WEBVTT

1
00:00:00.060 --> 00:00:12.660
On y'all. Welcome back to Blockspace Live, presented by CleanSpark. Packed docket today, specifically with guests. We will kick off with a story of Hut 8's Logan Prairie $5 billion

2
00:00:13.600 --> 00:00:18.050
data center build hitting a stalling point with a local commission.

3
00:00:18.800 --> 00:00:34.589
Following that, we have Michael San Miguel of Luxer on to talk about the GPU secondary market and what they are seeing from their GPU compute desk, as we see rental rates continue to skyrocket, Charlie, and demand for models as old as the A100.

4
00:00:35.440 --> 00:00:48.060
After that, we will have Matt Black on to talk about the recent sweep of funds from cold cards. I saw figures anywhere from 600 to 1,000 Bitcoin stolen in this, Charlie.

5
00:00:48.120 --> 00:00:54.320
So taking a little bit of a divergence from the compute route, but we felt like this was an absolutely necessary thing to cover.

6
00:00:54.400 --> 00:01:05.950
So Matt Black, co-founder of Lagos Finance, is going to give us a breakdown, and our fearless co-host, Charlie Spears, will be furnishing some hot takes and questions about this incident. After

7
00:01:07.220 --> 00:01:19.220
m- Matt Black, we have a note on how Kimi K3 was trained on-- reportedly trained on 20,000 Nvidia Hoppers, and what this means for the current export control regime.

8
00:01:19.260 --> 00:01:24.760
We're gonna untangle some of the differences between what's actually banned for Chinese companies and where they have wiggle room.

9
00:01:25.400 --> 00:01:39.540
And then we will have Carmen Lee of Compute Exchange on to talk about compute forwards, rental prices, and the token forwards that they are working on over at Compute Exchange. That's right.

10
00:01:39.660 --> 00:01:51.700
Blockspace goes live every single weekday at 1:00 p.m. Eastern. We are compute's daily live stream featuring quick hits on AI, data centers, emerging technology, and markets.

11
00:01:52.760 --> 00:01:58.020
If you like the live stream, this turns into a podcast. You can find that anywhere podcasts are found.

12
00:01:58.120 --> 00:02:13.900
And if you can't get enough of the podcast or the live stream, you can continue discovering more of the Blockspace content powerhouse at our website, blockspace.media. The URL is blockspace.media. Head there.

13
00:02:14.100 --> 00:02:26.840
We have a lot of written content too. This show is brought to you by CleanSpark, NASDAQ-listed ticker CLSK. More on CleanSpark later on in the show. Before we go to the news column, gotta kick it off.

14
00:02:27.360 --> 00:02:40.500
Little hash rate index. Gotta do a little hash rate index update from Luxer here. [lips smack] And you know what, Charlie? The more things change, the more they stay the same, my friend.

15
00:02:40.510 --> 00:02:56.150
[chuckles] We're looking at having a roughly negative seven, 0.74% difficulty adjustment. We're almost halfway through the current epoch, and we're coming off of a negative adjustment of roughly 1% from the last epoch.

16
00:02:57.120 --> 00:03:10.460
And this comes at a time, Charlie, where hash price looking a little better because of these difficulty adjustments. We're holding the line at $30 per hash per day. It's currently $31.59 per hash per day.

17
00:03:10.520 --> 00:03:20.080
But Bitcoin taking a hit right now. Don't wanna read too far into the tea leaves. I don't think that this hack with the cold card has breached containment yet.

18
00:03:20.860 --> 00:03:36.920
Maybe some folks are selling because of this, but we are starting to see more volatility in the markets. Bitcoin has held up remarkably well since cratering to below 60,000 in recent months. But Bitcoin down to 62,840.

19
00:03:37.580 --> 00:03:47.560
Hash rate currently stands at 886.91 exahashes. And s- new week, same story, Charlie.

20
00:03:47.820 --> 00:04:00.710
I don't see much wiggle room for hash rate right now with four CP season in full swing, with Bitcoin's price still relatively depressed and hash price depressed as a result. And I'm depressed. I'm just kidding.

21
00:04:00.710 --> 00:04:10.890
[chuckles] Everyone's depressed. Funnily enough, you know, it, it takes a tragedy to bring people together. Uh, Bitcoin Twitter has been popping off over the last 24 hours with this cold card hack.

22
00:04:11.580 --> 00:04:20.320
Uh, no one talking about hash rate though, because quite frankly, there's not too much to unpack here, just a very subdued environment.

23
00:04:20.700 --> 00:04:28.100
That being said, if you have low enough power cost, high efficiency machines, you're kinda liking what you're seeing right now.

24
00:04:28.180 --> 00:04:37.060
There's really no- Or, or maybe you're okay with just burning money to keep the lights on- [chuckles]... because you're a true believer, of which there are increasingly few left.

25
00:04:37.960 --> 00:04:42.000
But I guess my, my last comment on hash price is calling no.

26
00:04:42.040 --> 00:04:53.080
Like, with this cold heart, this, this cold card hack, which we will talk about later on in the show, um, there is cause for people sweeping their funds to move them to more secure places.

27
00:04:53.120 --> 00:05:06.940
This could result in, uh, sweep transactions or coin joins, uh, very urgent high fee transactions. However, we really have not seen that on Bitcoin yet, and it's unclear how much those could, those could contribute.

28
00:05:07.260 --> 00:05:18.460
The other one I'll put on your radar is, due to the exposure of the public keys or the private keys, there could be a scenario. Again, I'm talking gobbledygook to a lot of the miners here, but this is important.

29
00:05:18.540 --> 00:05:30.200
Put this on your radar. That we could see RBF sniping wars. Now, if that happens and it becomes a major theme, we'll cover it 'cause there's very few people who can. But other than that, uh, very quiet.

30
00:05:30.520 --> 00:05:41.720
It's too quiet, one might say. All quiet on the Blockspace front. I like that angle there though, Charlie. We might see a pop in transaction fees for admittedly a very sad reason- Yeah...

31
00:05:41.860 --> 00:05:55.940
as potential victims try to sweep their funds to new addresses and attackers try to front run that by RBF- Yep... transactions. But so far, uh, you know, transaction fees not really doing too much. About 1%

32
00:05:57.100 --> 00:06:08.444
of the total, uh, o- o- o- of, uh, total block rewards in the last 24 hours. Okay. Let's go ahead and move on to our first story of the day, Charlie.

33
00:06:08.454 --> 00:06:18.524
Uh, let's do a quick story 'cause we've got [laughs] multiple guests here today. Yeah, so this is an update on a data center for Hut 8 that we've covered in the past.

34
00:06:18.744 --> 00:06:28.184
They have had s- they've been stonewalled by this local council. This is Hut 8's 500 megawatt Illinois data center that was having some trouble, trouble with approving.

35
00:06:28.964 --> 00:06:41.264
Uh, earlier in the year, we covered the local council's meeting on the data center, and right now there has been a cancellation for the vote for zoning this data center.

36
00:06:41.304 --> 00:06:52.324
That doesn't mean that it is dead, I wanna get that out in front of the story. But it is stalled, and I believe the earliest time that it might be reconsidered is in September.

37
00:06:52.844 --> 00:07:01.604
So we've got about a month or so before this might reenter the docket for zoning approval. But the details are as such.

38
00:07:01.644 --> 00:07:17.684
Hut 8's proposed 500 megawatt data center, $5 billion projected spend for the Logan Prairie project hit another permitting wall when the village of La- Latham, Illinois, canceled an August 3rd vote on annexing and rezoning the site.

39
00:07:18.264 --> 00:07:27.324
The officials pulled the vote because they determined the proposal lacked the votes to pass. Again, crucially, the application has not been withdrawn.

40
00:07:27.524 --> 00:07:38.864
That application was filed by Logan Prairie Data Center LLC, a subsidiary of Hut 8. The attorney, the village attorneys called it postponed with a new date TBD, so it's stalled, not dead.

41
00:07:40.864 --> 00:07:48.374
What's really interesting about this to me, Charlie, is the fact of the matter that s- there are very few people in this township,

42
00:07:49.364 --> 00:08:02.434
roughly 350 residents, and that their complaints were enough to stonewall this data center in a trend that we have covered extensively here at Blockspace, the local pushback to some of these sites.

43
00:08:02.504 --> 00:08:14.244
And the residents have concerns which don't override, according to them, the perceived benefits that Hut 8 is pitching for their locality.

44
00:08:14.544 --> 00:08:29.204
So Hut 8 says that they plan to spend roughly $8 to $11 million per year for 15 years for community benefit programs. That's against the annual revenue of this township of $291,000.

45
00:08:29.264 --> 00:08:40.464
So [laughs] anywhere from 30 to- Get that bag. Yeah, g- I mean, get that... They do it. I mean, look, what is it? Y- your $5 billion project, I think the size of this is what Asher or Hut 8 estimates.

46
00:08:40.544 --> 00:08:51.094
So, um, if you, if it takes you under eight figures to, to make the county [laughs] like you, do that. [laughs] You know, like buying friends.

47
00:08:51.544 --> 00:08:59.094
I mean, you have to do this community engagement, and Hut 8's actually been one of the standard-bearers of this in their deals. Yeah.

48
00:08:59.124 --> 00:09:13.624
They've made a huge point of trying to educate and engage the community, and making legitimate financial stakes in the community to show that they are trying to be a good partner to these areas.

49
00:09:14.144 --> 00:09:29.784
But not only is there that $8 to $11 million per year contingent, Charlie, Hut 8 estimates that the property could generate $65 million a year on average in property taxes for a projected 30 years.

50
00:09:29.864 --> 00:09:37.444
I guess that's the lifespan they're putting on this data center, or at least that's how far they see it moving out into the future or in the midterm, right?

51
00:09:38.204 --> 00:09:55.504
So you're talking about $75-plus billion that will enter this community of 350 people, and that's real money that they could use for enriching schools in the area, for infrastructure, for parks, all of these things.

