracted, so- Right.
115
00:21:27.360 --> 00:21:36.720
We are gonna s- go, uh, and talk about Jensen's latest letter with, uh, Jay Patel, who's in the wings.
116
00:21:36.740 --> 00:21:44.420
We'll bring him up here, and we'll talk about the latest in AI HPC financing after a word from our sponsor, CleanSpark.
117
00:21:44.490 --> 00:21:59.440
[gentle music] 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.
118
00:22:01.260 --> 00:22:12.660
This is our proof of work, and we are setting the standard for what's next. Learn more about the intersection of energy and Bitcoin at cleanspark.com.
119
00:22:17.340 --> 00:22:32.289
All right, so let's bring on Jay Patel of Lagos Finance. Welcome, Jay, back to the show. How you guys doing? Fantastic. Good to see you. Um, there's always some-- Whenever Jensen drops a letter, I think we just...
120
00:22:32.870 --> 00:22:43.260
That's, that's what we've gotta talk about for the day. So [laughs] um- Well, at this point it's weekly, right? It's weekly, yeah. Uh, he went from zero to millions of followers on X overnight.
121
00:22:43.840 --> 00:22:56.880
And what he put out this morning was this, um, securing the infrastructure of intelligence, land power shell, the next critical resource for AI factories. He calls it LPS.
122
00:22:57.460 --> 00:23:06.640
Can you give me a, like, a just high-level overview of what he, of the points he lays out in this, uh, letter here, and then we can jump off from there? Yeah.
123
00:23:06.760 --> 00:23:17.180
Um, so you guys are probably actually a better fit to talk about the- [laughs]... the nuance of, of the actual deal than I am. But, um, you know, at a high level, I think the points he made were basically that...
124
00:23:17.980 --> 00:23:28.740
And, and I, I really think he's just referring to Anthropic and OpenAI, because I don't know about any other frontier labs with massive CapEx needs that don't have high credits.
125
00:23:29.260 --> 00:23:42.270
Um, but you know, uh, the, the deal they're working on with SB Energy is basically that, uh, NVIDIA will effectively provide a backstop for a portion of the leases, and then what would seem to be creatively worded, like minimum or
126
00:23:43.620 --> 00:23:51.360
maximum guaranteed capacity and minimum residual value in the case that OpenAI kind of backs out as a, as a tenant on the lease.
127
00:23:51.480 --> 00:24:02.900
Um, but you know, the case he made is basically that in the vast majority of their, you know, the vast majority of their business won't change, but there are, you know, in his words, limited circumstances.
128
00:24:03.240 --> 00:24:09.756
I'm thinking that, you know, OpenAI and Anthropic are probably the- Two largest consumers of compute are definitely up there.
129
00:24:10.376 --> 00:24:24.616
Um, that there are certain circumstances with the Frontier labs where their compute needs outpace their credit and balance sheet capacity, and in those situations, Nvidia might do more of these deals, which he says is clearly not circular financing.
130
00:24:24.676 --> 00:24:30.756
We can all, you know, uh, w-you know, make our own circular- We can debate the merits of that [laughs] Yeah.
131
00:24:30.916 --> 00:24:55.116
Um, but I, I did think that this was interesting just because it kind of goes back to a point that we talked about, it's gotta be almost a year ago, where, um, you know, I, I don't know, Charlie, if it was, if it was you, but, you know, we brought up the idea that, like, Nvidia has somewhat of a moat against other chip providers in the sense that they have capital, right?
132
00:24:55.216 --> 00:25:00.836
Uh, whether it's their balance sheet or the ability to backstop with their equity or credibility with investors.
133
00:25:00.936 --> 00:25:25.356
Um, and I do think that these kind of deals, along with the stuff we saw last week, um, probably reinforce the kind of Nvidia Lindy, if, if you will, in the sense that it becomes cheaper to finance the build-out of data centers that have Nvidia chips versus, you know, Cerebras or some of the dedicated TPUs or, or, or other inference-dedicated chips.
134
00:25:25.976 --> 00:25:40.546
Um, um, so I mean, I-- look, I think it's, it's just another sign that the capital that this space needs is way beyond anything we've seen, and obviously I think this is just the beginning of more and more creative structures.
135
00:25:42.696 --> 00:25:58.056
So, uh, his whole thesis, 'cause, like, Nvidia's had a, had a technical moat with the CUDA standard for a long time, and that's been instrumental to why they're in the position they currently are. Um, but, uh, does...
136
00:25:59.396 --> 00:26:08.436
You know, you say that capital is now this moat, and to me it's a little unclear what form that capital takes, because we have the GPU hour.
137
00:26:08.456 --> 00:26:17.255
We've been talking about it with various, like, compute options, uh, market, you know, makers and, uh, products.
138
00:26:17.296 --> 00:26:33.376
And, uh, to me it's, like, really difficult to identify what that asset is that is ba- that, that is the cap- the capital asset here. Um, I'm not really sure how I'm asking the right question, but, like, um,
139
00:26:34.496 --> 00:26:40.556
it-- there, like, what exactly are, like, the capital intent? What, w- where does the money come from?
140
00:26:40.656 --> 00:26:51.916
[laughs] What, is it gonna be GPU hours that we finance against, or is it gonna be specifically just Nvidia hardware, uh, that we're financing against? So I think [sighs]
141
00:26:51.976 --> 00:27:04.176
I'm, I'm wary to think that there's going to be just a broad GPU hours compute market that has a big financing structure behind it anytime soon, because of the fact that you have standardization.
142
00:27:04.206 --> 00:27:13.756
And then the other thing, you know, I think last week's news kind of reiterated this point, is that there's plenty of capital that wants to lend in this space. You saw Brookfield, and KKR- Yeah...
143
00:27:13.766 --> 00:27:24.696
and Goldman, uh, you know, I think there's three other names on that list last week. Um, but obviously if you're a, if you're a lender into this space, you have limited upside, right?
144
00:27:25.076 --> 00:27:28.756
Um, uh, you know, we, we were all familiar with the Bitcoin mining space last cycle.
145
00:27:29.136 --> 00:27:40.076
Um, if, if there's any sort of, like, cyclical space that you're lending into and you don't have upside, you want really strong credit protections. Um, and I think what Nvidia basically helps with here is two things.
146
00:27:40.196 --> 00:27:52.456
One, you know, last week, uh, you know, my understanding is, like, there's certain reference designs and certain implementation, you know, certain requirements for the data centers for them to provide this kind of, quote unquote, backstop.
147
00:27:52.905 --> 00:28:04.276
And then the other thing is that, um, you know, the question in all of these cases, and you know, you mentioned, you know, A100 spot prices are up, you know, maybe the depreciation of these assets isn't what we thought it was and they have a longer useful life.
148
00:28:05.256 --> 00:28:12.796
But I think over the medium to long term, it's hard to see GPUs not depreciating.
149
00:28:12.836 --> 00:28:23.416
Like, you know, we might be in this short-term scarcity where prices go up, but I don't think anyone will tell you that in seven years' time an H100 is gonna cost more than it does today. That just, like, makes no sense.
150
00:28:23.535 --> 00:28:42.216
Um, and so, you know, I think from a, from a investor's perspective, if you're, you know, KKR or Goldman or, you know, whoever else trying to lend in this space, having Nvidia be the one providing the kind of minimum residual valance-- uh, value guarantee versus someone else helps in two way, ways.
151
00:28:42.316 --> 00:28:52.736
One is I think there is the sense that Nvidia has a better idea of what's going to happen over the next six or seven years with regards to GPU development progress than anyone else, because they make the GPUs.
152
00:28:52.776 --> 00:29:04.976
And so if they say that they expect this to have a certain useful life and they're willing to guarantee a certain value for it at a certain point in time, you know, I think y- their word is worth more than, you know, some third party.
153
00:29:04.996 --> 00:29:12.936
And then the other is obviously the beneficiary from the GPU sales and the networking and everything else from day one is them, right?
154
00:29:12.976 --> 00:29:32.456
So they have, you know, this massive cashflow engine, and so they can use a part of, part of that, you know, that's left on their balance sheet against, you know, potential future liabilities versus, you know, smaller players, you know, their credit guarantee is not as meaningful if you don't think that they're going to produce, you know, hundreds of billions in cashflow over the next couple years.
155
00:29:34.316 --> 00:29:46.496
I mean, do you think that this means then that whatever Nvidia does with these backstops and what the risk they're willing to take will kind of set a standard vis-à-vis that look at how long these GPUs actually have useful life?
156
00:29:48.156 --> 00:29:49.876
Yeah, my sense is that there...
157
00:29:51.126 --> 00:30:05.256
[sighs] You know, I'd, I'd say the closest analog, and I, and I don't wanna be, like, too conspiracy theorist here, is Nvidia kind of wants to become the Fannie Mae or some equivalent of GPU financing, right?
158
00:30:05.816 --> 00:30:09.706
They're not... And, and actually this may be an even more balance sheet-light approach. Um-
159
00:30:11.084 --> 00:30:19.803
I heard it somewhere, but basically, you know, someone was saying that maybe they're even more like the Federal Reserve, where they're just setting some policy and some standards, right? "Hey, these are conforming loans.
160
00:30:20.154 --> 00:30:25.164
These are data centers with the reference design and implementation that we're willing to provide a backstop to."
161
00:30:25.204 --> 00:30:32.704
And then the private market, and I know NVIDIA's part of the private market, but, you know, in this analogy, NVIDIA, you know, can be the Federal Reserve or whatever you want it to be.
162
00:30:32.804 --> 00:30:40.284
Um, you know, all these asset managers they parked with- partnered with make the loans, right? NVIDIA's not actually going off balance sheet for any of this.
163
00:30:40.504 --> 00:30:48.764
But essentially, like, this only works because everyone has the understanding that if things go wrong, NVIDIA will come and clean up the mess, and they have the capital to do so.
164
00:30:49.464 --> 00:31:00.904
Um, now, I, you know, I guess o- one kind of hole in all of this is how does that, you know, like, what, what does their existing customer base think of this, right?
165
00:31:00.984 --> 00:31:08.444
You know, they've always had this tension between, you know, helping the Neo cloud scale versus satisfying hyperscalers who are longtime customers.
166
00:31:08.464 --> 00:31:15.344
And now, you know, with this OpenAI deal, you know, I, I'd say that's maybe even somewhat competitive to some of the Neo cloud ambitions. Um,
167
00:31:16.264 --> 00:31:22.624
uh, but Jensen's pretty good at politics, so I feel like there's some balance he'll find to make everyone happy.
168
00:31:25.464 --> 00:31:40.184
Um, I was looking at, uh, from the Wall Street Journal, I believe, that the, this, uh, like what exactly is being secured. And, um, a lot of it is not GPUs.
169
00:31:41.144 --> 00:31:53.864
Um, and which is real interesting because it, if, if NVIDIA is now backstopping, like, land and power, then that seems like a pretty different,
170
00:31:55.504 --> 00:32:06.484
that seems like a pretty different risk profile because now they're not just producing silicon. They're, now their balance sheet is now secured by, like, real assets which have a very different depreciation life cycle.
171
00:32:07.304 --> 00:32:12.024
Thoughts or takes on this. Yeah. I, I'd say, like, the, you know,
172
00:32:13.124 --> 00:32:22.913
a year or two ago, you might have thought, oh, NVIDIA will just be able to grow compute demand by financing demand and, like, capital on the purchaser side was the pure constraint.
173
00:32:23.424 --> 00:32:38.584
But clearly, you know, as you guys cover basically every week on this show, even if I had a delivery of, you know, a 100 giga- or 100 megawatts of GB200s or whatever, like, there's, you know, a multiyear backlog to try to find somewhere to actually power those.
174
00:32:38.924 --> 00:32:50.524
Um, and so I think from NVIDIA's perspective, it's like they gotta move further down the stack because the ability to buy their chips, while it is part of the limitation on demand, that's not the ultimate limitation.
175
00:32:50.884 --> 00:33:01.604
Um, and now they actually need to help their customers secure, you know, land, power, and shells so that they can plug in the NVIDIA chips that they're gonna buy. Um, but I wasn't, I was kind of...
176
00:33:02.204 --> 00:33:11.844
I mean, maybe I shouldn't be shocked. I, I'd be interested to hear what you guys think, that the OpenAI deal here was 20, a 20-year lease. Um, so it seems like NVIDIA's,
177
00:33:13.624 --> 00:33:24.464
I, I, I don't know, like, forecasting pretty far out into the future the useful life of the non-silicon assets. And I mean, you know, land and power definitely depreciate a lot slower than GPUs, but yeah.
178
00:33:25.284 --> 00:33:35.894
My take is absolutely that's a great, that's a great play. I don't wanna be trying to backst- I don't, I, I don't wanna guarantee that OpenAI's gonna be around in 20 years. I, uh, the, those power and land assets
179
00:33:36.964 --> 00:33:43.564
absolutely will. Um, I would wanna, you know, we'd, we'd floated maybe talking about, like, yield varying stablecoins. We're gonna table that.
180
00:33:44.084 --> 00:33:50.204
I think wanna throw one kinda last question to you, which is one thing that I was kinda confused at in, like, reviewing this story.
181
00:33:50.264 --> 00:33:57.344
We've covered it over the past few weeks, is that we've had, like, three different backstop numbers. I don't know if this is, like, a big deal or not.
182
00:33:57.724 --> 00:34:10.344
There was initial, like, 500 billion, and then I think NVIDIA came out and clarified that it was 250. And then on Friday, the Wall S- Wall Street Journal said it was, like, 150. Um, I don't know.
183
00:34:10.424 --> 00:34:22.644
Like, to me, I see a bunch of different numbers, which are hundreds of billions of dollars different. Um, should I be, like, concerned about this, or is this just, uh, a messaging thing that I'm not understanding?
184
00:34:23.444 --> 00:34:31.883
I think it's, like, I, I think it's broadly a messaging thing, but I also think it's one of those, like, um, depending on who the audience is, you wanna say a different number.
185
00:34:31.943 --> 00:34:43.604
So my understanding is the big number, the 500 billion, is the total, you know, for this initial program, the total size of the financing that the partners can provide, right?
186
00:34:43.634 --> 00:34:49.684
And so that's not NVIDIA putting up $500 billion of their own balance sheet or $500 billion of compute or whatever.
187
00:34:50.164 --> 00:34:57.364
It's just, hey, we've worked with these six asset managers to stand up some sort of program that can scale to $500 billion in financing,
188
00:34:58.644 --> 00:35:11.714
um, which I think NVIDIA has an incentive on the press side of things, on the initial side, to say, like, as big a number as possible because, you know, $500 billion, some decent portion of that is going to come to, to NVIDIA and NVIDIA equity holders.
189
00:35:12.144 --> 00:35:19.224
I think the clarification later in the week, you know, which I think people were getting to, is like, okay, if there's some backstop, what is the risk that NVIDIA's taking on?
190
00:35:19.284 --> 00:35:24.584
And I think that's what they're trying to minimize the appearance of, right? It's like, how much balance sheet risk is NVIDIA actually taking?
191
00:35:25.204 --> 00:35:37.844
Um, and my understanding for those initial deals is that most of the backstop comes in the form of, like, a minimum residual value guarantee. Um, and so it would be significantly less than,
192
00:35:38.984 --> 00:35:46.844
you know, the initial loan amount. You know, it's really what is the GPU worth in the out years for any financing deals, you know, year five, six, seven.
193
00:35:46.964 --> 00:36:04.584
Um, so I think the risk to NVIDIA is probably closer to the 100, 150, but the total size is maybe closer to the 500. All right. Jay, thank you so much. Illuminating the half a trillion dollar deal up in Ohio.
194
00:36:05.164 --> 00:36:16.944
Quite an insane deal. Thank you so much for your time. Thank you. We'll catch you next week again, and maybe we'll talk, we'll, we'll get around to a state of DeFi conversation, so. [laughs] Yeah, hopefully.
195
00:36:16.984 --> 00:36:28.764
Take care, guys. Cool. Thanks, Jay. All right, we have Dan Hilary of UTXO Management in the wings. We'll bring him on up here right after a word from our sponsor, Luxor.
196
00:36:30.724 --> 00:36:41.384
[whooshing sound] This episode of Blockspace Live is brought to you by Luxor's Commander, Bitcoin miner management software for enterprise operations.
197
00:36:41.904 --> 00:36:47.484
Luxor's Commander gives you real-time fleet monitoring, bulk remote commands across your fleet, and Intelligent Miner.
198
00:36:48.084 --> 00:36:55.433
That's an automated profitability engine that runs every five minutes and tests your fleet's power settings against live energy and hash rate markets.
199
00:36:56.064 --> 00:37:02.484
In fact, ERCOT backtests show 10% improved profitability with intelligent mining versus old-fashioned binary mining.
200
00:37:03.184 --> 00:37:11.444
Commander Pro is roughly half the cost of competition, $100 per megawatt, or a 25 basis point pool fee adder, but you can also try it free for 60 days.
