WEBVTT

1
00:00:00.140 --> 00:00:13.080
[upbeat music] What's up, y'all? Welcome back to the Blockspace Podcast. We are coming at y'all live from Bitcoin Vegas twenty twenty-six. I'm with Taylor Monnig of CleanSpark.

2
00:00:13.200 --> 00:00:21.940
Taylor, thank you for carving out some time in what is a very busy and bustling conference to record this for us. Hey, thanks for having me on my favorite podcast. Look forward to the conversation.

3
00:00:22.000 --> 00:00:24.280
Yeah, man, it's a fun one.

4
00:00:24.360 --> 00:00:40.080
Um, I'm happy to be talking about AI and HPC retrofits, or rather, even just greenfields, um, with a guy like yourself because we were talking before this, and you really manage or oversee almost every aspect of that for CleanSpark.

5
00:00:40.560 --> 00:00:45.540
So I wanna ask you a question that I asked off camera that I think sets the stage well.

6
00:00:45.600 --> 00:00:56.080
Like, what has been one of the biggest revelations about designing a data center for HPC and AI workloads rather than Bitcoin mining? Like, what's one of the biggest differences?

7
00:00:56.280 --> 00:01:05.860
I mean, I think every part of the build is, is different. Um, but really when you come down to the basis of it, right, it's a, it's a warehouse full of computers. And so at the end of it, y- that's what you're building.

8
00:01:06.220 --> 00:01:11.980
In the Bitcoin mining space, uh, you know, we talked about it a little bit earlier, y- you buy a miner, and they give you a little bit of guidance, right?

9
00:01:12.040 --> 00:01:17.720
Maybe a little pamphlet, a little bit of engineering support on, "Hey, this is gonna be a successful environment." And it's really been the Wild Wild West.

10
00:01:17.840 --> 00:01:21.130
Uh, you know, when me and you started mining, air cooling was the only thing that was around.

11
00:01:21.480 --> 00:01:30.290
We saw a rapid increase in immersion cooling and the adoption of that in our tenure in Bitcoin mining over the last ten years, the introdu- introduction of hydro. So it's been like a steady progression, right?

12
00:01:30.300 --> 00:01:36.880
And everybody's kind of been able to build whatever they want. Miner manufacturers adapt a little bit to what the customer needs are. You go down that path, right?

13
00:01:37.300 --> 00:01:44.340
But I think the biggest difference for me is, is when we're talking with the hyperscalers, the neo clouds, you know, just our general customers, it's no longer the Wild Wild West.

14
00:01:44.700 --> 00:01:48.860
In some cases, they come in with a 30-page pamphlet that says, "This is our basis of design.

15
00:01:49.180 --> 00:01:59.700
This is every mission-critical milestone you guys must hit," whether it's the strength of the slab, whether it's the humidity control, whether it's the GPM of the water flow. These are highly subscribed assets, right?

16
00:01:59.720 --> 00:02:07.760
These guys know exactly what they want built. They're working endlessly on the other side to get every little bit out of those GPUs as they're racing against the market of AI.

17
00:02:07.860 --> 00:02:18.890
So the biggest revelation for me is while the buildings are much more complicated in the sense of, like, network topology, you have more networking, uh, fiber in one rack than you do in a whole entire Bitcoin mine.

18
00:02:19.420 --> 00:02:29.140
Um- That's crazy. Yeah, it- it's much more crazy of a build, as in there's much more items. But it's almost easier in a way because you, you're not building something that you're not so much unsure of but is unknown.

19
00:02:29.220 --> 00:02:36.769
A lot of these sites and these basis of design have been built by other builders, other developers, and you're stepping in and building on something that you know is, is concrete. There's a good basis.

20
00:02:36.769 --> 00:02:41.940
There's a clear blueprint. There's a clear blueprint. Like what you need for, for the specs that you need to meet- Yeah... and for the conditions that you have.

21
00:02:42.100 --> 00:02:51.100
Yeah, and it differs whether it's a, a TPU or a GPU or whatever the end, uh, you know, compute load is. It's all gonna vary, but you get a very concrete building block.

22
00:02:51.120 --> 00:03:02.170
And not to mention the level of support that you see from companies like NVIDIA, Intel, and some of the major hitters is overwhelming compared to the level of support that we would have received from any of the Bitcoin mining manufacturers.

23
00:03:02.220 --> 00:03:05.400
They're really there with you and the customer hand-in-hand to be successful.

24
00:03:05.840 --> 00:03:16.280
That's interesting because I would have thought that it would be very challenging to go from, you know, very simple, just we're putting this shoebox miner here. We need to make sure that it's powered correctly.

25
00:03:16.380 --> 00:03:23.300
Obviously, you wanna make sure you have everything to spec. But you don't have to worry about all the other auxiliary infrastructure to support it as much as you do with AI HPC.

26
00:03:23.620 --> 00:03:32.200
I thought that that would be a steep learning curve. It sounds like from what you're saying, it's almost refreshing from an engineering perspective to have clear rules and a clear guidebook to follow.

27
00:03:32.280 --> 00:03:38.180
Yeah, and I'm, I'm certainly not saying it, that it's easy, right? And that we didn't have any pitfalls and there were no problems. Like, that's certainly not true, right?

28
00:03:38.580 --> 00:03:40.890
We had different steps, different hurdles that we had to get through.

29
00:03:40.980 --> 00:03:47.910
But to your point, when you have a clear path of where you need to get to go, it's a lot easier to build the, the steps, hire the right people, bring in the right thing, right?

30
00:03:47.940 --> 00:03:51.360
And it's not honestly a, a big surprise to a lot of us at CleanSpark, right?

31
00:03:51.700 --> 00:03:59.260
We have multiple members of the staff that have experience in both high-performance computing, standard data center operations, co-location, right?

32
00:03:59.320 --> 00:04:02.740
Like, we have a lot of guys because that was part of our thesis on Bitcoin mining, right?

33
00:04:03.040 --> 00:04:13.860
Is, okay, let's go find guys that operate highly complex facilities, bring them to our not so highly complex facilities, and they're gonna be, uh, overqualified to run a Bitcoin mine, but it's gonna be ran really well.

34
00:04:14.300 --> 00:04:22.720
When you look at the networking topology of redundancy and some of the stuff that we built into our big mega locations like Sandersville, it follows a pretty typical data center blueprint, right?

35
00:04:23.100 --> 00:04:30.680
So not everything was perfect. It definitely is a learning curve. But to your point, when you got something that's concrete that you can work towards, a lot easier to fill those gaps.

36
00:04:31.020 --> 00:04:39.540
And having the privir- pri- previous experience of a lot of our staff's been extremely helpful. And they've also been super interested in getting back into HPC, getting back in the state of the data center.

37
00:04:39.560 --> 00:04:46.760
So it's not like we're having to pull teeth, you know? The guys are just naturally excited about it, continuing to evolve their career and, you know, go into the future of, you know, AI and HPC.