52
00:09:55.514 --> 00:10:11.264
But that's falling on deaf ears because there are a few key worries that the locality has. Uh, first of all, or, or first up, they're really worried about Hut 8 gobbling up farmland.

53
00:10:11.944 --> 00:10:14.944
It's one of the things that was mentioned in this.

54
00:10:14.964 --> 00:10:25.364
It, there was less of a worry about water, but they're worried about farmland being occupied, they're worried about energy prices rising, and they're worried about noise.

55
00:10:25.424 --> 00:10:33.434
So that, those were the key pushbacks on this site, and a few other things holding it up. There's an annexation dependency.

56
00:10:33.504 --> 00:10:42.374
The benefit agreement is contingent on Latham annexing not just the project property, but intervening parcels whose owners would each separately have to petition for annexation.

57
00:10:42.384 --> 00:10:51.284
So there's like a multi-party consent problem that's at the center of this too. There's also problems with a parallel county in terms of a data center moratorium.

58
00:10:51.384 --> 00:11:03.184
Logan County passed a 12-month data center moratorium on May 19th, but the county state's attorney later said that the moratorium was invalid because it skipped the required zoning process.

59
00:11:03.304 --> 00:11:13.364
Um, the county officials also found technical defects in Hut 8's application, according to the reporting. Hut 8 fixes and resubmits. County hearings can't happen before December.

60
00:11:14.084 --> 00:11:27.724
So again, there's no meeting on the books for this, but the earliest we could see any sort of re-upping of this zoning push would be in September. There's also a question about the power f- that was needed for this site.

61
00:11:28.284 --> 00:11:38.444
The 500 megawatts would draw from a, a local substation, but the c- utility, Ameren, nor MISO, has publicly confirmed an approved interconnection.

62
00:11:39.184 --> 00:11:51.444
So according to the reporting that we saw, the power is not even guaranteed for this site yet. 500 megawatts, big chunk, especially for some of these rural areas.

63
00:11:51.504 --> 00:12:00.204
I would imagine that was probably the largest consideration with the local council with regards to whether or not they wanted to push zoning through.

64
00:12:00.604 --> 00:12:11.524
They might ask Hut 8, "Hey, you need to actually have interconnection approval before we can take this seriously." That's just speculation on my part, though. I'm not sure if that, those conversations are underway.

65
00:12:12.564 --> 00:12:25.372
But That is the TLDR for this news story. The reason we bring it up, this site in Illinois is the largest new site in Hut 8's pipeline.

66
00:12:25.392 --> 00:12:36.792
It's the only one that they have disclosed or has been disclosed through reporting as being worked on, and it's equal to the phase, a 500-megawatt phase that they are working on to extend Beacon Point.

67
00:12:36.992 --> 00:12:40.512
So a pretty notable piece of the pipeline. Yep.

68
00:12:40.912 --> 00:12:53.412
Uh, not a lot of comms except this will be a great case study of a company who is really proactive with trying to court the community, uh, and we'll see how the community responds, so.

69
00:12:53.502 --> 00:13:07.032
'Cause we've seen the op- we've seen the other side of being surprised by community pushback for good and bad reasons. So we'll h- eyes on this one. Before-- So we're gonna keep on going.

70
00:13:07.132 --> 00:13:18.452
We got Mike San Miguel from Luxur in the wings. We'll bring him on up here, we'll talk about GPUs and GPU racks specifically. Before we bring Mike up, a word from our sponsor, CleanSpark.

71
00:13:18.542 --> 00:13:33.392
[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.

72
00:13:35.252 --> 00:13:46.732
This is our proof of work. We are setting the standard for what's next. Learn more about the intersection of energy and Bitcoin at cleanspark.com.

73
00:13:49.712 --> 00:14:01.292
All righty, we've got Mike San Miguel, the Senior Business Development Manager at Luxur's hardware desk. We are going to get him on the stage. Mike, welcome to the show, sir. Thanks for joining.

74
00:14:01.952 --> 00:14:14.852
Hey, Charlie and Colin, always a pleasure. Good to see y'all. Likewise. Good to see you too, man. So I wanna kick off today's segment with a piece that you co-published with Ian Filpot on Hashrate Index.

75
00:14:15.452 --> 00:14:28.892
Specifically- Yeah... NVIDIA's Vera Rubin, why the rack is the new GPU. The excerpt here, "NVIDIA's Vera Rubin isn't just a faster GPU, it's a rack-scale reference architecture that changes what you buy."

76
00:14:29.452 --> 00:14:40.222
And so my first question with this, can you break down how this form factor is changing the, you know, consideration for how we view GPU rollouts?

77
00:14:40.272 --> 00:14:51.212
And also, how much demand are we seeing from buyers on the compute side of things for this new form factor? Gotcha. Yeah, yeah, great question.

78
00:14:51.772 --> 00:15:05.872
So, you know, by and large, what-- There's a couple trends that factored into this rack-scale rollout and sort of this pivot towards rack scale. So, um, it's not a new thing in NVIDIA's portfolio.

79
00:15:06.292 --> 00:15:16.432
Um, within the current Blackwell generation, there's the GB300, before that the GB200, uh, which is a, uh, another rack-scale solution.

80
00:15:16.832 --> 00:15:31.272
Um, the technical trend is that they started integrating Arm-based processors into these. So as opposed to your x86, your Intel and AMD procs for your CPU side, uh, these are using the Arm-based processors.

81
00:15:31.412 --> 00:15:39.622
Uh, it hasn't matured quite, uh, you know, before to the point it is now. Um, that has to do with software compatibility.

82
00:15:40.272 --> 00:15:51.232
Uh, but a lot more of these models are being made to, to run on, uh, the Arm-based, uh, procs and servers. Um, the, the other trend is a financial model, right?

83
00:15:51.352 --> 00:16:04.572
So at GTC, uh, s- in San Jose this year, uh, Jensen announced sort of this move more towards the, uh, the token factory, and that's really what these are, are by and large, uh, designed to do.

84
00:16:04.612 --> 00:16:13.552
So you've got the first, uh, name, Vera, in the new generation, which is the CPU, the Arm CPU. Rubin, which is the GPU.

85
00:16:13.892 --> 00:16:23.592
Um, the next thing with Vera Rubin is that they're integrating an AI accelerator and AI ASIC specifically for token generation on the inferencing side.

86
00:16:24.192 --> 00:16:32.252
So those are the two trends that are sort of, uh, pushing the way for these. Uh, probably won't take over the whole portfolio.

87
00:16:32.472 --> 00:16:44.032
You know, x86 platforms are probably not gonna go anywhere, but that's what's pushing, uh, this, this trend. It seems to me, like with each new advancement,

88
00:16:45.552 --> 00:16:59.472
I mean, you can have incredible efficiency gains, but a lot of these neo clouds and powered shells are outfitting their data centers for a very specific architecture, and it's not very easy to just backwards-- uh, make them backwards compatible, right?

89
00:16:59.552 --> 00:17:12.072
And you have billions of dollars invested in these things, but ultimately y- you can't just rip all of that out without having a huge capital expense. Yeah. Well, and especially for the Vera Rubin stacks.

90
00:17:12.112 --> 00:17:25.882
So, um, not to get too technical, but there's essentially five different racks that go into these. It's 80 different manufacturers. The, the weight of these is completely, uh, you know, a, a new [chuckles]

91
00:17:26.452 --> 00:17:35.172
scale that we, we haven't seen before, even in the previous Grace Blackwells. Uh, so just the cement slabs, the floors that these have to roll on has to be really built up.

92
00:17:35.892 --> 00:17:47.992
Uh, and the, the management, uh, for this even goes down to the grid level for managing power. So it's, it's a new ball game, but it, you know, to your point, it's not gonna replace previous installs.

93
00:17:48.772 --> 00:17:53.412
Um, you know, i- eventually those will migrate down, uh, stream.

94
00:17:54.122 --> 00:18:02.312
But currently, you know, we're, we're seeing older generation GPUs at close to a hundred percent utilization, so this isn't gonna be a rip and replace of the whole infrastructure.

95
00:18:03.012 --> 00:18:11.811
I'm, I'm glad you mentioned that, Mike, because that kind of leads well into my next question, and I have a tweet here from Carmen Lee, who will be joining us at the end of the show.

96
00:18:12.622 --> 00:18:21.652
She says here, quote, "We've seen fifteen to thirty percent price increases year to date across our A100, H100, H200, and B200 NeoCloud on-demand indices."

97
00:18:22.540 --> 00:18:27.360
And if you look at this, every single industry year to date is just up and to the right.

98
00:18:27.920 --> 00:18:45.260
A100's up 20% year to date, H100's up 25%, A- B200's up nearly 30%, and H2, uh, GPU index H200 is up 14.4%. Supply is clearly very tight right now.

99
00:18:45.780 --> 00:18:53.680
So I'm curious, with regards to what y'all are seeing on the compute desk and in the secondary market, how much demand are we really seeing for those older models?

100
00:18:53.700 --> 00:19:05.940
And also, how much is this pushing consumers down the curve towards, you know, little brother AMD rather than getting their hands on the latest from Nvidia? Yeah, great question.

101
00:19:06.260 --> 00:19:10.820
Uh, so on the, the used front, we're seeing a lot of movement there.

102
00:19:10.960 --> 00:19:24.630
So, um, H100 back in April, uh, price point depending on the configuration and the host, probably anywhere from about one fifty to, to one seventy, um, k per, per node in US dollars.

103
00:19:25.300 --> 00:19:32.480
Uh, now they're moving, uh, similar specs, uh, probably two sixty to two eighty, and that's a good deal.

104
00:19:33.300 --> 00:19:44.220
Uh, occasionally you'll find new old stock that for some reason hadn't been deployed that are almost 300K for an H100. Um, so it's, it's all over the place.