201
00:37:11.524 --> 00:37:24.984
So if you'd like to learn more, go to luxor.tech/commander to get started. All right, we're gonna talk a different kind of credit, the digital kind, with Dan Hilary. I'll bring him up, him up here.
202
00:37:25.964 --> 00:37:38.264
Dan, welcome to the show. Uh, I think you gotta unmute yourself on the Hey, guys. Great to be here. Thanks for having me on. Yeah. Super excited for this conversation. Thank you for joining, man. Yeah, cool. Yeah.
203
00:37:38.844 --> 00:37:50.984
So, uh, I believe recently added VP of credit, UTXO Management. Uh, great, uh, team, friends of ours. Um, and we're gonna talk some digital credit. So, uh,
204
00:37:52.144 --> 00:37:59.764
I've been noti- I've been seeing Michael Saylor's been selling the Bitcoin little bit below the cost basis to defend the preferred stock.
205
00:38:00.764 --> 00:38:17.584
Um, and, uh, I wasn't paying super close attention, but it looks like Saylor also, uh, flagged, uh, uh, into the first week in September as the date he wants Stretch back to $100. So,
206
00:38:19.104 --> 00:38:25.144
what's going on there? Do you think we'll hit $100, and do you think it actually matters if we hit $100?
207
00:38:26.104 --> 00:38:36.364
Look, I mean, I think, you know, this is where I focus all my time on these digital credit instruments, Stretch being the largest by market capitalization of all the preferred equities, including, you know, the fixed rate preferred equities.
208
00:38:36.404 --> 00:38:45.964
Even larger now than the convertible debt, which was, is kind of the old instrument by which strategy created leverage in their balance sheet. Now it's referred to as amplification.
209
00:38:46.604 --> 00:38:53.504
And obviously, you know, Stretch traded up within about three months, uh, from its IPO price of $85.
210
00:38:53.704 --> 00:39:04.584
Once it was IPO'd, it traded down to a low of 80, roughly, and then traded up to 100, and stayed around $100 for a, for a long clip of time, about, you know, just shy of nine months. And [clears throat]
211
00:39:05.024 --> 00:39:13.144
it, it's designed to trade around par, and there's a bunch of mechanisms now that Strategy uses to maintain that par pricing.
212
00:39:13.244 --> 00:39:20.904
And over the, you know, the first nine months of Stretch's history, it was this variable rate monthly dividend.
213
00:39:20.944 --> 00:39:32.794
So for anyone who's not familiar, they're able to adjust the dividend rate on STRC on a month-to-month basis based on the average trading price of the prior 30 days.
214
00:39:33.284 --> 00:39:39.464
So they've been ratcheting up the dividend, it was staying around par, and then it traded off, and that was a large function of the cash reserve.
215
00:39:39.924 --> 00:39:47.944
And, you know, obviously, the duty to shareholders, there's not a fiduciary obligation for it to trade at $100.
216
00:39:48.364 --> 00:39:53.824
However, management's put, put forth guidance such that they want to, they're targeting that $100 price point.
217
00:39:54.444 --> 00:40:04.844
So what they've done recently after Stretch traded off was rebuild the cash reserve to support dividend payments and also increase the execution of buybacks.
218
00:40:05.084 --> 00:40:33.184
And I think people are really discounting the effect that buybacks will have on price over a long period of time, and that's the m- that's the one thing, you know, we're really looking closely at, are how is the Stretch price affected by not only the dividend changes, which right now the dividend's at 12% annual based on $100 par and it's staying there, but how, how does the cash reserve size relative to the total Bitcoin assets, the senior debt, and then the execution of Stretch buybacks affect its trading price?
219
00:40:33.244 --> 00:40:46.114
And, you know, my three-second summary on that is cash reserve puts a floor in the Stretch price, and buybacks are more of a micro adjustment that are able to get it up to its, its desired trading range.
220
00:40:46.204 --> 00:41:00.084
Much like the sale of STRC could, and has, pushed the price down over short periods of time. So, uh, Saylor wants to get it back to 100.
221
00:41:00.584 --> 00:41:11.664
Do you think, like, if he gets close, you know, what is it, September 8th, September 9th, I- yeah, September 8th, do you think it, like, then that's sufficient? Will people... Will, will the product be...
222
00:41:11.944 --> 00:41:21.504
Will, will, will people, will people be happy? Uh, I don't know. And at what rate do you think it probably trades at, you know, on September 8th? Yeah.
223
00:41:22.004 --> 00:41:37.464
I think one chart for viewers to put into their trading view to look at is the price of STRC with dividends reinvested since inception. And so that chart has hit all-time highs over the past week or, or two weeks.
224
00:41:38.004 --> 00:41:47.244
So that's really the cumulative return for STRC investors over time, and that's really, really important. And so I think right now the rate's 12.
225
00:41:47.404 --> 00:41:51.614
I actually don't expect the rate to decrease in any near-term future, right? Like,
226
00:41:53.144 --> 00:42:04.794
you know, we can argue about credit spreads, which is the difference between the risk-free interest rate and a certain fixed income instrument, but Strategy's trying to grow the market capitalization of Stretch.
227
00:42:05.124 --> 00:42:14.724
To do that, they have to have a spread or a yield that is significantly higher than other instruments in the market. So I expect it to stay kind of where it is.
228
00:42:15.344 --> 00:42:22.323
Then to your, to your question of, like, will it get to 100, the stated trading range is 99 to 100. Right now we're at, like, 95.
229
00:42:24.244 --> 00:42:32.764
Clearly the buybacks are what are being used on a weekly basis to get it back to that hundred dollar price point. And I think long term it will get there.
230
00:42:32.884 --> 00:42:39.324
Um, but cl- you know, the volatility spooked a lot of people and, you know, I like to be transparent in my analysis.
231
00:42:39.384 --> 00:42:52.864
I'm definitely a Saylor fan, but having a 12% yielding instrument trade from a hundred dollars to $73 is very nerve-wracking, and I think there will be a lot of, kind of, rotation of the instrument.
232
00:42:53.704 --> 00:43:01.134
And, you know, that's kinda what... The fund I'm running at UTXO is designed to buffer that for different types of investor classes, right?
233
00:43:01.184 --> 00:43:04.664
There's a lot of people who want all this yield and believe the dividends will get paid.
234
00:43:05.104 --> 00:43:13.144
There's a lot of retirees, credit investors, and just, you know, institutional balance sheets that can't weather that volatility, and they want a high yield.
235
00:43:13.433 --> 00:43:23.384
And so I think there's gonna be products built on top of Stretch that tranche this, sort of, yield or volatility profile and make it more attractive to different investors.
236
00:43:23.464 --> 00:43:26.764
Right now, the volatility's high for a lot of, a lot of people.
237
00:43:28.004 --> 00:43:39.444
I have a, a ki- I think maybe a challenging question because, again, I'm not really, I don't really study digital credit, but, um, you know, you mentioned that there, the, uh, reinvestment strategy back into Stretch.
238
00:43:40.384 --> 00:43:43.744
Mm-hmm. Uh, that, that strategy has hit all-time highs, but, um,
239
00:43:44.824 --> 00:43:55.364
I, as a user, thought that this was primarily, like, a yield instrument so I could, like, put a bunch of money in and have it fund my, you know, my endeavors or retirement or whatever.
240
00:43:56.084 --> 00:44:04.264
Um, maybe are there, like, different strategies as the, as this, as these preferreds get up and running and more stable over time?
241
00:44:04.384 --> 00:44:18.784
Or, uh, or does this challenge, like, this idea that this is a, a simple, less volatile instrument that investors can be lured into? Um, yeah, how do you tackle this? Yeah, totally.
242
00:44:18.904 --> 00:44:30.124
I think that's a great question because, you know, there's obviously other preferreds issued by Strategy, and they have these fixed rate yields. So Strife, STRF, is senior to Stretch.
243
00:44:30.724 --> 00:44:36.584
It's, it's heavily over-collateralized and yields roughly 10% annually f- forever, as a perpetual preferred.
244
00:44:37.384 --> 00:44:42.844
That would be, you know, if you're an investor and you're looking for an investment and you're not trying to speculate on the principal value of that,
245
00:44:43.924 --> 00:44:57.244
I would look at that as a great way to lock in the amount of money every year I receive from the instrument and use that as, you know, dividend proceeds to fund my life. STRC is the same sort of idea.
246
00:44:57.484 --> 00:45:07.204
You have a fixed yield. If the principal of STRC falls, so if STRC falls from 100 to 90, you're still getting your 12% annualized return at par.
247
00:45:07.684 --> 00:45:14.804
So your dividend distributions every month or twice a month didn't decrease, right? So that, that's an important thing for people to understand.
248
00:45:15.364 --> 00:45:24.104
But, you know, with regards to the principal, I think the idea of Stretch was and, and continues to be that it's a shorter duration instrument.
249
00:45:24.124 --> 00:45:34.004
You can buy it at, uh, at 99, you can sell it at 99 a few months later, and that's the biggest sort of differentiator between Stretch and the other preferred instruments, and that's the value proposition.
250
00:45:34.094 --> 00:45:40.724
So at the moment, the buybacks are attempting to recover that sort of ethos in instrument.
251
00:45:40.764 --> 00:45:54.304
It's supposed to be where you can park your capital short term or where you can park your capital and not experience the principal fluctuations that you may experience in a traditional fixed income product, in a long duration bond, or any other preferred equities that Strategy issues.
252
00:45:55.504 --> 00:46:05.444
Dan, I have, I have a quick question with regards to one of the other, I should say, looming financial considerations for Strategy, specifically as it relates to the converts.
253
00:46:06.124 --> 00:46:09.424
So it seems like the fires for Stretch are somewhat put out at this point.
254
00:46:09.944 --> 00:46:23.773
I think that the difficulties with Stretch kind of belie a larger, uh, financing wall though, or rather a, um, a liability wall for Strategy, 'cause they've got the 2028 converts, and the holder put date is September 15th, 2027.
255
00:46:24.524 --> 00:46:39.944
Um, the strike on that is $183 per share, $183.20 per share. And then they've got a v... You know, even greater wall in 2028. They've got a $2 billion convert with a put date of March 1st, 2028.
256
00:46:41.304 --> 00:46:54.044
Um, they've got, uh, 1.5 billion with a put of June 1st, 2028. Um, and then they've got September 15th, uh, 2028, um, that two issues on the same day.
257
00:46:54.124 --> 00:47:08.024
They've got, um, roughly 1.4 billion in combined cash put exposure, and the strike prices for all of these are, um, higher than where s- Strategy is today. One of the highest is, uh, for the 2029 notes is 672.
258
00:47:08.804 --> 00:47:20.804
Uh, I'm just curious, kind of a multi-part question here. Do you think that they've done enough so far this year to, uh, bulwark themselves against those potential puts?
259
00:47:21.384 --> 00:47:33.144
And how would, how do you, as you look at trading Strategy, price that risk when you evaluate the company and you evaluate whether or not you wanna, uh, deploy more capital or wait to see what happens with Bitcoin and Strategy's price?
260
00:47:34.244 --> 00:47:43.624
Yeah, definitely. And I'll, I'll approach that question from a credit perspective, right? A- about two months ago, Strategy was in a tougher position.
261
00:47:43.724 --> 00:47:51.744
They were about, they were down to, like, 700 million or sub a billion dollars of cash reserve because they used part of the cash reserve to pay off the converts.
262
00:47:52.184 --> 00:48:09.164
Bitcoin was tanking, and there was this kind of looming, what you're saying is effectively 30% dilution of the common relative to total market capitalization that would have to, to go through to finance the, the payment of the converts, the redemption of the converts, if they didn't convert.
263
00:48:09.984 --> 00:48:14.474
Now, Strategy has, I think it's about a 4.6 or $4.8 billion cash reserve,
264
00:48:15.444 --> 00:48:25.984
and there's 6.7 billion of total outstanding convertible notes, uh, due, you know, one year before the actual expiration because of the put dates if they don't convert. So-
265
00:48:27.412 --> 00:48:37.822
I personally, what Strategy is doing is they're building their cash reserve, and their cash reserve is ser- serving two purposes, right? One, it provides financing or payment for the dibs on the preferreds.
266
00:48:38.332 --> 00:48:46.362
Two, it serves as cash that doesn't need to be raised at some point in the future to pay off the converts if they don't convert, right?
267
00:48:47.052 --> 00:48:57.172
Right now, the board approval is 18 months minimum of dividend coverage for the preferreds, so any cash reserve above and beyond that can be used to finance or to pay off those converts.
268
00:48:57.792 --> 00:49:06.832
So, you know, in terms of a credit profile, I think they pose, pose a very, very small threat to the preferreds. Like, those preferred dividends will continue to get paid.
269
00:49:07.252 --> 00:49:14.492
There's enough cash, and they've demonstrated their ability to raise enough cash to pay off the converts when they come due. Why are they not paying them off early?
270
00:49:14.852 --> 00:49:32.912
Well, simply, Strategy is a leveraged Bitcoin equity, so if they were to go and, and s- pay off the converts today, their amplification or the leverage in the capital structure would decrease, and they'd have to either raise more Stretch or raise more of the other preferred securities or issue more debt to reup that sort of leverage.
271
00:49:33.212 --> 00:49:44.922
So I think what they're doing is reducing the risk of converts not converting by hoarding cash in order to finance them and pay them off or even refinance them into the future.
272
00:49:44.992 --> 00:49:55.032
And two, they're keeping their Bitcoin upside or the leverage exposure to Bitcoin by not paying them off early. So I think it's actually a, a wise strategy, and I don't see a lot of credit risk.
273
00:49:55.422 --> 00:50:13.872
The, the biggest risk has actually already happened, and that was the MSTR equity sales that then get piled into cash or into, into the cash reserve, which for a lot of diehard Bitcoiners or people looking for levered Bitcoin exposure, that would be, you know, non-favorable, uh, for their positions.
274
00:50:15.952 --> 00:50:27.862
Yeah, MSTR holders, uh, maybe not super happy over the past few months. Okay, so I wanna zoom out really far, Dan, and just have you kinda, like,
275
00:50:29.152 --> 00:50:39.402
um, assess the, the success of this, like, the overall strategy. Like, if I can go back, and I remember, like, Saylor saying,
276
00:50:40.572 --> 00:50:53.921
uh, "We're going after, like, the fixed income markets," this massive amount of money that a lot of Bitcoiners who don't study finance didn't really realize, like, this, this, this, like, untapped market. Mm-hmm.
277
00:50:54.012 --> 00:51:05.421
How effective do you think this strategy, and then the, kinda the other types of pref- uh, the, the wi- the, the general universe of, like, preferred equities, but, like, in particular, is Saylor's strategy working?
278
00:51:05.512 --> 00:51:10.352
Um, is it, you know, qualify it a bit, is it working somewhat here, somewhat not here?
279
00:51:10.512 --> 00:51:18.652
And, like, are they on a path to, like, actually creating, um, something which can suck in this, like, black hole of fixed income market?
280
00:51:19.812 --> 00:51:29.612
Yeah, I think that's a great, great point, and I mean, it's difficult because Strategy's pioneering this sort of preferred equity market in a way that it really hasn't been pioneered.
281
00:51:29.652 --> 00:51:38.972
Strategy's STRC trades roughly $400 million of volume a day. That's more than the entire preferred equity market combined as it exists.
282
00:51:39.572 --> 00:51:47.472
So th- this is a really, really small, and historically the preferred market's been a, a terrible underperformer relative to both common equity and debt instruments.
283
00:51:47.852 --> 00:52:00.891
So they're breaking into the preferred market with an eye into fixed income. Preferreds give them infinite optionality and long duration credit, which they want to finance the Bitcoin. Bitcoin's a forever duration asset.
284
00:52:00.992 --> 00:52:13.812
So th- that's the reason they tapped into that market. I think there needs to be a lot more emphasis on creating structures around STRC that are applicable to fixed income mandates, right?
285
00:52:13.872 --> 00:52:31.972
Like, most of the debt, the US sovereign debt, uh, corporate debt, it's owned by pension funds, endowments, large fixed income funds, and they have mandates where they can only buy certain instruments, certain types of credit by certain issuers, and S&P Fi- S&P committee rating is a really big part of that decision-making process.
286
00:52:32.532 --> 00:52:43.602
So until we get strong ratings on Strategy's preferred equities and there are wrappers on top of those preferred equities that allow them to fix, fit institutional mandates,
287
00:52:44.792 --> 00:52:54.432
we're still kind of a nascent, um, asset class. So are we, are we attacking the fixed income, the, the debt markets yet? No.
288
00:52:55.312 --> 00:53:06.612
Are we making the base layer of a Bitcoin-backed credit instrument that will be used to target the fixed income markets? I believe so, but, you know, it's a, it's a decade-long process in my opinion.
