1
00:00:01,452 --> 00:00:02,933
Welcome to Let's Get Digital.

2
00:00:02,933 --> 00:00:04,354
This is Keri Charles.

3
00:00:04,354 --> 00:00:08,338
I'm your host and I am absolutely thrilled that you are here.

4
00:00:08,338 --> 00:00:11,270
You are in for a phenomenal show.

5
00:00:11,270 --> 00:00:13,532
I have with me today Wes Cummins.

6
00:00:13,532 --> 00:00:17,936
He is the founder and the CEO of Applied Digital.

7
00:00:17,936 --> 00:00:20,077
Thank you for coming on the show, Wes.

8
00:00:20,374 --> 00:00:21,400
Kerry, thanks for having me.

9
00:00:21,400 --> 00:00:22,573
I appreciate it.

10
00:00:22,752 --> 00:00:25,882
Yes, yes, uh let's get right into it.

11
00:00:25,882 --> 00:00:32,209
I have heard that you have a uh very interesting and inspiring story.

12
00:00:32,209 --> 00:00:37,953
So I'd love to hear a little bit about your journey and what got you here today.

13
00:00:38,572 --> 00:00:45,912
Yeah, I don't know if I have an inspiring story, but I appreciate you saying that.

14
00:00:46,952 --> 00:00:57,132
know, I think some people know, but it's an interesting start to the story because of
where it finishes, hopefully not finished yet.

15
00:00:57,852 --> 00:01:04,692
But so I grew up in a small town of 200 people in Idaho on a potato farm.

16
00:01:04,692 --> 00:01:08,526
And then I always make a joke of the

17
00:01:08,526 --> 00:01:19,506
One thing, besides working really hard, we worked on the farm, it still family farm times,
but the one thing I learned from working on a farm besides hard work was that I didn't

18
00:01:19,506 --> 00:01:23,586
want to be a farmer when I grew up because it is a lot of hard work.

19
00:01:24,686 --> 00:01:29,726
So I left, went to college, then I found myself into the investing world.

20
00:01:29,726 --> 00:01:35,506
I was a tech investor, been a tech investor for 25 plus years now.

21
00:01:36,206 --> 00:01:38,398
And as I spent a lot of time,

22
00:01:38,398 --> 00:01:42,301
Mostly in tech hardware, semiconductors, networking, lot of infrastructure.

23
00:01:42,301 --> 00:01:51,983
And then I saw an opportunity come along, which was actually in Bitcoin mining and it was
building data centers for Bitcoin mining back in late 20s.

24
00:01:51,983 --> 00:01:58,093
Started the company in 21, went aggressively into building data centers in the US for
Bitcoin miners.

25
00:01:58,454 --> 00:02:03,868
And then in 22, we decided to start to add high-performance computing.

26
00:02:03,868 --> 00:02:08,742
was a much different industry at that time in the early part of 22.

27
00:02:08,742 --> 00:02:19,029
So we hired some guys that were building actually data centers for Metta at the time,
designed a new purpose-built high-performance computing facility, much smaller scale.

28
00:02:19,129 --> 00:02:22,632
And we started building it in North Dakota, in our site in Jamestown, North Dakota.

29
00:02:22,632 --> 00:02:34,660
And the idea was we would add small increments of high-performance computing alongside our
Bitcoin facilities, high power density, high power needs, cooling, which is great for

30
00:02:34,660 --> 00:02:35,900
North Dakota.

31
00:02:36,081 --> 00:02:38,262
And so we were building 10 megawatts

32
00:02:38,262 --> 00:02:39,494
Jamestown.

33
00:02:39,564 --> 00:02:47,104
By December of 22, we were running some GPUs out of our facility in Jamestown.

34
00:02:47,404 --> 00:02:54,904
we had a few customers, mostly universities that were doing machine learning and deep
learning on the GPUs.

35
00:02:54,904 --> 00:02:56,484
And then ChatGPT happened.

36
00:02:56,484 --> 00:03:00,024
so ChatGPT didn't really change anything immediately when it happened.

37
00:03:00,024 --> 00:03:03,264
It was great to play around with.

38
00:03:04,124 --> 00:03:09,264
And it was novelty in that it went to a million users faster than any app.

39
00:03:09,518 --> 00:03:15,698
going, but then when Nvidia introduced the H100 in March of 23, our world started to
change.

40
00:03:15,698 --> 00:03:18,378
And then so I was out marketing data center capacity.

41
00:03:18,378 --> 00:03:20,638
We'd happened to have built some capacity.

42
00:03:20,638 --> 00:03:25,338
It was a pretty good fit for H100, 50 kilowatt rack densities.

43
00:03:25,338 --> 00:03:35,038
And then we started getting introduced into Frontier Model Companies, AI Labs, and what
they wanted was just GPU capacity.

44
00:03:35,038 --> 00:03:36,538
They want to deal with data center.

45
00:03:36,538 --> 00:03:39,451
So we started offering GPU capacity by July of that.

46
00:03:39,451 --> 00:03:52,388
year we had a little over a thousand H100 GPUs online and we had customers for that and
but it at that time because of what I saw in the market I went back to my data center

47
00:03:52,388 --> 00:03:58,573
design team and I said guys look I know I said we were gonna do 10 or 15 megawatts at
these sites

48
00:03:58,573 --> 00:04:02,453
I think we need to go back to building hundreds of megawatts of capacity.