105
00:19:44.480 --> 00:19:57.360
H200 is even rarer, but, uh, you know, low to mid three hundreds per, uh, HGX node. And then for the, uh, the A100s, those are hitting the market a little bit more, um,

106
00:19:58.360 --> 00:20:04.720
uh, probably in the, uh, you know, seventies to nineties depending on the specification per, per hardware node.

107
00:20:06.520 --> 00:20:28.680
So I'm thinking about like GPU depreciation cycles and, um, maybe, and I haven't been dialed into, you know, high, uh, you know, compute at this level for a while, but I w- i- am I correct in assuming that previous upgrade cycles have not required like full stack replacements of the entire node, the entire system?

108
00:20:29.340 --> 00:20:42.180
And then with the migration to like the rack, we have this logistics challenge, I imagine, of when we want to upgrade, now we have to pull the entire rack, or do we?

109
00:20:42.280 --> 00:20:53.800
And like, w- in light of this, what, you know, one, am I right about this? And two, what does this mean for like GPU depreciation or like the upgrade cycle of this high performance compute? Yeah. Yeah, good question.

110
00:20:54.380 --> 00:21:08.540
So, um, you know, there's, there's a couple things that I think are happening here. Uh, number one, we're starting to see software, um, users, AI software users actually starting to understand what they actually need.

111
00:21:09.380 --> 00:21:23.820
Uh, this is sort of a, a broader trend that was happening with, uh, you know, data center IT infrastructure before then, um, where most users aren't actually consuming all the hardware that they buy or that they rent.

112
00:21:23.980 --> 00:21:39.580
Uh, I spent, you know, seven, eight years at Dell, uh, running sizing on, uh, companies', um, uh, you know, data center infrastructure, and by and large, their compute, uh, consumption was maybe 10% of the CPU that they had in there.

113
00:21:39.800 --> 00:21:54.560
So not all inferencing is built equally. Um, and I think that's what has really happened with the secondary market is that, uh, users are seeing that they can get more bang for their buck with a older GPU.

114
00:21:55.300 --> 00:22:10.500
Um, the other thing that usually drives hardware, uh, refresh cycles has to do with the, um, uh, software compatibility. So that will probably push, uh, more of these to phase out rather than any actual demand.

115
00:22:10.620 --> 00:22:24.080
Um, now if you're doing like a full AI factory and not just bare metal nodes, everything within that AI factory reference architecture has to be compatible. So, uh, that would be more of a rip and replace.

116
00:22:24.200 --> 00:22:38.000
The shared storage, the networking, the management nodes, uh, head nodes, GPU servers, backend networking, all that has to be compatible. So, uh, when you go to upgrade that, you're gonna have to rip and replace.

117
00:22:39.640 --> 00:22:54.800
Mike, kind of building on that with regards to depreciation, it's one of the key tension points when analysts and investors look at some of these neo clouds and hyperscalers and anyone running this compute.

118
00:22:54.840 --> 00:23:03.850
I'm curious your thoughts on whether or not you think the depreciation schedules, roughly five years, if that's too aggressive for what we're seeing right now.

119
00:23:03.850 --> 00:23:12.380
And it's kind of hard to say because ultimately even A- A100, the oldest model that's still in the conversation is like, what, just from like twenty twenty-two, twenty twenty-three?

120
00:23:13.040 --> 00:23:27.000
So we don't even really have enough data yet, I think, to have a clear answer to this question, but I would love your opinion on whether or not you think that five-year average that analysts typically use is too aggressive or too conservative.

121
00:23:29.660 --> 00:23:43.160
Yeah, I, I agree with you that there's probably not enough data to really feel a hundred percent one way or another. Um, you know, if what we've seen so far continues, uh, then it's definitely not accurate.

122
00:23:43.380 --> 00:23:51.020
Um, you know, earlier this year, GTC, uh, CFO of CoreWeave said that H100 life cycle's probably eight to ten years out.

123
00:23:51.800 --> 00:24:01.580
Um, you know, something that, that I saw, uh, in more traditional hardware infrastructure is that people run hardware until the wheels fall off and the software stops working.

124
00:24:02.500 --> 00:24:13.020
Um, most workloads for inferencing are probably not gonna be ultra-latency specific, uh, need the latest and greatest for every single workload.

125
00:24:13.480 --> 00:24:24.480
It's gonna be more of a conversation about how do I get the most affordable tokens out, uh, that have the performance needs that I need rather than, uh, how do I always rent a Ferrari?

126
00:24:25.100 --> 00:24:36.140
And to that effect, I do think that, you know, probably five to seven years is-- that's in line with pre-LLM hardware. This is just another workload. It's a little bit more complex.

127
00:24:36.360 --> 00:24:43.240
It probably won't be that different going forward. Um, I guess last question is, um,

128
00:24:44.680 --> 00:24:59.460
recently, again, on the GPU life cycle, uh, Amazon shortened their GPU useful life cycle predictions, I believe, and Meta extended them, and there seems to be, like, disagreement across the industry. Um,

129
00:25:00.500 --> 00:25:13.500
you know, chime in here. How do you reconcile everybody kind of searching for their own depreciation life cycle timelines? Uh, it comes down to use case, which I know is kind of a great non-answer.

130
00:25:14.220 --> 00:25:25.500
Um, to me, I would be perfectly fine having a 1950s Land Rover Series I, but I'm not rock crawling up the side of, uh, you know, Denali anytime soon.

131
00:25:26.060 --> 00:25:34.500
Uh, getting around New Orleans is just fine in, in something nice and old. Now, uh, if I'm a serious rock crawler, I'm gonna get the latest and greatest, right?

132
00:25:34.960 --> 00:25:39.660
Uh, it's no different with these GPUs and what they're being used for.

133
00:25:39.900 --> 00:25:53.240
The other thing to understand with those hyperscalers is that they've been developing their own dedicated, uh, training and inferencing ASICs, uh, in-house for the past 10 years, and they've been in production for about as long.

134
00:25:53.860 --> 00:26:10.920
So, uh, you know, their use case and the way that they utilize their GPUs, their ASICs, all of their AI hardware is very different from the way, uh, you know, a, a typical neo cloud or a minor converting will use them.

135
00:26:10.960 --> 00:26:25.040
Uh, I don't necessarily think that, um, you know, it's, it's a apples to apples comparison there. Um, Matt, uh, o- one, one or two more questions from me, and I'm asking you to look into the crystal ball here.

136
00:26:25.680 --> 00:26:35.060
Um, so, um, please bear with me. And sorry, Mike, I, I'm scrambled today. You got too many M's on the show. [laughs] We got a lot of M's on the show today.

137
00:26:35.820 --> 00:26:40.940
So quick question with regards to where you see hardware prices throughout the rest of the year.

138
00:26:41.040 --> 00:26:53.220
Given the demand you're seeing, given the deal flow that we're seeing on the operator side in terms of CapEx, do you expect prices to compress, expand, or flatline throughout the rest of the year?

139
00:26:55.680 --> 00:27:11.540
I believe that due to, uh, Nvidia lead times, you're gonna see, uh, higher prices for hardware needed immediately. Uh, I think you're gonna pay a premium, uh, for what's immediately available.

140
00:27:12.020 --> 00:27:30.080
If you're willing to wait six months, which is the average, uh, lead time right now for a bulk amount of B300s, uh, from any Nvidia-certified OEM, uh, you know, those prices will probably flatten a bit, uh, especially as people wait to see how Vera Rubin actually gets deployed.

141
00:27:30.780 --> 00:27:39.340
Currently, Vera Rubin is in the, uh, you know, they're testing prototype stage. Marketing's starting to go out to the OEMs. These things haven't hit the open market.

142
00:27:40.080 --> 00:27:51.340
Um, so we don't really know what demand's gonna look like for Vera Rubin. Uh, there'll probably be a market, um, uh, hiccup when Vera Rubin finally launches.

143
00:27:51.680 --> 00:28:01.680
But keep in mind, any of those newest generation chips, they're already pretty much spoken for, right? Your hyperscalers are gonna get them in the first six months. Uh,

144
00:28:02.740 --> 00:28:16.040
maybe, you know, some, some big neo clouds after them, and it'll probably hit the public in about 12 months or eight, uh, depending on how friendly you get with your Nvidia rep. Um, now, with, uh, used market,

145
00:28:16.980 --> 00:28:29.400
if there's anything I've seen, um, it keeps going up. I think there will be a break-even point, uh, around profitability with those, and that's going to cause a flat line.

146
00:28:30.140 --> 00:28:42.400
Um, every time I see it, I think we're there, but they keep going up. Well, we'll be keeping an eye on that. Have to have an update towards the end of the year, see where things are at. Mike, thank you for joining, man.

147
00:28:42.440 --> 00:28:56.740
Really appreciate it and love the insights on hashrate index. Keep it up, and have a great weekend. Y'all too. Thanks for the time. Thank you. All right, we're gonna roll on, and we haven't totally abandoned it.

148
00:28:56.760 --> 00:29:08.760
We're gonna have a Bitcoin conversation, 'cause it's all hands on deck. For the Coldcard hat, we have Matt Black in the audience. We'll bring him on up here right after a word from our sponsor, Luxur.

149
00:29:09.210 --> 00:29:21.360
[fire blasts] This episode of Block Space Live is brought to you by Luxur's Commander, Bitcoin miner management software for enterprise operations.

150
00:29:21.940 --> 00:29:28.360
Luxur's Commander gives you real-time fleet monitoring, bulk remote commands across your fleet, and Intelligent Miner.

151
00:29:28.400 --> 00:29:35.480
That's an automated profitability engine that runs every five minutes and adjusts your fleet's power settings to live energy and hashrate markets.

152
00:29:36.060 --> 00:29:47.320
In fact, ERCOT back tests show 10% improved profitability with intelligent mining over binary mining, and you need every single sat with hash price compressed at where it is now.

