1
00:00:03,160 --> 00:00:17,680
Speaker 1: Media. Hi, I'm ed Zetran and welcome to Better Offline.

2
00:00:19,800 --> 00:00:21,959
Speaker 2: That's right, folks, were back for the third part of

3
00:00:22,000 --> 00:00:24,319
Speaker 2: our four part series about the terminal stage of the

4
00:00:24,320 --> 00:00:29,680
Speaker 2: incitification process, the in shitification of the financial markets. So far,

5
00:00:29,720 --> 00:00:32,640
Speaker 2: we've talked about the incertification of the stock market, the

6
00:00:32,640 --> 00:00:35,400
Speaker 2: complicity of the analysts and the media, and how venture

7
00:00:35,400 --> 00:00:37,400
Speaker 2: capital will be the Canarian the coal mine for any

8
00:00:37,440 --> 00:00:40,960
Speaker 2: eventual collapse. Venture capital is both the perpetrator and the

9
00:00:41,040 --> 00:00:43,599
Speaker 2: victim of the air bubble, and it's worth exploring how

10
00:00:43,600 --> 00:00:47,000
Speaker 2: it came to this point. Now, hang on the gentle voice, folks.

11
00:00:47,000 --> 00:00:49,239
Speaker 2: We haven't seen values this big and alone time. These

12
00:00:49,280 --> 00:00:52,120
Speaker 2: the biggest numbers we've ever seen. They're simply tremendous open

13
00:00:52,159 --> 00:00:55,120
Speaker 2: airs maybe with eight hundred and thirty billion. Can you

14
00:00:55,160 --> 00:00:57,360
Speaker 2: believe that they lose so much money? But folks, we

15
00:00:57,360 --> 00:01:00,640
Speaker 2: don't mind. Clamisam Morbourn. They call him Klamiel because everybody

16
00:01:00,680 --> 00:01:02,440
Speaker 2: he's going to give them one billion dollars and where

17
00:01:02,480 --> 00:01:04,280
Speaker 2: we love him dead. Centers are going to have the

18
00:01:04,319 --> 00:01:07,319
Speaker 2: biggest deals we've ever seen, even even if we have

19
00:01:07,400 --> 00:01:10,240
Speaker 2: to work with Dario. Sorry to those of you who

20
00:01:10,280 --> 00:01:12,720
Speaker 2: don't like that voice. It was only about thirty seconds,

21
00:01:13,160 --> 00:01:16,280
Speaker 2: you see right Now, AI startups are big exciting news

22
00:01:16,280 --> 00:01:19,759
Speaker 2: for the limited partners funding large language model companies. Things

23
00:01:19,800 --> 00:01:22,360
Speaker 2: feel exciting because the value of the assets under management

24
00:01:22,560 --> 00:01:26,280
Speaker 2: aum are going up, which is nothing dodgy, per se,

25
00:01:26,319 --> 00:01:28,679
Speaker 2: but it's just how vcs value things. And if they're

26
00:01:28,760 --> 00:01:31,839
Speaker 2: valuing AI stocks, well that's how their fees are getting paid,

27
00:01:31,840 --> 00:01:34,480
Speaker 2: because the thing is they've invested in are worth more money,

28
00:01:34,480 --> 00:01:36,880
Speaker 2: and now limited partners are paying them a percentage so

29
00:01:36,920 --> 00:01:38,679
Speaker 2: that they can sit in an apartment that costs too

30
00:01:38,760 --> 00:01:42,000
Speaker 2: much not see friends because they just sit online posting

31
00:01:42,040 --> 00:01:44,480
Speaker 2: for eleven hours a day. Yeah, and then they can

32
00:01:44,560 --> 00:01:47,160
Speaker 2: lose all the money in the future. I guess investing

33
00:01:47,160 --> 00:01:49,520
Speaker 2: early in open Ai allows a bench capitalist or even

34
00:01:49,560 --> 00:01:52,280
Speaker 2: an asset manager like Blackstone, which invested in twenty twenty

35
00:01:52,280 --> 00:01:54,480
Speaker 2: four to say it as a big holding and a

36
00:01:54,480 --> 00:01:58,720
Speaker 2: big increase in its aum AI stocks, make venture capitalists

37
00:01:58,720 --> 00:02:00,840
Speaker 2: who bet on them two years ago, Like genius is

38
00:02:00,880 --> 00:02:05,680
Speaker 2: on paper, you got in early on open Ai, Athropic, Cursor, Cognition, Perplexity,

39
00:02:05,920 --> 00:02:08,360
Speaker 2: or any other company that loves to burn several dollars

40
00:02:08,360 --> 00:02:11,040
Speaker 2: per dollar of revenue. You have a big, beautiful number,

41
00:02:11,080 --> 00:02:13,440
Speaker 2: the biggest you've ever seen, and your limited partners to

42
00:02:13,480 --> 00:02:16,400
Speaker 2: pay you a feed just to manage it. Venture capital

43
00:02:16,440 --> 00:02:18,919
Speaker 2: has not seen valuations like this in a long time,

44
00:02:19,120 --> 00:02:21,519
Speaker 2: and on paper it feels a lot like a lot

45
00:02:21,520 --> 00:02:24,120
Speaker 2: of vcs got in on companies worth billions of dollars

46
00:02:24,240 --> 00:02:28,160
Speaker 2: right on paper, that is, on paper, Cognition is worth

47
00:02:28,200 --> 00:02:32,239
Speaker 2: ten point two billion, Perplexit eighteen billion, Cursor twenty nine

48
00:02:32,280 --> 00:02:36,079
Speaker 2: point three billion, Lovable six point six billion, Coher six

49
00:02:36,080 --> 00:02:39,720
Speaker 2: point eight billion, Repic three billion, Glean seven point two billion.

50
00:02:40,440 --> 00:02:43,320
Speaker 2: Massive valuations to companies that all basically do products that

51
00:02:43,360 --> 00:02:45,800
Speaker 2: open AI or Anthropic or Amazon or Google or any

52
00:02:45,880 --> 00:02:48,760
Speaker 2: number of Chinese companies are already working to clone. They're

53
00:02:48,760 --> 00:02:50,799
Speaker 2: all losing lots of money too, and have no path

54
00:02:50,880 --> 00:02:54,760
Speaker 2: the profitability of any kind. But right now the numbers

55
00:02:54,800 --> 00:02:58,160
Speaker 2: are simply tremendous. Folks. I've heard venture capitalists tell me

56
00:02:58,160 --> 00:03:00,320
Speaker 2: that there are times when they have to agree to

57
00:03:00,360 --> 00:03:03,280
Speaker 2: invest with little to no information or know that they'll

58
00:03:03,280 --> 00:03:08,239
Speaker 2: lose the opportunity to another sucker, I mean, investor. I've

59
00:03:08,280 --> 00:03:11,480
Speaker 2: heard venture capitalists say they don't have any insight into finances,

60
00:03:11,639 --> 00:03:14,600
Speaker 2: but we love the numbers. Folks, they're simply tremendous. We

61
00:03:14,760 --> 00:03:17,520
Speaker 2: love it. Let me tell you how not tremendous they are.

62
00:03:17,960 --> 00:03:20,959
Speaker 2: I had a source recently suggests that nobody who invested

63
00:03:21,160 --> 00:03:24,640
Speaker 2: pasted open AI's one hundred billion dollar valuation have any

64
00:03:24,680 --> 00:03:28,400
Speaker 2: information rates. And what that means is they literally do

65
00:03:28,440 --> 00:03:31,280
Speaker 2: not know anything about the company. They don't know anything

66
00:03:31,320 --> 00:03:32,800
Speaker 2: other than what the company will tell them. That the

67
00:03:32,800 --> 00:03:35,440
Speaker 2: company doesn't have to tell them shit, which means they

68
00:03:35,440 --> 00:03:37,680
Speaker 2: don't know the revenues, they don't know the costs. They

69
00:03:37,760 --> 00:03:42,280
Speaker 2: only know what new start new ridiculous stat Sarah Friar,

70
00:03:42,400 --> 00:03:45,680
Speaker 2: CFO of open ais, farted out. They posted recently that

71
00:03:45,760 --> 00:03:49,440
Speaker 2: twenty billion dollars annualized revenue. Yeah. I think that's complete nonsense,

72
00:03:49,480 --> 00:03:51,360
Speaker 2: But they want to do that so that you go

73
00:03:51,440 --> 00:03:54,000
Speaker 2: open ai made twenty billion dollars in twenty twenty five.