49
00:04:02,613 --> 00:04:04,473
And so we worked through design.

50
00:04:04,473 --> 00:04:11,093
We actually worked with Nvidia's data center team, which I think at that point, a lot of
people didn't know that really existed.

51
00:04:11,093 --> 00:04:15,933
We worked with the Cray unit over at HPE that has a long history in supercomputing.

52
00:04:15,933 --> 00:04:18,733
And we designed a facility really around Blackwell.

53
00:04:18,733 --> 00:04:22,433
wasn't called Blackwell when we designed it, but we were looking at the Nvidia roadmap.

54
00:04:23,513 --> 00:04:25,333
so we designed the facility around that.

55
00:04:25,333 --> 00:04:28,536
We started building it in 23 because we

56
00:04:28,536 --> 00:04:32,429
We just thought Nvidia is the clear leader here and we want to just map to that.

57
00:04:32,429 --> 00:04:36,131
And you that facility turned on last month.

58
00:04:36,131 --> 00:04:38,213
The first capacity turned on last month.

59
00:04:38,213 --> 00:04:42,355
The remaining capacity in that building turned on this month.

60
00:04:42,355 --> 00:04:47,619
And then we'll have another building following that and another one following that that
we've contracted all the CoreWeave.

61
00:04:47,619 --> 00:04:57,375
And then we have another campus that looks similar to our Ellendale North Dakota campus,
also in North Dakota that we've contracted to a uh investment grade hyperscaler.

62
00:04:57,375 --> 00:04:58,466
So we target CoreWeave.

63
00:04:58,466 --> 00:05:05,169
We've plus the five investment-grade hyperscalers, are uh Microsoft, Amazon, Oracle, Meta,
Google.

64
00:05:05,210 --> 00:05:06,280
And so it's one of those.

65
00:05:06,280 --> 00:05:08,711
We haven't been able to disclose which one of those it is.

66
00:05:08,711 --> 00:05:14,885
uh So we're off and running and we're building at large scale and there's a lot of work
and we can talk about what went around that.

67
00:05:14,885 --> 00:05:17,036
But to bring this full circle.

68
00:05:17,428 --> 00:05:25,615
is we're doing this in small agricultural based towns, uh both of them right now in North
Dakota.

69
00:05:25,615 --> 00:05:26,856
We're looking at other towns.

70
00:05:26,856 --> 00:05:36,304
But for me, it's a really personal journey and interesting for me to be able to bring the
absolute leading edge of technology.

71
00:05:36,309 --> 00:05:39,751
to a town that's similar in size to what I grew up in, right?

72
00:05:39,751 --> 00:05:46,914
mean, Ellendale's five times the size of what I grew up in, but still a very small town,
great people, great community.

73
00:05:46,914 --> 00:05:53,797
We work really closely with the community and really try to be instrumental in positive
change.

74
00:05:53,797 --> 00:06:02,701
And the projects that we're doing are just such a perfect fit for these communities in
that it's a bit of a headache when we're doing the construction part of it.

75
00:06:02,701 --> 00:06:06,222
But at the end, you bring a few hundred jobs, maybe three, four hundred

76
00:06:06,222 --> 00:06:09,982
jobs into a location you don't change the face of the community.

77
00:06:09,982 --> 00:06:14,962
You create a massive amount of tax revenue for the community.

78
00:06:14,962 --> 00:06:19,682
now Allendale, North Dakota becomes the wealthiest school district in the state.

79
00:06:19,682 --> 00:06:22,142
The community, things that you don't think of.

80
00:06:22,142 --> 00:06:32,022
So you go to these towns and now all of sudden they were doing special tax assessments to
fix their sewer system, to do their fresh water, to fix a road.

81
00:06:32,022 --> 00:06:36,262
And now all of a sudden they have a budget that they can fund all of these programs
without tax

82
00:06:36,262 --> 00:06:45,158
the residents because you will find in these small towns typically property tax is higher
than it would be you know in a larger city because you have to provide these same

83
00:06:45,158 --> 00:06:56,225
essential services but and then it also has been rewarding in that it has actually allowed
people who grew up in Ellendale North Dakota left because there was no real job prospect

84
00:06:56,225 --> 00:07:06,062
in Ellendale North Dakota and they've actually come back to work at our facility and their
families there and it's really interesting to me and rewarding and how it's kind of

85
00:07:06,062 --> 00:07:12,182
went very much full circle coming from a small town where, you know, where the people left
behind.

86
00:07:12,182 --> 00:07:14,302
We're the last ones to get the internet.

87
00:07:14,462 --> 00:07:21,702
I think my parents stopped using dial-up internet in like 2018, and they still live there.

88
00:07:21,822 --> 00:07:25,102
so it is a good full circle story.

89
00:07:25,850 --> 00:07:29,293
I am hearing about Applied Digital everywhere.

90
00:07:29,293 --> 00:07:38,991
And in fact, I was in a meeting this morning with entrepreneurs, and it's just a smaller
mastermind, and somebody in the meeting said, I just bought the Applied Digital stock

91
00:07:38,991 --> 00:07:39,522
today.

92
00:07:39,522 --> 00:07:42,834
And I said, well, that's interesting because I'm interviewing the CEO today.

93
00:07:42,834 --> 00:07:46,928
So congratulations on all your success.

94
00:07:46,928 --> 00:07:52,471
And are you talking about that, the Polaris Forge 2?