289
00:53:09.432 --> 00:53:17.072
Um, yeah, do you wanna go, Colin? I don't really have anything else. Uh, yeah, I guess my last question is, like, um,
290
00:53:18.372 --> 00:53:34.952
so I'm looking at that STRC, uh, yields, what, 600-something bips over high-yield bonds right now. Do you think that that's enough? Uh, do you think that's compelling, uh, to the alternatives?
291
00:53:35.532 --> 00:53:43.732
Um, and maybe, uh, where do you, where do you think that goes, just to kinda round out this conversation? Yeah.
292
00:53:43.792 --> 00:53:49.472
I think that's the question Strategy had to answer when Stretch traded off and the yield blew out to, like, 15%, right?
293
00:53:49.652 --> 00:54:05.812
S- with STRC trading at $73 and the yield, the effective yield, cash on cash, is much higher than the 12% stated, 'cause that's at $100 par. Clearly the increase in yield wasn't increasing the principal price of STRC.
294
00:54:06.092 --> 00:54:13.092
So the market's pricing STRC risk more so based on balance sheet construction and management decisions regarding the cash reserve.
295
00:54:13.792 --> 00:54:21.752
So I actually don't think, and it, and they, they've clearly taken this to heart, that Strategy should continue increasing the divid- dividends on Stretch.
296
00:54:22.252 --> 00:54:29.632
So my long answer to your question is, yeah, I think 600 bips is plenty high relative to high-yield corporate debt.
297
00:54:30.220 --> 00:54:38.680
Um, and the most important step will be consistency of the cash reserve management and execution of buybacks to reduce the volatility in STRC.
298
00:54:38.710 --> 00:54:45.500
'Cause STRC has a much different mandate than a traditional fixed income bond. All right? The principle, the principle never gets returned [clears throat]
299
00:54:45.580 --> 00:54:54.740
to holders, so a buyback is effectively a strategy using their balance sheet to put a put or a floor in the STRC price.
300
00:54:54.900 --> 00:54:59.680
If the div- if the principle never gets returned over a saturation of time, it's very difficult to value these instruments.
301
00:55:00.360 --> 00:55:14.800
By executing open market transactions that are financed by the Bitcoin on the balance sheet and not by new capital raise, the market and, you know, sophisticated participants are starting to see, oh, you know, Strategy can use their existing assets to support the price of this instrument.
302
00:55:15.040 --> 00:55:28.000
That's kind of similar, kind of similar to a debt instrument where ultimately the principle's returned in the form of a, you know, the, the note coming due. Dan, thank you so much for your time and your insight.
303
00:55:28.340 --> 00:55:38.200
Um, as I hope, I hope Stretch continues to get really interesting, 'cause it's more fun to talk about when things happen to it. So- [laughs]... cheers. But thank you so much for your time.
304
00:55:38.360 --> 00:55:42.540
We'll, uh, we'll catch you around. Thanks, Dan. Bye, guys. Thank you. Great talking with you.
305
00:55:45.640 --> 00:55:56.980
The STRC, uh, rips, uh, are surprisingly, uh, popular on YouTube whenever we put them out, so excited to see, uh, how the people like that. If you have comments,
306
00:55:57.880 --> 00:56:13.860
drop them in the video replies below, positive and/or negative. So on that note, we're gonna go to gas fire and new s- uh, small modular reactors right after a word from our sponsor, Ligos.
307
00:56:18.560 --> 00:56:30.240
[static] Bitcoin took a little bit of a dive over the weekend, and if you're using a shady Bitcoin-backed lender, you might be asking yourself, do they still have my coins? Well, I'm gonna tell y'all something.
308
00:56:30.560 --> 00:56:42.500
If you're using Ligos Finance, our preferred Bitcoin-backed lender here at Blockspace, you don't have to ask that question, because you always control the keys to your Bitcoin when you are using them for a Bitcoin-backed loan.
309
00:56:43.120 --> 00:56:53.360
That's right, you heard that correctly. With Ligos Finance, they use Bitcoin native smart contracts to make sure you always control the Bitcoin that you are collat- using for collateral for your loan.
310
00:56:53.980 --> 00:57:04.820
With Ligos Finance, there's no bridging, no wrapping, no rehypothecation, no funny business, just competitive rates as low as 8.5 to 10% APR. Go to ligos.finance to learn more.
311
00:57:06.620 --> 00:57:17.840
All right, let's bring it on home with a little nuke, baby, nuke. Nukes. Guess kind, not the warfare kind- The good kind of, the good kind of nukes. The, the white pill... the good nuke. The happy math.
312
00:57:18.880 --> 00:57:30.880
This headline coming at us from Data Center Dynamics, Blue Energy, GE Vernova, Hitachi advance 2.5 gigawatt gas plus nuclear plant in Texas to power up to 1.5 gigawatt data center.
313
00:57:31.800 --> 00:57:46.740
Intended to supply one gigawatt of power to the Crusoe data center by 2030. Now, this is not a new project. I just wanna get that out of the way in front. This was announced back in, let me see, May,
314
00:57:47.880 --> 00:58:01.040
and the most recent development is that, um, Blue Energy and GE Vernova Hitachi have signed the next stage agreement to perform engineering, licensing, and safety analysis work for the site.
315
00:58:01.420 --> 00:58:15.500
The project still requires a final investment decision in 2027. There is a lot to unpack here, Charlie. So this will be in the Port of Victoria area, Texas. Crusoe's campus is in Calhoun County, Texas.
316
00:58:16.320 --> 00:58:26.980
Crusoe, the data center developer, 1,600 acre site. The power developer is Blue Energy. Ver- GE Vernova are supplying the gas equipment,
317
00:58:28.060 --> 00:58:38.920
and also, uh, going to supply the nuclear technology as well with BWRX-300 small modular reactors. The nuclear build
318
00:58:40.260 --> 00:58:55.140
is, believe, five SMRs at 300 megawatts each for up to 1.5 gigawatts. For the initial gas generation, they're targeting one gigawatt for 2030, and first nuclear power at, by 2032.
319
00:58:55.660 --> 00:59:10.460
An incredibly aggressive timeline for this. Again, the final investment decision's coming in 2027, and also the Nuclear Regulatory Commission's construction permit application is planned for 2027 as well.
320
00:59:10.500 --> 00:59:20.260
The timeline here is incredibly aggressive, Charlie. Uh, fir- first of all, as far as I understand it, we actually don't have any small modular reactors that- No, we don't...
321
00:59:20.270 --> 00:59:32.140
at least, well- We have, we have zero SMRs, zero small modular reactors in the West. There are some in the East, but there's none in the USA.
322
00:59:33.380 --> 00:59:44.260
Yeah, and so a lot of jawing as to getting those online, but so far none have been put online. But they're targeting 2032, so six years, uh, out for this.
323
00:59:45.240 --> 00:59:58.340
And they said that it could take roughly 24 to 36 months from first nuclear concrete to fuel loading for these. It's part of the bull case for the small modular reactors. They're much quicker to get online.
324
00:59:58.820 --> 01:00:06.200
And little point is, I had to do a little... I was like, "What do you mean nuclear concrete?" It's a special type of concrete you gotta pour and, like, seal and case stuff in.
325
01:00:06.260 --> 01:00:16.200
So I was like, [laughs] "What do you mean nuclear concrete?" Okay, so yeah. Well, this is part of the reason why these things take so long. The, the permitting and the regulation process for this is so painstaking.
326
01:00:17.000 --> 01:00:31.960
I mean, if you... So for that nuclear concrete, if you are an inch short of what is mandated by regulations, you have to tear out everything and start over. So the, the whole process takes a long time.
327
01:00:32.000 --> 01:00:44.556
It's absolutely choked and mired in red tape. But they're pretty-- They're saying here that they are somewhat confident they can get this done by 2032. Um, that being said, it's pretty aggressive.
328
01:00:44.596 --> 01:01:00.156
There are also questions about f- uh, how this is gonna be funded. Nuke is not cheap, neither is nat gas. Um, and so b- I mean, Blue Energy, um, let's see. Blue Energy raised 380 million earlier in 2026.
329
01:01:00.236 --> 01:01:12.316
Constellation Energy has made a strategic equity investment. But- Which is pretty notable 'cause Constellation Energy has the largest nuclear fleet in the United States. Yes. So they are like the nuclear company, and so
330
01:01:13.856 --> 01:01:19.956
this seems like validation to me. Yeah. Yeah, 100%. But that being said, that shouldn't be conflated with what they're doing here.
331
01:01:20.036 --> 01:01:26.076
This is gonna require billions of dollars worth of financing, and n- none of that has been buttoned up yet.
332
01:01:26.096 --> 01:01:32.466
And I don't think they will get a clear idea of what that actually looks like until this final investment decis- decision comes in 2027.
333
01:01:33.716 --> 01:01:41.396
Uh, but needless to say, though, part of the reason we chose this is that it's very interesting to see this dual gas and nuclear strategy.
334
01:01:41.436 --> 01:01:53.656
We've seen gas generation becoming the stop gap for projects that are waiting for grid interconnection. You know, I mean, we just talked about it with, um, the OpenAI project that we just discussed.
335
01:01:53.696 --> 01:02:01.736
I mean, they're, they're talking about deploying like 9.7 gigawatts of gas, I believe, for that Ohio project. Just an insane amount of energy.
336
01:02:01.776 --> 01:02:10.216
Natural gas will be the base load that is either permanent or a stop gap on the way to things like nuclear or interconnection to the grid for a lot of these projects.
337
01:02:10.816 --> 01:02:13.406
The other reason I wanted to bring this up, Charlie, [lip smack]
338
01:02:13.596 --> 01:02:30.196
the novel dual fuel approach notwithstanding, is the fact that, you know, when you look at this project, it becomes-- and, and others like it, it becomes increasingly clear that the nuclear renaissance in the US is being driven entirely by hyperscalers and data center companies.
339
01:02:30.706 --> 01:02:45.156
The, the, the-- I think the, the first shot across the bow with this was Microsoft with their 20-year PPA restart of, uh, the Three Mile Island unit one in Pennsylvania with Constellation Energy, 835 me- five megawatts of capacity.
340
01:02:45.956 --> 01:02:56.016
Uh, 2027 restart for that. That was announced, I believe, last year, and it made headlines because it was the first time we had seen this, and it made, I think, most people realize, oh,
341
01:02:57.056 --> 01:03:05.496
uh, th- [chuckles] there are some serious bottlenecks here, and also the big companies, the Max Seven are serious about deploying capital to get this done. Another notable one, Google,
342
01:03:06.516 --> 01:03:17.856
with NextEra Energy restarting the Duane Arnold Iowa nuclear plant for 650 megawatts, targeted for Q1 2029 on a 25-year agreement.
343
01:03:18.456 --> 01:03:31.196
Meta also with Constellation Energy seeking to restart over 1.1 gigawatts of nuclear load in Clinton, Illinois. Um, or sorry, preserve and expand this. It's not a restart.
344
01:03:31.256 --> 01:03:44.936
Me- Meta's 20-year deal, um, begins in June 2027. It keeps Clinton operating after state subsidies expire. 30 megawatts of, uh, a, a 30-megawatt uprate planned as well. And we can continue going down the list.
345
01:03:45.036 --> 01:03:57.596
Meta has another preserve and expand agreement for four- 400 and-- for like 2.2 gigawatts worth of load roughly in, um, Beaver Valley [lip smack] with Vistra.
346
01:03:58.276 --> 01:04:10.296
Uh, that's a 20-year agreement, 433 megawatts of additional reactor output is expected through uprates in early 2030. Google's got more. Amazon's got a few in Virginia and Washington.
347
01:04:11.096 --> 01:04:24.896
Meta's got a bunch of them, one in Pike County, Ohio with Oklo. This is for new advanced reactors of up to 2 point-- 1.2 gigawatts, targeting early 2030 for potential first energization there.
348
01:04:25.476 --> 01:04:32.296
Switch has a deal with Oklo. Equinix has a deal with Oklo. Um, Crusoe's got this one.
349
01:04:32.436 --> 01:04:44.716
I mean, you just go down this list and all of the nuclear reactors that are either being restarted or being supported by this or, um, new reactors being built, they're all contracts with these data center companies.
350
01:04:46.576 --> 01:04:56.156
Yeah. Every hyperscaler has their pet nuclear project. Um, there's not that many that aren't pet. I mean, I will point out to Microsoft restarting Three Mile Island.
351
01:04:56.235 --> 01:05:07.016
That was a pretty significant, uh, thing culturally as well as from a actual, um, business and operations, uh, power standpoint. So,
352
01:05:08.376 --> 01:05:20.136
um, I don't have like a coherent, uh, train of thought here, but I will say like I was kinda looking into like what are the different SMRs. Like are they all the same type thing? And there's a bunch of different designs.
353
01:05:20.316 --> 01:05:37.096
I'm not deep enough in the weeds to qualify them except that GE Vernona's design, they, uh, they say is the simplest. Which to me- I'm sure they all say that [laughs] I don't know. I don't-- I'm like, okay. Yeah.
354
01:05:37.116 --> 01:05:45.176
[chuckles] I mean, uh, I would be bullish on the most simple one. These seems like incredibly complex, uh, things to build.
355
01:05:45.276 --> 01:06:02.136
So whoever can convince me they have the simplest design, uh, probably has the most compelling, uh, thesis to me. Um, I think another angle that I have is that everybody's got like an early 2030s timeline.
356
01:06:02.556 --> 01:06:19.986
GE Vernona has the fastest, obviously. But, um, if it, if p-- it like-- because we-- this is such an unproven exploratory like field of, of entirely new energy like, uh, generation,
357
01:06:21.036 --> 01:06:24.756
um, right, we can't-- I don't wanna be betting on this.
358
01:06:24.836 --> 01:06:36.476
However, if in the early 2030s we start seeing these turn on and they actually do, you know, power 200, 300 plus megawatts of power, like this is an-- that's an energy revolution.
359
01:06:36.536 --> 01:06:50.387
Like that totally completely changes, um, pretty much everybody's assumptions about... what the grid looks like in, uh, the US and the West for, like, the rest of my entire life. Yeah.
360
01:06:50.548 --> 01:07:04.908
It, it, it would be an incredible, like you said, revolution in terms of capacity in the US. My one commie adjacent gripe with this is- Give me that commie take, Colin... it's all gonna go to the data centers.
361
01:07:05.468 --> 01:07:17.768
Now, granted, this will free up generation for oth- for residences and other commercial and industrial uses that aren't going to be taxed by these data centers, and by, I mean taxing the grid in terms of stressing it.
362
01:07:18.328 --> 01:07:30.168
So you'll free up load elsewhere. So it's not like there's not a problem with the fact that, uh, there's not a problem i- in the sense that, well, you're just gonna have load available for other uses.
363
01:07:30.668 --> 01:07:41.538
But when I think about nuclear energy be- the, the fact that it should be the backbone of base load for most US states at this point, considering we are so far removed from the breakthroughs of the '40s and '50s,
364
01:07:43.248 --> 01:07:54.988
I, I would, I would like to see some ways in which we can also start reworking this into the grid at large. Now, some of this might produce excess power that will go to the grid, which would be great.
365
01:07:55.428 --> 01:08:06.388
It does also raise an interesting question, though. What happens in 10 years when, or 20 years, when we have a breakthrough in computing density to where we don't need these data centers to be as large anymore?
366
01:08:06.488 --> 01:08:11.428
Maybe that, that happens, maybe it doesn't. But if this, if these data centers disappear,
367
01:08:12.388 --> 01:08:21.608
will we actually be able to run these nuclear plants properly for the remaining capacity on the grid, for the remaining load on the grid?
368
01:08:22.188 --> 01:08:29.918
That's one question I have, 'cause when we had, um, we, we had a nuclear expert on, uh, at the show beginning of the year, uh, Dr.
369
01:08:30.148 --> 01:08:41.908
Hashamian, and I asked him why haven't we seen more nuclear around the US, and part of his answer was just the fact that it wasn't economically viable in certain areas. And so I'm just imagining, you know, kind of
370
01:08:43.248 --> 01:08:57.028
playing the tape through in my head, or what does a future look like where, you know, you have these tech companies coming in restarting nuclear, building more nuclear, and then suddenly they're, they don't need it anymore, or they go bankrupt, or those data centers disappear, what happens to that nuclear load?
371
01:08:57.848 --> 01:09:07.208
Can we actually get... Can we actually connect it to the rest of the grid? Can it be furnished profitably to service everyone else? Yeah.
372
01:09:07.988 --> 01:09:19.348
I guess all of that to say is I, I hope that there's a way that this can then snowball into building grid for, or building nuke for the rest of the grid in addition to these data centers. I'm a big nuclear bull.
373
01:09:19.488 --> 01:09:29.788
I think that we should have more of it as long as it makes sense economically. In terms of clean energy and driving down cost, there's n- you know, there, there's no, there's no second best.