38
00:04:47.580 --> 00:04:51.860
So are you mentioning Sandersville? And that's a very good segue because I wanted to get to that at some point.

39
00:04:52.800 --> 00:05:04.820
Really interesting site, you know, a prime location, right, in Georgia, um, in terms of being close to metropolitan areas and also just having access to reliable baseload with all the nuclear that Georgia has.

40
00:05:05.560 --> 00:05:10.120
W- what, what have been the considerations for going about a retrofit?

41
00:05:10.260 --> 00:05:20.080
And then we'll kind of tease this out later versus, um, or, uh, for, for Sandersville, are, are y'all greenfielding or are you, are you-- Like what, what's, what's gonna be the process, I guess?

42
00:05:20.140 --> 00:05:29.520
Because when I look at the ones in Texas, like clear greenfield. Sandersville, y'all already have operations. Yeah. So what's the approach for how y'all are thinking about taking that facility and repurposing it for AI?

43
00:05:29.600 --> 00:05:34.350
Kinda nailed it in your question, so we'll step through that, right? So, you know, Sandersville, highly valuable asset for us, right?

44
00:05:34.380 --> 00:05:41.140
To your point, close to a metropolitan city or what the data center industry looks at as the NFL cities, right? Very close to downtown Atlanta.

45
00:05:41.200 --> 00:05:49.520
To your point, we already have reliable power, and the bit- biggest, biggest part or best part for us is it's proven power. We can bring a customer to our location.

46
00:05:49.560 --> 00:05:57.400
There's two hundred and fifty megawatts of power being drawn from the substations that are there. We're not waiting on a substation build. We're not waiting on a PPA. That power is live and current, right?

47
00:05:57.960 --> 00:06:06.490
And then you, you bring up the next point. Do we retrofit our, our, our chicken coop style facility that we have there, or do we retrofit our containers that we have there? 'Cause we got a mix, right?

48
00:06:06.520 --> 00:06:13.750
We have containers, chicken coop, and actually a large amount of immersion on that site.Um, so we looked at the financials and the processes of, of retrofit versus greenfield.

49
00:06:13.779 --> 00:06:19.870
And what we did is we worked with the city and said, "Hey, would you guys be acceptable if we move this power to another plot of land right down the road?"

50
00:06:20.210 --> 00:06:28.570
Development authority there, "Absolutely, we'd love to work with you guys." We got a tremendous relationship with the guys in Sandersville, all the way up to stack, to the mayor, the city council, right?

51
00:06:28.630 --> 00:06:35.810
We've, we've gone in there, we put our best foot forward. So we actually acquired another plot of land, um, uh, up the road across the street to our current facility.

52
00:06:35.970 --> 00:06:39.470
We've worked with all the right partners there in the city, including MIAG and everybody.

53
00:06:39.560 --> 00:06:45.770
So we are actually going to mine at our current location until the day the data center flips over, which is strategic for us, right?

54
00:06:45.850 --> 00:06:54.730
We have some miners there that by that, you know, let's call it a year and a half build cycle, the end of life, we can naturally retire those machines. Yeah. They did their job and we paid them back, made our money.

55
00:06:54.830 --> 00:07:04.760
Those we'll naturally retire, like S19 Xps. If the hash price continues where it is, we'll move our S21s to a location where we have Xps, retire those, backfill the fleet. Um, so yeah. Yeah.

56
00:07:04.790 --> 00:07:13.630
We're not going to retrofit those. We're actually gonna mine with those as long as we possibly can. And we've already started, uh, flattening the land, getting all the trees, stumps, all that kind of stuff out.

57
00:07:14.070 --> 00:07:25.570
By now it's actually probably done, clear and flat. And then we'll continue on our construction process in a completely greenfield new environment. So what, what will, uh, the, uh, let's say Sandersville 1.0- Yeah...

58
00:07:25.590 --> 00:07:33.010
for the old site, what, what eventually will come of it? Um, so we'll take our, uh, immersion modules that we put in, we'll take those to another location.

59
00:07:33.410 --> 00:07:42.350
Um, if the air-cooled fleet is in a place where the, the margins are still correct, the hash price is participating, take those air-cooled units out, move those to brand new- Those will retire once the AI data center- Yeah.

60
00:07:42.370 --> 00:07:49.630
And then we'll, we'll retire our chicken coops and we'll, we'll take those down if the hash price is applicable. We'll go stand those up somewhere else and continue to mine on them.

61
00:07:49.690 --> 00:07:53.310
But everything that we've put into Sandersville is something that we can take apart and move.

62
00:07:53.810 --> 00:08:00.270
Um, but we'll still have a large, uh, office presence there to obviously support the construction of the data center and then support the data center in the future.

63
00:08:00.730 --> 00:08:11.610
And then we're actually gonna refurbish some of the land into storage and, uh, doing a major increase on our miner repair center that we're gonna build out there. We are CleanSpark, America's Bitcoin miner.

64
00:08:12.230 --> 00:08:26.190
A publicly traded company with the largest operating hash rate. Powered entirely by self-operated infrastructure across four states. This is our proof of work.

65
00:08:26.870 --> 00:08:40.119
And we are setting the standard for what's next. Learn more about the intersection of energy and Bitcoin at cleanspark.com. So when you go about greenfielding one of these facilities,

66
00:08:41.530 --> 00:08:52.770
what's, what are some of the first steps that you take? Like, how do you, how do you lay the groundwork and plan for taking on this monumental task of setting up one of these massive data centers? Yeah.

67
00:08:52.850 --> 00:09:01.990
So I mean, breaking it down step by step, you know, starting at the core with a, with a Bitcoin asset, I think we all honestly look at it as a 5 to 10 year asset, and we plan around that.

68
00:09:02.050 --> 00:09:08.810
So the first major change you're making is you're now looking at a 25 to 30 year asset, right? So you need to make sure that the leases coincide with that.

69
00:09:08.890 --> 00:09:16.220
I know I'm getting basic and boring here, but I want to get into the weeds, right? Gotta make sure the leases are good. You gotta make sure the power contract's, um, good for that term, right?

70
00:09:16.290 --> 00:09:23.970
You need to make sure that the redundancies that the customer is gonna wanna be is, is in place. And then from there it's honestly, you know, kind of a standard construction build.

71
00:09:24.010 --> 00:09:32.650
You know, you go in, you work with your customer, they're gonna give you a basis of design or a set of requirements that we talked about earlier. From that basis of design, you're able to architect, "All right.

72
00:09:32.710 --> 00:09:37.750
Fantastic. The, the slab I need to build is this, the building I need to build is this, the office space is this."

73
00:09:37.830 --> 00:09:43.619
You start to lay that basis of design over the land, see where the topology cooks out and everything like that.

74
00:09:43.630 --> 00:09:57.890
Once that's all agreed and set, start marking your fire entrances, your fiber entrances, your roads, all those little minor details that maybe you don't, you know, care so much or think about on the Bitcoin site 'cause you're just, you know, bringing in a Starlink or you don't really need that so much support on a Bitcoin site.