74
00:03:54,480 --> 00:03:57,440
Speaker 2: Fuck that, I don't even think they made ten. Now,

75
00:03:57,520 --> 00:04:01,400
Speaker 2: venture capitalists would argue, of course that they'd say I'm insane.

76
00:04:01,200 --> 00:04:04,440
Speaker 2: They would say they would think me mad, and they

77
00:04:04,480 --> 00:04:07,240
Speaker 2: would say that the growth is obviously there, while pointing

78
00:04:07,240 --> 00:04:09,760
Speaker 2: to whatever startup has made one hundred million dollars in

79
00:04:09,880 --> 00:04:13,560
Speaker 2: annualized recurring revenue, which is eight point three million dollars

80
00:04:13,600 --> 00:04:16,560
Speaker 2: a month, by the way, all while not discussing the

81
00:04:16,640 --> 00:04:19,960
Speaker 2: underlying operating expenses. The idea I believe is that the

82
00:04:19,960 --> 00:04:22,839
Speaker 2: current state of AI spending is only set to increase

83
00:04:22,880 --> 00:04:27,479
Speaker 2: the next year, and that will somehow lead to fixing

84
00:04:27,520 --> 00:04:32,120
Speaker 2: the margins. I think venture capitalists staunchly refuse to learn

85
00:04:32,160 --> 00:04:34,760
Speaker 2: anything other than investing growth and then profit from growth,

86
00:04:34,800 --> 00:04:38,520
Speaker 2: even if profiting from growth doesn't appear to be happening anymore.

87
00:04:39,120 --> 00:04:41,760
Speaker 2: In reality, venture capital shouldn't have touched LMS with a

88
00:04:41,800 --> 00:04:45,240
Speaker 2: shitty stick because the margins were obviously blatantly bad from

89
00:04:45,279 --> 00:04:48,480
Speaker 2: the very beginning. We knew open ai would lose five

90
00:04:48,480 --> 00:04:51,200
Speaker 2: million dollars in the middle of twenty twenty four. The

91
00:04:51,240 --> 00:04:54,159
Speaker 2: same venture capital climate would have fucking panicked, but instead

92
00:04:54,240 --> 00:04:57,480
Speaker 2: chose to double dribble and kadruple down. I believe that

93
00:04:57,520 --> 00:05:01,040
Speaker 2: massive valuation drawdowns are a certainty for AI companies, and

94
00:05:01,080 --> 00:05:03,800
Speaker 2: by that I mean these companies will end up being

95
00:05:03,839 --> 00:05:06,479
Speaker 2: revalued and the number will be smaller than the big

96
00:05:06,560 --> 00:05:12,159
Speaker 2: number everybody fell in love with. Look, venture capitalists, I

97
00:05:12,200 --> 00:05:15,320
Speaker 2: gotta ask you a question, what happens if open AI dies?

98
00:05:15,680 --> 00:05:18,400
Speaker 2: Do you think this will make investors interested in funding

99
00:05:18,480 --> 00:05:21,440
Speaker 2: or acquiring other AI startups. How much longer are we

100
00:05:21,480 --> 00:05:24,120
Speaker 2: going to do this? When will Venture Capital realize it's

101
00:05:24,120 --> 00:05:27,279
Speaker 2: setting itself up for a disaster And what exactly is

102
00:05:27,320 --> 00:05:31,360
Speaker 2: the plan? Open AI and anthropic will cuck the lake's

103
00:05:31,440 --> 00:05:34,640
Speaker 2: dry like an Nvidia GPU named after Nancy Reagan? How

104
00:05:34,720 --> 00:05:36,560
Speaker 2: is this meant to continue and what will be left

105
00:05:36,600 --> 00:05:39,719
Speaker 2: when it does? The answer is simple, there won't be

106
00:05:39,800 --> 00:05:42,760
Speaker 2: any money for venture capital for a while. Those AI

107
00:05:42,839 --> 00:05:45,000
Speaker 2: holdings are going to be worth at best fifty percent

108
00:05:45,200 --> 00:05:48,400
Speaker 2: if they retain any value at all. Once one of

109
00:05:48,400 --> 00:05:51,440
Speaker 2: these startups die, Once there's a major casualty, a panic

110
00:05:51,480 --> 00:05:54,200
Speaker 2: will ensue, sending venture capitalists scrambling to get their holdings

111
00:05:54,240 --> 00:05:57,479
Speaker 2: acquired by which I mean sold to someone else, until

112
00:05:57,480 --> 00:06:00,960
Speaker 2: there's little or no investor interest left. Why would limited

113
00:06:01,000 --> 00:06:04,279
Speaker 2: partners ever trust venture capital after this? Why would anybody?

114
00:06:04,600 --> 00:06:06,760
Speaker 2: Because based on the past four years, it doesn't appear

115
00:06:06,760 --> 00:06:09,279
Speaker 2: that venture capital is actually good at investing money. It

116
00:06:09,360 --> 00:06:11,520
Speaker 2: just got lucky year after year until there were a

117
00:06:11,520 --> 00:06:13,600
Speaker 2: few ideas that they could sell for one hundreds of

118
00:06:13,600 --> 00:06:17,760
Speaker 2: billions or billions of dollars, And now we're out of them,

119
00:06:18,080 --> 00:06:20,880
Speaker 2: we're out. I think we're out. I think there's probably

120
00:06:21,000 --> 00:06:23,800
Speaker 2: still some, but I think the era of the unicorn

121
00:06:23,880 --> 00:06:27,000
Speaker 2: might be dead. For a while, venture capital believed it

122
00:06:27,040 --> 00:06:30,039
Speaker 2: knew better as it turned its back on basic business fundamentals,

123
00:06:30,040 --> 00:06:34,360
Speaker 2: starting with Clubhouse, Crypto, the Metaverse, and now Generative AI.

124
00:06:34,760 --> 00:06:37,000
Speaker 2: Yet they're far from the only fuck wits on the

125
00:06:37,000 --> 00:06:40,800
Speaker 2: Dickhead Express. Because yes, I am going to talk about

126
00:06:40,920 --> 00:06:44,599
Speaker 2: data centers. Per Bloomberg, there were at least one hundred

127
00:06:44,600 --> 00:06:46,920
Speaker 2: and seventy eight point five billion dollars in data center

128
00:06:46,960 --> 00:06:49,120
Speaker 2: credit deals and in the US in twenty twenty five,

129
00:06:49,480 --> 00:06:51,880
Speaker 2: rivaling the two hundred and fifteen point four billion dollars

130
00:06:51,880 --> 00:06:54,400
Speaker 2: invested in US venture capital in twenty twenty four and

131
00:06:54,440 --> 00:06:56,559
Speaker 2: the one hundred and ninety seven point two billion dollars

132
00:06:56,640 --> 00:06:59,359
Speaker 2: invested in US venture capital through August seventh to twenty

133
00:06:59,360 --> 00:07:02,680
Speaker 2: twenty five, and over one hundred billion dollars more than

134
00:07:02,720 --> 00:07:05,360
Speaker 2: the sixty point sixty nine billion dollars of data center

135
00:07:05,440 --> 00:07:09,680
Speaker 2: deals done in twenty twenty four. That's that's really worrying.

136
00:07:10,560 --> 00:07:13,440
Speaker 2: It's like more than more than more than one hundred

137
00:07:13,440 --> 00:07:15,480
Speaker 2: percent more. I'm very worried, and I'm going to tell

138
00:07:15,520 --> 00:07:19,160
Speaker 2: you why, using a company called core Weave, core Weave

139
00:07:19,280 --> 00:07:22,760
Speaker 2: that I've been actively warning about, warning people about since March.

140
00:07:22,800 --> 00:07:26,280
Speaker 2: And I get a little crazy around corwaf because I

141
00:07:26,760 --> 00:07:30,360
Speaker 2: have had the core Weave conversation a lot, and I've

142
00:07:30,360 --> 00:07:33,000
Speaker 2: seen people straight up just steal the title of my

143
00:07:33,120 --> 00:07:36,160
Speaker 2: articles as they come to this realization six to seven

144
00:07:36,160 --> 00:07:39,480
Speaker 2: months later. But cor Weave is a mess, and corwy

145
00:07:39,680 --> 00:07:42,880
Speaker 2: was obviously always a mess. And everyone has been saying

146
00:07:42,880 --> 00:07:47,080
Speaker 2: Corwave's fine because they went public. Core Weave is not fine.