374
01:09:29.828 --> 01:09:38.628
There really is no second best. Who isn't a nuclear, who isn't pro-nuclear at this point? Oh. 'Cause it used to be- Oh, buddy. Like the entire green energy lobby?
375
01:09:39.048 --> 01:09:50.688
Most li- I- Most people in the- Yeah, but I think a lot of the greenies, I think a lot of, like, the, the actual grassroots green folks have, have come around to be pro, uh, nuclear or, like, less anti-nuclear.
376
01:09:50.728 --> 01:10:01.318
I think it's really just kinda like the, the, the walking dead husk of, like, green energy lobby that has found itself- I don't know. Yeah. I'm gonna disagree on this one. Yeah.
377
01:10:01.348 --> 01:10:06.677
I think a lot of people who are on the environmental left think that it is polluting.
378
01:10:07.288 --> 01:10:13.928
They, they, they don't actually understand how the fuel is either recycled or stored, and they think that there's no safe way to store it.
379
01:10:14.888 --> 01:10:26.388
Um, and so I think that you're right that there are some on the environmental left who are a little more enlightened to this, but a lot of people have no idea how the actual supply chain of nuclear fuel works.
380
01:10:27.008 --> 01:10:35.508
They don't know- Yeah... that part of it can be recycled. They don't know that it's often stored on site in these massive steel and concrete silos, right?
381
01:10:36.048 --> 01:10:49.008
So- Or it's stored as, uh, the proposal for the Oakla- this has been burning a hole in my pocket this whole time. Oklahoma nuclear hub plans deep borehole waste disposal.
382
01:10:49.588 --> 01:11:00.918
So they're gonna, in, in the, in Oklahoma they're pl- they're, uh, planning on, um, drilling a big-ass hole and putting the nuclear waste down there. I, sorry to interrupt you, but I had- No, I, I-... to bring that up...
383
01:11:00.938 --> 01:11:12.238
appreciate that. Yeah. I mean, I think that that really go... Like, I legitimately think that a lot of people, when they think about nuclear waste, like The Simpsons plays in their head- Yeah...
384
01:11:12.268 --> 01:11:20.318
where, like, Homer is, like, going down to, like, Lake Springfield and just dumping a, like, corn silo full of green liquid into the lake.
385
01:11:20.808 --> 01:11:32.448
Or they legitimately imagine, like, oil tank, you know, like oil tanks of toxic waste sloshing around in a cave somewhere. Yeah. Like, they, I, I legitimately mean this.
386
01:11:32.568 --> 01:11:37.268
I think Harry Sudock was the first one to mention this on a What Bitcoin Did interview a long time ago.
387
01:11:38.008 --> 01:11:50.498
I think The Simpsons and Matt Groening unknowingly did the most cultural damage to the nuclear debate out of literally any meltdown, like more than any of the meltdowns that we've seen.
388
01:11:50.498 --> 01:11:54.728
Well, yeah, that and Chernobyl, dude. I mean- Yeah... but yes. Yeah.
389
01:11:55.108 --> 01:12:07.568
Um, and then you can't, like, you know, if you talk about Chernobyl with most of these people, you can't get into how the cooling design was faulty and we don't use graphite anymore, and how there were a number of warnings that went dismissed.
390
01:12:07.668 --> 01:12:20.128
I mean, the whole thing is difficult to unpack for the average person. Uh, that being said, I do think that we are probably entering a new phase where the economics are such to where
391
01:12:22.148 --> 01:12:29.128
they just don't care about the critics anymore, and, like, these companies are gonna go and do it. We've seen that with the green energy debate writ large. Yeah.
392
01:12:29.168 --> 01:12:43.048
I mean, we're about to re- we're about to, you know, reverse a decade or two of building solar and wind on, like, a single plant or a single data center from OpenAI in, in states like Ohio, right?
393
01:12:43.108 --> 01:12:55.406
Like, they're just gonna start building more nat gas- Oh, it's, it is real- oh, it's just, it's, uh, it is conspicuous how all the, all the EI whatever, uh- ESG...
394
01:12:55.456 --> 01:13:08.956
ESG folks were like, "And now we're pivoting to, uh, energy resilience." I'm like, "You know why we had to pivot to energy resilience? [laughs] Because of your dang private equity fund and having its directives."
395
01:13:09.036 --> 01:13:17.196
Okay, for all those of you who are done with my ranting, um, this is how we, I feel about, uh, nuclear. Uh, here's Homer Simpson.
396
01:13:17.616 --> 01:13:38.676
If you're familiar with one of the, uh, intros to The Simpsons and a common recurring theme, Homer Simpson sits in the control room [laughs] of the nuclear plant as his job, and, uh, on the wall there is like, uh, you know, some like PSA safety thing that says, "Don't forget to something something," but he's obviously covered it with pictures of his daughter.
397
01:13:39.096 --> 01:13:51.656
And so, uh, he's changed the words to do it for her. So do it for your kids. Pro-nuclear, pro-energy, do it for the future generations of United States, and maybe some data centers.
398
01:13:52.296 --> 01:14:05.626
On that note, thank you so much for listening to Block Space Live. We do this every single weekday at 1:00 PM Eastern. We are Compute's daily live show covering quick hits on AI, data centers, energy [laughs]
399
01:14:05.626 --> 01:14:21.036
, emerging technology, and markets. If you like the stream but miss it, you can catch it in podcast form anywhere podcasts are found, and hit our website blockspace.media for even more content, including our newsletter.
400
01:14:21.076 --> 01:14:36.956
That's blockspace.media. This show is brought to you by CleanSpark, NASDAQ listed at ticker CLSK. I'll catch you tomorrow. I'm Charlie. I'm Colin. Peace. [dramatic music]
1
00:00:00.080 --> 00:00:10.960
On y'all. Welcome back to Blockspace Live, presented by CleanSpark. Charlie, we had earnings week the last two weeks, but we've got one more earnings to cover.
2
00:00:11.060 --> 00:00:19.080
That is Hive Digital's Q1 fiscal year 2027 earnings, which dropped at the end of last week after market close.
3
00:00:19.120 --> 00:00:24.000
They just had their earnings call this morning, so we, we will be unpacking those results for our first story.
4
00:00:24.120 --> 00:00:38.960
Following that, we've got Lagos Finance CEO Jay Patel on to talk about Jensen Huang in NVIDIA, specifically their new backstop of OpenAI, and just generally how the credit outlook is looking for the powered shell at AI CapEx bubble.
5
00:00:40.060 --> 00:00:50.460
Second for interviews today, we've got Dan Hillary of UTXO Management on to talk about digital credit and whether or not we think that stretch and strategy are in the clear.
6
00:00:50.570 --> 00:01:01.240
And then we will cap off the show with news that Blue Energy and GE Vernova are doing a little dual energy strategy for a Caruso campus in Texas.
7
00:01:01.300 --> 00:01:14.940
They are going to be deploying nat gas as they look ahead to deploying small modular reactors to power the roughly 1.5 gigawatt site. Blockspace goes live every weekday at 1:00 p.m. Eastern.
8
00:01:15.120 --> 00:01:25.360
We are compute's daily live show, featuring quick hits and stories on the latest in AI data center deals, markets, and emerging technology.
9
00:01:26.080 --> 00:01:36.560
If you like the live stream but you miss it, you can catch it anywhere podcasts are found later in the day when we turn it around and put it onto all platforms. Check out our RSS feed. Also,
10
00:01:37.780 --> 00:01:49.740
hit our website, blockspace.media. We have a newsletter that goes out every single day, as well as a bunch of other stories written and distributed in your favorite outlets, such as Yahoo Finance.
11
00:01:50.880 --> 00:02:04.660
This show is brought to you by CleanSpark, NASDAQ-listed ticker CLSK. More on CleanSpark later on in the show. Yes, we did earnings the past two weeks, Colin, but there are a few stragglers.
12
00:02:04.720 --> 00:02:17.040
The, uh, the first one today is Hive. I believe there's Iron later this month. Um, but there's a couple more of these that we will cover as they happen. The bulk of earnings, though, happened in the past two-week period.
13
00:02:17.180 --> 00:02:27.440
I would recommend anybody, um, who missed some of these, we had some very stellar CEO C-suite executive interviews, um, with pretty much everybody did their earnings. So make sure to go check those out.
14
00:02:27.500 --> 00:02:36.880
Um, check out the executive interviews playlist on our YouTube channel. And if you want to hear the earnings themselves, we've also got that on YouTube.
15
00:02:37.400 --> 00:02:48.540
So if there's anything you missed, you want the Q&A section, wanna see what management has to say about their strategies, go check that out. But we will go ahead and kick it off with Hive's latest earnings.
16
00:02:48.560 --> 00:03:05.720
I've got our coverage at Blockspace here. Hive Q1 2027 fiscal year revenue rises to 74%, rises 74% as GPU cloud ARR earnings reaches 110 million.
17
00:03:07.100 --> 00:03:19.160
Funnily enough, their cloud ARR is actually higher than that considering a new contract they signed, signed and, and, uh, announced during earnings. But at the end of Q1, that was the number.
18
00:03:19.170 --> 00:03:27.700
We're gonna go ahead and dive into the details here. There's a lot to unpack here, Charlie, and Hive, one of the more interesting case studies, because they're really kind of dividing their strategy.
19
00:03:27.720 --> 00:03:37.620
They're going cloud and colo, and they had a few questions about that on the earnings call. But before we get to the qualitative, let's focus on the quantitative really quickly.
20
00:03:38.180 --> 00:03:48.040
Revenue came in for the quarter ending in June 20- at 79.1 million. That was up 73% year over year and percent quarter over quarter.
21
00:03:48.049 --> 00:03:58.100
The biggest bump here was from mining, uh, revenue, 72.1 million coming in at 77% year over year and 7% quarter over quarter.
22
00:03:58.220 --> 00:04:15.740
HPC revenue, though, nearly doubling over the, uh, quarter at 7.1 million, up 47% year over year. Gross operating margin came in at 31%, and HPC margin at 44%.
23
00:04:15.780 --> 00:04:30.500
More on that later during the analyst Q&A. They had a GAAP net loss of 142.9 million. So what's behind the revenue increase? I've got one word for you, Charlie. Paraguay.
24
00:04:31.220 --> 00:04:44.800
Now, those of you who have been tuning in for a while will remember that Hive purchased a facility in Paraguay from Bitfarms, formerly Bitfarms, now Keel, uh, a little while ago.
25
00:04:44.860 --> 00:04:56.180
That facility accounts for about 18.9 exahashes of their total operational capacity out of 23.6 exahashes. Their hash rate has absolutely surged as a result of that.
26
00:04:56.260 --> 00:05:02.300
And they mentioned on the analyst call when they were asked, "Well, what are you gonna do with those assets in Paraguay?"
27
00:05:02.390 --> 00:05:12.040
And they kind of demurred on this point in, in with regards to how far they would go, um, in, you know, maybe deciding whether or not to turn that into an AI facility.
28
00:05:12.070 --> 00:05:21.420
They did say for the time being, though, that they would keep that as Bitcoin mining as a cash flow cow for the business. And it is the largest cash flowing arm right now.
29
00:05:21.520 --> 00:05:27.990
Uh, you know, two or three years from now, I doubt that will be the case given their AI deal flow. You would actually-- you'd hope not if you're a Hive investor.
30
00:05:28.520 --> 00:05:32.550
But they're going to keep that facility as pure play Bitcoin mining as it stands right now.
31
00:05:33.800 --> 00:05:44.030
As for their GPU line, the acceleration for that revenue largely came from a contract deployment with Bell Canada at their Merit facility.
32
00:05:44.880 --> 00:06:00.320
504 NVIDIA B200 GPUs were deployed, um, in May, and the-- so the June quarter did not include a, the, a full three-month contribution from that, and you would actually see a significant revenue bump, I would imagine, if that were included.
33
00:06:01.220 --> 00:06:13.192
Of the current clusters that cloud ha- or that, uh, um, that Hive has for cloud, uh, right now- They've got roughly 4,300 A series. Look at that, Charlie.
34
00:06:13.432 --> 00:06:24.212
The A100s and A series GPUs continue to hit the news cycle- That's been the theme this past week, is that everyone's just saying, "Yeah, we've got these, and they're still useful." It's been like the theme. Yeah.
35
00:06:24.232 --> 00:06:40.892
D- uh, you know, depreciation bears in shambles. They've got roughly 4,300 of those and another 340, uh, 344 H100s and 504 H200s in Montreal and Quebec. Um, about two or three megawatts estimated there.
36
00:06:41.412 --> 00:06:56.452
They have roughly 4,300 A series GPUs in a one megawatt, roughly one megawatt deployment in Stockholm, Sweden. Uh, the Bell AI Fabric in Manitoba, they have 204 B200s and a few others for roughly one megawatts.
37
00:06:58.192 --> 00:07:10.072
And in Asuncion, Paraguay, they've got a very small undisclosed, uh, deployment as well. Give me one second here, y'all. Google thought it would be a great time to boot me out of my prep sheet.
38
00:07:10.392 --> 00:07:15.412
[laughs] Gotta love it when you just randomly get kicked out of, out of the cloud. Okay.
39
00:07:15.792 --> 00:07:26.252
So let's move on quickly to the, the GAAP loss here because this is really im- this is, uh, quite important, and, uh, they, they got some questions about it during the Q&A.
40
00:07:27.232 --> 00:07:38.012
That loss is substantially less if you strip out a 804-- or, sorry, $84.65 million non-cash regulatory provision that they list as a liability.
41
00:07:38.022 --> 00:07:47.192
Specifically, they're in a legal battle right now with the Swiss government over that. Value added tax, the bane of many a business in Europe.
42
00:07:47.252 --> 00:07:57.202
There's 76.6 million in disputed input VAT, 1.5 million tax supplements on top of that, and 6.6 million in interest. This is a liability. They have not paid it yet.
43
00:07:57.272 --> 00:08:06.432
And in fact, fielding a question from KBW Steven Gladgola, they said they have no intention to pay it. They are going to ke- continue to fight this through the court system.
44
00:08:06.442 --> 00:08:17.912
They've already lost, I believe, uh, two cases with the Swedish Tax Agency on this. And the TLDR is that the Swedish tax authorities are saying that they didn't pay the right VAT for this.
45
00:08:18.452 --> 00:08:29.432
Hive, as I understand it, is saying, "We don't understand why these wouldn't be considered under the same VAT classifications as a data center." This is for the Bitcoin mining operations they have in Sweden.
46
00:08:30.141 --> 00:08:37.692
And they're basically telling the Swedish authorities, "Why would you classify this as anything else? Seems punitive to up that for no reason."
47
00:08:38.452 --> 00:08:46.361
And Aydin Kilic, CEO, said that they have no plans to pay that unless they were forced to from a, from a new court order.
48
00:08:46.392 --> 00:08:59.182
They're moving it up to, I believe, the Supreme Administrative Court of Sweden, or they're m- uh, working on doing that right now. Um, moving on to a few other items before I get to some of the meat from the Q&A.
49
00:08:59.812 --> 00:09:11.492
Cash sits at 208 million on the balance sheet. This comes from 150 million in principal, 130 million in principal of convertible notes. The first issued in April, the second issued in June.
50
00:09:11.652 --> 00:09:18.762
Um, they had 109 point- 109.8 million proceeds from the April one and 124.9 million proceeds from the June.
51
00:09:19.322 --> 00:09:32.172
They also issued thirty million in at the market, uh, or, or they, they drew 130-- uh, they drew 30 million from an ATM f- throughout the quarter. And they have 265.9 million in total debt.
52
00:09:32.232 --> 00:09:42.792
Only 3.2 million of that is current, and 245 million of that comes from the converts. So unclear whether or not that will be converted to equity or paid down through debt.
53
00:09:44.322 --> 00:09:56.992
[lip smack] They have 174.5 million in purchase commitments sitting on the balance sheet right now, as well as a $66.3 million lease liability for their Merit facility [sniffs] in Canada.
54
00:09:58.012 --> 00:10:07.452
They also, big news from the earnings call, they announced a 350 million five-year deal with a, quote, "investment-grade enterprise customer," end quote.
55
00:10:07.952 --> 00:10:22.972
This brings ARR to 180 million, up from the 110 million from what was booked during the quarter. And they're expecting a deployment by end of year 2026 at the Merit facility with 2016 Blackwell GB300s.
56
00:10:23.692 --> 00:10:32.682
Now, the first question they got during the Q&A, Charlie, was on this site and how they specifically will finance the $185 million deployment. And Kilic was pretty straightforward.
57
00:10:33.152 --> 00:10:49.912
He said he expects a three-year payback, 20% cost down on GPUs to start, and they are going to equipment finance the rest, and those payments will start hitting h- or the, um, the interest and the terms for that will start hitting once the shipment is imminent.