75
00:09:58.430 --> 00:10:01.910
Um, and then, you know, you flatten the land out and make sure it's all buildable.

76
00:10:01.950 --> 00:10:13.179
And then you just start from the concrete up of building out that basis of design with all the, you know, prescribed requirements that that person has, whether it's liquid cooled or air cooled, that'll obviously change your equipment and design philosophy.

77
00:10:13.440 --> 00:10:19.090
And then you build it like a general construction project. Get it to a, kind of a ready for service date.

78
00:10:19.120 --> 00:10:26.550
And depending on your contract, you might be responsible for building hot aisle containment into the, the white space. You might be responsible for bringing water lines into the white space.

79
00:10:27.090 --> 00:10:32.110
Customer might want you to also install the racks and, and go forward. So varies a little bit on contract.

80
00:10:32.200 --> 00:10:40.730
But outside of just making sure that you're secured for a much longer, um, uh, term of that asset, rather than that, it- it's, it's a lot similar to a Bitcoin mine.

81
00:10:40.770 --> 00:10:49.650
But to our point earlier, it's a much more complicated Bitcoin mine, uh, with a lot more redundancies, a lot more cable, a lot more fiber, a lot more of, of everything that we know. Then you go from there.

82
00:10:49.710 --> 00:10:57.070
But it's not exactly rocket science, you know. People have been building data centers since the early '60s, and they'll change and evolve just like we did in Bitcoin mining.

83
00:10:57.170 --> 00:11:08.750
But, uh, yeah, nothing major past just making sure you're set up for a much longer period of time and, and maybe being, uh, more secure than you would be in a Bitcoin mine with a, a little bit less risk than you'd be willing to take just on that shorter timetable.

84
00:11:08.790 --> 00:11:13.710
Right. Right. You know, you said it's more complicated obviously because of all the infrastructure that goes into it.

85
00:11:14.130 --> 00:11:21.069
Also seems more complicated in a good way as well because you can, uh, work with these people and have synergies with them.

86
00:11:21.130 --> 00:11:30.230
But, uh, it seems like all the different, uh, counterparties in this, i- in this business are more really involved in the data center process. Yeah.

87
00:11:30.390 --> 00:11:42.030
Like, like you mentioned, getting support from the, the folks at Dell, some of the OEMs. But imagine also, uh, your partner Submer also, um, have been helpful in terms of thinking about how to, to devise cooling.

88
00:11:42.930 --> 00:11:53.430
On, on that note, what are some of the, uh, considerations you have to take for cooling these computers versus Bitcoin miners? Like, how, how much different is that infrastructure? Oh.

89
00:11:53.440 --> 00:11:59.590
Oh, well, I mean, the infrastructure is certainly different. I mean, at our, what I would call some of our premier locations, all-- Sandersville, for example, right?

90
00:12:00.150 --> 00:12:08.530
We have these chicken coop style designs that we worked endlessly on the CFD, and they run a PUE of, of zero, right? Or, or one, however you wanna describe it.

91
00:12:08.790 --> 00:12:16.570
There's no entry fans, there's no exhaust fans, there's no air conditioning, there's nothing, right? The miner sucks in fresh air from the outside.

92
00:12:16.870 --> 00:12:26.306
We came up with some tricks and methods to make sure that that hot air on the other side lasted far enough away from the cold air intake that it can run through a 104-degree dayAnd have no input, right?

93
00:12:26.346 --> 00:12:35.526
But a standard data center's gonna have chillers and, and, and all these things that are built in, and all of those are what's gonna end up doing the cooling and making sure that those chips are at the correct level.

94
00:12:35.606 --> 00:12:53.906
So from a engineering, like, basis of the, like, base layer, Bitcoin mining, we almost looked at it as what's the most to keep it alive, versus the Bitcoin industry, it, like we talked about, has these very strict set of s- just set of, like, specifications that they say, "Hey, we want it to be 72 degrees in the data center plus or minus one degree.

95
00:12:54.206 --> 00:13:04.036
We want the humidity to be exactly this," right? So the engineering perspective is instead of building to, to survive, you're building to thrive, is almost like the way to look at it, right? Yeah.

96
00:13:04.066 --> 00:13:12.636
And so all of those things make it more complicated. But to your point, the data center companies are willing to spend money on redundancy. So if you have a chiller go down, you're gonna have a backup chiller.

97
00:13:12.646 --> 00:13:20.406
You got a CDU go down, you're gonna have a backup CDU. So there's redundancies and better maintenance processes that go into all these facilities that make them function at a much higher level.

98
00:13:20.446 --> 00:13:27.666
And then a little bit off-topic here, but kind of on the same level, the quality of the infrastructure going in is monumental, right?

99
00:13:28.146 --> 00:13:39.086
The, the difference of a, a, a, a Bitcoin mining, a immersion container versus the level of quality that we're gonna receive from a, a Vertiv, right, an Integra or Submer is gonna be through the roof, right?

100
00:13:39.166 --> 00:13:48.086
So not only is the building more complicated, it's more redundant, but, like, the core product that you're putting in there is at a much higher quality than what we were seeing in, in Bitcoin mining infrastructure.

101
00:13:48.386 --> 00:13:58.066
Yeah, I mean, you've been referring to them as chicken shacks. Yeah. Right? You know, and it's, that's, uh, that's not to denigrate the fact that these are well run and clean, but you just don't need that much.

102
00:13:58.126 --> 00:14:05.686
You know what I mean? For the Bitcoin mines. You can- Oh, and I don't mean to degrade the chicken coop at all, right? From a engineering perspective, it's a beautiful- Yeah... it's a work of art, right?

103
00:14:05.766 --> 00:14:14.146
The fact that we can support 150 megawatts of computers with zero cooling, you tell that to somebody in the data center industry, they're gonna be really impressed, right? Yeah.

104
00:14:14.226 --> 00:14:20.896
But to your point, it comes with some drawbacks of not having the humidity control and the temperature control and those things. Yeah. So I don't, I don't mean to degrade what we've built at all. No, no.

105
00:14:20.896 --> 00:14:27.826
I'm very proud of it, actually. Yeah. Um, but to your point, yeah, you just get much more sophistication in the data center, and, and it's for good reason, right?

106
00:14:27.926 --> 00:14:36.346
The margins are much higher on the data center on the end user product, right? Yeah. Versus in Bitcoin mining, the margins are tight. We're trying to maximize everything that we possibly can.

107
00:14:36.406 --> 00:14:46.766
So spending 10% and having a 1.1 PUE is really impactful in a negative way for us, versus on a data center having a PUE of 1.3, they're still making plenty of money, and that's absolutely no issue.

108
00:14:46.846 --> 00:14:54.836
So they're able to have that extra capital and funding to make these facilities more sophisticated, more resilient, ultimately, you know, a better quality product. Right, 100%.