147
00:07:48,000 --> 00:07:52,960
Speaker 2: All right, Let's talk about core Weave. Corwave is something

148
00:07:52,960 --> 00:07:56,440
Speaker 2: called a neocloud. It's a company that builds data centers

149
00:07:56,560 --> 00:07:59,119
Speaker 2: by renting out, and they build data centers they film

150
00:07:59,160 --> 00:08:01,200
Speaker 2: full of GPUs and they rent them to people. They're

151
00:08:01,240 --> 00:08:03,760
Speaker 2: in something called AI Compume, And as I explained a

152
00:08:03,760 --> 00:08:06,240
Speaker 2: few months ago, they do so by building data centers

153
00:08:06,240 --> 00:08:10,840
Speaker 2: backed by endless debt. And I quote myself saying, up

154
00:08:10,840 --> 00:08:13,520
Speaker 2: a neocloud is expensive. Even if the company in question

155
00:08:13,560 --> 00:08:15,680
Speaker 2: already has data centers, as core We've did with its

156
00:08:15,680 --> 00:08:20,200
Speaker 2: scryptocurrency mining operation, AI requires completely new data center infrastructure

157
00:08:20,200 --> 00:08:22,880
Speaker 2: to house and call the GPUs, and those GPUs also

158
00:08:22,960 --> 00:08:26,360
Speaker 2: need paying for, and then there's the other stuff like power, water,

159
00:08:26,440 --> 00:08:28,760
Speaker 2: and the other bits of the computer, the CPU, the motherboard,

160
00:08:28,800 --> 00:08:30,280
Speaker 2: the memory of the storage, and the housing and the

161
00:08:30,320 --> 00:08:33,319
Speaker 2: power supplying all that good stuff. As a result, these

162
00:08:33,360 --> 00:08:35,680
Speaker 2: neoclouds are forced to raise billions of dollars in debt,

163
00:08:35,800 --> 00:08:39,160
Speaker 2: which they collateralize using the GPUs they already have, along

164
00:08:39,200 --> 00:08:44,079
Speaker 2: with contracts from customers which they use to buy more GPUs. Corwei,

165
00:08:44,160 --> 00:08:46,720
Speaker 2: for example, has twenty five billion dollars in debt on

166
00:08:46,760 --> 00:08:49,319
Speaker 2: an estimated five point three five billion dollars in revenue,

167
00:08:49,440 --> 00:08:52,200
Speaker 2: losing hundreds of billions of dollars in a quarter. You

168
00:08:52,280 --> 00:08:57,160
Speaker 2: know who also invests in these neoclouds in Video, the

169
00:08:57,200 --> 00:09:00,160
Speaker 2: people who make the GPUs in VideA is also one

170
00:09:00,160 --> 00:09:02,920
Speaker 2: of Corwey's largest customers, accounting for fifty percent of its

171
00:09:02,960 --> 00:09:05,079
Speaker 2: revenue in twenty twenty four, and just signed a deal

172
00:09:05,120 --> 00:09:07,480
Speaker 2: to buy six point three billion dollars of any capacity

173
00:09:07,600 --> 00:09:09,800
Speaker 2: the care we can't otherwise sell to someone else through

174
00:09:09,800 --> 00:09:12,640
Speaker 2: twenty thirty two, an extension of a one point three

175
00:09:12,720 --> 00:09:15,360
Speaker 2: billion dollar deal from twenty twenty three reported by The

176
00:09:15,360 --> 00:09:18,520
Speaker 2: Information oh And in Video was the anchor investor of

177
00:09:18,559 --> 00:09:21,640
Speaker 2: Corweav's IPO, putting in two hundred and fifty million dollars.

178
00:09:21,960 --> 00:09:25,160
Speaker 2: Corwave is also one of the largest providers AI compute

179
00:09:25,200 --> 00:09:28,040
Speaker 2: in the world, and its business model is indicative of

180
00:09:28,080 --> 00:09:31,200
Speaker 2: how most data center companies make money. In fact, i'd

181
00:09:31,280 --> 00:09:33,600
Speaker 2: argue that they are probably doing better because they have

182
00:09:33,640 --> 00:09:36,560
Speaker 2: the backing of everyone. Now you may think that's a

183
00:09:36,559 --> 00:09:38,720
Speaker 2: too big to fail thing. Not really, It just means

184
00:09:38,720 --> 00:09:43,439
Speaker 2: that they're kind of the village bicycle of AI compute grossness. Aside,

185
00:09:43,800 --> 00:09:46,760
Speaker 2: let me explain how this works. First, core we've signed

186
00:09:46,800 --> 00:09:49,360
Speaker 2: contracts such as it's fourteen billion dollar deal with Meto

187
00:09:49,440 --> 00:09:51,400
Speaker 2: or it's twenty two point five billion dollar deal with

188
00:09:51,440 --> 00:09:54,480
Speaker 2: open Ai before it has the physical infrastructure to actually

189
00:09:54,520 --> 00:09:57,800
Speaker 2: service them. It then raises debt using this contract as

190
00:09:57,800 --> 00:10:01,680
Speaker 2: collateral orders GPUs from VideA. Those take three months to

191
00:10:01,840 --> 00:10:04,600
Speaker 2: arrive and then another three months to install, at which

192
00:10:04,600 --> 00:10:07,880
Speaker 2: point monthly client payments begin. And I should also add

193
00:10:07,920 --> 00:10:10,120
Speaker 2: that that's kind of taken out of deferred revenue that

194
00:10:10,160 --> 00:10:13,839
Speaker 2: the customers already put down. To really simplify this, data

195
00:10:13,840 --> 00:10:16,800
Speaker 2: center developers are raising money months up to a year

196
00:10:16,840 --> 00:10:19,360
Speaker 2: before they ever expect to make a penny of revenue,

197
00:10:19,800 --> 00:10:23,319
Speaker 2: not profit. In fact, I can find no consistent answer

198
00:10:23,520 --> 00:10:25,480
Speaker 2: to how long a data center takes to build, and

199
00:10:25,520 --> 00:10:28,080
Speaker 2: the answer here is pretty important because that's how the

200
00:10:28,080 --> 00:10:29,800
Speaker 2: money is going to get made from the data center's

201
00:10:29,800 --> 00:10:33,520
Speaker 2: for building fucking everywhere. You may also notice that monthly

202
00:10:33,559 --> 00:10:35,840
Speaker 2: payments in the chart that I linked, which I should

203
00:10:35,880 --> 00:10:38,959
Speaker 2: have mentioned, begin at six to thirty months, a curious

204
00:10:38,960 --> 00:10:41,800
Speaker 2: and broad blob of time. You see, data centers are

205
00:10:41,800 --> 00:10:44,280
Speaker 2: extremely difficult to build, and the concept of an AI

206
00:10:44,400 --> 00:10:46,640
Speaker 2: data center is barely a few years old, with the

207
00:10:46,679 --> 00:10:49,800
Speaker 2: concepts of hundreds of megawats in one data center campus

208
00:10:49,920 --> 00:10:53,680
Speaker 2: entirely made up of aigpus barely two years old jpus,

209
00:10:54,120 --> 00:10:57,120
Speaker 2: which means basically everybody building one is doing so for

210
00:10:57,160 --> 00:11:00,440
Speaker 2: the first time, and even experienced developers are running into problems.

211
00:11:01,720 --> 00:11:05,839
Speaker 2: For example, core Scientific, Corweave's weird partner organization it tried

212
00:11:05,880 --> 00:11:08,440
Speaker 2: and failed to buy that is truly different despite the

213
00:11:08,440 --> 00:11:10,960
Speaker 2: similar name, has been trying to convert it's Denton, Texas

214
00:11:11,000 --> 00:11:13,960
Speaker 2: scriptocurrency mining data center into an AI data center since

215
00:11:14,000 --> 00:11:17,080
Speaker 2: November twenty twenty four, specifically so that Corewave can rent

216
00:11:17,080 --> 00:11:21,080
Speaker 2: it to Microsoft for Open AI. This hasn't gone well,

217
00:11:21,320 --> 00:11:23,959
Speaker 2: and this is when things get weird. The Wall Street

218
00:11:24,000 --> 00:11:26,360
Speaker 2: Journal reported a few weeks ago that Denton had been

219
00:11:26,400 --> 00:11:29,160
Speaker 2: racked with several months of delays thanks to rainstorms preventing

220
00:11:29,200 --> 00:11:34,200
Speaker 2: contractors and pooring contry contrete. I'm going with it. It's contrete.