95
00:07:52,471 --> 00:07:54,689
Was that what you were just talking a little bit about?

96
00:07:54,689 --> 00:07:57,020
Polaris Forge 1 is in Ellendale, North Dakota.

97
00:07:57,020 --> 00:08:00,091
Polaris Forge 2 is in Harwood, North Dakota.

98
00:08:00,452 --> 00:08:05,634
in Polaris Forge 1 is where we've contracted CoreWeave for 400 megawatts of critical IT
load.

99
00:08:05,634 --> 00:08:11,727
And in Polaris Forge 2 is where we've contracted the investment grade hyperscaler, uh
building the same design.

100
00:08:11,727 --> 00:08:15,388
our building design, we put a lot of work into this.

101
00:08:15,388 --> 00:08:18,750
We're on our fourth generation of data center design.

102
00:08:18,750 --> 00:08:24,262
So we've made a lot of uh improvements, I would say, both in the state

103
00:08:24,322 --> 00:08:34,629
scale the flexibility, the speed at which we can build these now and the cost at which we
can build them and then efficiency and the flexibility is key for the future, right?

104
00:08:34,629 --> 00:08:45,957
Because we're like, I always make a joke that operating in the AI uh ecosystem is like
operating in dog years and it truly is and the change is so fast.

105
00:08:45,957 --> 00:08:50,740
And so we're building assets that we expect to last for 30 to 50 years.

106
00:08:50,740 --> 00:08:53,331
Then we need to try to future proof those as much as possible.

107
00:08:53,331 --> 00:08:54,622
So we give those

108
00:08:54,702 --> 00:08:58,644
as much flexibility as we possibly can in these.

109
00:08:58,644 --> 00:09:00,155
But yeah, we've made a lot of progress.

110
00:09:00,155 --> 00:09:06,528
We started the company five years ago, now, so we still operate 300 megawatts of Bitcoin
data centers.

111
00:09:06,668 --> 00:09:15,868
But now we have 700 megawatts of critical IT load under construction, 16 billion of
contracted revenue between two customers over 15 years.

112
00:09:15,868 --> 00:09:18,568
So it's been an eventful year.

113
00:09:18,568 --> 00:09:24,528
mean, it's been a lot of work and a lot of stress, but it's been a really good, rewarding
year so far.

114
00:09:24,788 --> 00:09:27,668
But the year's not finished yet.

115
00:09:27,668 --> 00:09:32,008
So let's see how it finishes out and see what else we can do.

116
00:09:32,906 --> 00:09:35,830
What I mean, that's incredible and five years.

117
00:09:35,830 --> 00:09:38,453
What is your signature advantage?

118
00:09:38,453 --> 00:09:40,295
How did you how did you do this?

119
00:09:40,295 --> 00:09:41,696
Give me a secret

120
00:09:41,901 --> 00:09:42,609
You

121
00:09:42,829 --> 00:09:53,389
There's no real secret other than we were early so then you know in 20 2023 I know it was
only two years ago, but again dog years, right?

122
00:09:53,389 --> 00:10:06,229
And so when when I saw what was happening and I saw the demand level and at that time the
real focus was Getting could you procure H100s from Nvidia, right?

123
00:10:06,229 --> 00:10:12,782
There was no ability to procure these there was such so if you place orders you weren't
getting delivery for six plus months

124
00:10:12,782 --> 00:10:25,531
uh And so everyone was very focused on procuring the GPUs and me with my history and
background of as a hardware investor, I thought to myself that, uh you know, they're going

125
00:10:25,531 --> 00:10:28,462
to fix wafer throughput at TSMC.

126
00:10:28,462 --> 00:10:32,988
They're going to fix the co-op packaging, the advanced packaging for the semiconductors.

127
00:10:32,988 --> 00:10:36,748
They're going to fix bottlenecks in transceiver manufacturing.

128
00:10:36,748 --> 00:10:42,722
And so like they'll fix all of this, but then what will happen is you'll the GPU supply
chain itself.

129
00:10:42,722 --> 00:10:45,824
will balance out with the demand.

130
00:10:45,824 --> 00:10:51,888
But I was looking at the data center space and I'm going, where are we going to put all of
these GPUs?

131
00:10:51,888 --> 00:10:59,313
And eventually, I think I tweeted about this in the summer of 23, it said data center
space is going to be the issue.

132
00:10:59,313 --> 00:11:00,944
Eventually, that's going to be the issue.

133
00:11:00,944 --> 00:11:05,217
And I thought it might be all the way to 25, it could happen in 24.

134
00:11:05,217 --> 00:11:12,622
But there was so little of this high power density data center capacity in the market
because the vast majority of data center capacity was built for streaming.

135
00:11:12,622 --> 00:11:21,942
and all of our video apps and pictures and messaging and all of those are really CPU based
data centers.

136
00:11:22,060 --> 00:11:31,860
and they only need like I think the average density power density is like seven kilowatts
a rack and then if you want to be efficient with H100 you need to be at 50 and then when

137
00:11:31,860 --> 00:11:40,420
we look at Blackwell now where we're deploying it's at 135 kw a rack and so the power
densities went up and so now you need different style of cooling because you like air

138
00:11:40,420 --> 00:11:51,020
cooling works to about 50 kw and then you need liquid cooling after that and so I was just
working through this in my mind and going how are we going to get power how are we going

139
00:11:51,020 --> 00:11:52,327
to get permits

140
00:11:52,327 --> 00:11:55,870
and how we're going to build and deal with the supply chain.