153
00:29:47.340 --> 00:30:03.619
Commander Pro is $100 per megawatt or a 25 basis point pool fee adder, but you can also try it free for 60 days. So if you'd like to learn more, go to luxur.tech/commander. All right, let's talk Bitcoin.

154
00:30:04.540 --> 00:30:18.100
About time, and under duress do we do this. Time, let's bring up Matt Black. Matt, welcome to the show. Thank you, gents. Thanks for having me. Uh, my goodness, this is, this is crazy to see what's going on.

155
00:30:18.180 --> 00:30:30.380
I can't, I can't believe it. Um- I, yeah, it's... Matt, what, e- explain to me, what is going on? Again, like, our, our audience has kind of shifted to more data center REITs, AI folks.

156
00:30:31.160 --> 00:30:46.580
What is roiling the Bitcoin ecosystem right now? So there was, uh, for those that haven't been listening, uh, there was a vulnerability found in the seed and entropy generation of Coldcard devices, MK2, MK3.

157
00:30:46.700 --> 00:31:01.468
First off, if you have a Cold- If you're listening right now and you have a ColdCard device, MK2, MK3, you generated that seed anytime 2021 or later, and actually the later devices too are affected, you should be looking at moving your funds immediately.

158
00:31:02.088 --> 00:31:07.628
Um, the entropy is the process of, you know, how does Bitcoin work? Bitcoin has private keys.

159
00:31:07.768 --> 00:31:21.047
It's a random number in a very large, uh, search space, but if you don't generate that random number properly, uh, someone can find your private keys and they can take your Bitcoin. So, uh, yeah, that's what's going on.

160
00:31:22.828 --> 00:31:38.608
Um, you know, maybe talk through, like, the implications of this. Um, because I, I don't think the average person realizes, like, that it's not Bitcoin that broke, but it's, like, a major

161
00:31:40.228 --> 00:31:53.548
piece of hardware from, like, a prominent manufacturer, and the implications- And, and not just prominent, but, like, gold stamp Bitcoin Maximus approved We've given out, we've given out ColdCards to listeners, so.

162
00:31:54.728 --> 00:32:03.508
Well, well, I, you know, and I've, I've gone and I've recommended ColdCard to so many, so many folks over the years. It's like, okay, well, you know, I want to go and I want...

163
00:32:03.588 --> 00:32:15.268
I'm buying Bitcoin, and I want to go and I want to self custody. How am I gonna go do that? Um, the recommendation has always been ColdCard. It's always been, hey, like, ColdCard is the gold standard.

164
00:32:15.308 --> 00:32:26.128
You know, you could, you g- you generate your seed on it. You use, you can have the device so that it's, uh, you know, not even, not even, you know, attached to your computer. It's air gapped.

165
00:32:26.828 --> 00:32:38.068
Um, and it's just really sad to see that this is what happened. Um, and so I, [sighs] I don't know. Like, for me, you know, I've always told...

166
00:32:38.208 --> 00:32:44.688
You know, and I guess to, to, so people understand a little bit more, like, the actual issue was just one line of code, right?

167
00:32:44.748 --> 00:32:51.388
It was one line of code that was affected, which was basically like, hey, how are we actually generating the seed? Um,

168
00:32:52.428 --> 00:33:01.828
and for, you know, and I think this is a larger discussion too, which is like, okay, what is the right way to actually go and secure your Bitcoin? Like, what is the, the safest way to do that?

169
00:33:01.888 --> 00:33:12.728
It's always been multi-sig is too complicated. You should use a single sig. Well, the irony is, you know, that's really biting a lot of people in the, you know, in the, in the butt today.

170
00:33:12.848 --> 00:33:17.628
You can say ass because we're getting- [laughs] Yeah, it's, it's rough today. Okay.

171
00:33:17.648 --> 00:33:29.848
Well, before we get into, like, the existential refl- like, uh, reflection on the ecosystem, let's touch a few more points about what we know. Ballpark, what is the scale of the attack, and is it ongoing?

172
00:33:29.968 --> 00:33:40.728
Uh, toss it to you. Yeah. So far, I, so what I've seen so far is that, and just looking at the mempool, it's been at, um, at least 1,000 Bitcoin that's been affected. What is that?

173
00:33:40.868 --> 00:33:49.888
You know, uh, $65 million in Fiat dollars. Um, we've got... And, and that's, that's just the start because now, you know, Pandora's box is open.

174
00:33:50.008 --> 00:33:55.208
So, that was likely folks that had MK2s, MK3s, which were the most vulnerable.

175
00:33:55.248 --> 00:34:07.028
So, generally, when you generate a seed, you need a certain amount of entropy, and, you know, if you have a 12-word phrase, that's usually 128 bits. If you have a 24-word phrase, that's, you know, 256 bits.

176
00:34:07.508 --> 00:34:14.288
But the problem there is, like, if there's... And what the issue with the ColdCard specifically was there, there was a fallback, right?

177
00:34:14.348 --> 00:34:22.108
So, they didn't realize that one of the values used, uh, in the, in the code was basically just checking, does this value exist, not is it zero.

178
00:34:22.788 --> 00:34:35.468
And so it ended up having a fallback, which means, like, you know the difference between having 128 bits of randomness of entropy and 45 bits is, you know, forever and ever being able to find it and you're screwed.

179
00:34:36.128 --> 00:34:43.988
Um, so [laughs] um, so it's been 1,000 Bitcoin that's been affected, and that's just from the folks that are the most at risk.

180
00:34:44.028 --> 00:34:55.708
But now I think it's likely gonna be a situation where we have, you know, the folks that have MK4s, MK5s. Well, they probably haven't got all the Bitcoin of the M- MK3s, to be honest, right?

181
00:34:55.768 --> 00:35:00.768
Like, if you look at the blockchain, there's people who had like half of their wallet stolen. Well, how does that happen?

182
00:35:00.808 --> 00:35:11.348
Well, it's probably 'cause it was someone using AI that was, like, trying to, trying to grab those funds, and they only grabbed, they only grabbed this first X number of UTXOs and, and grabbed them.

183
00:35:11.428 --> 00:35:20.528
So now, and now it's a free-for-all for, for getting the rest of them. Um, but the thing is, it doesn't just affect people who just did a, you know, a single sig wallet.

184
00:35:21.448 --> 00:35:30.568
You know, if you set up a, say, say you set up a ColdCard multisig, right? You, maybe you did two of three. You had three ColdCards. You have two of three, and you set this up in 2022.

185
00:35:30.848 --> 00:35:38.908
Well, now you're affected, and now people are gonna be looking at the blockchain to see you spending, so, you know, it's, it's a really, really tough time.

186
00:35:38.948 --> 00:35:51.128
We haven't seen the end of where this is going to go, um, and how many people are gonna be affected. So Matt, a quick question from me. I saw a lot of jawing on social media about

187
00:35:52.108 --> 00:36:05.538
this may have surfaced, and I'm not as plugged into this as Charlie, so just caveat up front. This may have surfaced from the fact that ColdCard used to base some of their design off of Foundation's model. Ooh.

188
00:36:05.568 --> 00:36:15.198
And then they came under a lot of flack for that, and then... Well, I mean, allegedly, but they came under a lot of flack for that. Well, it wasn't, it was... I mean, declare, I'll, I'll clarify. Yeah, clarify it.

189
00:36:15.198 --> 00:36:25.628
Foundation used some open source licensing from ColdCard. It caused a whole thing. Um, I don't know if you want to speak to that, Matt, but it, yeah. You can do it.

190
00:36:25.698 --> 00:36:35.968
But, but, uh, my question is, they changed something when all of this came to a head, and some people were insinuating that this may have been the root cause of the flaw that we're seeing here.

191
00:36:36.148 --> 00:36:43.788
Can you speak to that at all? Am I just totally off base or misunderstanding? So, so my understanding was that in 2020,

192
00:36:44.748 --> 00:36:55.340
um, you know, Foundation devices was basically, you know, they were just getting off the ground and- Originally, um, the licenses of ColdCard were open source.

193
00:36:56.020 --> 00:37:08.160
And so obviously, you know, MIT license, you're able, you know, with attribution to the original author, you're able to use that code for commercial purposes. So that's, you know, that's what Foundation did.

194
00:37:08.260 --> 00:37:19.240
And, um, and basically ColdCard, you know, Coinkite was not happy about that. They were not happy that someone else was basically taking their code and then going and making money off of it.

195
00:37:20.040 --> 00:37:32.199
And so, you know, around 2021 they started having the conversation of, "Hey, we're gonna go and, um, you know, rewrite, you know, large parts of our code base, um, in order to...

196
00:37:32.290 --> 00:37:43.560
and then put it under a different license," such that people weren't able to go and, you know, use that for commercial purposes. Now, part- And then they closed sourced everything, right, after that? ColdCard did?

197
00:37:43.680 --> 00:37:52.990
Am I wrong? Uh, no, it was still, uh... What, what's the, what is the, what's the term? Open viewable or... Like, it's where you can see the code but you can't, like, use it. What's the term? Okay.

198
00:37:53.120 --> 00:37:59.460
Charlie, do you remember? Oh, I don't know, but yeah, this is... We're a little, we're going to be too deep down the rabbit hole here, but yes. I know, yeah.

199
00:37:59.520 --> 00:38:08.310
[laughs] But, but, but basically, like, so, a- a- and as part of those changes in 2021, they ended up, like, introducing this vulnerability.

200
00:38:08.520 --> 00:38:25.960
Whereas ironically, if they had just stayed, like, "Hey, we're gonna be open source, and yeah, people are gonna take our code and they're gonna improve on it or they're gonna change it," if they had been both using the same source code, well, it's like, it's likely the other company would've noticed a bug of this, of this nature.

201
00:38:26.080 --> 00:38:36.110
Um, s- which is kind of like the whole point of, uh, uh, of open source. So I don't know. Like, there's a lot of contention now, like al- also in the term- in terms of the time of AI, right?