221
00:11:34,240 --> 00:11:39,000
Speaker 2: Now we're calling it contry concrete. I originally, within this

222
00:11:39,080 --> 00:11:41,360
Speaker 2: script wrote down that this was a possible thing. I

223
00:11:41,480 --> 00:11:45,840
Speaker 2: thought that this was normal. However, I should add something,

224
00:11:46,240 --> 00:11:49,920
Speaker 2: this is this may actually be bollocks. So I'm a psycho,

225
00:11:50,040 --> 00:11:53,360
Speaker 2: and I went and I looked up whether. I looked

226
00:11:53,440 --> 00:11:56,960
Speaker 2: up weather in Abilene, Texas, and it turns out that

227
00:11:57,040 --> 00:12:01,439
Speaker 2: there were like eleven days with rain total, and there

228
00:12:01,520 --> 00:12:04,200
Speaker 2: was only I think on June third and September twenty second,

229
00:12:04,240 --> 00:12:07,120
Speaker 2: there were rainstorms or thunderstorms of any kind. I think

230
00:12:07,160 --> 00:12:09,000
Speaker 2: that there are only two days where rain was over

231
00:12:09,120 --> 00:12:13,520
Speaker 2: point one inches. The people I've talked to about this

232
00:12:13,640 --> 00:12:15,960
Speaker 2: deal are saying it wouldn't be weather, it would actually

233
00:12:16,040 --> 00:12:19,760
Speaker 2: just be money. Nevertheless, those are things that are stopping

234
00:12:19,760 --> 00:12:22,319
Speaker 2: them building a two hundred and sixty megawat data center,

235
00:12:22,760 --> 00:12:25,560
Speaker 2: and that's fucking nothing compared to the Giga. What's the

236
00:12:25,600 --> 00:12:28,959
Speaker 2: open Ai claims to want to build. And what this means,

237
00:12:29,000 --> 00:12:31,960
Speaker 2: by the way, is that call we've can't actually get

238
00:12:31,960 --> 00:12:34,920
Speaker 2: paid by open Ai because per its contract, customers don't

239
00:12:34,920 --> 00:12:37,559
Speaker 2: have to stop paying until the computer is actually available.

240
00:12:38,640 --> 00:12:40,920
Speaker 2: This is a very important deal to know for literally

241
00:12:40,960 --> 00:12:43,679
Speaker 2: any data center development you've ever heard of or seen.

242
00:12:55,160 --> 00:12:57,800
Speaker 2: As of cour we've SQ three twenty twenty five earnings,

243
00:12:57,880 --> 00:13:00,000
Speaker 2: they're sitting on one point one billion dollars into thirst

244
00:13:00,200 --> 00:13:03,040
Speaker 2: revenue that's income for service is not yet rendered are

245
00:13:03,080 --> 00:13:05,200
Speaker 2: from nine hundred and fifty one million in Q two

246
00:13:05,240 --> 00:13:08,160
Speaker 2: and four hundred and thirty six million in Q one

247
00:13:08,240 --> 00:13:11,080
Speaker 2: twenty twenty five. This means deposits have been made, but

248
00:13:11,120 --> 00:13:14,920
Speaker 2: the contract is yet to be serviced. Now, I'm also

249
00:13:15,040 --> 00:13:16,839
Speaker 2: a curious little critter, so I went and found the

250
00:13:16,880 --> 00:13:19,040
Speaker 2: nine hundred and twenty one page two point six billion

251
00:13:19,080 --> 00:13:23,840
Speaker 2: dollar DDTL delayed draw Term Loan three point zero loan

252
00:13:23,840 --> 00:13:28,280
Speaker 2: agreement between core Weave and banks including Morgan Stanley, Mitsubishi,

253
00:13:28,400 --> 00:13:31,080
Speaker 2: UFG and Gold and Sachs, and in doing so I

254
00:13:31,160 --> 00:13:34,360
Speaker 2: learned something well. First of all, open ai appears to

255
00:13:34,360 --> 00:13:36,960
Speaker 2: have net three sixty payment terms from Corwave, meaning that

256
00:13:37,000 --> 00:13:40,880
Speaker 2: it can literally pay a year from invoice. Second, Corweve

257
00:13:41,000 --> 00:13:44,760
Speaker 2: is required to maintain something called a contract realization ratio

258
00:13:44,760 --> 00:13:47,240
Speaker 2: of point eighty five times, meaning that core Weave has

259
00:13:47,280 --> 00:13:49,640
Speaker 2: to make at least eighty five cents of every expected

260
00:13:49,679 --> 00:13:52,360
Speaker 2: dollar or it is in default of its loan. This

261
00:13:52,480 --> 00:13:54,280
Speaker 2: is important to note because it means that if say

262
00:13:54,280 --> 00:13:56,200
Speaker 2: open Ai decides not to pay up in a year,

263
00:13:56,440 --> 00:13:59,480
Speaker 2: corwave will be up shit creek without a paddle. Now,

264
00:13:59,520 --> 00:14:02,280
Speaker 2: I apolog that suggests that core Weave is an already

265
00:14:02,360 --> 00:14:04,960
Speaker 2: up shit creek, or that it might have a paddle

266
00:14:05,000 --> 00:14:08,200
Speaker 2: of some sort buried inside in Vidia's latest earnings on

267
00:14:08,240 --> 00:14:11,920
Speaker 2: page seventeen, there's a little clue at home, and I

268
00:14:12,040 --> 00:14:15,760
Speaker 2: quote in the third quarter of fiscal year twenty twenty six,

269
00:14:15,800 --> 00:14:18,199
Speaker 2: we entered into an agreement to guarantee of partner's facility

270
00:14:18,280 --> 00:14:20,960
Speaker 2: lease obligations in the event of their default. The agreement

271
00:14:21,040 --> 00:14:24,200
Speaker 2: allows our partner to secure a limited availability facility lease

272
00:14:24,320 --> 00:14:28,000
Speaker 2: backed by our credit profile in exchange for issuing US warrants.

273
00:14:28,160 --> 00:14:30,680
Speaker 2: The maximum gross exposure is eight hundred and sixty million,

274
00:14:30,880 --> 00:14:32,880
Speaker 2: which is reduced as the partner makes payments to the

275
00:14:32,920 --> 00:14:35,480
Speaker 2: leaser of a five years. The partner has placed four

276
00:14:35,520 --> 00:14:38,200
Speaker 2: hundred and seventy million dollars in escrow and executed an

277
00:14:38,200 --> 00:14:41,000
Speaker 2: agreement to sell the data center cloud capacity, mitigating our

278
00:14:41,040 --> 00:14:44,920
Speaker 2: default risk credit where credit is due. Eaglide analysts Just

279
00:14:45,040 --> 00:14:47,760
Speaker 2: Dario caught this in November, but in core weaves condensed

280
00:14:47,760 --> 00:14:50,200
Speaker 2: to consolidated balance sheets, there sits a four hundred and

281
00:14:50,240 --> 00:14:53,000
Speaker 2: seventy seven point five million dollars line item under restricted

282
00:14:53,000 --> 00:14:56,400
Speaker 2: cash and cash equivalence non current, though this might not

283
00:14:56,520 --> 00:15:00,320
Speaker 2: be in Vidia's escrow. This number shifted from six hundred

284
00:15:00,320 --> 00:15:02,640
Speaker 2: and seventeen in Q one to three hundred and forty

285
00:15:02,680 --> 00:15:05,160
Speaker 2: million in Q two. It lines up all too precisely,

286
00:15:05,880 --> 00:15:10,800
Speaker 2: and who else would in video be guaranteeing In any case,

287
00:15:10,840 --> 00:15:12,760
Speaker 2: corweav is likely to get the best deals in data

288
00:15:12,760 --> 00:15:15,840
Speaker 2: centered debt outside of Oracle. It has top tier financiers

289
00:15:15,880 --> 00:15:17,880
Speaker 2: who I'll get to in a little bit, the fallbacking

290
00:15:18,280 --> 00:15:20,720
Speaker 2: in Video who is both an investor customer and apparent

291
00:15:20,760 --> 00:15:23,640
Speaker 2: financial backstop, and also Core with as a customer of

292
00:15:23,680 --> 00:15:27,360
Speaker 2: theirs and the ability to raise debt quickly. Corweb's deals

293
00:15:27,360 --> 00:15:30,160
Speaker 2: are likely indicative of how data center financing takes place,

294
00:15:30,160 --> 00:15:34,680
Speaker 2: and those top tier financiers it's they may basically been

295
00:15:34,720 --> 00:15:38,160
Speaker 2: in every single deal. It's actually it's actually really worrying

296
00:15:38,200 --> 00:15:40,680
Speaker 2: when you, I mean, you lay it all out, which

297
00:15:40,720 --> 00:15:43,400
Speaker 2: is exactly what I'm going to do. So who's actually

298
00:15:43,400 --> 00:15:45,720
Speaker 2: paying for this shit? I went and dug through a

299
00:15:45,760 --> 00:15:48,640
Speaker 2: pile of twenty six prominent data center loan deals, including

300
00:15:48,680 --> 00:15:51,880
Speaker 2: the proposed thirty eight billion dollar debt package that Oracle

301
00:15:51,920 --> 00:15:55,760
Speaker 2: Advantage data Center partners are raising for Stargate Shackelford and Wisconsin,

302
00:15:55,840 --> 00:15:59,880
Speaker 2: Stargate Ablene, New Mexico, and of course soft Banks fifty

303
00:16:00,040 --> 00:16:02,120
Speaker 2: billion dollar bridge loan, which I included for a reason

304
00:16:02,120 --> 00:16:04,760
Speaker 2: that will become obvious shortly, and multiple core wave loans.