141
00:11:55,870 --> 00:12:00,574
And I'm going the semiconductor supply chain knows how to ramp up really quickly.

142
00:12:00,574 --> 00:12:02,735
How do you deal with all of that?

143
00:12:02,735 --> 00:12:10,351
And so we did that early and we started early and then we went and built a team really
early before the demand kind of is what it is.

144
00:12:10,351 --> 00:12:15,965
And so we put a great team together and we stubbed our toe in how we did things.

145
00:12:15,965 --> 00:12:22,370
So we kind of started with, we're going to retrofit Bitcoin facilities and then we
abandoned that pretty quick.

146
00:12:22,711 --> 00:12:31,398
And then we were going to build a scaled down version and this was before the chat GPT had
showed up and so we kind of been through iterations and by the time we went to large scale

147
00:12:31,398 --> 00:12:43,407
we said look we have this design we worked with the leader for GPUs in the industry the
real leader in AI innovation from a hardware perspective and so we designed to that.

148
00:12:43,660 --> 00:12:45,880
And then we went out and secured supply chain around that.

149
00:12:45,880 --> 00:12:57,440
And so now we're sitting, I thought, you know, 18 months ago, I was super optimistic
securing, you know, six or 700 megawatts per year of MEP, mechanical, electrical and

150
00:12:57,440 --> 00:12:59,260
plumbing for our supply chain.

151
00:12:59,260 --> 00:13:01,600
And now I don't think I was optimistic enough.

152
00:13:01,600 --> 00:13:10,260
But we went and secured these so that we can actually build, we can deliver for our
customers because there's a lot of news and there's a lot of big contracts and a lot of

153
00:13:10,260 --> 00:13:10,740
announcements.

154
00:13:10,740 --> 00:13:13,654
And I've been saying lately, you know, it's really

155
00:13:13,654 --> 00:13:18,027
fun to make big announcements and signing contracts, but you have to deliver on those
contracts.

156
00:13:18,027 --> 00:13:19,678
Like that's the next step.

157
00:13:19,678 --> 00:13:21,259
And so we're actually delivering.

158
00:13:21,259 --> 00:13:31,706
We turned over the first half of our building on schedule, turned over the second half of
our building ahead of schedule, and we're working to really deliver uh that second

159
00:13:31,706 --> 00:13:37,250
building ahead of schedule as well because our customers are really needing the space.

160
00:13:37,250 --> 00:13:39,892
So we're trying to pull that ahead and I think we will pull that ahead.

161
00:13:39,892 --> 00:13:43,487
But the key here is, you know, not only signing the contracts,

162
00:13:43,487 --> 00:13:51,967
making the design, actually building and delivering a product on time and for the customer
so that our customers could be successful as well.

163
00:13:52,187 --> 00:14:02,407
So the real advantage here is we started early, we've made some mistakes, we've ironed a
lot of those out and we have a really streamlined process now for supply chain, for

164
00:14:02,407 --> 00:14:05,467
building, we have a lot of power from a site perspective.

165
00:14:05,467 --> 00:14:07,727
So we're just in a really good position.

166
00:14:08,436 --> 00:14:14,376
So there's another bottleneck in the data center world, and that is people and workforce.

167
00:14:14,376 --> 00:14:18,378
So is that affecting you or how is that affecting you?

168
00:14:18,378 --> 00:14:20,177
Yeah, so like I said with our.

169
00:14:20,192 --> 00:14:26,295
With our team that manages construction and does design and does operations, we did that
early.

170
00:14:26,295 --> 00:14:27,736
And so that was great.

171
00:14:27,736 --> 00:14:28,606
have all of them.

172
00:14:28,606 --> 00:14:31,717
We have really fantastic people in place for that.

173
00:14:31,717 --> 00:14:37,880
And then there's dealing with the local labor that actually builds the facility.

174
00:14:37,880 --> 00:14:44,443
And one of the things that we have done, so we've carved out this spot for ourselves in
North Dakota.

175
00:14:44,443 --> 00:14:47,485
And we looked at North Dakota and said, it's abundant power.

176
00:14:47,485 --> 00:14:50,126
They're generating eight to 10 times the amount of

177
00:14:50,126 --> 00:14:52,308
energy in the state versus what they use.

178
00:14:52,308 --> 00:15:04,516
The former governor, Doug Burgum, who's now Secretary of Interior, one of his biggest
agenda items in the state when he was governor was we want to use the power, the energy

179
00:15:04,516 --> 00:15:09,399
that the state generates for the economic benefit of the state rather than exporting it.

180
00:15:09,399 --> 00:15:10,430
Here we are, right?

181
00:15:10,430 --> 00:15:13,442
So we bring that use right into the state.

182
00:15:13,442 --> 00:15:16,284
And we started with Bitcoin and now we're doing it with AI data center.

183
00:15:16,284 --> 00:15:20,146
So you have that, you have a cold climate, which is phenomenal.

184
00:15:20,625 --> 00:15:25,186
And the cold climate allows us to run at a much higher efficiency.

185
00:15:25,186 --> 00:15:29,006
a PUE, so the power that we use goes mostly to the IT.

186
00:15:29,006 --> 00:15:39,726
we published a white paper back in June because of the location with the abundant low cost
energy, because of the weather, we actually can save our customers versus Southern states

187
00:15:39,726 --> 00:15:45,126
on 100 megawatts, roughly 50 to $65 million per year in operating costs.