202
00:38:36.160 --> 00:38:47.000
Because, like, now, you know, it brings up the question of like, hey, we have open source, but you know, someone who's vibe coding can find vulnerabilities and, and hack things. Uh- Yeah.

203
00:38:47.160 --> 00:38:55.820
And, uh, just a quick mea culpa for me there. Totally flipped the, [laughs] the relationship there on the ColdCard- No, I, I corrected... came out. Yeah, I appreciate that, Charlie.

204
00:38:55.940 --> 00:39:05.520
But I, I was just curious whether or not how much of this could be attributable to that spat because of the changes ColdCard made. It certainly feels somewhat.

205
00:39:05.560 --> 00:39:16.940
It's re- you know, we're still in the, like, discovery period of this. Like, it was, I mean, there were spaces last night going well into the wee hours, and information is still coming out.

206
00:39:17.720 --> 00:39:36.530
Matt, you talked about AI, and I actually think this is a major role to play because e- everyone has been t- everybody, you know, who talks about AI and security and cryptography knows that with meet those level, level capabilities, and now with Kimi being cheap and less constrained,

207
00:39:37.520 --> 00:39:51.900
we're in a new era for AI vulnerabilities, where even like script kiddies, people who are maybe less experienced but maybe wi- uh, wildcatting, able to start exploiting things. What does this mean?

208
00:39:52.020 --> 00:40:01.240
Is this the begin- is this just, just the first shot across the bow? What does this mean for like the landscape of Bitcoin and self-custody?

209
00:40:02.380 --> 00:40:20.740
Well, I think the, like, the reality is that it's just gonna be that much easier for folks to be able, you know, your, your average vibe coder, to be able to sit there, you know, whether they're in North America or they're in China, using whatever model, they're gonna be able to look at all of this open source code and just run, you know...

210
00:40:21.220 --> 00:40:31.380
A- a- and, and what's, what's, what's really wild about this is with the ColdCard example, it wasn't that hard to find the vulnerability. A- anyone could have pointed an AI, you know, AI model at that.

211
00:40:31.480 --> 00:40:39.840
Um, you know, one of the gu- the guys on our team that's, you know, non-technical ran Codex on this and found the vulnerability in no time on the MK3 code base.

212
00:40:40.360 --> 00:40:47.660
So it's, it's almost a thing of like, you just need to point your code, or you just need to point your model at something and you're gonna find vulnerabilities.

213
00:40:48.480 --> 00:40:55.360
Um, but I think the bigger question is, like, what does this mean for open source, right? Is this the end of open source? Do people stop using it?

214
00:40:55.380 --> 00:41:04.820
But I think it's more likely that it's just a situation where we need to catch up, where, you know, there's, it's a cat and mouse game, where, you know, now anyone can point their model at something and find a vulnerability.

215
00:41:05.620 --> 00:41:24.720
But, you know, in two, three, four years' time, you know, if we don't, if, if, unless we get another exponential growth in, um, you know, AI capabilities, then, then it's likely that, okay, we're gonna still end up, you know, building in open source land, but it's just that, like, now we're gonna be checking.

216
00:41:24.780 --> 00:41:34.460
You know, everyone's gonna be running a security analysis on every piece of code before they go and put that out in the, in the public. Um, so I don't know. That feels like the new paradigm to me.

217
00:41:35.860 --> 00:41:43.840
Final question here from me, Matt. What are the ramifications for the rest of the hardware wallet manufacturers here? I mean, right now this is a ColdCard issue.

218
00:41:43.860 --> 00:41:55.910
I think everyone who has their funds in a single sig hardware wallet, like Trezor or Ledger, is probably thinking, "Ooh, like how, is there a risk for my coins to be swept as well?"

219
00:41:55.980 --> 00:42:05.740
Like are, is it just a matter of time before other security vulnerabilities are found? You know, it's impossible to say, obviously, what's going to happen in the future, but like,

220
00:42:06.900 --> 00:42:16.100
what are the shock waves for the rest of the self-custody ecosystem here, and whether or not other users should be worried about similar attacks?

221
00:42:16.160 --> 00:42:37.220
W- well, I feel like, okay, to be honest, I feel like after the news came yesterday, likely what was occurring was every Bitcoin company, whether they were a hardware wallet company, whether they were software wallet, mobile wallet, whether they're a custodian, was sitting there with every s- every AI model they can get and double-checking, "Hey, how is my entropy generated?"

222
00:42:37.820 --> 00:42:42.240
So, um, to be honest, I think, you know, this is a terrible, this is heart-wrenching.

223
00:42:42.390 --> 00:42:52.320
This is a heart-wrenching thing to occur, but it also means that everyone's like double, triple, quadruple-checking their work now, which I think is, is incredibly important.

224
00:42:53.252 --> 00:42:59.292
But I, I think it also, I don't know, for me personally, it brings up a bigger question of, like, the multi-sig discussion.

225
00:42:59.352 --> 00:43:12.352
You know, I remember those conversations back in 2020 where people would always say, "Don't do a multi-sig because you're gonna screw up your setup and you're gonna lose your own funds." And that was the discussion.

226
00:43:13.042 --> 00:43:24.972
And it turns out the folks that, you know, were considered crazy at the time, who are, who were considered paranoid, are the ones whose funds are now safe. And so I don't know, I think this... Well, there's two things.

227
00:43:25.072 --> 00:43:42.412
I think it brings up a bigger question of, "Hey, everyone's gotta check their code with models now?" And triple- Uh-

228
00:43:46.152 --> 00:44:00.272
I'm, I'm... Why'd you mute me? I'm just saying we need to wait to see what's going on with Matt. Yeah. Oh, I didn't mute you. Um, I mean, dang. Okay, he froze. Uh- One, one thing really quickly here, Charlie.

229
00:44:00.392 --> 00:44:12.582
I actually want to, uh, I wanna just fla- Oh, he's back. Oh, he's back. We're gonna bring him back up. Put him back up there to just wrap up. Matt, are you back? Sorry, guys. Yeah, you're good. Um- Okay. Just to wrap up.

230
00:44:12.582 --> 00:44:21.792
Jack you're back in. Yeah. You guys hear me okay? Perfect. Yep. Um, so yeah, as I was saying, like, you know, what, what does this mean for the larger ecosystem?

231
00:44:21.972 --> 00:44:30.452
We're gonna have folks that are quadruple, you know, quadruple ch- [audio glitches] -cking all of their work, number one, for all of the hardware wallets out there.

232
00:44:30.772 --> 00:44:39.332
But I think the second thing is, like, the discussion around multi-sig, right? We're gonna have a larger discussion around how can we make multi-sig simpler for folks?

233
00:44:39.592 --> 00:44:48.362
How can we set it up such that it's, you know, multi-vendor, right? Like, and then how are we gonna make it easier for folks to go and generate their own entropy, right? People wanna be able to...

234
00:44:49.032 --> 00:45:02.131
You know, now people know, "Hey, I should probably go and roll those dice so that I'm not at risk of this." But I think the final thing as well is the whole Bitcoin community has been talking a lot about spam.

235
00:45:02.692 --> 00:45:16.942
And I think it's time to get back to a discussion around covenants about not only, you know, reactive security, but proactive security so that we have better ways of people being able to secure their Bitcoin. Go off.

236
00:45:16.942 --> 00:45:25.512
I think if anything, hopefully this reignites that discussion, although I've been pretty black pilled about whether or not we're gonna get, you know, fruitful discussion in the technical community

237
00:45:26.692 --> 00:45:32.712
over the last few months. Matt, before we leave, before we, uh, sign you off, I just wanted to flag this tweet from you.

238
00:45:33.812 --> 00:45:46.412
You had a great breakdown on Twitter about what you need to do to address this if you have a cold card and how you can protect your stack for the future. So go check out @matthewj- J.A.

239
00:45:46.452 --> 00:45:57.951
Black on X if you wanna learn more about this vulnerability and what you can do to address it if you are potentially affected. Matt, thank you so much for your time and insight.

240
00:45:58.252 --> 00:46:13.012
Uh, this is probably gonna be an ongoing topic, so we may touch on it. God forbid, more Bitcoin doesn't get hacked, but we know it's coming. So fingers crossed. Matt, thank you so much. Cheers, Matt. Adios.

241
00:46:14.052 --> 00:46:21.652
Thanks, every- [audio cuts out] [laughs] Whoops. You know- But super sharp guy. Um- Very sharp. Yeah.

242
00:46:21.752 --> 00:46:32.432
I re- I'm glad he ended it with covenants because that, that was, like, the beat we covered on the show, uh, for, like- Yeah, and if, if proof-... you're not here... that actually we need these things, what else? Yeah.

243
00:46:32.472 --> 00:46:47.612
You know what I mean? The whole, "Oh, it's just an education, dude. No one's gonna use them." I don't know. Could've saved a lot of Bitcoin here if we had them, so. Well, you know, again, it's a deep, complex topic.

244
00:46:47.852 --> 00:47:01.432
Um, I'm unashamedly, uh, pro-covenants and soft forks, but not the soft fork coming in nine days. But we're not gonna dwell on that because we are [laughs] gonna go back to AI, Colin.

245
00:47:01.482 --> 00:47:16.492
We're gonna go back to AI and compute. Specifically, gonna keep on the Kimi train. We'll cover the Kimi, the, uh, moonshot, NVIDIA story here in a moment after a word from our sponsor, Ligos.

246
00:47:17.352 --> 00:47:30.232
[whooshing sound] The sad fact is a black hat hacker with Kimi K3 doesn't need to be the thing that takes down your Bitcoin stack.

247
00:47:30.292 --> 00:47:42.612
If you're using a Bitcoin-backed lender that is not respecting your stack, you could become the next Celsius or BlockFi victim. But with Ligos Finance, you don't have to worry about that.