58
00:10:50.672 --> 00:11:03.472
The customer for this, unnamed customer, also made a 10% upfront payment of 35 million. They're expecting a yield of 28 to 30% annual EBITDA/CapEx before financing costs on this deployment.
59
00:11:03.482 --> 00:11:12.772
This is at the Merit facility in Canada, the one that we just covered or talked about just a moment ago. Hive doesn't own this facility. They just run GPUs out of it.
60
00:11:13.712 --> 00:11:29.442
And that's important to keep into context with regards to their strategy of colo versus cloud computing. Mike Rondall of Northland asked directly, um, about, uh-- Or sorry, that's, uh, another, uh...
61
00:11:29.442 --> 00:11:38.712
Mike was asking another question. He was asking what's holding up their Big Boden campus in terms of signing a lease. Uh, Kilic said the lease is already being drafted and revised between parties.
62
00:11:38.752 --> 00:11:49.482
Hive is essentially fine-tuning commercial terms here. Management says it hopes to update investors sometime in September. [lip smack] Um, regarding colo versus cloud, to-- per my earlier point,
63
00:11:50.652 --> 00:11:57.482
th-this is kind of, uh, w-was one of the things that was brought up a few times in the earnings call, going back to the Merit facility.
64
00:11:57.552 --> 00:12:07.392
And a lot of o-other facilities that Hive has, like Big Boden, w-with regards to what they're going to do, whether or not outfitting it with cloud or run- doing the powered shell colocation route-
65
00:12:08.388 --> 00:12:19.468
Hive said it has roughly 400 megawatts of Canadian utility capacity in its development pipeline with own sites theoretically supporting more than 120,000 GPUs.
66
00:12:20.948 --> 00:12:31.988
And management basically said, "Look, we are going to do what makes the most sense depending on the site." It was a safe answer, but the fact of the matter is they have quite a bit of optionality here.
67
00:12:32.688 --> 00:12:41.208
So they've got their Bell.ai Fabric site in Winnipeg, Manitoba. Um, they've got enterprise customers for that, unnamed.
68
00:12:41.248 --> 00:12:53.408
They also have the Bell.ai Fabric site in Merritt, British Columbia, [lips smack] uh, with- assigned with Cohere via Bell.ai Fabric. Again, these are cloud deployments for Hive. They don't own the facilities.
69
00:12:53.868 --> 00:13:07.348
They also have this new deal that with an unnamed tenant at the Merritt facility. Bigbot in Sweden is 32 megawatts utility, 25 megawatts critical IT. Um, they have a non-binding LOI, not definitive.
70
00:13:07.388 --> 00:13:09.988
Like we said, we might get an update there in September.
71
00:13:09.998 --> 00:13:23.348
But they also have a number of these other sites, like their Grand Falls, New Brunswick site, which is an a- a Bitcoin mining campus that they actually own, and they are thinking about converting it into a tier-three-plus AI center for 70 megawatts gross, 50 megawatts critical IT.
72
00:13:23.888 --> 00:13:32.608
I think, I don't know for sure, Charlie, but if you were going to see, um, if you were going to see any sort of retrofit where they do colocation,
73
00:13:33.668 --> 00:13:40.828
it's probably going to be in places like that Grand Falls, New Brunswick site. That's the route they're going with Bigbot in Sweden.
74
00:13:40.908 --> 00:13:49.408
That is the template, I think, that we will see with Bigbot and, and with other sites that they own outright as well, that they will probably go the colocation route.
75
00:13:49.808 --> 00:14:01.128
The other site in their portfolio that they are currently thinking about standing up a pretty big operation at is what they call the Greater Toronto Area AI Gigafactory.
76
00:14:01.608 --> 00:14:14.568
It's 320 megawatts of utility, 240 megawatts of critical IT, um, and it could h- house 100,000 GPUs at full build-out. There's currently no customer for that site, [lips smack]
77
00:14:14.588 --> 00:14:24.868
and they were asked questions about whether or not that will be colo or cloud. They did not explicitly say anything during the call about that.
78
00:14:25.278 --> 00:14:40.828
There will be three phases to this, 80 megawatts of critical IT each, and they said this could include a government colocation contract, hyperscaler colocation contract, multiple separate colo customers, maybe even some Hive-owned GPU cloud capacity.
79
00:14:40.848 --> 00:14:52.028
They're targeting integerization at the end of 2027 with GPUs online early 2028. Um, you know, it, it bears mentioning there that for a site that large,
80
00:14:53.128 --> 00:14:58.208
you would imagine that they're probably moving more towards the colo route. That will be very expensive.
81
00:14:58.748 --> 00:15:11.368
And they were asked about this directly of like how do they weigh colo versus, uh, versus cloud in terms of, uh, return on invested capital, and,
82
00:15:12.808 --> 00:15:21.188
uh, they, they really didn't, uh, th- uh, they really did not give an answer as to what exactly would happen at each individual site.
83
00:15:21.668 --> 00:15:33.828
But Killick did give some numbers that kind of illustrate how much you could expect from either, uh, either deployment with regards to how much the, how much revenue they could earn.
84
00:15:34.228 --> 00:15:41.088
For colo, roughly 1.8 million per megawatt per year. GPU cloud, 14 million per megawatt per year.
85
00:15:41.208 --> 00:15:54.348
The illustrative revenue multiple for colo is 10 X, for GPU cloud 3.5 X, with an implied, um, enterprise value density of $18 million per megawatt for colo and 49 to 50 million megawatts for GPU cloud.
86
00:15:54.848 --> 00:16:02.728
The takeaway here, obviously GPU cloud, as they were asked by analysts, this is the more lucrative one. Fact of the matter is, though, incredibly CapEx intensive.
87
00:16:03.148 --> 00:16:16.488
I would imagine that Hive for the larger megawatt sites is going to go for a more colo-centered strategy because there's less risk involved. There's less reward, but there's less risk, and it's easier to finance, so.
88
00:16:18.368 --> 00:16:33.258
So, uh, Hive up pretty big today. Actually, I would say it's the first, um, stock, Colin, the first equity that we've covered, where it's, uh, really soared on its earnings the first day. We saw CoreWeave.
89
00:16:33.268 --> 00:16:47.288
We saw some of the Neo clouds do it, but, uh, they floundered a bit, um, you know, during the intraday trading, whereas Hive, uh, earnings dropped Friday. Uh, uh, call was this morning. Now up 20 per- percent.
90
00:16:47.668 --> 00:16:59.008
Up 50% of the year. I'm looking at DI metrics here, dimetrics.ai. Link in the, in video description. Colin, can I say something without everybody getting mad? You can try. [laughs] Okay.
91
00:16:59.068 --> 00:17:12.528
Okay, we spent the whole time just covering their, their AI HPC deals. Um, but if we look at Hive's real, like, revenue-generating business today, it's really overweight their Paraguay Bitcoin mining operation.
92
00:17:12.568 --> 00:17:27.688
Now, they do use language such as looking forward to Boden, looking forward to these AI HPC deals. But, um, I kinda have a bit of an interesting insight here, which is, uh, so their, their Paraguay site in...
93
00:17:28.168 --> 00:17:42.348
How do you say the name? It's, uh- Iguazu... Igu- Iguazu. Iguazu. Um, yeah, they're the largest client of the Paraguayan, uh, power monopoly elec- uh, you know, grid monopoly down there, ANDE.
94
00:17:42.788 --> 00:17:48.448
I won't, you know, I won't explain what that... It's just the abbreviated version of the utility down there.
95
00:17:48.548 --> 00:18:03.688
Um, and, uh, their power contract goes up for renewal the end of 2027, and they're adding, I believe, another 100 megawatts this quarter of Bitcoin mining. Um, they have a ton of negotiating power with Paraguay.
96
00:18:05.080 --> 00:18:18.970
Uh, because of this. End of 2027 is very, very far away if we consider the, the speed and scale of build-outs as well as just the, the churn of markets. Where will the AI trade be in 2027?
97
00:18:19.400 --> 00:18:34.540
Where will the Bitcoin mining risk-reward trade be in 2027? We're looking at an interesting scenario where all of the North American miners are as quickly as they possibly can converting their Bitcoin mining sites to AI.
98
00:18:34.620 --> 00:18:47.440
If Hive is in a position where they're, they have added and are the largest client of one of the, like, lowest cost energy, uh, grids in the world in Paraguay, and they have 400 megawatts of Bitcoin mining,
99
00:18:48.540 --> 00:19:02.180
there could be a scenario if Bitcoin turns, everyone else in the world is locked into long-term duration HPC contracts, who's got hash rates online? Hive. So this is, uh, this is me trying to...
100
00:19:02.720 --> 00:19:08.500
I'm grasping for straws for bullish Bitcoin mining narrative. And this is what I can put out, so yeah. Yeah.
101
00:19:08.580 --> 00:19:17.940
And just going back to the Paraguay site, and I'll cap it after a few more comments with regards to the margins for Bitcoin mining versus HPC.
102
00:19:18.640 --> 00:19:25.780
If they could deploy this for, like, a powered shell colo for AI, I think they would've announced that already and tried to- I think so too, yeah.
103
00:19:26.120 --> 00:19:34.740
There, there has to be something either with investor risk of being in that region. Perhaps it's just not well-suited for this. You know, it's, it's, it's in the middle of the Amazon.
104
00:19:34.860 --> 00:19:47.180
How many of your clients, how many clients near Paraguay need inference? Because look, I mean, the closest major city is San Juan, I believe. Um, yeah, so. Uh, yeah, potentially. I'm not, I don't know the geography.
105
00:19:47.280 --> 00:19:56.620
I would've considered it, would've thought that it was the capital, but I could be wrong. Um- Yeah, but I mean, do you need 400 megawatts of, of inference, Hive needs in- But, you know- Yeah. Right. Right.
106
00:19:56.780 --> 00:20:02.300
I mean, that's a good... You know, but you... There are other things we could use it for, potentially, like training, right? Yeah. I just, I wonder how much...
107
00:20:02.320 --> 00:20:16.520
We saw this with the reason why Keel got out of the region is that, you know, quite frankly, most North American investors don't really want what they perceive as the political risk or other risk of being in, in Central or South America.
108
00:20:17.160 --> 00:20:22.719
They, they don't know how to price it properly. They don't understand the governments, they don't understand the local cultures.
109
00:20:22.800 --> 00:20:31.160
So I think that if they could, they would've done something more with this already, and the margins are kind of proving that out. I'll just leave on this point.
110
00:20:31.200 --> 00:20:44.860
Chris Brendler of Rosenblatt asked about future margin expectations. Hive didn't give any future expectations, um, but they did mention that, you know... So right now it's 44% on the HPC business line.
111
00:20:45.420 --> 00:21:02.560
There's no target, but Killick said that their A Series fleet of GPUs is running out for 40 cents per GPU hour right now, while the 844 H100s and H200s they have a- are driving 20 million of the current 35 million ARR for the contracts that are already active.
112
00:21:02.589 --> 00:21:06.570
That's not contracted. Um, potential ARR, that's what is running today.
113
00:21:07.320 --> 00:21:15.710
So when the Blackwells hit, you could imagine that the margins would potentially be even juicier, and the fact that they're getting that much out of these older models right now really tells you something.
114
00:21:16.280 --> 00:21:26.080
I will say spot price for A100 80 gigabyte models are, is, like, close to $1, so 40 cents would be lower than that, but that's, like, for spot- Yeah... um, and not for contracted, so- Right.
115
00:21:27.360 --> 00:21:36.720
We are gonna s- go, uh, and talk about Jensen's latest letter with, uh, Jay Patel, who's in the wings.
116
00:21:36.740 --> 00:21:44.420
We'll bring him up here, and we'll talk about the latest in AI HPC financing after a word from our sponsor, CleanSpark.
117
00:21:44.490 --> 00:21:59.440
[gentle music] 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.
118
00:22:01.260 --> 00:22:12.660
This is our proof of work, and we are setting the standard for what's next. Learn more about the intersection of energy and Bitcoin at cleanspark.com.
119
00:22:17.340 --> 00:22:32.289
All right, so let's bring on Jay Patel of Lagos Finance. Welcome, Jay, back to the show. How you guys doing? Fantastic. Good to see you. Um, there's always some-- Whenever Jensen drops a letter, I think we just...
120
00:22:32.870 --> 00:22:43.260
That's, that's what we've gotta talk about for the day. So [laughs] um- Well, at this point it's weekly, right? It's weekly, yeah. Uh, he went from zero to millions of followers on X overnight.
121
00:22:43.840 --> 00:22:56.880
And what he put out this morning was this, um, securing the infrastructure of intelligence, land power shell, the next critical resource for AI factories. He calls it LPS.
122
00:22:57.460 --> 00:23:06.640
Can you give me a, like, a just high-level overview of what he, of the points he lays out in this, uh, letter here, and then we can jump off from there? Yeah.
123
00:23:06.760 --> 00:23:17.180
Um, so you guys are probably actually a better fit to talk about the- [laughs]... the nuance of, of the actual deal than I am. But, um, you know, at a high level, I think the points he made were basically that...
124
00:23:17.980 --> 00:23:28.740
And, and I, I really think he's just referring to Anthropic and OpenAI, because I don't know about any other frontier labs with massive CapEx needs that don't have high credits.
125
00:23:29.260 --> 00:23:42.270
Um, but you know, uh, the, the deal they're working on with SB Energy is basically that, uh, NVIDIA will effectively provide a backstop for a portion of the leases, and then what would seem to be creatively worded, like minimum or
126
00:23:43.620 --> 00:23:51.360
maximum guaranteed capacity and minimum residual value in the case that OpenAI kind of backs out as a, as a tenant on the lease.
127
00:23:51.480 --> 00:24:02.900
Um, but you know, the case he made is basically that in the vast majority of their, you know, the vast majority of their business won't change, but there are, you know, in his words, limited circumstances.
128
00:24:03.240 --> 00:24:09.756
I'm thinking that, you know, OpenAI and Anthropic are probably the- Two largest consumers of compute are definitely up there.
129
00:24:10.376 --> 00:24:24.616
Um, that there are certain circumstances with the Frontier labs where their compute needs outpace their credit and balance sheet capacity, and in those situations, Nvidia might do more of these deals, which he says is clearly not circular financing.
130
00:24:24.676 --> 00:24:30.756
We can all, you know, uh, w-you know, make our own circular- We can debate the merits of that [laughs] Yeah.
131
00:24:30.916 --> 00:24:55.116
Um, but I, I did think that this was interesting just because it kind of goes back to a point that we talked about, it's gotta be almost a year ago, where, um, you know, I, I don't know, Charlie, if it was, if it was you, but, you know, we brought up the idea that, like, Nvidia has somewhat of a moat against other chip providers in the sense that they have capital, right?
132
00:24:55.216 --> 00:25:00.836
Uh, whether it's their balance sheet or the ability to backstop with their equity or credibility with investors.
133
00:25:00.936 --> 00:25:25.356
Um, and I do think that these kind of deals, along with the stuff we saw last week, um, probably reinforce the kind of Nvidia Lindy, if, if you will, in the sense that it becomes cheaper to finance the build-out of data centers that have Nvidia chips versus, you know, Cerebras or some of the dedicated TPUs or, or, or other inference-dedicated chips.
134
00:25:25.976 --> 00:25:40.546
Um, um, so I mean, I-- look, I think it's, it's just another sign that the capital that this space needs is way beyond anything we've seen, and obviously I think this is just the beginning of more and more creative structures.
135
00:25:42.696 --> 00:25:58.056
So, uh, his whole thesis, 'cause, like, Nvidia's had a, had a technical moat with the CUDA standard for a long time, and that's been instrumental to why they're in the position they currently are. Um, but, uh, does...
136
00:25:59.396 --> 00:26:08.436
You know, you say that capital is now this moat, and to me it's a little unclear what form that capital takes, because we have the GPU hour.
137
00:26:08.456 --> 00:26:17.255
We've been talking about it with various, like, compute options, uh, market, you know, makers and, uh, products.
138
00:26:17.296 --> 00:26:33.376
And, uh, to me it's, like, really difficult to identify what that asset is that is ba- that, that is the cap- the capital asset here. Um, I'm not really sure how I'm asking the right question, but, like, um,
139
00:26:34.496 --> 00:26:40.556
it-- there, like, what exactly are, like, the capital intent? What, w- where does the money come from?
140
00:26:40.656 --> 00:26:51.916
[laughs] What, is it gonna be GPU hours that we finance against, or is it gonna be specifically just Nvidia hardware, uh, that we're financing against? So I think [sighs]
141
00:26:51.976 --> 00:27:04.176
I'm, I'm wary to think that there's going to be just a broad GPU hours compute market that has a big financing structure behind it anytime soon, because of the fact that you have standardization.
142
00:27:04.206 --> 00:27:13.756
And then the other thing, you know, I think last week's news kind of reiterated this point, is that there's plenty of capital that wants to lend in this space. You saw Brookfield, and KKR- Yeah...