188
00:15:45,126 --> 00:15:46,386
That's savings to our customers.

189
00:15:46,386 --> 00:15:48,686
That's a huge benefit over time.

190
00:15:48,686 --> 00:15:49,989
And so we're signing 15 years.

191
00:15:49,989 --> 00:15:59,066
contracts so you can just do the math and even you know that's on 100 megawatts 750
million dollars on the low end that you're saving over that over that contract length so

192
00:15:59,066 --> 00:16:08,223
big big deal there so and then a lot of fiber in the state from a connectivity standpoint
because you have a lot of cross-country routes that run through North Dakota but you're

193
00:16:08,223 --> 00:16:20,242
back to the point of the question uh we're creating a data center region there and we're
the only ones doing it so we have a really good handle on the local labor force for

194
00:16:20,623 --> 00:16:30,329
and even better, we're not competing with 20 other data center sites like we would be if
we were operating in Texas or if we were operating in other states that are much bigger

195
00:16:30,329 --> 00:16:31,790
data center states.

196
00:16:31,790 --> 00:16:33,361
And we're creating the region ourselves.

197
00:16:33,361 --> 00:16:39,495
We're creating the data center region in North Dakota and now we have two hyperscalers in
North Dakota.

198
00:16:39,495 --> 00:16:41,279
I think we'll add more to North Dakota.

199
00:16:41,279 --> 00:16:49,812
We'll add some to South Dakota as well and just create that region, but have that labor
force that's more captive to apply digital instead of us.

200
00:16:49,812 --> 00:16:55,966
plus an electrician or a plumber can just drive 20 or 30 miles down the road and get a 10
% raise.

201
00:16:55,966 --> 00:17:03,701
Now, we do spend a lot of time, you'll laugh at this, but we treat the construction people
on site really well.

202
00:17:03,701 --> 00:17:15,359
I was having a conversation two weeks ago with Brad Barton, who leads construction for us
about the catering and curating menus so that the food is really good on site.

203
00:17:15,359 --> 00:17:19,802
so we do these things and I'm like, there's a lot of things on my bingo card.

204
00:17:19,802 --> 00:17:29,550
that I would have never thought that I would be doing, know, going alongside of building
data centers and curating menus and what the catering looks like on site was one of those

205
00:17:29,550 --> 00:17:30,430
things.

206
00:17:30,591 --> 00:17:38,357
But we do, you know, we want to be a great place to work when you're on site so that we
retain the labor that's there and is already done.

207
00:17:38,357 --> 00:17:43,261
If they're doing building one, we want them for building two and building three and
building four, so it's just easier to retain that labor.

208
00:17:43,261 --> 00:17:46,904
But it is, in a lot of areas, it's a huge issue, huge issue.

209
00:17:46,904 --> 00:17:47,684
oh

210
00:17:48,070 --> 00:17:49,791
Yes, I agree.

211
00:17:50,792 --> 00:18:01,980
Performance drives culture and obviously everything you're doing is in alignment with
that, including feeding the construction people.

212
00:18:01,980 --> 00:18:07,424
Tell me a little bit about your culture and you are scaling fast.

213
00:18:07,424 --> 00:18:15,084
What does it look like and what does leadership look like inside of this rapidly scaling
environment?

214
00:18:15,084 --> 00:18:29,931
Yeah, so the day I started the company, my number one goal from a culture perspective was
I wanted to create a company where people were excited on Sunday night.

215
00:18:29,931 --> 00:18:30,421
Right?

216
00:18:30,421 --> 00:18:32,212
That was number one for me.

217
00:18:32,212 --> 00:18:38,945
I want people to be excited on Sunday night and I want them to, you know, feel like we're
making a difference.

218
00:18:38,945 --> 00:18:45,048
They get to see their impact inside the company and we're a small, you know, we've got to
be much bigger, but we're still, my view, a small

219
00:18:45,048 --> 00:18:52,003
company, we very much are to create an underdog mentality, ah which is, you know, it's
always great to have that.

220
00:18:52,003 --> 00:18:53,203
I'm a big sports person.

221
00:18:53,203 --> 00:18:59,607
you know, I always talk to people like I feel like I'm now the head coach of the football
team, right?

222
00:18:59,728 --> 00:19:03,911
And so it's, but we have an underdog mentality.

223
00:19:03,911 --> 00:19:08,994
And then I would just say that the culture outside of that is we just get stuff done,
right?

224
00:19:08,994 --> 00:19:09,965
We make it happen.

225
00:19:09,965 --> 00:19:12,076
I have a lot of problem solvers.

226
00:19:12,672 --> 00:19:16,974
inside the org, we've hired just the right people.

227
00:19:17,114 --> 00:19:20,636
We've been growing fast, so you don't always hire the right ones.

228
00:19:20,636 --> 00:19:22,837
We've had some mistakes there for sure.

229
00:19:22,837 --> 00:19:30,521
But we hire the right people that can get the job done, put them in the place, in the
seat, and just let them run.

230
00:19:30,521 --> 00:19:35,833
And so I try to remove those roadblocks for people to let them do what they do best.

231
00:19:35,833 --> 00:19:40,175
And we hired Todd Gale, who runs Data Center Design for us.