109
00:14:55.406 --> 00:15:12.526
This episode of Blockspace Live is brought to you by Ligos. Hedge funds are reeling asset managers, and lenders are in the throes of the 101025 liquidation events.

110
00:15:13.446 --> 00:15:24.506
But with Ligos, you always know that your Bitcoin is safe. Counterparty risk is rampant, so it's more important than ever to understand who actually controls your Bitcoin, and with Ligos, that is always yourself.

111
00:15:24.896 --> 00:15:27.586
Don't be the next FTX or Celsius victim.

112
00:15:27.626 --> 00:15:39.646
With Ligos, you can use Bitcoin-native smart contracts to always self-custody the Bitcoin that you are lending out through the platform, and you always know that it's there because Ligos makes it super easy to verify.

113
00:15:39.886 --> 00:15:48.866
They are our preferred Bitcoin-backed lender, and we recommend you check out their services. With them, you always know where your Bitcoin is. Hold your keys.

114
00:15:49.086 --> 00:15:58.486
There's no wrapping, bridging, or rehypothecation, and you can get competitive rates as low as 10% APR. Go to ligos.finance to learn more.

115
00:15:59.546 --> 00:16:09.306
We were talking before recording a little bit about the decision to focus on a power shell- Yeah... CleanSpark versus going into the NIO Cloud. Can you speak to that decision?

116
00:16:09.326 --> 00:16:25.766
Because to me, it's one of the more interesting, uh, divisions right now with the public Bitcoin miners that are verging into this business because you have basically Irin is the, the, I would say, the blue chip here, and then Hive is also, uh, making their way into it as well.

117
00:16:25.786 --> 00:16:32.616
But most of the rest of the industry is saying, "We're just gonna run the infrastructure. We're just gonna manage the power." Why did CleanSpark also decide to take that approach?

118
00:16:32.746 --> 00:16:39.616
So it's a multitude of stuff, but I'll kind of break it down into, into two key points. Something that we're really good at is build and operation, right?

119
00:16:39.646 --> 00:16:47.586
We're one of the only few Bitcoin miners that actually built everything that we installed. We didn't third-party it out. We have a construction team at the time that was ran by Scott Garrison, right?

120
00:16:47.956 --> 00:16:56.706
And he built Sandersville, he built Washington. He built almost, you know, every megawatt that we have that we didn't purchase through acquisition, right? So we're really good at it. We're really repeatable about it.

121
00:16:56.786 --> 00:17:04.386
We, we understand that core competency, right? And then on the operation side, I think we've done a really good job, and I hope the market's seen it. I think we've been the leader in uptime.

122
00:17:04.426 --> 00:17:14.026
We've been the leader in efficiency, right? We've, we've really shown our prowess, and we can own and operate facilities. We talked about it. It is a little bit more complicated, but we've got the base fundamentals down.

123
00:17:14.066 --> 00:17:22.606
There's no difference for us to maintain a, a, a chiller rather than a, a, a dry cooler or, you know, the crossovers that come from our big basis in immersion cooling.

124
00:17:22.646 --> 00:17:27.786
A lot of that stuff transfers over, like, with the chip. We have a good understanding of chemistry and corrosion, how to maintain the pipes.

125
00:17:27.806 --> 00:17:36.366
Like, all the little minute details, right, are something that we're, we feel really good about. And the, the, the second half is obviously the rat race, right?

126
00:17:36.406 --> 00:17:47.466
You've been in Bitcoin mining since probably the S9, maybe a little bit earlier, and you've seen all the companies chase S9, S17, S19, S19 XP, S21, and it just never ends, right?

127
00:17:47.486 --> 00:17:55.386
And the capital outlay is obviously massive. The efficiencies are, are radically growing. And we just don't think that, one, we want to participate in that.

128
00:17:55.516 --> 00:18:01.426
And I also think that we're very, very early in AI, and there could be a massive change in AI compute over the next four to five years.

129
00:18:01.546 --> 00:18:13.206
And do you want to beat several billion dollars into some old GPU that has absolutely, you know, no efficiency because X, Y, or Z technology came out? I mean, look at the amount of capital that is going into AI and HPC.

130
00:18:13.246 --> 00:18:20.486
It's crazy. There is going to be innovation. And so for us, we feel really comfortable that we can provide a really meaningful, really high quality product.

131
00:18:20.526 --> 00:18:35.160
We can take advantage of the land and the power that we have, not only that we've acquired through Bitcoin mining-But to your point, our new greenfield builds that we have, and then we actually see a road long-term where, you know, we got a lot of kind of flak about having 32 locations instead of having mega sites, right?

132
00:18:35.170 --> 00:18:44.490
For right now, we're doing one part of HPC. Eventually we're gonna have to get the HPC to the end user. We think that there's gonna be a lot of validity in our 10, 20, 30, 50 megawatt sites.

133
00:18:44.510 --> 00:18:50.690
While they might not be the great for a hyperscaler model company trying to train something, the future of edge computing to me looks extremely bright.

134
00:18:51.030 --> 00:18:55.580
We're excited about repurposing some of those facilities in the future for a different type of compute that's coming up on the horizon.

135
00:18:55.890 --> 00:19:02.770
It, it seems to me that it makes a lot of sense to focus on the bread and butter of what... Because Bitcoin miners ultimately,

136
00:19:04.490 --> 00:19:16.560
kind of to put a cap on a lot of what we talked about, you're- there's not so much hands-on, uh, management of the compute itself. So make sure the computers don't overheat basically, right? Yeah.

137
00:19:16.590 --> 00:19:25.110
Uh, you don't have to network them or anything. You just point them to a straight up off you go to your miner. And so it makes sense to me that, like, you know, Bitcoin miners are basically doing power management.

138
00:19:25.530 --> 00:19:35.030
It's just really what it boils down to. That's all. Why they're going to AI, HPC, uh, 'cause they can make more bang for their buck. I'm curious for your thoughts on what do you think it would look like?

139
00:19:35.070 --> 00:19:45.870
Like place yourself in the shoes of a Bitcoin miner that was deciding to go the DO Clouds out. Yeah. Where would you even start trying to hire out for something like a, someone who's gonna manage the cluster?

140
00:19:46.390 --> 00:19:51.310
Like, do you think you could train for that in-house? Do you think you would almost certainly have to hire out for something like that?

141
00:19:51.810 --> 00:20:03.250
Like how, how, how would you even go about building a team to source GPUs, run the, network them, find someone to buy the compute? It, it seems almost like an entirely different business in its own right. Oh, no.

142
00:20:03.290 --> 00:20:10.770
It absolutely is a entirely different business in its own right. And, and that's a completely fair statement. Um, you know, if I was gonna do it, I, I think there's two paths, right?

143
00:20:11.230 --> 00:20:18.370
Um, you can obviously go in and incentivize people at industry leading companies with, with capital and, and stock, and incentivize them to come over.

144
00:20:18.450 --> 00:20:25.170
Uh, but more than likely what I would do is try and find a young, a young startup and acquire them and bring them into the CleanSpark fold.