305
00:16:04,760 --> 00:16:07,080
Speaker 2: I've found a few commonalities, and forgive me, I'm about

306
00:16:07,080 --> 00:16:08,640
Speaker 2: to say a lot of names and numbers. You might

307
00:16:08,680 --> 00:16:10,560
Speaker 2: want to pull up the newsletter to read along with

308
00:16:10,600 --> 00:16:14,640
Speaker 2: it like a sing along. Now, Blue Owl, Blue ol,

309
00:16:14,680 --> 00:16:17,080
Speaker 2: We love blue ol folks. They were present in every

310
00:16:17,080 --> 00:16:20,120
Speaker 2: single Stargate deal other than the thirty eight billion dollar

311
00:16:20,160 --> 00:16:23,000
Speaker 2: deal that's being raised by Advantage that is yet to close.

312
00:16:23,960 --> 00:16:26,040
Speaker 2: Blue Hour was also involved in a one point three

313
00:16:26,040 --> 00:16:28,680
Speaker 2: billion dollar Australian data set and a debt package by

314
00:16:28,760 --> 00:16:33,960
Speaker 2: Virtue of owning Stack infrastructure. Remember that name MUFG MUG

315
00:16:34,400 --> 00:16:37,280
Speaker 2: also Mitsubishi UFJ Financial Group is their full name, but

316
00:16:37,320 --> 00:16:40,160
Speaker 2: I like saying. MUG was present in seventeen out of

317
00:16:40,160 --> 00:16:43,359
Speaker 2: twenty six of the deals, including three separate core weafinancing,

318
00:16:43,560 --> 00:16:47,000
Speaker 2: Stargate New Mexico an eighteen billion dollar deal, the alleged

319
00:16:47,040 --> 00:16:50,600
Speaker 2: Stargate thirty eight billion dollar deal, SoftBank's bridge loan which

320
00:16:50,640 --> 00:16:52,720
Speaker 2: they had to raise by the way, to fund open AI,

321
00:16:52,960 --> 00:16:55,520
Speaker 2: and a five billion dollar green loan package for Vantage

322
00:16:55,600 --> 00:16:59,240
Speaker 2: Data Centers, who you might have just heard me say. JP.

323
00:16:59,360 --> 00:17:02,080
Speaker 2: Morgan Chase was involved in eight deals, and they were

324
00:17:02,120 --> 00:17:04,600
Speaker 2: involved in some of the largest. Two cory was October

325
00:17:04,600 --> 00:17:08,560
Speaker 2: twenty twenty four financing their third delayed draw Tone loan

326
00:17:08,920 --> 00:17:12,359
Speaker 2: and November financing as well the funding behind Stargate, Apple

327
00:17:12,400 --> 00:17:14,879
Speaker 2: in the thirty eight billion dollar Oracle deal, and Blue

328
00:17:14,880 --> 00:17:18,480
Speaker 2: Owl's acquisition of IPI Partners in twenty twenty four. And yes,

329
00:17:18,520 --> 00:17:20,960
Speaker 2: they were also part of soft banks bridge loan. Now,

330
00:17:21,040 --> 00:17:24,000
Speaker 2: let's not forget Deutsche Bank, who've never done anything dodgy. Ever,

331
00:17:24,160 --> 00:17:27,000
Speaker 2: do not google what Deutsche Bank has done with banks

332
00:17:27,040 --> 00:17:30,320
Speaker 2: or particular fellas Deutsche Bank, they were involved in soft

333
00:17:30,320 --> 00:17:34,000
Speaker 2: Bank's Bridge loan and three smaller deals data center in Seoul,

334
00:17:34,520 --> 00:17:37,480
Speaker 2: Cory's twenty twenty four debt couise, November financing, and a

335
00:17:37,560 --> 00:17:40,080
Speaker 2: data center deal in Latin America, as well as a

336
00:17:40,119 --> 00:17:42,320
Speaker 2: six hundred and ten million dollar data center project in

337
00:17:42,400 --> 00:17:44,840
Speaker 2: Virginia and a billion euro of data center project in

338
00:17:44,880 --> 00:17:49,479
Speaker 2: Germany that was invested in within Video BNP Perribus seven deals.

339
00:17:49,600 --> 00:17:53,199
Speaker 2: Cory was delayed draw Tone loan three Stargate New Mexico,

340
00:17:53,320 --> 00:17:57,040
Speaker 2: Stargate Wisconsin and Texas IPI partners of Blue Owl, that

341
00:17:57,200 --> 00:17:59,920
Speaker 2: data center deal in Seoul, and another data center in Chile.

342
00:18:01,119 --> 00:18:04,000
Speaker 2: Morgan Stanley. You remember Morgan Stanley, don't you? While they

343
00:18:04,000 --> 00:18:07,040
Speaker 2: were in eight of these fucking things. Core's October twenty

344
00:18:07,080 --> 00:18:11,280
Speaker 2: twenty four loan loan, three November loans stargeting New Mexico

345
00:18:11,440 --> 00:18:14,040
Speaker 2: Stargate with Advantage. They're thirty eight billion dollar one I

346
00:18:14,160 --> 00:18:17,960
Speaker 2: just talked about, and one with EQT Edge Connects where

347
00:18:18,000 --> 00:18:20,160
Speaker 2: they I don't I don't need to go into detO.

348
00:18:20,280 --> 00:18:22,320
Speaker 2: It's another fucking data center. Oh and they were also

349
00:18:22,359 --> 00:18:25,880
Speaker 2: in SoftBank's Bridge loan, which they needed to to find

350
00:18:25,920 --> 00:18:28,159
Speaker 2: open AI. Now here's another name that you're going to

351
00:18:28,200 --> 00:18:32,520
Speaker 2: hear in the future, SNBC, Sumi Tomo, Mitsui Banking Corporation,

352
00:18:32,640 --> 00:18:35,720
Speaker 2: and I must say the Japanese nyler name of fucking company. Anyway,

353
00:18:35,960 --> 00:18:39,240
Speaker 2: they were in seven deals, all notable cor Weeve's DDTL

354
00:18:39,320 --> 00:18:42,399
Speaker 2: three point zero and November Financing loans, Stargate, New Mexico,

355
00:18:42,480 --> 00:18:45,960
Speaker 2: Stargate Texas, Wisconsin. That's that thirty eight billion dollar one

356
00:18:46,040 --> 00:18:49,600
Speaker 2: A data center in Rowan, Maryland, also involving mug TD

357
00:18:49,680 --> 00:18:52,359
Speaker 2: Securities and HSBC, as well as data center deals in

358
00:18:52,440 --> 00:18:54,960
Speaker 2: Chile and Latin America you've already heard about. Oh and

359
00:18:55,000 --> 00:18:58,240
Speaker 2: guess what Soft Banks Bridge loan which they used to

360
00:18:58,280 --> 00:19:01,720
Speaker 2: invest in open Ai. There are a lot more data

361
00:19:01,720 --> 00:19:04,480
Speaker 2: centered deals than these, but I wanted to tell you

362
00:19:04,560 --> 00:19:08,520
Speaker 2: exactly how centralized they are. I also really need you

363
00:19:08,560 --> 00:19:11,480
Speaker 2: to know how worrying that is, because literally every part

364
00:19:11,560 --> 00:19:14,800
Speaker 2: of this pie is being funded by like eight to

365
00:19:14,880 --> 00:19:18,800
Speaker 2: nine banks. If open ai is raising money for a date, well,

366
00:19:18,920 --> 00:19:21,520
Speaker 2: they wouldn't raise it would be Oracle or core Weave

367
00:19:21,600 --> 00:19:23,439
Speaker 2: or have you raising for a data center. One of

368
00:19:23,440 --> 00:19:25,840
Speaker 2: these people is in it or companies. They're not people.