232
00:19:40,175 --> 00:19:42,626
That was a huge hire, and he's just been an absolute

233
00:19:43,320 --> 00:19:45,161
home run for the company.

234
00:19:45,161 --> 00:19:49,384
Brad was at DPR, was building data centers for hyperscalers.

235
00:19:49,585 --> 00:19:56,109
On the construction side, we brought in Laura Latrello, who came to us from Honeywell, was
at Lenovo.

236
00:19:56,150 --> 00:19:58,492
Prior to that, is doing operations for us.

237
00:19:58,492 --> 00:19:59,262
She's been great.

238
00:19:59,262 --> 00:20:07,419
Then my co-founder, uh Jason, has just really come up to speed and he's been, he's a fast
learner.

239
00:20:07,419 --> 00:20:09,780
He went to Harvard, so he's smart guy.

240
00:20:09,820 --> 00:20:11,821
But he's been really

241
00:20:11,982 --> 00:20:14,013
instrumental in getting the company to where it is.

242
00:20:14,013 --> 00:20:22,349
And there's just, there's so many people I can't list, but getting the right people in the
right seats and then removing roadblocks from them and everyone's excited.

243
00:20:22,349 --> 00:20:27,173
uh You know, we're building some of the largest scale projects.

244
00:20:27,173 --> 00:20:35,609
We're building the largest scale project in the history of North Dakota for sure, but some
of the largest scale projects in the country and in the world, and it just creates a lot

245
00:20:35,609 --> 00:20:37,480
of excitement inside the company.

246
00:20:37,898 --> 00:20:42,104
and a lot of excitement throughout the entire country and world.

247
00:20:42,104 --> 00:20:44,266
And that's what we're hearing right now.

248
00:20:44,427 --> 00:20:50,914
So what do you see in the way of trends where the industry is going?

249
00:20:51,596 --> 00:20:56,441
What are we underestimating, if anything, about where the space is headed?

250
00:20:56,566 --> 00:21:00,186
So I think one of the biggest kind of...

251
00:21:00,212 --> 00:21:11,234
pieces of information that I don't think is correct that I think flows around in the
market is still this thought process around we're going to build some training sites in

252
00:21:11,234 --> 00:21:21,432
more remote locations and then all the inference is going to be built out in population
centers like traditional data center and when we're working with our customers and when

253
00:21:21,432 --> 00:21:30,196
we're working directly with the largest frontier models uh in the world, you know, we're
hearing the exact opposite, you know, our

254
00:21:30,196 --> 00:21:33,508
is a dual use training and inference site.

255
00:21:33,508 --> 00:21:42,524
Our next site is also dual use and we expect them all to be dual use training and
inference uh because we have to drive cost down in the industry.

256
00:21:42,524 --> 00:21:50,300
And so I do think there will be some inferencing that's done, you know, like CDN style
content delivery network style in metro areas.

257
00:21:50,300 --> 00:22:00,546
But I think a vast majority is going to be in sites like we're building and other people
are building because when you think about uh efficiency on a site and driving cost down,

258
00:22:00,779 --> 00:22:04,719
what's the most expensive part of this entire ecosystem?

259
00:22:04,719 --> 00:22:06,419
It's the hardware itself, right?

260
00:22:06,419 --> 00:22:13,279
So the Nvidia GPUs, they could be AMD, but we're mostly Nvidia GPUs, what we focus on.

261
00:22:13,319 --> 00:22:15,399
So that's the most expensive part.

262
00:22:15,399 --> 00:22:21,979
And so that means you need to get as close to 100 % utilization as you possibly can out of
those GPUs.

263
00:22:21,979 --> 00:22:27,159
So you want them in a location where if they're not being used for training, they can be
used for inference.

264
00:22:27,159 --> 00:22:29,299
And if they're not being used for inference, it can be used for training.

265
00:22:29,299 --> 00:22:30,122
And so

266
00:22:30,122 --> 00:22:43,248
you don't want a location where you have tens of thousands of GPUs sitting around and the
entire country is asleep and so the usage goes from 95 % down to 5 % and you don't have

267
00:22:43,248 --> 00:22:45,990
anything else to do with those and that's 10 hours out of the day.

268
00:22:45,990 --> 00:22:51,052
uh It drives your cost up way too high and so...

269
00:22:51,146 --> 00:22:54,677
we see a lot more sensitivity around cost and the whole ecosystem.

270
00:22:54,677 --> 00:23:00,148
And that's just been tech, you know, from the history of tech, because you've got to get
the cost down over time.

271
00:23:00,148 --> 00:23:08,786
It's how, you know, we're sitting with laptops on our desks that are, you know,
significantly more powerful than mainframes were at a fraction of the cost.

272
00:23:08,786 --> 00:23:10,837
uh But we're doing the same here.

273
00:23:10,837 --> 00:23:12,708
But one of those key steps is

274
00:23:12,756 --> 00:23:17,598
you need a hundred percent as close to a hundred percent utilization as you can get out of
these GPUs.

275
00:23:17,598 --> 00:23:23,921
And so I think the trend is really for these large scale sites and they're not going to be
built in metro areas.

276
00:23:23,921 --> 00:23:28,723
uh And this is where most of the inferencing is coming from anyways at this point.

277
00:23:28,963 --> 00:23:34,246
So that's probably the biggest one in the trend that I expect to continue is just more and
more scale.

278
00:23:34,246 --> 00:23:40,729
Like we're talking with some customers at our sites about like, I'm a, I'm a really
optimistic guy.