145
00:20:25.230 --> 00:20:29.990
The ones that have the competencies that we need, whether it's networking, procurement, that type of strategy.

146
00:20:30.130 --> 00:20:38.530
So yeah, you're either gonna have to go out and steal them from the big guys, which is gonna be hard because like we talked about, so much capital in this industry, they're not gonna let their high-performing employees go.

147
00:20:38.550 --> 00:20:48.170
Or you gotta find an opportunity where there's a startup with a bunch of guys that came together and maybe they're lacking funding or want to accelerate or wanna be a part of something bigger and make a strategic move to bring those resources in-house.

148
00:20:48.190 --> 00:20:57.060
But, um, training from scratch is gonna be really hard 'cause, I mean, you've interacted with both types of compute, not a lot of crossover between logging into your, your Bitmain miner root, root and- Right...

149
00:20:57.090 --> 00:21:09.330
you know, putting in a pool and setting it up, versus to your point, just, just networking, uh, topology alone in, in Bitcoin and, and HPC is, uh, astronomically complicated and it varies upon different chipsets and how people wanna ar- ar- architect things.

150
00:21:09.670 --> 00:21:17.130
Then- That's another reason why we wanna set- And it's a burgeoning form of compute. You know what I'm saying? Like finding the ways to optimize these computers for these AI workloads.

151
00:21:17.150 --> 00:21:25.970
They're evolving in tandem with each other, the use cases and the optimizations for how to actually provide your computing power. So we- I imagine it's a fast-moving thing.

152
00:21:26.310 --> 00:21:35.070
Yeah, and we think that's another strength of the core and shell, right? If we're building core and shells that are high performing, it allows our end customers to focus on exactly what you just said.

153
00:21:35.130 --> 00:21:38.830
How do they extract as much as they can out of their models, GPUs, wherever they're at in the stack?

154
00:21:39.130 --> 00:21:49.180
It allows them to focus on that, and then we wanna be a strategic partner working with very particular infrastructure providers to say over time, "Hey, how can we make this facility 1%, 2%, 3% better?"

155
00:21:49.210 --> 00:21:55.730
They can work on what they're really good at. We can work out on what we're really good at and meet in the middle on ultimately having like, you know, world leading products.

156
00:21:55.830 --> 00:22:04.210
It, it makes a lot of sense to split up the business, uh, or the business focuses I think because it's basically either side taking a different risk and shouldering a different burden, you know? Yeah.

157
00:22:04.690 --> 00:22:14.830
Um, and I think that for some people outside looking in, I've often had people, uh, my everyday life when, you know, I talk about this 'cause they're, they're pretty fascinated with the whole, you know, AI transition.

158
00:22:15.110 --> 00:22:20.080
Yeah. So like why wouldn't one company just do it all? And it's like, well, it's [laughs] really complicated. Oh, yeah.

159
00:22:20.180 --> 00:22:28.150
I mean, you started the chip level with what Nvidia's doing, right, and then you look at the rack level of, of Dell and all the other providers, and then above that you have the infrastructure providers.

160
00:22:28.210 --> 00:22:35.930
One company trying to do all of that, it's gonna be very hard for them to be, you know, leading experts in every step. You know, I guess we'll count Elon out of this. I think he's gonna end up going full stack.

161
00:22:36.390 --> 00:22:41.520
Um, but everybody else, no, I, I definitely get your point. You gotta basically build on everybody. It's, it's almost like building a team, right?

162
00:22:41.870 --> 00:22:46.160
Like a baseball team's not gonna win 'cause one guy's really good or whatever it is, right?

163
00:22:46.160 --> 00:22:52.180
But if you've got a team member where you're pulling expertise from every single different player, that's how you're gonna win a championship, and I kind of look at it the same as- Yeah...

164
00:22:52.190 --> 00:22:58.910
building a really good data center. You're gonna want the best chip, the best rack, the best infrastructure, the best land and power provider, the best construction team.

165
00:22:59.370 --> 00:23:02.550
You take all those expertise, put 'em in one area, you're gonna get something really good.

166
00:23:03.030 --> 00:23:14.290
So going back to the power shell model, the one- another big difference between setting up these data centers versus for AI versus buying Bitcoin is now you actually do have to go find a customer for the space.

167
00:23:14.870 --> 00:23:23.010
Uh, I mean, if you were hosting Bitcoin miners, that was always the case, but actually find someone with the parked computers for compute. I- I'm curious, uh, how do you go about that?

168
00:23:23.130 --> 00:23:29.330
You say, "Okay, we're, we're, uh, we're greenfielding these sites. We've got the power. We're- we've got some financing. We're going to build it."

169
00:23:29.970 --> 00:23:39.530
How do you start with business development for fi- finding a tenant for those locations? Um, you know, on a basis level, um, we, we went out and hired somebody, uh, his name's Boyd.

170
00:23:39.990 --> 00:23:50.150
Um, he was a, a 20-year veteran over at TierPoint, so he's been selling, building data centers for, you know, a majority of his career. Um, so that's one way. Obviously, he's got a lot of connections in the market.

171
00:23:50.210 --> 00:23:57.790
He goes out, you know, we get in the weeds of making the one-pagers, putting the data rooms together, you know, and then he takes all that ammunition and he goes out to his contacts and starts to sell them.

172
00:23:58.330 --> 00:24:06.350
But to be honest, and, and not to sound, you know, odd here, but given the, the heat of the market, we've also had a lot of inbounds, you know what I mean? Uh, being- I was wondering about that. Yeah.

173
00:24:06.390 --> 00:24:15.550
I mean, we're a big publicly traded entity. We've been very vocal about our Bitcoin mining, right? We made a CEO transition last August and, you know, brought back Matthew Schultz as our CEO and chairman now.

174
00:24:15.570 --> 00:24:24.300
And he's been very vocal that, and very clear that we have HPC endeavors and we are going to build data centers. And, you know, from, you know, that publication and, and that media- Yeah...

175
00:24:24.350 --> 00:24:27.030
um, we've been getting a lot of inbounds from a lot of amazing customers.

176
00:24:27.090 --> 00:24:38.582
So we've tackled it the traditional way of going outHiring a, you know, a veteran that has the experience, has the knowledge and can, you know, do his part, partnered with this just unrivaled need for compute.

177
00:24:38.642 --> 00:24:46.582
You know, we've been able to drive a lot of clients just naturally. Um, and then some of our mild contacts in the data center space have been reaching out. You know, Matt has some and Gary has some.

178
00:24:46.682 --> 00:24:56.982
So yeah, it's been a, it's been a mix of both, but for-- luckily for us, it's been a lot of natural inbounds. Uh, yeah. It seems to me like there's just so little power available- Yeah...

179
00:24:57.062 --> 00:25:06.622
that, um, if you have large blocks and you're, uh, a reputable serious company, I was wondering if they're gonna come to you because I've heard that from a lot of other Bitcoin miners, and it makes a lot of sense.