248
00:47:42.652 --> 00:47:51.272
They are our preferred Bitcoin-backed lender here at Blockspace. They use Bitcoin native smart contracts to make sure you are always in control of your coins. You heard that right.

249
00:47:51.332 --> 00:47:57.631
With Ligos Finance, you always have control of your keys when you are pledging collateral for a Bitcoin-backed loan.

250
00:47:58.092 --> 00:48:08.072
There's no funny business, no rehypothecation, no bridging, no wrapping, just competitive rates as low as 8.5 to 10% APR. Go to ligos.finance to learn more.

251
00:48:09.852 --> 00:48:25.552
All right, Charlie, another headline in the export-import wars for the heartbeats and the brains of the AI revolution. This one specifically is coming from Bloomberg,

252
00:48:26.632 --> 00:48:34.812
and it is a report that claims that Kimi K3, specifically Moonshot, the lab behind it,

253
00:48:36.472 --> 00:48:46.252
had a computing power agreement with Alibaba for roughly 20,000 NVIDIA chips, forming a substantial chunk of the compute behind its Kimi models.

254
00:48:47.112 --> 00:48:54.072
Sources say the chips are Hopper generation H200s, though Alibaba disputes that specific claim.

255
00:48:54.572 --> 00:49:08.392
What's interesting is, though- According to the reporting, Alibaba pushed back on the stipulation that it was Hopper generation, but they didn't necessarily totally dismiss that they were providing compute for Moonshot.

256
00:49:09.632 --> 00:49:20.012
What's interesting about this, number one, is that the United States doesn't want these Chinese companies to have cutting-edge chips.

257
00:49:21.032 --> 00:49:31.912
We'll get into some of the export controls, because actually Hopper generations are not totally barred as long as they have White House sign-off. But Beijing also doesn't want them in their country either, allegedly.

258
00:49:32.352 --> 00:49:39.352
That's their posturing, because they want to show that they are protecting domestic producers of compute.

259
00:49:40.952 --> 00:49:59.332
So there's this weird push and pull here on both sides of the ocean in terms of the United States doesn't want NVIDIA shipping GPUs, specifically Blackwells, to China, but China is also saying that they don't want their companies using them.

260
00:49:59.652 --> 00:50:10.972
That's what they're saying, but if this story is true, then clearly there's some wiggle room here, or Alibaba was actually just caught with its pants down doing something that Xi Jinping does not want happening.

261
00:50:13.032 --> 00:50:23.632
[lip smack] So Mike, uh-- they're-- a White House official, Michael Kratsios, accused Moonshot last week of illegally acquiring NVIDIA's most advanced Blackwell chips. That's another curveball here.

262
00:50:24.132 --> 00:50:27.852
The government hasn't specified how many it believes Moonshot has.

263
00:50:28.692 --> 00:50:42.792
In the X post, he claimed that Moonshot distilled Anthropic's Fable model to develop K-3, building a, quote, "sophisticated internal platform to conduct large-scale distillation against US model- models."

264
00:50:42.832 --> 00:50:54.872
Critically, Bloomberg notes he provided no concrete evidence for the distillation claim. You know, maybe this is just, this is just posturing on our own government side.

265
00:50:55.552 --> 00:51:06.202
I would wager a guess that someone on the inside of the White House probably has better intelligence than Bloomberg. Call me naive, but I think there's probably a good chance that that's happening.

266
00:51:06.232 --> 00:51:12.132
But this distillation allegation ha- is something that's been flagged across social media and across reporting.

267
00:51:12.592 --> 00:51:27.072
In fact, it's well assumed, if not well known, that the only way that models like Kimi K-3 and DeepSeek could actually have the rapid advancements that they have portrayed is if they're distilling frontier models here in the US.

268
00:51:27.972 --> 00:51:40.932
Charlie, I'll toss it to you for second takes before I get into the legal nuances of what exactly the White House's export controls do to prevent the latest generation GPUs from being in Chinese hands. Uh, yeah.

269
00:51:41.072 --> 00:51:44.512
I mean, that was probably gonna be my take, but I'll let you, uh, run with it.

270
00:51:45.092 --> 00:52:23.441
Um, my kind of, like, subjective experience, given the fact that we just talked about the, this, this hack on Bitcoin hardware device, um, is that, um, in, in just recent memory, in the past 24 hours, in some of these chats that I've had with developers who are performing security audits on their own hardware, they were unable to use largely the frontier models in the United States, specifically Fable, but also ChatGPT's sole, uh, level, uh, LL, uh, models, did not, like, want to play ball and do these audits from white hat hackers.

271
00:52:23.592 --> 00:52:37.952
And what they found is, in multiple cases, were they had to turn immediately to Kimi. So, um, this does, like, bring into existential question, what even are these export bans functionally doing?

272
00:52:38.572 --> 00:52:50.872
Um, are they functionally preventing the GPUs, uh, from being physically inside China? Or, uh, are they simply just kind of a friction for domestic American users?

273
00:52:51.252 --> 00:53:03.112
Kind of almost like the, the, the days of trading crypto with VPN on Binance. Like, it didn't really stop anybody. You had to pay your taxes, but like, you know, you weren't actually prevented. So some analogies there.

274
00:53:03.172 --> 00:53:16.211
Um, I'll toss it back to you to get into the nuances of, like, what are, uh, where- What are we doing here? What, where can Kimi run according to obscure- Yeah, I think this is im-... US export control?

275
00:53:16.492 --> 00:53:21.372
I think this is important because kind of like with the tariff argument or, or the tariff discourse,

276
00:53:22.492 --> 00:53:38.852
there's no gray area in people's minds on these things, and the actual facts of what's happening get obscured by sensational headlines of, you know, Trump bans China from having access to GPUs or, you know, there are 100% tariffs on everything now.

277
00:53:39.492 --> 00:53:52.012
So for the specifics here, the US rules actually do allow for access to resi- restric- res- restricted chips via compute rental agreements, renting not buying.

278
00:53:53.232 --> 00:54:04.432
And that's what Kratsios actually says Moonshot did through an unnamed party in Thailand. And this is a workaround that we've actually covered for other things like Bloom Energy's procurement of scandium.

279
00:54:05.052 --> 00:54:20.112
There is a, there's, there's kind of a plausible deniability element here where Bloom Energy's CEO can say, "We don't rely on China for scandium," because they're buying from offshore companies in Southeast Asia that are importing the scandium from China.

280
00:54:20.732 --> 00:54:34.092
But there's enough of a layer there to where what you're saying is factually true, even if it's not directionally true. And so these, most of these Chinese companies do have access to

281
00:54:35.392 --> 00:54:47.872
compute in places like Thailand, I would imagine Indonesia, and other Southeast Asian hubs for commerce. The physical location s- on that note of these Alibaba

282
00:54:48.772 --> 00:54:58.092
GPUs is unclear, and that matters legally because the rent-restricted chip offshore workaround hinges entirely on whether, where the hardware physically sits.

283
00:54:58.532 --> 00:55:09.632
Now, that being said GPUs themselves aren't wholly banned from being exported to China, but there are big catches. The Blackwell line are completely banned.

284
00:55:09.952 --> 00:55:19.352
Again, whi- the White House and Uncle Sam do not want Chinese corporations having access to bleeding edge tech. But c- the Hopper generation are conditionally allowed.

285
00:55:19.432 --> 00:55:30.752
These were recently unbanned in December 2025, and Trump announced that H200s could be sold to China on the condition that they get a rubber stamp from Washington.

286
00:55:31.232 --> 00:55:44.612
So if Washington approves the sale, they can be exported and purchased by these Chinese companies. So the older, less capable variants are actually somewhat approved. But

287
00:55:45.952 --> 00:56:05.262
there's this other consideration here, like I said at the front, Charlie, where the dynamics here are very odd because ultimately Beijing has postured that they don't want their companies purchasing these, because they want to see as actually having leading GPU development of th- on their own right, and they want their models to be trained and inferenced with Chinese hardware.

288
00:56:05.772 --> 00:56:15.472
Whether or not that plays out in reality obviously is up for debate, but the TLDR for the actual restrictions, rentals themselves, a Chinese company can rent GPU capacity.

289
00:56:15.512 --> 00:56:19.192
I mean, the, the United States can't tell them that they can't, and they're just gonna do it, right?

290
00:56:19.672 --> 00:56:28.562
If they're renting it outside of the United States, if it's, or outside of China and it's an Nvidia chip, the United States has thrown their hands up and say, "We can't really control that."

291
00:56:28.572 --> 00:56:36.772
But they are restricting Blackwells and bleeding edge units from being shipped to the mainland. So- Yeah... a decent amount of nuance here.

292
00:56:37.912 --> 00:56:50.312
Um, and I'll probably throw this topic as a first question to our next guest who, by the way, is in the backstage, and we're gonna roll to our final segment and guest of the stream.

293
00:56:50.692 --> 00:57:05.812
Got Carmen Li, gonna bring her on up here. All right. Carmen, welcome back to the show. Thank you. It's good to be back here. Yeah. So, uh, we were just talking about the Kimi Nvidia deal. Um, where are these GPUs going?

294
00:57:05.892 --> 00:57:11.472
And it kind of gets into, like, the export control conversation. Um, what do you think?

295
00:57:11.552 --> 00:57:27.192
Like, is, is the US just kind of, uh, tying a bow on and pretending like they are functionally prohibiting Chinese companies from running, uh, y- uh, you know, using GPUs, or is there actually, is it somewhat effective, uh, in a way?

296
00:57:27.272 --> 00:57:36.212
I don't know, what are your thoughts on this, this story and topic? Um, so I heard the last five minutes of your guys' conversation, sorry I came in a little late, um, so I didn't hear the whole argument.

297
00:57:36.312 --> 00:57:46.302
But in general I feel like what is the angle, right? If the angle is to make sure China is always six months behind, it's not working. Kinda it's not working, right?