143
00:27:13.766 --> 00:27:24.696
and Goldman, uh, you know, I think there's three other names on that list last week. Um, but obviously if you're a, if you're a lender into this space, you have limited upside, right?
144
00:27:25.076 --> 00:27:28.756
Um, uh, you know, we, we were all familiar with the Bitcoin mining space last cycle.
145
00:27:29.136 --> 00:27:40.076
Um, if, if there's any sort of, like, cyclical space that you're lending into and you don't have upside, you want really strong credit protections. Um, and I think what Nvidia basically helps with here is two things.
146
00:27:40.196 --> 00:27:52.456
One, you know, last week, uh, you know, my understanding is, like, there's certain reference designs and certain implementation, you know, certain requirements for the data centers for them to provide this kind of, quote unquote, backstop.
147
00:27:52.905 --> 00:28:04.276
And then the other thing is that, um, you know, the question in all of these cases, and you know, you mentioned, you know, A100 spot prices are up, you know, maybe the depreciation of these assets isn't what we thought it was and they have a longer useful life.
148
00:28:05.256 --> 00:28:12.796
But I think over the medium to long term, it's hard to see GPUs not depreciating.
149
00:28:12.836 --> 00:28:23.416
Like, you know, we might be in this short-term scarcity where prices go up, but I don't think anyone will tell you that in seven years' time an H100 is gonna cost more than it does today. That just, like, makes no sense.
150
00:28:23.535 --> 00:28:42.216
Um, and so, you know, I think from a, from a investor's perspective, if you're, you know, KKR or Goldman or, you know, whoever else trying to lend in this space, having Nvidia be the one providing the kind of minimum residual valance-- uh, value guarantee versus someone else helps in two way, ways.
151
00:28:42.316 --> 00:28:52.736
One is I think there is the sense that Nvidia has a better idea of what's going to happen over the next six or seven years with regards to GPU development progress than anyone else, because they make the GPUs.
152
00:28:52.776 --> 00:29:04.976
And so if they say that they expect this to have a certain useful life and they're willing to guarantee a certain value for it at a certain point in time, you know, I think y- their word is worth more than, you know, some third party.
153
00:29:04.996 --> 00:29:12.936
And then the other is obviously the beneficiary from the GPU sales and the networking and everything else from day one is them, right?
154
00:29:12.976 --> 00:29:32.456
So they have, you know, this massive cashflow engine, and so they can use a part of, part of that, you know, that's left on their balance sheet against, you know, potential future liabilities versus, you know, smaller players, you know, their credit guarantee is not as meaningful if you don't think that they're going to produce, you know, hundreds of billions in cashflow over the next couple years.
155
00:29:34.316 --> 00:29:46.496
I mean, do you think that this means then that whatever Nvidia does with these backstops and what the risk they're willing to take will kind of set a standard vis-à-vis that look at how long these GPUs actually have useful life?
156
00:29:48.156 --> 00:29:49.876
Yeah, my sense is that there...
157
00:29:51.126 --> 00:30:05.256
[sighs] You know, I'd, I'd say the closest analog, and I, and I don't wanna be, like, too conspiracy theorist here, is Nvidia kind of wants to become the Fannie Mae or some equivalent of GPU financing, right?
158
00:30:05.816 --> 00:30:09.706
They're not... And, and actually this may be an even more balance sheet-light approach. Um-
159
00:30:11.084 --> 00:30:19.803
I heard it somewhere, but basically, you know, someone was saying that maybe they're even more like the Federal Reserve, where they're just setting some policy and some standards, right? "Hey, these are conforming loans.
160
00:30:20.154 --> 00:30:25.164
These are data centers with the reference design and implementation that we're willing to provide a backstop to."
161
00:30:25.204 --> 00:30:32.704
And then the private market, and I know NVIDIA's part of the private market, but, you know, in this analogy, NVIDIA, you know, can be the Federal Reserve or whatever you want it to be.
162
00:30:32.804 --> 00:30:40.284
Um, you know, all these asset managers they parked with- partnered with make the loans, right? NVIDIA's not actually going off balance sheet for any of this.
163
00:30:40.504 --> 00:30:48.764
But essentially, like, this only works because everyone has the understanding that if things go wrong, NVIDIA will come and clean up the mess, and they have the capital to do so.
164
00:30:49.464 --> 00:31:00.904
Um, now, I, you know, I guess o- one kind of hole in all of this is how does that, you know, like, what, what does their existing customer base think of this, right?
165
00:31:00.984 --> 00:31:08.444
You know, they've always had this tension between, you know, helping the Neo cloud scale versus satisfying hyperscalers who are longtime customers.
166
00:31:08.464 --> 00:31:15.344
And now, you know, with this OpenAI deal, you know, I, I'd say that's maybe even somewhat competitive to some of the Neo cloud ambitions. Um,
167
00:31:16.264 --> 00:31:22.624
uh, but Jensen's pretty good at politics, so I feel like there's some balance he'll find to make everyone happy.
168
00:31:25.464 --> 00:31:40.184
Um, I was looking at, uh, from the Wall Street Journal, I believe, that the, this, uh, like what exactly is being secured. And, um, a lot of it is not GPUs.
169
00:31:41.144 --> 00:31:53.864
Um, and which is real interesting because it, if, if NVIDIA is now backstopping, like, land and power, then that seems like a pretty different,
170
00:31:55.504 --> 00:32:06.484
that seems like a pretty different risk profile because now they're not just producing silicon. They're, now their balance sheet is now secured by, like, real assets which have a very different depreciation life cycle.
171
00:32:07.304 --> 00:32:12.024
Thoughts or takes on this. Yeah. I, I'd say, like, the, you know,
172
00:32:13.124 --> 00:32:22.913
a year or two ago, you might have thought, oh, NVIDIA will just be able to grow compute demand by financing demand and, like, capital on the purchaser side was the pure constraint.
173
00:32:23.424 --> 00:32:38.584
But clearly, you know, as you guys cover basically every week on this show, even if I had a delivery of, you know, a 100 giga- or 100 megawatts of GB200s or whatever, like, there's, you know, a multiyear backlog to try to find somewhere to actually power those.
174
00:32:38.924 --> 00:32:50.524
Um, and so I think from NVIDIA's perspective, it's like they gotta move further down the stack because the ability to buy their chips, while it is part of the limitation on demand, that's not the ultimate limitation.
175
00:32:50.884 --> 00:33:01.604
Um, and now they actually need to help their customers secure, you know, land, power, and shells so that they can plug in the NVIDIA chips that they're gonna buy. Um, but I wasn't, I was kind of...
176
00:33:02.204 --> 00:33:11.844
I mean, maybe I shouldn't be shocked. I, I'd be interested to hear what you guys think, that the OpenAI deal here was 20, a 20-year lease. Um, so it seems like NVIDIA's,
177
00:33:13.624 --> 00:33:24.464
I, I, I don't know, like, forecasting pretty far out into the future the useful life of the non-silicon assets. And I mean, you know, land and power definitely depreciate a lot slower than GPUs, but yeah.
178
00:33:25.284 --> 00:33:35.894
My take is absolutely that's a great, that's a great play. I don't wanna be trying to backst- I don't, I, I don't wanna guarantee that OpenAI's gonna be around in 20 years. I, uh, the, those power and land assets
179
00:33:36.964 --> 00:33:43.564
absolutely will. Um, I would wanna, you know, we'd, we'd floated maybe talking about, like, yield varying stablecoins. We're gonna table that.
180
00:33:44.084 --> 00:33:50.204
I think wanna throw one kinda last question to you, which is one thing that I was kinda confused at in, like, reviewing this story.
181
00:33:50.264 --> 00:33:57.344
We've covered it over the past few weeks, is that we've had, like, three different backstop numbers. I don't know if this is, like, a big deal or not.
182
00:33:57.724 --> 00:34:10.344
There was initial, like, 500 billion, and then I think NVIDIA came out and clarified that it was 250. And then on Friday, the Wall S- Wall Street Journal said it was, like, 150. Um, I don't know.
183
00:34:10.424 --> 00:34:22.644
Like, to me, I see a bunch of different numbers, which are hundreds of billions of dollars different. Um, should I be, like, concerned about this, or is this just, uh, a messaging thing that I'm not understanding?
184
00:34:23.444 --> 00:34:31.883
I think it's, like, I, I think it's broadly a messaging thing, but I also think it's one of those, like, um, depending on who the audience is, you wanna say a different number.
185
00:34:31.943 --> 00:34:43.604
So my understanding is the big number, the 500 billion, is the total, you know, for this initial program, the total size of the financing that the partners can provide, right?
186
00:34:43.634 --> 00:34:49.684
And so that's not NVIDIA putting up $500 billion of their own balance sheet or $500 billion of compute or whatever.
187
00:34:50.164 --> 00:34:57.364
It's just, hey, we've worked with these six asset managers to stand up some sort of program that can scale to $500 billion in financing,
188
00:34:58.644 --> 00:35:11.714
um, which I think NVIDIA has an incentive on the press side of things, on the initial side, to say, like, as big a number as possible because, you know, $500 billion, some decent portion of that is going to come to, to NVIDIA and NVIDIA equity holders.
189
00:35:12.144 --> 00:35:19.224
I think the clarification later in the week, you know, which I think people were getting to, is like, okay, if there's some backstop, what is the risk that NVIDIA's taking on?
190
00:35:19.284 --> 00:35:24.584
And I think that's what they're trying to minimize the appearance of, right? It's like, how much balance sheet risk is NVIDIA actually taking?
191
00:35:25.204 --> 00:35:37.844
Um, and my understanding for those initial deals is that most of the backstop comes in the form of, like, a minimum residual value guarantee. Um, and so it would be significantly less than,
192
00:35:38.984 --> 00:35:46.844
you know, the initial loan amount. You know, it's really what is the GPU worth in the out years for any financing deals, you know, year five, six, seven.
193
00:35:46.964 --> 00:36:04.584
Um, so I think the risk to NVIDIA is probably closer to the 100, 150, but the total size is maybe closer to the 500. All right. Jay, thank you so much. Illuminating the half a trillion dollar deal up in Ohio.
194
00:36:05.164 --> 00:36:16.944
Quite an insane deal. Thank you so much for your time. Thank you. We'll catch you next week again, and maybe we'll talk, we'll, we'll get around to a state of DeFi conversation, so. [laughs] Yeah, hopefully.
195
00:36:16.984 --> 00:36:28.764
Take care, guys. Cool. Thanks, Jay. All right, we have Dan Hilary of UTXO Management in the wings. We'll bring him on up here right after a word from our sponsor, Luxor.
196
00:36:30.724 --> 00:36:41.384
[whooshing sound] This episode of Blockspace Live is brought to you by Luxor's Commander, Bitcoin miner management software for enterprise operations.
197
00:36:41.904 --> 00:36:47.484
Luxor's Commander gives you real-time fleet monitoring, bulk remote commands across your fleet, and Intelligent Miner.
198
00:36:48.084 --> 00:36:55.433
That's an automated profitability engine that runs every five minutes and tests your fleet's power settings against live energy and hash rate markets.
199
00:36:56.064 --> 00:37:02.484
In fact, ERCOT backtests show 10% improved profitability with intelligent mining versus old-fashioned binary mining.
200
00:37:03.184 --> 00:37:11.444
Commander Pro is roughly half the cost of competition, $100 per megawatt, or a 25 basis point pool fee adder, but you can also try it free for 60 days.
201
00:37:11.524 --> 00:37:24.984
So if you'd like to learn more, go to luxor.tech/commander to get started. All right, we're gonna talk a different kind of credit, the digital kind, with Dan Hilary. I'll bring him up, him up here.
202
00:37:25.964 --> 00:37:38.264
Dan, welcome to the show. Uh, I think you gotta unmute yourself on the Hey, guys. Great to be here. Thanks for having me on. Yeah. Super excited for this conversation. Thank you for joining, man. Yeah, cool. Yeah.
203
00:37:38.844 --> 00:37:50.984
So, uh, I believe recently added VP of credit, UTXO Management. Uh, great, uh, team, friends of ours. Um, and we're gonna talk some digital credit. So, uh,
204
00:37:52.144 --> 00:37:59.764
I've been noti- I've been seeing Michael Saylor's been selling the Bitcoin little bit below the cost basis to defend the preferred stock.
205
00:38:00.764 --> 00:38:17.584
Um, and, uh, I wasn't paying super close attention, but it looks like Saylor also, uh, flagged, uh, uh, into the first week in September as the date he wants Stretch back to $100. So,
206
00:38:19.104 --> 00:38:25.144
what's going on there? Do you think we'll hit $100, and do you think it actually matters if we hit $100?
207
00:38:26.104 --> 00:38:36.364
Look, I mean, I think, you know, this is where I focus all my time on these digital credit instruments, Stretch being the largest by market capitalization of all the preferred equities, including, you know, the fixed rate preferred equities.
208
00:38:36.404 --> 00:38:45.964
Even larger now than the convertible debt, which was, is kind of the old instrument by which strategy created leverage in their balance sheet. Now it's referred to as amplification.
209
00:38:46.604 --> 00:38:53.504
And obviously, you know, Stretch traded up within about three months, uh, from its IPO price of $85.
210
00:38:53.704 --> 00:39:04.584
Once it was IPO'd, it traded down to a low of 80, roughly, and then traded up to 100, and stayed around $100 for a, for a long clip of time, about, you know, just shy of nine months. And [clears throat]
211
00:39:05.024 --> 00:39:13.144
it, it's designed to trade around par, and there's a bunch of mechanisms now that Strategy uses to maintain that par pricing.
212
00:39:13.244 --> 00:39:20.904
And over the, you know, the first nine months of Stretch's history, it was this variable rate monthly dividend.
213
00:39:20.944 --> 00:39:32.794
So for anyone who's not familiar, they're able to adjust the dividend rate on STRC on a month-to-month basis based on the average trading price of the prior 30 days.
214
00:39:33.284 --> 00:39:39.464
So they've been ratcheting up the dividend, it was staying around par, and then it traded off, and that was a large function of the cash reserve.
215
00:39:39.924 --> 00:39:47.944
And, you know, obviously, the duty to shareholders, there's not a fiduciary obligation for it to trade at $100.
216
00:39:48.364 --> 00:39:53.824
However, management's put, put forth guidance such that they want to, they're targeting that $100 price point.
217
00:39:54.444 --> 00:40:04.844
So what they've done recently after Stretch traded off was rebuild the cash reserve to support dividend payments and also increase the execution of buybacks.
218
00:40:05.084 --> 00:40:33.184
And I think people are really discounting the effect that buybacks will have on price over a long period of time, and that's the m- that's the one thing, you know, we're really looking closely at, are how is the Stretch price affected by not only the dividend changes, which right now the dividend's at 12% annual based on $100 par and it's staying there, but how, how does the cash reserve size relative to the total Bitcoin assets, the senior debt, and then the execution of Stretch buybacks affect its trading price?
219
00:40:33.244 --> 00:40:46.114
And, you know, my three-second summary on that is cash reserve puts a floor in the Stretch price, and buybacks are more of a micro adjustment that are able to get it up to its, its desired trading range.
220
00:40:46.204 --> 00:41:00.084
Much like the sale of STRC could, and has, pushed the price down over short periods of time. So, uh, Saylor wants to get it back to 100.
221
00:41:00.584 --> 00:41:11.664
Do you think, like, if he gets close, you know, what is it, September 8th, September 9th, I- yeah, September 8th, do you think it, like, then that's sufficient? Will people... Will, will the product be...
222
00:41:11.944 --> 00:41:21.504
Will, will, will people, will people be happy? Uh, I don't know. And at what rate do you think it probably trades at, you know, on September 8th? Yeah.
223
00:41:22.004 --> 00:41:37.464
I think one chart for viewers to put into their trading view to look at is the price of STRC with dividends reinvested since inception. And so that chart has hit all-time highs over the past week or, or two weeks.
224
00:41:38.004 --> 00:41:47.244
So that's really the cumulative return for STRC investors over time, and that's really, really important. And so I think right now the rate's 12.
225
00:41:47.404 --> 00:41:51.614
I actually don't expect the rate to decrease in any near-term future, right? Like,
226
00:41:53.144 --> 00:42:04.794
you know, we can argue about credit spreads, which is the difference between the risk-free interest rate and a certain fixed income instrument, but Strategy's trying to grow the market capitalization of Stretch.
227
00:42:05.124 --> 00:42:14.724
To do that, they have to have a spread or a yield that is significantly higher than other instruments in the market. So I expect it to stay kind of where it is.
228
00:42:15.344 --> 00:42:22.323
Then to your, to your question of, like, will it get to 100, the stated trading range is 99 to 100. Right now we're at, like, 95.
229
00:42:24.244 --> 00:42:32.764
Clearly the buybacks are what are being used on a weekly basis to get it back to that hundred dollar price point. And I think long term it will get there.