180
00:25:07.062 --> 00:25:17.202
I think it also kind of pours cold water on a, a criticism some of the miners had who announced pivots a little later. It's like, oh well, you missed out on all the deals.

181
00:25:17.782 --> 00:25:24.382
And it's like, no, dude, we're still in the first inning of this thing. Oh, I couldn't agree more, man. We are absolutely in the first inning of this thing.

182
00:25:24.442 --> 00:25:31.102
I mean, even looking at the progression of verbiage out of the, let's just call it the model companies, right? At first it was a hundred megawatts, five hundred megawatts, a gigawatt.

183
00:25:31.162 --> 00:25:40.502
Now they're talking about, you know, endless amounts of power and, and, and again, like we have no idea where AI's gonna go. AI has been great. It's been a really useful tool and it's doing all this kind of stuff.

184
00:25:40.582 --> 00:25:50.882
But what happens when it makes its first like major, major breakthrough on cancer or replacement for oil or one of these like, you know, mega thought projects that society ultimately needs to figure out?

185
00:25:51.392 --> 00:25:56.472
What's the investment gonna look like then when they prove that, hey, we hit this milestone, it's just gonna double. Is it gonna triple?

186
00:25:56.922 --> 00:26:04.662
I mean, the projections for these companies are absolutely, absolutely astronomical. So did we miss out on some of the first deals? Yes, but we're also learning from those first deals.

187
00:26:04.702 --> 00:26:10.702
What SLAs do they have that are pitfalls? What construction schedules really didn't make sense? Like, we're learning from all the first kind of movers.

188
00:26:11.162 --> 00:26:20.282
And so yeah, obviously there's, there's the argument to say you, you missed the initial boat, but it might be better to be on the second boat, learn from the first boat and not have what happened- Right... happen, right?

189
00:26:20.582 --> 00:26:28.622
And have a better outcome. So we try and look at it from a positive perspective. And, and to your point, it's endless opportunity. There's more than enough to go around for, for everybody.

190
00:26:28.722 --> 00:26:37.122
There's just not enough power, and the people that have secured land and power are, you know, in my opinion, really in the driver's seat. Yeah. And so power really is the bottleneck- Yeah... at this point, right?

191
00:26:37.402 --> 00:26:50.182
I mean, how do you think about, uh, the US' grid in terms of putting a certain cap on the AI and HPC, uh, expansion that we've-- you know, 'cause we've seen voracious appetite.

192
00:26:50.602 --> 00:27:02.662
But if you think about obviously we're gonna continue to build, uh, you know, more generation, but if you look at like the schedule for megawatts that are supposed to come online across the country over the next like ten years, a lot of it is renewable.

193
00:27:02.722 --> 00:27:11.502
Yeah. Like there's, I mean, there, there's some ga-- natural gas plants planned, but there's very-- there's like one nuclear facility. Um, and who knows when that'll even come online.

194
00:27:12.282 --> 00:27:21.042
So I'm just curious if you think that-- well, I guess a better way to ask this question, sorry, it's a little rambly, but like what does the rat race for megawatts look like in a few years?

195
00:27:21.642 --> 00:27:27.862
You know, I, I've been being told over and over for the last few years, you know, there's no more grid power, there's no more grid power, you know, X, Y, and Z.

196
00:27:28.382 --> 00:27:36.622
Um, I wanna give a huge shout-out to our land and energy team, Duncan Poe. We, we basically gave him the initiative in, in let's call it November. We said, "Hey, go find us power for HPC."

197
00:27:37.062 --> 00:27:47.262
And over, you know, a few short months, we're upwards of nine hundred megawatts secured all grid, no behind the meter, no gas, nothing. Just purely, purely off the grid, right? So I think there's still power to be found.

198
00:27:47.962 --> 00:27:57.822
I think humans are going to see this national need for power, and they're gonna double and triple down on it and start building generation to catch up, right? 'Cause like we're gonna continue to build, right?

199
00:27:57.922 --> 00:28:05.142
Humans just like don't stop evolving, right? Like we just continue to evolve, evolve, evolve, make things better. So I think the power ramp's gonna catch up. Not completely, right?

200
00:28:05.542 --> 00:28:11.302
I think a lot of the data center products they, or projects that you hear about may, may or may not be real, right?

201
00:28:11.382 --> 00:28:20.202
We see the cancellation rate of a lot of these facilities, like I think forty percent of them are stopped throughout the country. Um, but also you, you know, you heard from President Trump directly.

202
00:28:20.222 --> 00:28:25.762
He wants people to start putting on behind the meter and start paying for their own load. So I think that's an obvious, you know, place where it's gonna go.

203
00:28:26.302 --> 00:28:34.802
And then you have the, the, the modular reactors that have been, you know, kind of in the vaporware for [laughs] ten years, and people are super excited about the thought experiment that it is.

204
00:28:34.862 --> 00:28:42.842
But I think those are actually starting to come to market, truly be landed- They're building one in East Tennessee. They're building one- Yeah... near, uh, near Alcoa. Yeah.

205
00:28:42.922 --> 00:28:51.202
So behind, behind the meter I think is gonna be a huge driver. I think if this nuclear thing catches on in the way that it's been, uh, proliferated in the market, that could be another huge driver, right?

206
00:28:51.542 --> 00:28:59.242
But I also don't think the grid's tapped out yet. We keep finding power and, and maybe we just have a really special team or whatever it is, right? But we keep finding it.

207
00:28:59.442 --> 00:29:07.002
People have been telling us for years and years, "There's no more, there's no more." Well, CleanSpark's scaled from fifty to well over a gigawatt of power all on the, all on the grid.

208
00:29:07.202 --> 00:29:13.902
So- How much of that, not to toot y'all's horn too much, but we can extrapolate this out to the whole Bitcoin mining industry.

209
00:29:13.922 --> 00:29:20.022
How much of that is because that was the whole business for Bitcoin miners for the last, you know, five, six years in the US?

210
00:29:20.102 --> 00:29:29.692
It's like the reason why they've been so well-positioned for this AI boom is because they've been the only ones buying up- Yeah... a hundred megawatt blocks out, right, all over the US.

211
00:29:30.242 --> 00:29:38.182
I mean, is part of the success that y'all are seeing is because you just have teams that are dedicated to this and like know the game? Oh, absolutely.

212
00:29:38.242 --> 00:29:44.562
Like, I don't wanna get into the, the secrets of how our team does stuff, right? [laughs] Yeah. But from a high level, we call it the front door approach, right?

213
00:29:44.662 --> 00:29:52.602
We go into the city, we tell them exactly what we're gonna do, we show them exactly the infrastructure is, we show them sound studies, we show them that we don't use any water.

214
00:29:52.922 --> 00:30:02.242
Like we go in the front door and say, "Hey, transparently, this is exactly what we wanna do. This is the amount of power we wanna take. This is the land that we wanna build it on. Is this zoned correctly for you guys?