298
00:57:46.832 --> 00:58:01.592
If the angle is say, "Hey, we just want to make sure that, um, people developing on stud- the Nvidia cluster, Nvidia ecosystem," that's kinda working, 'cause right now the most efficient chips deploy K3 is Nvidia chips, right?

299
00:58:01.641 --> 00:58:09.492
Depends what's the angle, which I, again, don't want to speculate too much. Um, but you look at K3 deployment right now, which just came out a few days ago, right?

300
00:58:09.532 --> 00:58:21.032
There's, what, two, three providers, they public claim they can support K3, which is kinda exciting 'cause it's, it's very hard model to support. Um, you need to have a lot of capacity obviously.

301
00:58:22.052 --> 00:58:25.652
And, uh, and it's a complete different licensing model, right?

302
00:58:25.692 --> 00:58:37.051
I don't, you know, if you guys chat about this already, you take a look at term, terms and conditions for K3, it's not like the MIT license which they had previously. You have to freaking pay for licensing.

303
00:58:37.632 --> 00:58:49.952
Y- so it's, you, you can argue we're in the way between open source and closed source models now, right? Obviously if closed source models, which game is off, right? If you are token factory in your cloud, you can host.

304
00:58:50.612 --> 00:59:01.772
But you have K3 of the world where sure you can, you have all the weights, you have everything you need, but you have to pay Kimi for licensing fee for API. And they set a price. They set a floor price, right?

305
00:59:01.912 --> 00:59:06.732
So if you're token factory, you're bound. You can't just charge whatever you, whatever you want, right? There's a price for that.

306
00:59:07.252 --> 00:59:20.152
I'm not saying it's bad or good for the token factories, but obviously there's, there's rules supposedly. But then if you are saying that I'm gonna deploy my own cluster on, on premise, then you can do whatever you want.

307
00:59:20.292 --> 00:59:28.482
You, you're not API based, right? But again, if you look at terms and conditions, depends what kind of enterprise you are. Who are you serving? What kind of products, right?

308
00:59:28.532 --> 00:59:34.052
There's all those arrangements that people need to, like, go through. That's not, I think things change, right?

309
00:59:34.092 --> 00:59:39.872
If you look at closed source model, other open weights models, you look at Llama, there's con- terms and conditions.

310
00:59:40.432 --> 00:59:50.312
You know, uh, your enterprise adoption, I think no enterprises, I like to think, want to do the wrong thing. So you have to be, be really thoughtful about how, who, how you want to use the model.

311
00:59:50.852 --> 01:00:00.732
You know, commercially what makes sense. If AP, does API make sense, so there's only a clone cluster that gets someone manage it for you makes sense, right? Well- So, yeah.

312
01:00:01.032 --> 01:00:06.572
So, um, that, you get into tokens and GPU prices, and that's what we want to talk about here.

313
01:00:06.652 --> 01:00:21.252
And you had this tweet recently noticing, saying, "We've seen a 15% to 30% price increases year to date across A100s, H100s, H200s and B200s, new cloud on-demand indices.

314
01:00:21.832 --> 01:00:37.172
But at the same time we've also seen that, uh, token prices have been down." I believe, uh, 18, down eigh- 18% in the past month or so. Um, what's going on here?

315
01:00:37.252 --> 01:00:51.532
Is there a mismatch between, uh, GPU prices and token prices, or is there some kind of, like, market dislocation happening? Um, I don't think... I think it's interesting, right?

316
01:00:51.692 --> 01:01:01.968
So again, our token indices measured the willingness to pay for a million dollar token doesn't measure total demand of the society Right?

317
01:01:02.528 --> 01:01:09.668
If you look at GPU prices, which is a reflection of the world, the total world supply-demand curve ex China where it meets.

318
01:01:10.248 --> 01:01:24.448
So the demand, so the reason why it goes up, usually the demand rise faster than the supply, so meaning people want more GPU, but they don't wanna [laughs] pay the premiums they were paying for the closed source models back then, right?

319
01:01:24.788 --> 01:01:39.428
So translate-ly to people using different model selection. However, obviously whatever the cutoff today is, not many people are using K3 yet, right? K3 is not that cheap if you look at the pri- API prices, right?

320
01:01:39.568 --> 01:01:50.568
If we see huge adoption of K3, people moving from the d- uh, the, the, uh, the, um, [lip smack] uh, the Z AI, the price may come back up. I have no idea, right? Depends on when people adopt it.

321
01:01:50.648 --> 01:01:54.978
Again, it's more expensive models. Um, so this is all willingness to pay, right?

322
01:01:55.428 --> 01:02:11.488
Do you wanna pay more, um, or le- le- less than the premium labs, but still more than the, the, uh, the open source models that's available. So this can change. So I'm, I'm curious from your perspective, um, you, on that

323
01:02:12.388 --> 01:02:20.288
tweet, Carmen, you mentioned obviously rental rates are increasing, and you said that y'all run, like, 17,000 reference points for your research.

324
01:02:21.308 --> 01:02:32.808
What trends and insights are you seeing that you think other people are missing from these rental rate increases? Like- So-... what's the read between the lines look at what's happening here?

325
01:02:33.428 --> 01:02:43.528
So one of the things that I think will be interesting is look at forward curve, right? So everything I post on that tweet is on demand to really give people a what's going on right now snapshot, right?

326
01:02:43.648 --> 01:02:56.288
If you're on demand, what's the price that you are paying. However, um, I would argue a majority of the world's capacity is sitting with, um, reserved capacity, meaning you reserve it and then that's it, right?

327
01:02:56.808 --> 01:03:03.078
Um, it's not part of on-demand supply, supply curve. Um, so those are the prices what impact a lot.

328
01:03:03.558 --> 01:03:13.228
If you RFQ anything today, um, for one year long contract or three year long contract, you know, and then, it, it can be contango or downward sloping.

329
01:03:13.657 --> 01:03:22.078
So you think about in general you would like to think the longer contract you signed, the cheaper it should be per GPU power, right?

330
01:03:22.348 --> 01:03:29.948
The more you sign up for, right, you're buying a, you know, 20,000 versus two, you would think the more you buy, the cheaper it should be per GPU wise.

331
01:03:30.448 --> 01:03:41.708
It's really not true sometimes in our world, where you see people- So the forwards contracts are con- in contango rather than backwardation? Exactly. That- Exactly right. So if you look at...

332
01:03:41.828 --> 01:03:55.128
I don't know, can, can I share a screen? Yes. Yes, you can. Okay. Let me see if we can share screen here. So I'm gonna... Okay, here. Can you see this? Yep. Yes. There we go.

333
01:03:56.048 --> 01:04:05.568
So if you look at H100, right, that's the, this is today or yesterday, um, we see a little contango here, right? So this, look at the blue line only.

334
01:04:06.138 --> 01:04:19.988
This translate to if you are entering contract today for 36-month long contract, you're paying about $2.36. But you enter twal- 24, which is two years, you're paying $2.30. So you see more contango here, right?

335
01:04:20.048 --> 01:04:25.968
That's crazy. So you will... This, this in commodity happens, right? So now, so a few things.

336
01:04:26.368 --> 01:04:36.908
Number one is a lot more people wanna pay, they're willing to pay premium just to lock in the resources than say, "Hey, you know, we'll, we'll, we'll at the mercy of demand supply curve any time."

337
01:04:37.548 --> 01:04:48.928
Or the supplier has the opt- have the p- market power saying that, "Guess what? We only wanna do three years. That's just the, we wanna do it," right? They can just move everybody to three years on curve.

338
01:04:49.788 --> 01:04:57.588
So if you look at last November, um, the price were going down last year across all different GPUs. It was downward sloping.

339
01:04:58.448 --> 01:05:04.997
So essentially this curve is a market consensus every day tell you what people think about the future, right?

340
01:05:05.028 --> 01:05:16.528
The orange line basically is a forward rate that Silicon Data calculates, implied rate for short-term rental beginning a given term. So you can essentially, oh, similar methodology as treasury futures curve, right?

341
01:05:16.888 --> 01:05:25.168
So people think the price is gonna come down, um, versus keep going up. If you look at B200 curve, again, it's completely different, right?

342
01:05:25.778 --> 01:05:31.018
And then you see a little, even three months you have to see all different kind of variation. This is really reason why

343
01:05:32.128 --> 01:05:44.228
I think dated futures, so have futures ex- have different expiration, will be precisely needed for the market. 'Cause you, every, so it's like every other commodity. You have a term structure risk, right?

344
01:05:44.388 --> 01:05:54.408
Either your bank, you have six long, six years long exposure. You know you have a client for first two or three years. What's gonna happen year three, four, five? You're concerned.

345
01:05:54.428 --> 01:06:04.588
Like, is, like, my rent gonna, gonna, my, my rent only gonna pay down my debt? That you, you have that year, that, that duration exposure you need to hedge. That's precisely future can help you do so.

346
01:06:05.548 --> 01:06:18.468
That, that's a really crazy insight, and kind of runs afoul of common sense for, like you said at the beginning, for the whole reason why you, you enter into these forwards, to lock in a potentially lower rate for revenue certainty.

347
01:06:18.478 --> 01:06:25.328
You kind of touched on this, but if you could unpack it a little bit further, Carmen, what do you think that's saying about the supply and demand story of compute?

348
01:06:25.808 --> 01:06:48.888
Because we're going to see gigawatts of capacity come online next year and into 2028, but to me it almost seems like the market is betting that that's not going to be enough, especially if we see token prices fall, and then Jevons Paradox kicks in to where people just start using more and more because the token prices are so cheap that you actually don't have a supply buffer that people are expecting.

349
01:06:49.528 --> 01:07:02.908
What is the market telling us with that contango in your mind? So, so the market obviously think the price will, will not come down. You, you know? So again, we all can calculate supply side.

350
01:07:03.028 --> 01:07:12.188
You know how many data centers supposed to be online globally. You know how many chips Nvidia can turn out. You know how much megawatt, gigawatt we can re- legitimately put to data centers, right?