230
00:42:32.884 --> 00:42:39.324
Um, but cl- you know, the volatility spooked a lot of people and, you know, I like to be transparent in my analysis.
231
00:42:39.384 --> 00:42:52.864
I'm definitely a Saylor fan, but having a 12% yielding instrument trade from a hundred dollars to $73 is very nerve-wracking, and I think there will be a lot of, kind of, rotation of the instrument.
232
00:42:53.704 --> 00:43:01.134
And, you know, that's kinda what... The fund I'm running at UTXO is designed to buffer that for different types of investor classes, right?
233
00:43:01.184 --> 00:43:04.664
There's a lot of people who want all this yield and believe the dividends will get paid.
234
00:43:05.104 --> 00:43:13.144
There's a lot of retirees, credit investors, and just, you know, institutional balance sheets that can't weather that volatility, and they want a high yield.
235
00:43:13.433 --> 00:43:23.384
And so I think there's gonna be products built on top of Stretch that tranche this, sort of, yield or volatility profile and make it more attractive to different investors.
236
00:43:23.464 --> 00:43:26.764
Right now, the volatility's high for a lot of, a lot of people.
237
00:43:28.004 --> 00:43:39.444
I have a, a ki- I think maybe a challenging question because, again, I'm not really, I don't really study digital credit, but, um, you know, you mentioned that there, the, uh, reinvestment strategy back into Stretch.
238
00:43:40.384 --> 00:43:43.744
Mm-hmm. Uh, that, that strategy has hit all-time highs, but, um,
239
00:43:44.824 --> 00:43:55.364
I, as a user, thought that this was primarily, like, a yield instrument so I could, like, put a bunch of money in and have it fund my, you know, my endeavors or retirement or whatever.
240
00:43:56.084 --> 00:44:04.264
Um, maybe are there, like, different strategies as the, as this, as these preferreds get up and running and more stable over time?
241
00:44:04.384 --> 00:44:18.784
Or, uh, or does this challenge, like, this idea that this is a, a simple, less volatile instrument that investors can be lured into? Um, yeah, how do you tackle this? Yeah, totally.
242
00:44:18.904 --> 00:44:30.124
I think that's a great question because, you know, there's obviously other preferreds issued by Strategy, and they have these fixed rate yields. So Strife, STRF, is senior to Stretch.
243
00:44:30.724 --> 00:44:36.584
It's, it's heavily over-collateralized and yields roughly 10% annually f- forever, as a perpetual preferred.
244
00:44:37.384 --> 00:44:42.844
That would be, you know, if you're an investor and you're looking for an investment and you're not trying to speculate on the principal value of that,
245
00:44:43.924 --> 00:44:57.244
I would look at that as a great way to lock in the amount of money every year I receive from the instrument and use that as, you know, dividend proceeds to fund my life. STRC is the same sort of idea.
246
00:44:57.484 --> 00:45:07.204
You have a fixed yield. If the principal of STRC falls, so if STRC falls from 100 to 90, you're still getting your 12% annualized return at par.
247
00:45:07.684 --> 00:45:14.804
So your dividend distributions every month or twice a month didn't decrease, right? So that, that's an important thing for people to understand.
248
00:45:15.364 --> 00:45:24.104
But, you know, with regards to the principal, I think the idea of Stretch was and, and continues to be that it's a shorter duration instrument.
249
00:45:24.124 --> 00:45:34.004
You can buy it at, uh, at 99, you can sell it at 99 a few months later, and that's the biggest sort of differentiator between Stretch and the other preferred instruments, and that's the value proposition.
250
00:45:34.094 --> 00:45:40.724
So at the moment, the buybacks are attempting to recover that sort of ethos in instrument.
251
00:45:40.764 --> 00:45:54.304
It's supposed to be where you can park your capital short term or where you can park your capital and not experience the principal fluctuations that you may experience in a traditional fixed income product, in a long duration bond, or any other preferred equities that Strategy issues.
252
00:45:55.504 --> 00:46:05.444
Dan, I have, I have a quick question with regards to one of the other, I should say, looming financial considerations for Strategy, specifically as it relates to the converts.
253
00:46:06.124 --> 00:46:09.424
So it seems like the fires for Stretch are somewhat put out at this point.
254
00:46:09.944 --> 00:46:23.773
I think that the difficulties with Stretch kind of belie a larger, uh, financing wall though, or rather a, um, a liability wall for Strategy, 'cause they've got the 2028 converts, and the holder put date is September 15th, 2027.
255
00:46:24.524 --> 00:46:39.944
Um, the strike on that is $183 per share, $183.20 per share. And then they've got a v... You know, even greater wall in 2028. They've got a $2 billion convert with a put date of March 1st, 2028.
256
00:46:41.304 --> 00:46:54.044
Um, they've got, uh, 1.5 billion with a put of June 1st, 2028. Um, and then they've got September 15th, uh, 2028, um, that two issues on the same day.
257
00:46:54.124 --> 00:47:08.024
They've got, um, roughly 1.4 billion in combined cash put exposure, and the strike prices for all of these are, um, higher than where s- Strategy is today. One of the highest is, uh, for the 2029 notes is 672.
258
00:47:08.804 --> 00:47:20.804
Uh, I'm just curious, kind of a multi-part question here. Do you think that they've done enough so far this year to, uh, bulwark themselves against those potential puts?
259
00:47:21.384 --> 00:47:33.144
And how would, how do you, as you look at trading Strategy, price that risk when you evaluate the company and you evaluate whether or not you wanna, uh, deploy more capital or wait to see what happens with Bitcoin and Strategy's price?
260
00:47:34.244 --> 00:47:43.624
Yeah, definitely. And I'll, I'll approach that question from a credit perspective, right? A- about two months ago, Strategy was in a tougher position.
261
00:47:43.724 --> 00:47:51.744
They were about, they were down to, like, 700 million or sub a billion dollars of cash reserve because they used part of the cash reserve to pay off the converts.
262
00:47:52.184 --> 00:48:09.164
Bitcoin was tanking, and there was this kind of looming, what you're saying is effectively 30% dilution of the common relative to total market capitalization that would have to, to go through to finance the, the payment of the converts, the redemption of the converts, if they didn't convert.
263
00:48:09.984 --> 00:48:14.474
Now, Strategy has, I think it's about a 4.6 or $4.8 billion cash reserve,
264
00:48:15.444 --> 00:48:25.984
and there's 6.7 billion of total outstanding convertible notes, uh, due, you know, one year before the actual expiration because of the put dates if they don't convert. So-
265
00:48:27.412 --> 00:48:37.822
I personally, what Strategy is doing is they're building their cash reserve, and their cash reserve is ser- serving two purposes, right? One, it provides financing or payment for the dibs on the preferreds.
266
00:48:38.332 --> 00:48:46.362
Two, it serves as cash that doesn't need to be raised at some point in the future to pay off the converts if they don't convert, right?
267
00:48:47.052 --> 00:48:57.172
Right now, the board approval is 18 months minimum of dividend coverage for the preferreds, so any cash reserve above and beyond that can be used to finance or to pay off those converts.
268
00:48:57.792 --> 00:49:06.832
So, you know, in terms of a credit profile, I think they pose, pose a very, very small threat to the preferreds. Like, those preferred dividends will continue to get paid.
269
00:49:07.252 --> 00:49:14.492
There's enough cash, and they've demonstrated their ability to raise enough cash to pay off the converts when they come due. Why are they not paying them off early?
270
00:49:14.852 --> 00:49:32.912
Well, simply, Strategy is a leveraged Bitcoin equity, so if they were to go and, and s- pay off the converts today, their amplification or the leverage in the capital structure would decrease, and they'd have to either raise more Stretch or raise more of the other preferred securities or issue more debt to reup that sort of leverage.
271
00:49:33.212 --> 00:49:44.922
So I think what they're doing is reducing the risk of converts not converting by hoarding cash in order to finance them and pay them off or even refinance them into the future.
272
00:49:44.992 --> 00:49:55.032
And two, they're keeping their Bitcoin upside or the leverage exposure to Bitcoin by not paying them off early. So I think it's actually a, a wise strategy, and I don't see a lot of credit risk.
273
00:49:55.422 --> 00:50:13.872
The, the biggest risk has actually already happened, and that was the MSTR equity sales that then get piled into cash or into, into the cash reserve, which for a lot of diehard Bitcoiners or people looking for levered Bitcoin exposure, that would be, you know, non-favorable, uh, for their positions.
274
00:50:15.952 --> 00:50:27.862
Yeah, MSTR holders, uh, maybe not super happy over the past few months. Okay, so I wanna zoom out really far, Dan, and just have you kinda, like,
275
00:50:29.152 --> 00:50:39.402
um, assess the, the success of this, like, the overall strategy. Like, if I can go back, and I remember, like, Saylor saying,
276
00:50:40.572 --> 00:50:53.921
uh, "We're going after, like, the fixed income markets," this massive amount of money that a lot of Bitcoiners who don't study finance didn't really realize, like, this, this, this, like, untapped market. Mm-hmm.
277
00:50:54.012 --> 00:51:05.421
How effective do you think this strategy, and then the, kinda the other types of pref- uh, the, the wi- the, the general universe of, like, preferred equities, but, like, in particular, is Saylor's strategy working?
278
00:51:05.512 --> 00:51:10.352
Um, is it, you know, qualify it a bit, is it working somewhat here, somewhat not here?
279
00:51:10.512 --> 00:51:18.652
And, like, are they on a path to, like, actually creating, um, something which can suck in this, like, black hole of fixed income market?
280
00:51:19.812 --> 00:51:29.612
Yeah, I think that's a great, great point, and I mean, it's difficult because Strategy's pioneering this sort of preferred equity market in a way that it really hasn't been pioneered.
281
00:51:29.652 --> 00:51:38.972
Strategy's STRC trades roughly $400 million of volume a day. That's more than the entire preferred equity market combined as it exists.
282
00:51:39.572 --> 00:51:47.472
So th- this is a really, really small, and historically the preferred market's been a, a terrible underperformer relative to both common equity and debt instruments.
283
00:51:47.852 --> 00:52:00.891
So they're breaking into the preferred market with an eye into fixed income. Preferreds give them infinite optionality and long duration credit, which they want to finance the Bitcoin. Bitcoin's a forever duration asset.
284
00:52:00.992 --> 00:52:13.812
So th- that's the reason they tapped into that market. I think there needs to be a lot more emphasis on creating structures around STRC that are applicable to fixed income mandates, right?
285
00:52:13.872 --> 00:52:31.972
Like, most of the debt, the US sovereign debt, uh, corporate debt, it's owned by pension funds, endowments, large fixed income funds, and they have mandates where they can only buy certain instruments, certain types of credit by certain issuers, and S&P Fi- S&P committee rating is a really big part of that decision-making process.
286
00:52:32.532 --> 00:52:43.602
So until we get strong ratings on Strategy's preferred equities and there are wrappers on top of those preferred equities that allow them to fix, fit institutional mandates,
287
00:52:44.792 --> 00:52:54.432
we're still kind of a nascent, um, asset class. So are we, are we attacking the fixed income, the, the debt markets yet? No.
288
00:52:55.312 --> 00:53:06.612
Are we making the base layer of a Bitcoin-backed credit instrument that will be used to target the fixed income markets? I believe so, but, you know, it's a, it's a decade-long process in my opinion.
289
00:53:09.432 --> 00:53:17.072
Um, yeah, do you wanna go, Colin? I don't really have anything else. Uh, yeah, I guess my last question is, like, um,
290
00:53:18.372 --> 00:53:34.952
so I'm looking at that STRC, uh, yields, what, 600-something bips over high-yield bonds right now. Do you think that that's enough? Uh, do you think that's compelling, uh, to the alternatives?
291
00:53:35.532 --> 00:53:43.732
Um, and maybe, uh, where do you, where do you think that goes, just to kinda round out this conversation? Yeah.
292
00:53:43.792 --> 00:53:49.472
I think that's the question Strategy had to answer when Stretch traded off and the yield blew out to, like, 15%, right?
293
00:53:49.652 --> 00:54:05.812
S- with STRC trading at $73 and the yield, the effective yield, cash on cash, is much higher than the 12% stated, 'cause that's at $100 par. Clearly the increase in yield wasn't increasing the principal price of STRC.
294
00:54:06.092 --> 00:54:13.092
So the market's pricing STRC risk more so based on balance sheet construction and management decisions regarding the cash reserve.
295
00:54:13.792 --> 00:54:21.752
So I actually don't think, and it, and they, they've clearly taken this to heart, that Strategy should continue increasing the divid- dividends on Stretch.
296
00:54:22.252 --> 00:54:29.632
So my long answer to your question is, yeah, I think 600 bips is plenty high relative to high-yield corporate debt.
297
00:54:30.220 --> 00:54:38.680
Um, and the most important step will be consistency of the cash reserve management and execution of buybacks to reduce the volatility in STRC.
298
00:54:38.710 --> 00:54:45.500
'Cause STRC has a much different mandate than a traditional fixed income bond. All right? The principle, the principle never gets returned [clears throat]
299
00:54:45.580 --> 00:54:54.740
to holders, so a buyback is effectively a strategy using their balance sheet to put a put or a floor in the STRC price.
300
00:54:54.900 --> 00:54:59.680
If the div- if the principle never gets returned over a saturation of time, it's very difficult to value these instruments.
301
00:55:00.360 --> 00:55:14.800
By executing open market transactions that are financed by the Bitcoin on the balance sheet and not by new capital raise, the market and, you know, sophisticated participants are starting to see, oh, you know, Strategy can use their existing assets to support the price of this instrument.
302
00:55:15.040 --> 00:55:28.000
That's kind of similar, kind of similar to a debt instrument where ultimately the principle's returned in the form of a, you know, the, the note coming due. Dan, thank you so much for your time and your insight.
303
00:55:28.340 --> 00:55:38.200
Um, as I hope, I hope Stretch continues to get really interesting, 'cause it's more fun to talk about when things happen to it. So- [laughs]... cheers. But thank you so much for your time.
304
00:55:38.360 --> 00:55:42.540
We'll, uh, we'll catch you around. Thanks, Dan. Bye, guys. Thank you. Great talking with you.
305
00:55:45.640 --> 00:55:56.980
The STRC, uh, rips, uh, are surprisingly, uh, popular on YouTube whenever we put them out, so excited to see, uh, how the people like that. If you have comments,
306
00:55:57.880 --> 00:56:13.860
drop them in the video replies below, positive and/or negative. So on that note, we're gonna go to gas fire and new s- uh, small modular reactors right after a word from our sponsor, Ligos.
307
00:56:18.560 --> 00:56:30.240
[static] Bitcoin took a little bit of a dive over the weekend, and if you're using a shady Bitcoin-backed lender, you might be asking yourself, do they still have my coins? Well, I'm gonna tell y'all something.
308
00:56:30.560 --> 00:56:42.500
If you're using Ligos Finance, our preferred Bitcoin-backed lender here at Blockspace, you don't have to ask that question, because you always control the keys to your Bitcoin when you are using them for a Bitcoin-backed loan.
309
00:56:43.120 --> 00:56:53.360
That's right, you heard that correctly. With Ligos Finance, they use Bitcoin native smart contracts to make sure you always control the Bitcoin that you are collat- using for collateral for your loan.
310
00:56:53.980 --> 00:57:04.820
With Ligos Finance, there's no bridging, no wrapping, no rehypothecation, no funny business, just competitive rates as low as 8.5 to 10% APR. Go to ligos.finance to learn more.
311
00:57:06.620 --> 00:57:17.840
All right, let's bring it on home with a little nuke, baby, nuke. Nukes. Guess kind, not the warfare kind- The good kind of, the good kind of nukes. The, the white pill... the good nuke. The happy math.
312
00:57:18.880 --> 00:57:30.880
This headline coming at us from Data Center Dynamics, Blue Energy, GE Vernova, Hitachi advance 2.5 gigawatt gas plus nuclear plant in Texas to power up to 1.5 gigawatt data center.
313
00:57:31.800 --> 00:57:46.740
Intended to supply one gigawatt of power to the Crusoe data center by 2030. Now, this is not a new project. I just wanna get that out of the way in front. This was announced back in, let me see, May,
314
00:57:47.880 --> 00:58:01.040
and the most recent development is that, um, Blue Energy and GE Vernova Hitachi have signed the next stage agreement to perform engineering, licensing, and safety analysis work for the site.
315
00:58:01.420 --> 00:58:15.500
The project still requires a final investment decision in 2027. There is a lot to unpack here, Charlie. So this will be in the Port of Victoria area, Texas. Crusoe's campus is in Calhoun County, Texas.