215
00:30:02.602 --> 00:30:09.141
Are you guys happy with this? Is the power company, you know, on board?" Like we try and do everything we can on the front end, and that builds a lot of really good relationships.

216
00:30:09.182 --> 00:30:15.742
And I, I'll highlight Sandersville again, right? We went into Sandersville. We took over a mine from a company that hadn't paid their, their power bill.

217
00:30:16.222 --> 00:30:22.842
We cleaned up the debt with the power bill, which was obvi- automatically like a good shoe-in with the city, right? We did everything that we told the city we would do.

218
00:30:22.962 --> 00:30:25.502
Every permit we would file, filed it exactly on time.

219
00:30:25.882 --> 00:30:38.474
Build and construction process, we used all local labor from that area, providing thousands of jobs throughout that process, right in the build.And when you go in through the front door, you're honest, you're transparent, you use the local trades, you help support the community.

220
00:30:38.554 --> 00:30:47.574
Like, we've worked with the schools, we worked with, you know, a wide variety of deals. You know, when that mayor gets a call from a mayor in another city saying, "Hey, CleanSpark's here. Should we let them in?"

221
00:30:47.614 --> 00:30:57.414
And this mayor's like, "Oh yeah, they've donated this, they've built that, they've done everything that they've said from start to finish," it helps grow your brand, you know, really organically, and I don't think a lot of people have taken that approach.

222
00:30:57.774 --> 00:31:07.903
But the reason why we're in the driver's seat is exactly what you said. Hyper-focused on securing as much land and power as we can, and then out of nowhere three years ago, HBC takes a foothold.

223
00:31:07.974 --> 00:31:14.194
All everybody wants is land and power, and the people that have it are the Bitcoin miners. A quick shout-out to our sponsor, Luxur.

224
00:31:14.654 --> 00:31:20.274
This episode is brought to you by Luxur Commander, Bitcoin miner management software for enterprise solutions.

225
00:31:20.494 --> 00:31:32.534
Commander gives you real-time fleet monitoring, bulk remote commands over your fleet, and Intelligent Miner, an automated profitability engine that runs every five minutes and adjusts power settings to live hash rate markets and energy markets.

226
00:31:32.574 --> 00:31:43.934
ERCOT back tests show 10% more profitability with intelligent mining versus binary mining. Commander Pro is $100 per megawatt or a 25 basis point pool fee adder.

227
00:31:44.274 --> 00:31:56.954
It's roughly half the price of competitors, but you can try it for free for 60 days. So if you want to learn more and are interested in getting started, go to luxur.tech/commander. So pivoting a little bit to

228
00:31:58.374 --> 00:32:12.514
adversarial thinking here, the, the AI CapEx boom has been almost unprecedented in a lot of ways. The billions of dollars, hundreds of billions of dollars have been invested and poured into building these data centers.

229
00:32:13.794 --> 00:32:17.474
Some of them have been delayed. Um, there have been a few announcements,

230
00:32:18.634 --> 00:32:33.234
and I'm just curious if you have any comments on what you see as being potential pitfalls for these AI builds and for these HBC data center rollouts. Another way to put it, what could throw kinks into the process?

231
00:32:33.314 --> 00:32:44.674
Not for CleanSpark specifically, but just for the whole industry. Because, you know, everyone's subject to laws, same laws of physics, you know, fuel costs rising, supply chains maybe failing in certain places.

232
00:32:44.714 --> 00:32:50.804
Walk us through how you think about planning for some of those hiccups. Yeah. So I think you hit a lot of the well-talked ones, right?

233
00:32:50.874 --> 00:32:55.594
You have supply chain, construction, like, you know, uh, just labor shortages, things like that.

234
00:32:55.614 --> 00:33:04.934
But one thing I don't think gets enough, uh, discussion, and it's something that's personally been taking up a lot of my time and I've been looking at is there is an anti-data center movement growing in this country, right?

235
00:33:05.394 --> 00:33:12.534
You see it in the canceled projects. You've seen it in, you know, some of the, the cities, uh... Austin city council members that have approved data centers, right?

236
00:33:12.594 --> 00:33:21.074
You see the Facebook groups, and even on, like, my own personal social media, kids that I went to school with and growing up with, I'm seeing a rise in anti-data center posts, right?

237
00:33:21.094 --> 00:33:29.614
And a lot of it comes from misinformation. You guys are gonna use all the water. Well, most data centers at this point are closed loop and use less water than a McDonald's does in any given day, right?

238
00:33:29.674 --> 00:33:39.074
You're gonna make the power rates go up. Well, in, in most cases, in the CleanSpark fashion, we've actually driven property taxes down, energy costs down in the communities that we operate in, right?

239
00:33:39.084 --> 00:33:44.714
Uh, so there's a ton of misinformation out there around data centers, and that is gaining traction, like, really, really rapidly.

240
00:33:44.754 --> 00:33:54.994
So we're trying to do everything that we can, not only for our own company, but for the general, you know, industry itself on properly educating people that just because we use the power doesn't mean your power bill's gonna increase, right?

241
00:33:55.694 --> 00:34:02.394
You think we're gonna use a million gallons of water. Actually, this whole entire system's closed loop. Some people promote that a AI data center has no jobs.

242
00:34:02.494 --> 00:34:11.054
We both know that there is a ton of maintenance, not only on the MEP side, but also in the white space of maintaining the GPUs, maintaining those chillers, pumps, air conditioners, right?

243
00:34:11.114 --> 00:34:19.174
Like, there's a lot of jobs and a lot of value creation that come out of this, and so we really need to fight a lot of that misinformation, provide people the truth on what's going on.

244
00:34:19.194 --> 00:34:21.034
You know, hopefully over time we can curb that.

245
00:34:21.064 --> 00:34:30.994
'Cause I think we can both agree AI in its different use cases is extremely important, not only for everything that is going on with the model companies and things like that, but also for natural defense, right?

246
00:34:31.074 --> 00:34:34.084
We do not want to be not market leader in AI, right? Right.

247
00:34:34.114 --> 00:34:45.174
Especially as you see some of these, like, amazing companies like Anduril and things like that are developing fully autonomous drones, submarines, planes flying right along F-35s, right?

248
00:34:45.294 --> 00:34:48.134
A lot of that relies on AI, AI compute, AI development.

249
00:34:48.534 --> 00:34:58.394
And as a country, you know, we really need to be on the forefront of leading AI, not only for, you know, our country to prosper, but also for natural defense, you know, keeping that together. Wow. So it's multifold.

250
00:34:58.534 --> 00:35:08.254
I, I agree with that 100%. And so you, you're saying that one of the greatest threats actually to this is pushback for these AI data centers in terms of impeding progress. Yeah, absolutely.

251
00:35:08.454 --> 00:35:16.274
Like, things getting tied up in local council or protests scaring, you know- Sure... people who would rubber stamp a project, right? 'Cause we're starting to see that from what we're seeing.

252
00:35:17.014 --> 00:35:25.994
Yeah, starting to see it, you know, all over the country, you know? And even in places like, uh, Virginia, you know, capital of data centers, there's starting to be a lot of pushback there.