369
00:19:26,800 --> 00:19:30,679
Speaker 2: It's really worrying. It's worrying how centralized this is, and

370
00:19:30,720 --> 00:19:33,800
Speaker 2: the largest deals, such as the thirty eight billion Stargate Texas,

371
00:19:33,880 --> 00:19:37,000
Speaker 2: Wisconsin deal or the eighteen billion dollar Stargate New Mexico

372
00:19:37,080 --> 00:19:40,120
Speaker 2: deal both involved Gold and Sachs b and Peter Ribas,

373
00:19:40,200 --> 00:19:43,800
Speaker 2: SMBC and fuf MUFG, and all four of those companies

374
00:19:43,840 --> 00:19:46,600
Speaker 2: have at some point funded core Weave. In fact, everybody

375
00:19:46,600 --> 00:19:51,680
Speaker 2: appears to a funded core Weave at some point. City Bank, Credit, Agricole, Society, General,

376
00:19:52,000 --> 00:19:55,560
Speaker 2: Wells Fargo, Carlile, Blackstone, black Rock, Barkley's, Magnatar, and Jeffrey

377
00:19:55,600 --> 00:19:58,399
Speaker 2: is the name of few, and if you've been watching

378
00:19:58,560 --> 00:20:02,679
Speaker 2: the news recently, I have heard that term term word

379
00:20:02,880 --> 00:20:06,560
Speaker 2: jeffries to refer to the company that invested in first

380
00:20:06,600 --> 00:20:25,199
Speaker 2: brands and their exposure anyway. Of the forty banks and

381
00:20:25,240 --> 00:20:28,840
Speaker 2: financial institutions I research twenty four have at some point

382
00:20:29,040 --> 00:20:32,879
Speaker 2: loan to or organized dead for Corwave. Of those institutions, Blackstone,

383
00:20:32,920 --> 00:20:36,560
Speaker 2: Deutsche Bank, JP, Morgan Chase, Morgan, Stanley, MUFG, and Wells

384
00:20:36,600 --> 00:20:39,720
Speaker 2: Fargo have done so multiple times. Core Weave is a

385
00:20:39,720 --> 00:20:43,520
Speaker 2: deeply unprofitable company, saddled with incredible debt and deteriorating margins.

386
00:20:43,600 --> 00:20:46,160
Speaker 2: With one of the largest clients paying net three sixty

387
00:20:46,320 --> 00:20:49,480
Speaker 2: and as I've said, it is arguably the best financed

388
00:20:49,560 --> 00:20:51,520
Speaker 2: data center company in the world with the best chances

389
00:20:51,560 --> 00:20:56,760
Speaker 2: of survival. What I'm getting at is that most data

390
00:20:56,800 --> 00:20:58,960
Speaker 2: centered deals are likely much worse than the terms that

391
00:20:59,000 --> 00:21:01,920
Speaker 2: corwav faces and are likely financed in a similar way

392
00:21:02,080 --> 00:21:04,880
Speaker 2: whereas where a client is signed for data center capacity

393
00:21:04,920 --> 00:21:07,400
Speaker 2: that doesn't exist, such as when nebuists raised four point

394
00:21:07,400 --> 00:21:10,199
Speaker 2: three billion dollars through a share sale and convertible notes

395
00:21:10,359 --> 00:21:13,280
Speaker 2: read loans to handle its seventeen point four billion dollar

396
00:21:13,400 --> 00:21:17,720
Speaker 2: data center contract with Microsoft and yees what Dobin Sachs

397
00:21:17,760 --> 00:21:19,960
Speaker 2: acted as the lead underwriter on the deal, with assistance

398
00:21:20,000 --> 00:21:24,000
Speaker 2: from Bank of America, City Group, Morgan Stanley, all three

399
00:21:24,000 --> 00:21:27,439
Speaker 2: of which have invested in core Weave. AI. Data centers

400
00:21:27,480 --> 00:21:29,960
Speaker 2: are expensive, required debt due to the massive cost of

401
00:21:30,000 --> 00:21:32,560
Speaker 2: construction in the cost of GPUs, and all take at

402
00:21:32,640 --> 00:21:35,280
Speaker 2: least a year, if not two, to start generating a

403
00:21:35,359 --> 00:21:38,359
Speaker 2: single dollar of revenue, at which point they also begin

404
00:21:38,400 --> 00:21:40,840
Speaker 2: losing money because it seems that renting out GPUs is

405
00:21:40,840 --> 00:21:44,480
Speaker 2: really unprofitable. Didn't think, no, no one, no one fucking

406
00:21:44,520 --> 00:21:47,679
Speaker 2: think to check that one Every single major bank and

407
00:21:47,720 --> 00:21:50,560
Speaker 2: financial institution has piled hundreds of millions, if not billions

408
00:21:50,560 --> 00:21:53,199
Speaker 2: of dollars into building data centers that take forever to

409
00:21:53,240 --> 00:21:56,120
Speaker 2: even start generating money, at which point they only seem

410
00:21:56,160 --> 00:21:59,680
Speaker 2: to lose it. Were still in Video sells GPUs and

411
00:21:59,720 --> 00:22:02,560
Speaker 2: a one near upgrade cycle, meaning that all of those

412
00:22:02,640 --> 00:22:05,040
Speaker 2: data centers being built right now are being filled with

413
00:22:05,080 --> 00:22:07,440
Speaker 2: Blackwell chips, and by the time they turn on and

414
00:22:07,560 --> 00:22:10,280
Speaker 2: Video will be selling its next generation via Ruben chips,

415
00:22:10,640 --> 00:22:14,680
Speaker 2: baking them obsolete. Now now now, now, no, no, no.

416
00:22:14,920 --> 00:22:18,480
Speaker 2: You've probably heard that Verra Ruben, the next GPU from

417
00:22:18,480 --> 00:22:21,920
Speaker 2: in Video, will use the same ranks called Oberon as Blackwell,

418
00:22:22,000 --> 00:22:23,960
Speaker 2: which is true to an extent, but won't be true

419
00:22:24,000 --> 00:22:26,159
Speaker 2: for long as in Video intends to shift to their

420
00:22:26,280 --> 00:22:28,960
Speaker 2: Kuiper racks in twenty twenty seven, hoping to build one

421
00:22:29,040 --> 00:22:33,720
Speaker 2: megawat it rax, which involve entire racks full of power supplies,

422
00:22:34,000 --> 00:22:36,080
Speaker 2: meaning that all of those data centers you see today,

423
00:22:36,800 --> 00:22:38,840
Speaker 2: whenever it is they get built, if they get built,

424
00:22:39,080 --> 00:22:41,560
Speaker 2: will be full of racks incompatible with the next generation

425
00:22:41,600 --> 00:22:44,439
Speaker 2: of GPUs. This will also decrease the value of the

426
00:22:44,440 --> 00:22:46,720
Speaker 2: assets inside the data centers, which will in turn decrease

427
00:22:46,760 --> 00:22:48,800
Speaker 2: the value of the assets held by the firms investing

428
00:22:49,160 --> 00:22:52,120
Speaker 2: Stargate Aplin, the one invested in by JP Morgan, Blue

429
00:22:52,119 --> 00:22:55,600
Speaker 2: Oul and Primary Digital Infrastructure and Society General, the one

430
00:22:55,600 --> 00:22:57,840
Speaker 2: that's been heavily delayed and won't be ready until the

431
00:22:57,920 --> 00:22:59,840
Speaker 2: end of twenty twenty six of the earliest fall to

432
00:22:59,880 --> 00:23:02,960
Speaker 2: the brim to the gills with two year old GB

433
00:23:03,040 --> 00:23:08,119
Speaker 2: two hundred Blackwell racks. Hell yeah, baby woo by the

434
00:23:08,119 --> 00:23:12,200
Speaker 2: beginning of twenty twenty seven, when this shithole eventually opens,

435
00:23:12,720 --> 00:23:15,959
Speaker 2: I should really say potentially opens, because let's be honest,

436
00:23:16,359 --> 00:23:19,600
Speaker 2: I'm not confident Stargate Apolene will be obsolete, as will

437
00:23:19,600 --> 00:23:22,199
Speaker 2: any and all data centers filled with Blackwell GPUs, as

438
00:23:22,240 --> 00:23:25,399
Speaker 2: will any and all data centers being built today. Every

439
00:23:25,600 --> 00:23:28,320
Speaker 2: single one takes one to three years and hundreds of

440
00:23:28,320 --> 00:23:32,280
Speaker 2: millions or billions of dollars to build, but probably raised

441
00:23:32,280 --> 00:23:35,119
Speaker 2: in damp, and every single one faces the same kinds

442
00:23:35,240 --> 00:23:37,919
Speaker 2: of construction delays and better yeah, almost all of them

443
00:23:37,920 --> 00:23:44,080
Speaker 2: will turn on in roughly the same timeframe. All right, Look, folks,

444
00:23:44,119 --> 00:23:46,119
Speaker 2: I gotta admit I don't really get our money works.