279
00:23:40,729 --> 00:23:42,663
I'd have to be, I have to be an optimistic guy.

280
00:23:42,663 --> 00:23:49,148
to be crazy enough to go build a $1.2 billion building in the middle of nowhere with no
customer, right?

281
00:23:49,148 --> 00:23:55,543
And even I underestimated the scale as to what we're talking about on single locations.

282
00:23:55,543 --> 00:24:00,817
I was thinking a gigawatt is a massive amount on a single location.

283
00:24:00,817 --> 00:24:12,096
We've been having discussions around five to 10 gigawatts on a single site, which is these
are theoretical discussions about what you could do from lowering the cost of the build

284
00:24:12,096 --> 00:24:12,526
and getting

285
00:24:12,526 --> 00:24:14,008
everything more efficient.

286
00:24:14,389 --> 00:24:20,439
But we're just seeing that kind of demand and that kind of thought process on the types of
sites that we're working on.

287
00:24:21,934 --> 00:24:24,918
So what drives you, Wes?

288
00:24:24,918 --> 00:24:31,948
What makes you just jump out of bed in the morning, on a Monday morning because you're so
excited on Sunday night, right?

289
00:24:32,502 --> 00:24:44,518
Well, so generally, uh I gave that for me, it's not quite that because as being said on a
Sunday night, because work never really stops on the weekend.

290
00:24:44,518 --> 00:24:47,199
uh You can check my wife on that.

291
00:24:47,199 --> 00:24:49,300
She'll confirm.

292
00:24:49,300 --> 00:24:54,522
uh for me, uh it's a couple of things.

293
00:24:54,522 --> 00:25:00,455
you know, obviously running the business day to day is, you know, got all of the
challenges and problem solving.

294
00:25:00,455 --> 00:25:02,446
And what I learned about myself years ago is

295
00:25:02,446 --> 00:25:05,839
oh when I was doing investing is why do I like this so much?

296
00:25:05,839 --> 00:25:08,021
It's like I like puzzles, right?

297
00:25:08,021 --> 00:25:18,550
And so you're just solving a puzzle all the time and you're doing that with the company as
well, whether it be operational, whether it be construction or on design or new customers,

298
00:25:18,550 --> 00:25:18,940
right?

299
00:25:18,940 --> 00:25:21,923
It's all of these things of trying to put all the pieces together.

300
00:25:21,923 --> 00:25:23,424
And so I like that.

301
00:25:23,556 --> 00:25:26,697
But it's been exciting on almost all aspects, right?

302
00:25:26,697 --> 00:25:32,659
So we're in a place of we're both delivering and winning new contracts.

303
00:25:32,659 --> 00:25:40,162
And so I'm kind of in the best of both worlds right now where we get to be doing both
because the market's just there for it right now.

304
00:25:40,162 --> 00:25:45,304
uh But probably the most excited I get is

305
00:25:45,589 --> 00:25:49,351
Just thinking through the possibilities of the future.

306
00:25:49,351 --> 00:25:58,935
I could, I don't have enough time for it these days, but I could do hours of discussion
around a lot of the what ifs and the trends and what we have to be ready for in the

307
00:25:58,935 --> 00:25:59,505
future.

308
00:25:59,505 --> 00:26:04,847
And we think a lot about this and we try to make sure that, you know, let's talk about
future proofing our data centers.

309
00:26:04,847 --> 00:26:15,802
There's so much new technology coming in, uh in a physical format, which is for me even
better versus, you know, just from an excitement perspective versus software.

310
00:26:15,802 --> 00:26:25,929
So there's a lot of problems that we're solving and there's going to be more problems, but
we're spending a lot of time on load balancing on site with batteries, with super

311
00:26:25,929 --> 00:26:26,870
capacitors.

312
00:26:26,870 --> 00:26:32,173
uh We're spending a lot of time on cooling, on power distribution.

313
00:26:32,173 --> 00:26:40,859
So I'm spending a fair amount of time on looking at compound semiconductor solutions for
power distribution inside of the data center for more efficiency.

314
00:26:40,859 --> 00:26:45,422
It's kind of similar to how uh EVs went from silicon to

315
00:26:45,422 --> 00:26:54,391
silicon carbide, I think we'll see something similar in the data center because we need to
go from 415 volts up to 800 volt distribution on the next generation.

316
00:26:55,293 --> 00:27:01,760
the new technology, because I've been a technology investor for a long time, that's always
probably the most exciting part for me.

317
00:27:02,218 --> 00:27:03,942
So how do you get your information?

318
00:27:03,942 --> 00:27:10,927
Do you have time where you just sit and read and subscribe to newsletters and talk to
people?

319
00:27:10,927 --> 00:27:13,561
mean, how are you seeing the future?

320
00:27:14,348 --> 00:27:17,530
ah Just to do a shameless plug here, chat GPT.

321
00:27:17,530 --> 00:27:30,419
uh I do ask chat a lot of things, um but I read constantly on new products that come out.

322
00:27:30,419 --> 00:27:35,521
I love listening to the...

323
00:27:35,657 --> 00:27:47,592
or the interviews or panels that engineers sit on that are in the industry, whether it be
specifically data center engineers or the data scientists that are working for some of

324
00:27:47,592 --> 00:27:49,523
these frontier model companies.

325
00:27:49,703 --> 00:27:59,968
I love listening to the way they're thinking through solving the future, whether it's with
large language models or with humanoids or whatever it might be.