253
00:35:26.034 --> 00:35:33.444
It ju- it's just a growing thing. Really interesting to, like, see where some of this misinformation's coming with. You know, I would assume that it's a big China influence of, you know- Right...

254
00:35:33.474 --> 00:35:36.134
trying to slow down our development so they could speed up their development.

255
00:35:36.574 --> 00:35:44.194
Spending a lot of money, you know, building a lot of coalition groups and things like that where, you know, people don't really look too much into where the funding for some of these groups comes in.

256
00:35:44.234 --> 00:35:52.514
And, you know, you start looking at some of the comments, and there may or may not be a lot of bots on a lot of the posts. Who knows where the bots are coming from. But yeah, it could definitely be something.

257
00:35:52.594 --> 00:36:01.154
We could have the best supply chain, all the right workers, all this kind of stuff, but be defeated by misinformation and ultimately having a city council or, you know, somebody step in and shut the build down.

258
00:36:01.854 --> 00:36:08.174
So just a few more questions, Taylor, and then we'll wrap up so you can get to your panel. What is the timeline for one of these creates?

259
00:36:08.694 --> 00:36:17.454
You could give, like, a rough estimate for how long something takes, like, something takes from, uh, announcing, planning, building, getting fence in, in there, booting up.

260
00:36:18.394 --> 00:36:24.414
If you're starting completely from scratch, the land is in prepared, you've never contacted a customer, could be upwards of two years, 24 months.

261
00:36:24.914 --> 00:36:34.074
Um, if you've already got a customer engaged, understand the basis of design, land's relatively flat, uh, you know, you're seeing accelerated construction schedules anywhere from 14 to 18 months.

262
00:36:34.134 --> 00:36:35.554
You know, people are getting better about these.

263
00:36:35.654 --> 00:36:44.090
The word modular is becoming more and more a thing, whether that's a, you know, actual modular build or building a-You know, still wall building and sliding the modules in to save time.

264
00:36:44.650 --> 00:36:57.110
Um, so yeah, you could range anywhere from a year if you're mega aggressive with everything in place and y- you started that day versus, you know, two years if you're a brand-new company coming out, having to figure out a basis of design, figure out a customer, you know, go through the whole marketing process.

265
00:36:57.650 --> 00:37:07.270
Yeah, it could be upwards two years. All right, Taylor, as a closing question, the annoying one. Uh, what is the future of CleanSpark's [coughs] Bitcoin mining?

266
00:37:07.510 --> 00:37:14.030
And I'm specifically interested in a lot of the smaller sites that y'all have around Billy Bucks. Oh, we're, we're still gonna continue to mine.

267
00:37:14.550 --> 00:37:20.650
Um, none of our leadership has ever publicly or privately said we're gonna stop mining. Um, we're at a really good fleet efficiency.

268
00:37:21.060 --> 00:37:27.610
Uh, we still actually have about sixty to seventy-five megawatts of infrastructure still to deploy. We have, uh, more immersion containers.

269
00:37:27.690 --> 00:37:38.650
We have S19 immersion XBs that we're actively deploying right now in, uh, Ripley, Tennessee. We're going into, uh, South Dakota. You know, we're actually, uh, energizing an expansion this week at our Campbell site.

270
00:37:38.730 --> 00:37:39.930
Uh, so yeah, we're gonna continue to mine.

271
00:37:39.970 --> 00:37:48.550
We have a relatively new fleet, and, uh, we'll naturally wind down, you know, starting with our S19 XBs, our oldest miners, and we'll take those as, you know, margins come and go.

272
00:37:48.650 --> 00:37:55.940
But, uh, yeah, Bitcoin is still an extremely part, big part of our business. Uh, we bifurcated our leader [coughs] bifurcated our leadership structure.

273
00:37:56.470 --> 00:38:05.590
That way we have a head of Bitcoin mining and a head of AI, and they both report up through me. I was curious about that, how you structure- Yeah... that internally in the company in terms of s- splitting those focuses.

274
00:38:05.850 --> 00:38:13.770
Yeah, I mean, we, we, we split the groups up. And so our head of AI is, is, uh, now Brad Ottos. He's worked under me as our SVP of mining for the last three years.

275
00:38:13.790 --> 00:38:22.460
There's literally no one else in the world I trust to run our operation. Um, and he's, he's taken the lead on that. We've hired a guy named Jeffrey Thomas. He runs our, our, our AI stack.

276
00:38:22.510 --> 00:38:35.430
And you know, I got the great honor and pleasure of being able to oversee both those departments and, and basically, like, watch how we're gonna, you know, use the strengths of, of one to help the other and, and where they should cross-pollinate and where they should separate and all those types of things.

277
00:38:35.830 --> 00:38:42.230
It's honestly been one of the funnest parts of my whole entire career is seeing all that come together. Um, but yeah, being super clear, we have no intent to stop mining.

278
00:38:42.590 --> 00:38:44.910
We're still currently deployed right around fifty exahash.

279
00:38:45.330 --> 00:38:55.070
Um, like I said, we got new sites coming along, and, you know, I think that represents, like, you know, anywhere thirty, forty million dollars of infrastructure, so we're putting our money where our mouth is, and we're out saying we're just gonna keep mining.

280
00:38:55.110 --> 00:39:03.670
You know, so we'll, we'll play it, uh, you know, year by year, but right now, definition is clear. CleanSpark's still a mining company. All right, I lied. Last question. Yes.

281
00:39:03.790 --> 00:39:15.470
Um, what, what's gonna be your biggest focus this year as you look forward to these green flows? What, what is the thing you are laser focused on for the rest of the year? Simple. Sign lease, build data center.

282
00:39:15.630 --> 00:39:19.450
You know what I mean? Uh, we've gotten our Bitcoin mining operation into an amazing spot.

283
00:39:19.490 --> 00:39:27.690
I've got more than enough confidence in Brad that he can take the helm up there and, and, and carry of, you know, if not all of the burden and load of our Bitcoin mining operation.

284
00:39:28.050 --> 00:39:37.070
Obviously, there, I'm still there to support and help in any shape, manner, or form. But right now, directive is clear. Sign lease, build data center. We're super excited about our HPC endeavor.

285
00:39:37.410 --> 00:39:43.810
I'm genuinely really excited about what AI can do, um, not only for our company and our shareholders, but obviously civilization for ourself.

286
00:39:43.870 --> 00:39:50.390
You know, we talked about those key societal problems that we have, replacements for oil, uh, you know, cures for cancer, all those types of stuff.

287
00:39:50.470 --> 00:39:59.370
If we don't build this compute, I don't really see us a way of, you know, getting there otherwise. Um, so yeah, that, that's my main focus. Keep it short and sweet. Sign lease, build data center.

288
00:39:59.830 --> 00:40:08.070
Taylor, thank you so much, man. That was a fun conversation. Yeah. Thank you. [upbeat music]