445
00:23:46,320 --> 00:23:50,840
Speaker 2: I'm not economist, but I do know that supply and

446
00:23:50,880 --> 00:23:53,760
Speaker 2: demand has an effect on pricing. What do you believe

447
00:23:53,840 --> 00:23:56,640
Speaker 2: happens to the price of renting at Blackwell GPU when

448
00:23:56,720 --> 00:23:59,760
Speaker 2: all of these data centers come on? Do you think

449
00:24:00,240 --> 00:24:02,919
Speaker 2: it will it will mean that the price goes up

450
00:24:03,000 --> 00:24:06,720
Speaker 2: or down. Also, while we're on the subject, what do

451
00:24:06,760 --> 00:24:11,320
Speaker 2: you think happens if there isn't sufficient demand? And demand

452
00:24:11,400 --> 00:24:14,240
Speaker 2: is a real question, by the way. Right now, open

453
00:24:14,240 --> 00:24:16,359
Speaker 2: ai makes up a large chunk of the global sale

454
00:24:16,359 --> 00:24:18,840
Speaker 2: of compute, at least eight point six seven billion dollars

455
00:24:18,880 --> 00:24:21,800
Speaker 2: of a zero revenue which is Microsoft's cloud platform through

456
00:24:21,840 --> 00:24:25,560
Speaker 2: September twenty twenty five, and they're part of about twenty

457
00:24:25,560 --> 00:24:28,639
Speaker 2: two point five billion dollars of Corway's backlog, thirty eight

458
00:24:28,680 --> 00:24:31,440
Speaker 2: billion dollars of Amazon's backlog, and so on and so forth,

459
00:24:31,720 --> 00:24:34,000
Speaker 2: and made based on my reporting from last year, just

460
00:24:34,040 --> 00:24:36,280
Speaker 2: over four point five billion dollars through the end of

461
00:24:36,280 --> 00:24:39,880
Speaker 2: September twenty twenty five. Open ai can't afford to pay anyone,

462
00:24:40,000 --> 00:24:42,160
Speaker 2: and nowhere is that more obvious than when it negotiated

463
00:24:42,240 --> 00:24:45,199
Speaker 2: year long payment terms of core with Otherwise, when you

464
00:24:45,240 --> 00:24:48,200
Speaker 2: remove the contracts signed by Hyperscale as an open Ai,

465
00:24:48,320 --> 00:24:50,840
Speaker 2: which I do not believe has the money to pay anybody.

466
00:24:51,680 --> 00:24:54,200
Speaker 2: Based on my analysis, there was less than a billion

467
00:24:54,240 --> 00:24:57,280
Speaker 2: dollars of AI compute revenue in twenty twenty five, or

468
00:24:57,359 --> 00:24:59,680
Speaker 2: zero point five eighty three percent of the money spent

469
00:24:59,680 --> 00:25:04,560
Speaker 2: on day center credit deals in the US. Is that good?

470
00:25:05,280 --> 00:25:09,159
Speaker 2: Hyperscalar revenue is also immediately questionable with Microsoft's steel with

471
00:25:09,200 --> 00:25:12,399
Speaker 2: Nebeus per their six K filing set to default in

472
00:25:12,440 --> 00:25:15,040
Speaker 2: the event that Nebius cannot provide the capacity it's sold

473
00:25:15,320 --> 00:25:18,280
Speaker 2: of its unfinished Fineland, New Jersey data center, which is

474
00:25:18,320 --> 00:25:20,720
Speaker 2: being built by Data One, a company that's never built

475
00:25:20,760 --> 00:25:23,119
Speaker 2: an AI data center, with a CEO that had his

476
00:25:23,240 --> 00:25:28,119
Speaker 2: LinkedIn location set to United Arab Emirates as a very

477
00:25:28,160 --> 00:25:31,720
Speaker 2: recently and indeed was in Arabic until very recently as well,

478
00:25:32,160 --> 00:25:34,560
Speaker 2: and they have funding from a concrete firm that is

479
00:25:34,560 --> 00:25:37,320
Speaker 2: also a vendor on the project. I also believe that

480
00:25:37,359 --> 00:25:40,359
Speaker 2: Microsoft is setting Nebeus up to fail based on discussions

481
00:25:40,359 --> 00:25:42,720
Speaker 2: with sources direct with direct knowledge of plans for the

482
00:25:42,760 --> 00:25:45,880
Speaker 2: violent New Jersey data center. Nebius has agreed to timelines

483
00:25:45,880 --> 00:25:48,320
Speaker 2: that involve having eighteen thousand video B two hundred and

484
00:25:48,359 --> 00:25:50,840
Speaker 2: B three hundred GPUs up and running by the end

485
00:25:50,840 --> 00:25:54,320
Speaker 2: of January for a total of fifty megawats, with another

486
00:25:54,359 --> 00:25:57,240
Speaker 2: eighteen thousand, b three hundreds due by the end of May.

487
00:25:57,640 --> 00:25:59,760
Speaker 2: On speaking with experts in the field about how viable

488
00:25:59,800 --> 00:26:01,600
Speaker 2: he is, the plans are two laugh and one to

489
00:26:01,760 --> 00:26:05,080
Speaker 2: me to fuck off. If Nebius fails to build the capacity,

490
00:26:05,119 --> 00:26:07,360
Speaker 2: Microsoft can walk away, much like open Ai can walk

491
00:26:07,400 --> 00:26:09,840
Speaker 2: away from Stargate in the event that Oracle fails to

492
00:26:09,840 --> 00:26:12,280
Speaker 2: build it on time, as reported by Anissa Gardisi of

493
00:26:12,320 --> 00:26:15,200
Speaker 2: The Information in April twenty twenty five, and I believe

494
00:26:15,240 --> 00:26:17,280
Speaker 2: that this is the case for literally any data center

495
00:26:17,320 --> 00:26:20,080
Speaker 2: provider that's building a data center for any signed up tenant.

496
00:26:20,400 --> 00:26:22,920
Speaker 2: This is another layer of risk to data center developers

497
00:26:22,920 --> 00:26:25,680
Speaker 2: that nobody bothers to fucking discuss, because everybody loves seeing

498
00:26:25,720 --> 00:26:29,160
Speaker 2: these big, beautiful numbers, Except the numbers might have become

499
00:26:29,200 --> 00:26:32,879
Speaker 2: a little too beautiful for sun. On December seventeenth, the

500
00:26:32,880 --> 00:26:35,280
Speaker 2: Financial Times report that the Blue Ou Capitol had pulled

501
00:26:35,280 --> 00:26:38,800
Speaker 2: out of the ten billion dollar Stargate Michigan data center project,

502
00:26:39,040 --> 00:26:41,879
Speaker 2: citing and I quote concerns about its rising debt and

503
00:26:41,960 --> 00:26:45,680
Speaker 2: artificial intelligence spending to quote the ft again. Blue Oul

504
00:26:45,680 --> 00:26:48,080
Speaker 2: had been discussions with Lender's and Oracle about investing in

505
00:26:48,080 --> 00:26:50,199
Speaker 2: the planned one gig what data center being built to

506
00:26:50,280 --> 00:26:55,040
Speaker 2: serve open AI in Saline Township, Michigan. What debt, you

507
00:26:55,119 --> 00:26:58,200
Speaker 2: may ask? While Blue Owl, formerly the loosest legs in

508
00:26:58,280 --> 00:27:01,880
Speaker 2: data center financing, was in always six hundred million dollar deal,

509
00:27:02,000 --> 00:27:03,960
Speaker 2: the seven hundred and fifty million dollar deal as well

510
00:27:03,960 --> 00:27:07,360
Speaker 2: for its planned Virginia data center with Teresa Technology Parks,

511
00:27:07,600 --> 00:27:09,879
Speaker 2: a four billion dollar corps with have data center project

512
00:27:09,920 --> 00:27:14,119
Speaker 2: in Pennsylvania, Stargate Abilene, Stargate Mexico, met As thirty billion

513
00:27:14,160 --> 00:27:17,080
Speaker 2: dollar hyperiod data centered deal, and a one point three

514
00:27:17,119 --> 00:27:21,200
Speaker 2: billion dollar data centered deal in Australia through Stack Infrastructure.