326
00:28:00,088 --> 00:28:05,110
But I love listening to a lot of those interviews because reading articles is great,

327
00:28:05,190 --> 00:28:15,326
but hearing from also very smart people that have digested a lot of information or sitting
on the front lines, that's one of my favorite places to get information.

328
00:28:15,758 --> 00:28:17,558
So it's 2035.

329
00:28:17,878 --> 00:28:20,218
Tell me about Applied Digital.

330
00:28:21,196 --> 00:28:28,244
uh So, 2035 again, reminder dog years for AI.

331
00:28:28,546 --> 00:28:33,251
But for us, when I look out to the future, you know, we've...

332
00:28:33,728 --> 00:28:44,493
We're executing and we're in discussions with almost every single hyperscaler and advanced
discussions in some cases for new campuses.

333
00:28:44,493 --> 00:28:49,385
And when I look out to the future, it was a big breakthrough when we landed Core Weave,
right?

334
00:28:49,385 --> 00:28:50,526
That's our first contract.

335
00:28:50,526 --> 00:28:51,676
That was a big relief for me.

336
00:28:51,676 --> 00:28:55,938
Again, that big risk on that building uh in Ellendale.

337
00:28:56,038 --> 00:28:57,389
And then we have our second one.

338
00:28:57,389 --> 00:28:59,260
We're just gaining momentum.

339
00:28:59,300 --> 00:29:02,701
And we've been qualified with basically every hyperscaler.

340
00:29:04,123 --> 00:29:13,092
different conversations now than I did a year ago where a year ago if I sat down with a
hyperscaler it would you know be 15 or 20 minutes of explaining who we are and what we do.

341
00:29:13,092 --> 00:29:21,571
Now we're known so we just get right down to what you know here's what's available and
what we can build for you and then you know and then the price talk always always shows

342
00:29:21,571 --> 00:29:21,801
up.

343
00:29:21,801 --> 00:29:25,124
uh

344
00:29:25,420 --> 00:29:31,540
We have the two campuses, we'll get, I think we'll get a third and a fourth and a fifth.

345
00:29:31,540 --> 00:29:37,740
And if I have five of these campuses amongst different customers, they all go to over a
gigawatt.

346
00:29:37,740 --> 00:29:47,860
So if I'm looking at what can we actually build, and I look through that pipeline through
2030 or 2031, we could build that out, right?

347
00:29:47,860 --> 00:29:54,980
We could build out five plus gigawatts of capacity for these tenants and they're all great
tenants.

348
00:29:55,327 --> 00:30:06,726
Stability contracted revenue so we get great visibility as as as much as I love the you
the building with with no customer Building with a customer already is a lot more fun.

349
00:30:06,726 --> 00:30:20,046
I found so the the uh You know just executing and then and then continuing on the forward
thinking from an infrastructure perspective and we've already started down the path of uh

350
00:30:20,046 --> 00:30:25,290
power generation and bringing more power to our locations uh over the next few years and

351
00:30:25,290 --> 00:30:26,791
that'll just continue.

352
00:30:26,791 --> 00:30:31,334
And so we're just, we have kind of our hands in every part of the ecosystem.

353
00:30:31,334 --> 00:30:42,001
We stay away from being in like an actual language model or doing any data scientists
ourselves, but we try to be the best at what we do and as forward looking as we can be.

354
00:30:42,001 --> 00:30:54,730
And we're seeing that really start to pay dividends for us as far as the interactions with
current customers, potential customers, and kind of the scale that now that I think our

355
00:30:54,730 --> 00:30:55,150
company

356
00:30:55,150 --> 00:31:02,770
can be by 2035, it's just wildly bigger than I thought it could be a year ago.

357
00:31:02,770 --> 00:31:09,969
When you're trying to do that breakthrough for just, I want to sell this 100 megawatt
building to someone, now my mindset is just very different.

358
00:31:09,969 --> 00:31:12,012
I really think that...

359
00:31:12,012 --> 00:31:21,029
we can be the leader in the industry and one of the largest companies, if not the largest
in the industry, just given the momentum that we have and what I see on the demand side

360
00:31:21,029 --> 00:31:29,295
for us and how good I feel about the team we have in place and the ability to uh on that
roadmap.

361
00:31:29,295 --> 00:31:37,672
It's one thing about just dreaming up a roadmap, but I feel really confident in our
ability to actually execute on a really big roadmap like that.

362
00:31:37,838 --> 00:31:41,798
And also don't forget Wes, your passion and your purpose.

363
00:31:41,798 --> 00:31:43,158
It's inspiring, right?

364
00:31:43,158 --> 00:31:45,418
That's gonna be a huge driver of where you're going.

365
00:31:45,598 --> 00:31:47,898
thank you so much for coming on the show.

366
00:31:47,898 --> 00:31:49,538
This has just been phenomenal.

367
00:31:49,538 --> 00:31:50,858
Congratulations on your success.

368
00:31:50,858 --> 00:31:55,552
I cannot wait to see where you're gonna be in 2035 and beyond.

369
00:31:55,552 --> 00:31:56,164
Thanks, Carrie.

370
00:31:56,164 --> 00:31:57,897
I really appreciate you having me.

371
00:31:57,899 --> 00:31:59,823
You too.

372
00:32:01,343 --> 00:32:03,359
All right, that was awesome.