316
00:58:16.320 --> 00:58:26.980
Crusoe, the data center developer, 1,600 acre site. The power developer is Blue Energy. Ver- GE Vernova are supplying the gas equipment,
317
00:58:28.060 --> 00:58:38.920
and also, uh, going to supply the nuclear technology as well with BWRX-300 small modular reactors. The nuclear build
318
00:58:40.260 --> 00:58:55.140
is, believe, five SMRs at 300 megawatts each for up to 1.5 gigawatts. For the initial gas generation, they're targeting one gigawatt for 2030, and first nuclear power at, by 2032.
319
00:58:55.660 --> 00:59:10.460
An incredibly aggressive timeline for this. Again, the final investment decision's coming in 2027, and also the Nuclear Regulatory Commission's construction permit application is planned for 2027 as well.
320
00:59:10.500 --> 00:59:20.260
The timeline here is incredibly aggressive, Charlie. Uh, fir- first of all, as far as I understand it, we actually don't have any small modular reactors that- No, we don't...
321
00:59:20.270 --> 00:59:32.140
at least, well- We have, we have zero SMRs, zero small modular reactors in the West. There are some in the East, but there's none in the USA.
322
00:59:33.380 --> 00:59:44.260
Yeah, and so a lot of jawing as to getting those online, but so far none have been put online. But they're targeting 2032, so six years, uh, out for this.
323
00:59:45.240 --> 00:59:58.340
And they said that it could take roughly 24 to 36 months from first nuclear concrete to fuel loading for these. It's part of the bull case for the small modular reactors. They're much quicker to get online.
324
00:59:58.820 --> 01:00:06.200
And little point is, I had to do a little... I was like, "What do you mean nuclear concrete?" It's a special type of concrete you gotta pour and, like, seal and case stuff in.
325
01:00:06.260 --> 01:00:16.200
So I was like, [laughs] "What do you mean nuclear concrete?" Okay, so yeah. Well, this is part of the reason why these things take so long. The, the permitting and the regulation process for this is so painstaking.
326
01:00:17.000 --> 01:00:31.960
I mean, if you... So for that nuclear concrete, if you are an inch short of what is mandated by regulations, you have to tear out everything and start over. So the, the whole process takes a long time.
327
01:00:32.000 --> 01:00:44.556
It's absolutely choked and mired in red tape. But they're pretty-- They're saying here that they are somewhat confident they can get this done by 2032. Um, that being said, it's pretty aggressive.
328
01:00:44.596 --> 01:01:00.156
There are also questions about f- uh, how this is gonna be funded. Nuke is not cheap, neither is nat gas. Um, and so b- I mean, Blue Energy, um, let's see. Blue Energy raised 380 million earlier in 2026.
329
01:01:00.236 --> 01:01:12.316
Constellation Energy has made a strategic equity investment. But- Which is pretty notable 'cause Constellation Energy has the largest nuclear fleet in the United States. Yes. So they are like the nuclear company, and so
330
01:01:13.856 --> 01:01:19.956
this seems like validation to me. Yeah. Yeah, 100%. But that being said, that shouldn't be conflated with what they're doing here.
331
01:01:20.036 --> 01:01:26.076
This is gonna require billions of dollars worth of financing, and n- none of that has been buttoned up yet.
332
01:01:26.096 --> 01:01:32.466
And I don't think they will get a clear idea of what that actually looks like until this final investment decis- decision comes in 2027.
333
01:01:33.716 --> 01:01:41.396
Uh, but needless to say, though, part of the reason we chose this is that it's very interesting to see this dual gas and nuclear strategy.
334
01:01:41.436 --> 01:01:53.656
We've seen gas generation becoming the stop gap for projects that are waiting for grid interconnection. You know, I mean, we just talked about it with, um, the OpenAI project that we just discussed.
335
01:01:53.696 --> 01:02:01.736
I mean, they're, they're talking about deploying like 9.7 gigawatts of gas, I believe, for that Ohio project. Just an insane amount of energy.
336
01:02:01.776 --> 01:02:10.216
Natural gas will be the base load that is either permanent or a stop gap on the way to things like nuclear or interconnection to the grid for a lot of these projects.
337
01:02:10.816 --> 01:02:13.406
The other reason I wanted to bring this up, Charlie, [lip smack]
338
01:02:13.596 --> 01:02:30.196
the novel dual fuel approach notwithstanding, is the fact that, you know, when you look at this project, it becomes-- and, and others like it, it becomes increasingly clear that the nuclear renaissance in the US is being driven entirely by hyperscalers and data center companies.
339
01:02:30.706 --> 01:02:45.156
The, the, the-- I think the, the first shot across the bow with this was Microsoft with their 20-year PPA restart of, uh, the Three Mile Island unit one in Pennsylvania with Constellation Energy, 835 me- five megawatts of capacity.
340
01:02:45.956 --> 01:02:56.016
Uh, 2027 restart for that. That was announced, I believe, last year, and it made headlines because it was the first time we had seen this, and it made, I think, most people realize, oh,
341
01:02:57.056 --> 01:03:05.496
uh, th- [chuckles] there are some serious bottlenecks here, and also the big companies, the Max Seven are serious about deploying capital to get this done. Another notable one, Google,
342
01:03:06.516 --> 01:03:17.856
with NextEra Energy restarting the Duane Arnold Iowa nuclear plant for 650 megawatts, targeted for Q1 2029 on a 25-year agreement.
343
01:03:18.456 --> 01:03:31.196
Meta also with Constellation Energy seeking to restart over 1.1 gigawatts of nuclear load in Clinton, Illinois. Um, or sorry, preserve and expand this. It's not a restart.
344
01:03:31.256 --> 01:03:44.936
Me- Meta's 20-year deal, um, begins in June 2027. It keeps Clinton operating after state subsidies expire. 30 megawatts of, uh, a, a 30-megawatt uprate planned as well. And we can continue going down the list.
345
01:03:45.036 --> 01:03:57.596
Meta has another preserve and expand agreement for four- 400 and-- for like 2.2 gigawatts worth of load roughly in, um, Beaver Valley [lip smack] with Vistra.
346
01:03:58.276 --> 01:04:10.296
Uh, that's a 20-year agreement, 433 megawatts of additional reactor output is expected through uprates in early 2030. Google's got more. Amazon's got a few in Virginia and Washington.
347
01:04:11.096 --> 01:04:24.896
Meta's got a bunch of them, one in Pike County, Ohio with Oklo. This is for new advanced reactors of up to 2 point-- 1.2 gigawatts, targeting early 2030 for potential first energization there.
348
01:04:25.476 --> 01:04:32.296
Switch has a deal with Oklo. Equinix has a deal with Oklo. Um, Crusoe's got this one.
349
01:04:32.436 --> 01:04:44.716
I mean, you just go down this list and all of the nuclear reactors that are either being restarted or being supported by this or, um, new reactors being built, they're all contracts with these data center companies.
350
01:04:46.576 --> 01:04:56.156
Yeah. Every hyperscaler has their pet nuclear project. Um, there's not that many that aren't pet. I mean, I will point out to Microsoft restarting Three Mile Island.
351
01:04:56.235 --> 01:05:07.016
That was a pretty significant, uh, thing culturally as well as from a actual, um, business and operations, uh, power standpoint. So,
352
01:05:08.376 --> 01:05:20.136
um, I don't have like a coherent, uh, train of thought here, but I will say like I was kinda looking into like what are the different SMRs. Like are they all the same type thing? And there's a bunch of different designs.
353
01:05:20.316 --> 01:05:37.096
I'm not deep enough in the weeds to qualify them except that GE Vernona's design, they, uh, they say is the simplest. Which to me- I'm sure they all say that [laughs] I don't know. I don't-- I'm like, okay. Yeah.
354
01:05:37.116 --> 01:05:45.176
[chuckles] I mean, uh, I would be bullish on the most simple one. These seems like incredibly complex, uh, things to build.
355
01:05:45.276 --> 01:06:02.136
So whoever can convince me they have the simplest design, uh, probably has the most compelling, uh, thesis to me. Um, I think another angle that I have is that everybody's got like an early 2030s timeline.
356
01:06:02.556 --> 01:06:19.986
GE Vernona has the fastest, obviously. But, um, if it, if p-- it like-- because we-- this is such an unproven exploratory like field of, of entirely new energy like, uh, generation,
357
01:06:21.036 --> 01:06:24.756
um, right, we can't-- I don't wanna be betting on this.
358
01:06:24.836 --> 01:06:36.476
However, if in the early 2030s we start seeing these turn on and they actually do, you know, power 200, 300 plus megawatts of power, like this is an-- that's an energy revolution.
359
01:06:36.536 --> 01:06:50.387
Like that totally completely changes, um, pretty much everybody's assumptions about... what the grid looks like in, uh, the US and the West for, like, the rest of my entire life. Yeah.
360
01:06:50.548 --> 01:07:04.908
It, it, it would be an incredible, like you said, revolution in terms of capacity in the US. My one commie adjacent gripe with this is- Give me that commie take, Colin... it's all gonna go to the data centers.
361
01:07:05.468 --> 01:07:17.768
Now, granted, this will free up generation for oth- for residences and other commercial and industrial uses that aren't going to be taxed by these data centers, and by, I mean taxing the grid in terms of stressing it.
362
01:07:18.328 --> 01:07:30.168
So you'll free up load elsewhere. So it's not like there's not a problem with the fact that, uh, there's not a problem i- in the sense that, well, you're just gonna have load available for other uses.
363
01:07:30.668 --> 01:07:41.538
But when I think about nuclear energy be- the, the fact that it should be the backbone of base load for most US states at this point, considering we are so far removed from the breakthroughs of the '40s and '50s,
364
01:07:43.248 --> 01:07:54.988
I, I would, I would like to see some ways in which we can also start reworking this into the grid at large. Now, some of this might produce excess power that will go to the grid, which would be great.
365
01:07:55.428 --> 01:08:06.388
It does also raise an interesting question, though. What happens in 10 years when, or 20 years, when we have a breakthrough in computing density to where we don't need these data centers to be as large anymore?
366
01:08:06.488 --> 01:08:11.428
Maybe that, that happens, maybe it doesn't. But if this, if these data centers disappear,
367
01:08:12.388 --> 01:08:21.608
will we actually be able to run these nuclear plants properly for the remaining capacity on the grid, for the remaining load on the grid?
368
01:08:22.188 --> 01:08:29.918
That's one question I have, 'cause when we had, um, we, we had a nuclear expert on, uh, at the show beginning of the year, uh, Dr.
369
01:08:30.148 --> 01:08:41.908
Hashamian, and I asked him why haven't we seen more nuclear around the US, and part of his answer was just the fact that it wasn't economically viable in certain areas. And so I'm just imagining, you know, kind of
370
01:08:43.248 --> 01:08:57.028
playing the tape through in my head, or what does a future look like where, you know, you have these tech companies coming in restarting nuclear, building more nuclear, and then suddenly they're, they don't need it anymore, or they go bankrupt, or those data centers disappear, what happens to that nuclear load?
371
01:08:57.848 --> 01:09:07.208
Can we actually get... Can we actually connect it to the rest of the grid? Can it be furnished profitably to service everyone else? Yeah.
372
01:09:07.988 --> 01:09:19.348
I guess all of that to say is I, I hope that there's a way that this can then snowball into building grid for, or building nuke for the rest of the grid in addition to these data centers. I'm a big nuclear bull.
373
01:09:19.488 --> 01:09:29.788
I think that we should have more of it as long as it makes sense economically. In terms of clean energy and driving down cost, there's n- you know, there, there's no, there's no second best.
374
01:09:29.828 --> 01:09:38.628
There really is no second best. Who isn't a nuclear, who isn't pro-nuclear at this point? Oh. 'Cause it used to be- Oh, buddy. Like the entire green energy lobby?
375
01:09:39.048 --> 01:09:50.688
Most li- I- Most people in the- Yeah, but I think a lot of the greenies, I think a lot of, like, the, the actual grassroots green folks have, have come around to be pro, uh, nuclear or, like, less anti-nuclear.
376
01:09:50.728 --> 01:10:01.318
I think it's really just kinda like the, the, the walking dead husk of, like, green energy lobby that has found itself- I don't know. Yeah. I'm gonna disagree on this one. Yeah.
377
01:10:01.348 --> 01:10:06.677
I think a lot of people who are on the environmental left think that it is polluting.
378
01:10:07.288 --> 01:10:13.928
They, they, they don't actually understand how the fuel is either recycled or stored, and they think that there's no safe way to store it.
379
01:10:14.888 --> 01:10:26.388
Um, and so I think that you're right that there are some on the environmental left who are a little more enlightened to this, but a lot of people have no idea how the actual supply chain of nuclear fuel works.
380
01:10:27.008 --> 01:10:35.508
They don't know- Yeah... that part of it can be recycled. They don't know that it's often stored on site in these massive steel and concrete silos, right?
381
01:10:36.048 --> 01:10:49.008
So- Or it's stored as, uh, the proposal for the Oakla- this has been burning a hole in my pocket this whole time. Oklahoma nuclear hub plans deep borehole waste disposal.
382
01:10:49.588 --> 01:11:00.918
So they're gonna, in, in the, in Oklahoma they're pl- they're, uh, planning on, um, drilling a big-ass hole and putting the nuclear waste down there. I, sorry to interrupt you, but I had- No, I, I-... to bring that up...
383
01:11:00.938 --> 01:11:12.238
appreciate that. Yeah. I mean, I think that that really go... Like, I legitimately think that a lot of people, when they think about nuclear waste, like The Simpsons plays in their head- Yeah...
384
01:11:12.268 --> 01:11:20.318
where, like, Homer is, like, going down to, like, Lake Springfield and just dumping a, like, corn silo full of green liquid into the lake.
385
01:11:20.808 --> 01:11:32.448
Or they legitimately imagine, like, oil tank, you know, like oil tanks of toxic waste sloshing around in a cave somewhere. Yeah. Like, they, I, I legitimately mean this.
386
01:11:32.568 --> 01:11:37.268
I think Harry Sudock was the first one to mention this on a What Bitcoin Did interview a long time ago.
387
01:11:38.008 --> 01:11:50.498
I think The Simpsons and Matt Groening unknowingly did the most cultural damage to the nuclear debate out of literally any meltdown, like more than any of the meltdowns that we've seen.
388
01:11:50.498 --> 01:11:54.728
Well, yeah, that and Chernobyl, dude. I mean- Yeah... but yes. Yeah.
389
01:11:55.108 --> 01:12:07.568
Um, and then you can't, like, you know, if you talk about Chernobyl with most of these people, you can't get into how the cooling design was faulty and we don't use graphite anymore, and how there were a number of warnings that went dismissed.
390
01:12:07.668 --> 01:12:20.128
I mean, the whole thing is difficult to unpack for the average person. Uh, that being said, I do think that we are probably entering a new phase where the economics are such to where
391
01:12:22.148 --> 01:12:29.128
they just don't care about the critics anymore, and, like, these companies are gonna go and do it. We've seen that with the green energy debate writ large. Yeah.
392
01:12:29.168 --> 01:12:43.048
I mean, we're about to re- we're about to, you know, reverse a decade or two of building solar and wind on, like, a single plant or a single data center from OpenAI in, in states like Ohio, right?
393
01:12:43.108 --> 01:12:55.406
Like, they're just gonna start building more nat gas- Oh, it's, it is real- oh, it's just, it's, uh, it is conspicuous how all the, all the EI whatever, uh- ESG...
394
01:12:55.456 --> 01:13:08.956
ESG folks were like, "And now we're pivoting to, uh, energy resilience." I'm like, "You know why we had to pivot to energy resilience? [laughs] Because of your dang private equity fund and having its directives."
395
01:13:09.036 --> 01:13:17.196
Okay, for all those of you who are done with my ranting, um, this is how we, I feel about, uh, nuclear. Uh, here's Homer Simpson.
396
01:13:17.616 --> 01:13:38.676
If you're familiar with one of the, uh, intros to The Simpsons and a common recurring theme, Homer Simpson sits in the control room [laughs] of the nuclear plant as his job, and, uh, on the wall there is like, uh, you know, some like PSA safety thing that says, "Don't forget to something something," but he's obviously covered it with pictures of his daughter.
397
01:13:39.096 --> 01:13:51.656
And so, uh, he's changed the words to do it for her. So do it for your kids. Pro-nuclear, pro-energy, do it for the future generations of United States, and maybe some data centers.
398
01:13:52.296 --> 01:14:05.626
On that note, thank you so much for listening to Block Space Live. We do this every single weekday at 1:00 PM Eastern. We are Compute's daily live show covering quick hits on AI, data centers, energy [laughs]
399
01:14:05.626 --> 01:14:21.036
, emerging technology, and markets. If you like the stream but miss it, you can catch it in podcast form anywhere podcasts are found, and hit our website blockspace.media for even more content, including our newsletter.
400
01:14:21.076 --> 01:14:36.956
That's blockspace.media. This show is brought to you by CleanSpark, NASDAQ listed at ticker CLSK. I'll catch you tomorrow. I'm Charlie. I'm Colin. Peace. [dramatic music]