351
01:07:12.268 --> 01:07:23.220
Which was the bottleneck. That will, that constrain our supply curve, right? In the long term, in the next two years, whatever. But demand side, then your speculation is as good as everyone else speculation, right?

352
01:07:24.440 --> 01:07:36.560
On, to a point, like, when people are gonna start to see adoption, right? 'Cause everything you see today, um, are with enterprises really in the beginning piloting production phase.

353
01:07:37.180 --> 01:07:47.940
W- I don't mean, e- even w- within US, right, you see some companies are more sort of cutting edge and maybe mostly coastal companies. You look at Europe, right, w- we're starting.

354
01:07:48.280 --> 01:07:57.580
Look at Japan, they haven't summoned the businesses, they're not even digitalized yet. Like, when they are gonna pick up the whole demand, so, uh, uh, uh, uh, use cases, right?

355
01:07:58.020 --> 01:08:13.200
So tho- those are, people are trying to figure out, when does the true enterprise ROI-backed, um, demand sustainably gonna come in, right? Obviously lower the cost will translate to people more willing to experiment.

356
01:08:13.460 --> 01:08:18.059
They're okay with not having gigantic ROI because they're spending, but not that much.

357
01:08:18.460 --> 01:08:27.550
They can try 100 things and pick maybe two or three that make sense to them versus, "Holy shit, it's so expensive, I can only try two, but make sure one works," right?

358
01:08:27.640 --> 01:08:37.900
Just by lowering the cost, you allow a more piloting program to start without too much a threshold. So at the end, yeah, it give people more, more optionalities to adopt, for sure.

359
01:08:39.580 --> 01:08:49.060
You know, one interesting thing is that, um, pretty much all the end users, and even, like, many of the application companies, uh, just care about token price.

360
01:08:49.160 --> 01:08:57.340
They don't care what, they're not even aware what GPU is running underneath. We shouldn't. But I mean, but, yeah, and they should. And, well, what are the implications long-term for the industry?

361
01:08:57.420 --> 01:09:06.640
Maybe you've already spelled it out, but, like, you know, the world's gonna consume tokens, they don't really care about the, the GPU model. Right. What does this mean long-term?

362
01:09:07.100 --> 01:09:15.180
So, which is, which is interesting, right? You look at our GPU indices, majority are NVIDIA-based, right? We're, we're agnostic, we do have MI300 indices.

363
01:09:15.840 --> 01:09:24.200
Um, but essentially if you do token, you and me couldn't care less what GPU's running on there. It can be some Nova, it can be anything, it can be different design houses, right?

364
01:09:24.299 --> 01:09:37.210
So that essentially token indices, token, um, price, can be more encompassing of the whole, um, uh, infrastructure, right? In system than just, hey, this is particular chips. Um,

365
01:09:38.120 --> 01:09:51.240
and the way I would think about this is we all need to use electricity every day, but we don't really care, I mean, a- aside from o- obviously, uh, you know, uh, green, being green and then the carbon footprint, right?

366
01:09:51.300 --> 01:09:56.220
We don't really care the energy source, it's how people generate electricity. We care about electricity prices at the end.

367
01:09:56.520 --> 01:10:01.720
So similarly to, to this, right, we care about token prices, we don't care about what GPU running underneath it.

368
01:10:02.180 --> 01:10:09.640
If you are average user, like, we don't need to hedge our electricity use at home, 'cause it's not that much money, right? We'll, we'll pay whatever on the spot, whatever they charge us.

369
01:10:10.120 --> 01:10:22.390
But you own a factory, electricity token is a essential productive mean, can be 10%, can be 80% of my cost when goods sold, yeah, I will care a lot, right? I wanna hedge the token prices.

370
01:10:22.950 --> 01:10:30.899
'Cause right now token prices are pretty much on demand. You're paying API, for example. You're paying whatever they charge API. You can't just say, "Hey, can we talk about it?

371
01:10:30.980 --> 01:10:40.760
Can we just say, you are gonna pay you this price [laughs] for the next six months?" You can't do that. So that's why Compute Exchange doing the token forward packages now, is you can lock in your price.

372
01:10:41.280 --> 01:10:54.520
So obviously even for the same model, okay, three, you have all different parameters, different precisions, different quantizations, and different sort of other, um, latencies. You, people will charge you differently.

373
01:10:54.580 --> 01:11:02.460
They should, right? So whatever you decide on, you, whatever the specs you decide on for that particular model, you wanna lock in your price.

374
01:11:02.940 --> 01:11:12.410
Say, "Hey, if I was dedicated to this particular, um, open source models, this is the price that I wanna pay, right, for next six months at least," and I locked in, right?

375
01:11:12.880 --> 01:11:36.520
You can, you, you, by th- in, in theory, you can do so, 'cause you know what, as from token factory point of view, you know what GPU you have, you know what's the throughput for that particular cluster, you know in theory how much tokens you can generate in probable tokens, and then slap a margin on top of bare metal, and you know how much you wanna charge people, and you can lock in the prices with that large enterprises, and then people both side have peace of mind.

376
01:11:36.860 --> 01:11:46.220
You know you're gonna be consumed on token factory point of view. From client, you know what price to lock in, you know exactly what you're gonna get. Then rely on API, which you don't even know the precision sometimes.

377
01:11:47.580 --> 01:11:57.370
Carmen, a closing question from me. So as Compute Exchange is designing these token forwards, the necessary counterpart to the GPU hour, uh, the GPU rental price forwards,

378
01:11:58.600 --> 01:12:07.080
it seems like there are more variables that would go into informing the price of those token hours versus GPU rental rates.

379
01:12:07.260 --> 01:12:19.070
I could be totally wrong about that- It- But I'm curious, when you're designing this product, what do you have to take into consideration for nailing how to specifically tailor these forwards for the token price model?

380
01:12:19.780 --> 01:12:23.500
So similar to GPUs, right? Even GPUs, they are heterogeneous.

381
01:12:23.740 --> 01:12:33.320
So even the H100, even the same C- CPU, same RAM, same geolocation can have different throughputs, and different latencies, and different memories, and different flops, right?

382
01:12:33.360 --> 01:12:38.160
So you need third-party benchmark to say, "Hey, this is what the spec says, that's the model itself.

383
01:12:38.640 --> 01:12:47.960
Here's the token, here's the, the throughputs te- a- availability latency we observe from third-party point of view, and here's SLA." Again, it's heterogeneous.

384
01:12:48.040 --> 01:12:56.140
You need third party to say, "Hey, this is the deliverables category." Again, it's a physically settled products, either token or GPU, right?

385
01:12:56.200 --> 01:13:06.880
You define the metrics, you have to meet that metrics, and you charge us prices, and that's end of story, right? So you have to ma- make sure third party benchmarking have all the observable, uh, results.

386
01:13:08.900 --> 01:13:21.820
Carmen, thank you so much for your time and your insights. Would love to have you on again, uh, as the contango continues, uh, token pricing and GPUs. Carmen, thank you so much. Yeah, thank you for having me.

387
01:13:21.840 --> 01:13:38.396
Love to come back. Have a great weekend, Carmen. Cheers. Bye. Have a good Friday. Cheers. Uh, glad I know what Contango is from hash price forwards. [laughs] So- It, you- That is wild to me. Yeah, it is wild.

388
01:13:38.696 --> 01:13:50.536
Nothing, nothing crystallizes the everyone is short compute narrative more than that. Yeah. Yeah, if you didn't, yeah. And it's crazy what a forward curve can reveal.

389
01:13:50.616 --> 01:14:02.526
It, it, it basically cuts through, like, will they, won't they? You know, is your favorite blogger right? Well, just look at the market. So, um, obviously asterisks, look at volume. Well, yeah- Other than...

390
01:14:02.536 --> 01:14:07.076
because that, that forward curve could collapse tomorrow. Yeah, I could be wrong. I mean, not tomorrow, but you know what I mean.

391
01:14:07.266 --> 01:14:22.476
It, all it takes is some shifts in how people are viewing the capacity environment, more capacity coming online, usage dropping, or another variable that I haven't mentioned, and that could go to backwardation, right?

392
01:14:22.516 --> 01:14:36.146
I mean, we've seen this in the Luxur, Luxur's forward curve a lot. A, a change in Bitcoin's price, a change in difficulty can completely throw that regime on its head. But- Yep... anywho. All right.

393
01:14:36.156 --> 01:14:52.916
Thank you so much for spending the week with us, finishing the week out strong with Blockspace Live. Uh, if you like our content, you can find all of it written and on demand at our website, blockspace.media.

394
01:14:52.996 --> 01:15:01.596
That's our website, blockspace.media. And if you got your Bitcoin on a cold cart, get it off. Never thought I'd be saying those words, but go do it now.

395
01:15:02.976 --> 01:15:15.616
Heart goes out to all my fellow Bitcoiners who are panicking or have opened up their wallets to chagrin. Panickins in control once again. Uh, yeah, [laughs] I know. Um, anyway, but there's always a brighter future.

396
01:15:16.116 --> 01:15:26.596
Um, don't know when, but sometime in the future. Thank you for listening. Uh, this show is brought to you by CleanSpark, NASDAQ listed ticker CLSK. I'm Charlie. I'm Colin.

397
01:15:26.636 --> 01:15:41.916
And quick note, next week we are going heavy into earnings. Yeah. We will be covering earnings exhaustively starting next week and into the following week as some of our favorite names release their quarterly earnings.

398
01:15:41.956 --> 01:15:55.256
We're talking Hut, we're talking Cipher, we're talking Riot, Mara, Keel, all the way down the line. We will be having coverage on the live stream of the numbers, and we will also have executive interviews lined up.

399
01:15:55.656 --> 01:16:08.236
So stick around for all-inclusive coverage of that. And with that, have a great weekend, y'all. Have a great weekend, y'all. [upbeat outro music]