515
00:27:21,200 --> 00:27:23,959
Speaker 2: You remember I mentioned those like ten minutes ago if

516
00:27:24,000 --> 00:27:27,040
Speaker 2: you if you email me for a fish biscuit, I

517
00:27:27,040 --> 00:27:29,879
Speaker 2: haven't gotten. I'm sorry. Anyway, they owned that company and

518
00:27:30,480 --> 00:27:33,520
Speaker 2: may I mentioned it earlier. Anyways, Let's keep going to

519
00:27:33,600 --> 00:27:35,800
Speaker 2: be clear. Blue Owl pulling out is not the same

520
00:27:35,840 --> 00:27:39,520
Speaker 2: as a regular deal. It's a BBC business development company.

521
00:27:39,640 --> 00:27:42,479
Speaker 2: Then invest both this money and rallies together various banks

522
00:27:42,480 --> 00:27:47,440
Speaker 2: in this case SMBCBMP, Peribus, MUFG, and Gold and Sachs

523
00:27:47,480 --> 00:27:51,760
Speaker 2: all part of Stargate, New Mexico, which makes me wonder

524
00:27:51,800 --> 00:27:57,199
Speaker 2: why it didn't happen. In fact, it makes me very

525
00:27:57,320 --> 00:28:01,159
Speaker 2: much worry about that. Per the Financial time, and I quote,

526
00:28:01,200 --> 00:28:03,600
Speaker 2: the private Capitol group had been has been the primary

527
00:28:03,640 --> 00:28:06,359
Speaker 2: backer for Oracle's largest data center projects in the US,

528
00:28:06,600 --> 00:28:08,960
Speaker 2: investing its own money and raising billions more in debt

529
00:28:09,000 --> 00:28:11,560
Speaker 2: to build the facilities. Blue Ol typically sets up a

530
00:28:11,560 --> 00:28:14,399
Speaker 2: special purpose vehicle which owns the data center and leases

531
00:28:14,440 --> 00:28:18,199
Speaker 2: it to Oracle. Blue Owl is incredibly well connected and

532
00:28:18,240 --> 00:28:20,920
Speaker 2: experienced in putting together these kinds of deals and very

533
00:28:21,040 --> 00:28:24,080
Speaker 2: likely went to many banks, many banks. It's worked with

534
00:28:24,160 --> 00:28:27,520
Speaker 2: banks that have been basically giving them blank checks who

535
00:28:27,560 --> 00:28:31,280
Speaker 2: apparently had and I quote concerns about Blue Owl's rising debt,

536
00:28:31,920 --> 00:28:35,960
Speaker 2: much of which it had issued them. While rumor suggests

537
00:28:35,960 --> 00:28:38,760
Speaker 2: that Blackstone may step in, I need to be clear,

538
00:28:38,800 --> 00:28:41,680
Speaker 2: and I've spoken to numerous people in private equity about this.

539
00:28:41,680 --> 00:28:45,480
Speaker 2: This is not a consumer mortgage. This is not a

540
00:28:45,520 --> 00:28:48,160
Speaker 2: young couple buying a start at home. This is a

541
00:28:48,320 --> 00:28:52,080
Speaker 2: giant deal. If it's ten billion dollars, Blue Owl probably

542
00:28:52,080 --> 00:28:54,720
Speaker 2: would put in two billion themselves. That'd be like, I

543
00:28:54,720 --> 00:28:58,960
Speaker 2: don't know, maybe eight billion dollars of credit. I mean,

544
00:29:00,000 --> 00:29:02,800
Speaker 2: pipping in would require billions of dollars in legal logistics

545
00:29:02,800 --> 00:29:05,600
Speaker 2: and likely Blackstone talking to the very same banks it

546
00:29:05,720 --> 00:29:10,720
Speaker 2: already said no, why are they not doing this? And indeed,

547
00:29:10,760 --> 00:29:14,480
Speaker 2: why are things looking shaky? Well remember that thing about

548
00:29:14,520 --> 00:29:17,000
Speaker 2: how this data cent would be least to Oracle. Well,

549
00:29:17,040 --> 00:29:20,440
Speaker 2: Oracle had a free cash flow of negative thirteen billion

550
00:29:20,520 --> 00:29:23,280
Speaker 2: dollars on revenues of sixteen billion dollars, with its most

551
00:29:23,360 --> 00:29:27,680
Speaker 2: recent earnings only beating as analyst estimates thanks to the

552
00:29:27,680 --> 00:29:29,840
Speaker 2: sale of its two point sixty eight billion dollar steak

553
00:29:29,880 --> 00:29:33,200
Speaker 2: in Ampare. And you remember that soft bank Bridge loan.

554
00:29:33,240 --> 00:29:35,760
Speaker 2: I mentioned that fifteen billion dollar one Yeah, half of

555
00:29:35,800 --> 00:29:37,720
Speaker 2: that went to open AI, half of that went to

556
00:29:37,720 --> 00:29:41,000
Speaker 2: buying Ampare, So that's a one off of them. Oracle's

557
00:29:41,040 --> 00:29:43,400
Speaker 2: date is exploding, with over a billion dollars in interest

558
00:29:43,400 --> 00:29:46,840
Speaker 2: payments in its last quarter alone. It's GPU gross margins

559
00:29:46,880 --> 00:29:49,880
Speaker 2: of fourteen percent, which does not mean profitable. Its latest

560
00:29:50,040 --> 00:29:53,200
Speaker 2: Nvidio GB two hundred GPUs have a negative one hundred

561
00:29:53,200 --> 00:29:55,520
Speaker 2: percent gross margin, and it has two hundred and forty

562
00:29:55,520 --> 00:29:58,640
Speaker 2: eight billion dollars in upcoming data center leases is yet

563
00:29:58,640 --> 00:30:02,640
Speaker 2: to begin, and thus all for the most part to

564
00:30:02,640 --> 00:30:05,960
Speaker 2: handle compute for one customer open Ai, which needs to

565
00:30:06,040 --> 00:30:10,000
Speaker 2: raise one hundred billion dollars and then probably much more

566
00:30:10,640 --> 00:30:17,160
Speaker 2: to survive. Is that good anyway, that's a good point

567
00:30:17,160 --> 00:30:19,360
Speaker 2: to jump off from. Next episode, we're gonna wrap all

568
00:30:19,440 --> 00:30:22,200
Speaker 2: this up. It's been a great time reading this for you,

569
00:30:22,560 --> 00:30:24,280
Speaker 2: and we're gonna talk about how the data center of

570
00:30:24,280 --> 00:30:35,720
Speaker 2: Apocalypse will start and what might come next. Thank you

571
00:30:35,720 --> 00:30:38,440
Speaker 2: for listening to Better Offline. The editor and composer of

572
00:30:38,480 --> 00:30:41,440
Speaker 2: the Better Offline theme song is Matasowski. You can check

573
00:30:41,480 --> 00:30:44,440
Speaker 2: out more of his music and audio projects at Matasowski

574
00:30:44,440 --> 00:30:47,960
Speaker 2: dot com, M A T T O. S O w

575
00:30:48,280 --> 00:30:51,840
Speaker 2: Ski dot com. You can email me at easy at

576
00:30:51,840 --> 00:30:54,520
Speaker 2: Better offline dot com, or visit Better Offline dot com

577
00:30:54,520 --> 00:30:56,960
Speaker 2: to find more podcast links and of course my newsletter.

578
00:30:57,360 --> 00:30:59,800
Speaker 2: I also really recommend you go to chat dot Where's

579
00:30:59,840 --> 00:31:02,360
Speaker 2: your dot am to visit the discord, and go to

580
00:31:02,400 --> 00:31:05,800
Speaker 2: our slash Better Offline to check out I'll Reddit. Thank

581
00:31:05,800 --> 00:31:09,240
Speaker 2: you so much for listening. Better Offline is a production

582
00:31:09,320 --> 00:31:11,960
Speaker 2: of cool Zone Media. For more from cool Zone Media,

583
00:31:12,080 --> 00:31:15,400
Speaker 2: visit our website Coolzonemedia dot com or check us out

584
00:31:15,440 --> 00:31:18,440
Speaker 2: on the iHeartRadio app, Apple Podcasts or wherever you get

585
00:31:18,440 --> 00:31:19,160
Speaker 2: your podcasts