1
00:00:03,160 --> 00:00:07,960
Speaker 1: Media. Greetings and salutations, traveler. I'm your host ed Xetron

2
00:00:08,039 --> 00:00:15,720
Speaker 1: and this is your better offline monologue for the week. Also,

3
00:00:18,239 --> 00:00:21,000
Speaker 1: blood is running through the stock market as everybody realizes

4
00:00:21,040 --> 00:00:23,680
Speaker 1: that once shit I've been saying since twenty twenty four,

5
00:00:23,960 --> 00:00:27,280
Speaker 1: the generative AI costs too much, loses everybody money, and

6
00:00:27,320 --> 00:00:29,120
Speaker 1: doesn't have the growth potential to make any of this

7
00:00:29,240 --> 00:00:32,760
Speaker 1: shit worth it. Yet this has been a week for

8
00:00:32,800 --> 00:00:36,680
Speaker 1: the more sophisticated haters, with Oracle stock plummeting on the

9
00:00:36,720 --> 00:00:38,760
Speaker 1: news that it had to raise another forty five to

10
00:00:38,760 --> 00:00:42,080
Speaker 1: fifty five billion dollars to fund data centers for open Ai,

11
00:00:42,360 --> 00:00:45,040
Speaker 1: which timed poorly with a story running that in video

12
00:00:45,120 --> 00:00:47,800
Speaker 1: was no longer doing its one hundred billion dollar investment

13
00:00:47,840 --> 00:00:50,480
Speaker 1: in open Ai, something you will know was the case

14
00:00:50,560 --> 00:00:52,640
Speaker 1: because it was on this podcast months ago and has

15
00:00:52,720 --> 00:00:57,040
Speaker 1: been multiple times since. To catch you one. Last September,

16
00:00:57,080 --> 00:01:00,520
Speaker 1: Oracle announced the three hundred billion dollar, five year long

17
00:01:00,600 --> 00:01:03,800
Speaker 1: deal to provide compute to open Ai, powered by four

18
00:01:03,800 --> 00:01:06,720
Speaker 1: point five gigawatts of compute capacity that at the time

19
00:01:06,760 --> 00:01:09,600
Speaker 1: and at this time did not and does not exist

20
00:01:09,959 --> 00:01:12,400
Speaker 1: as far as I can tell, only two hundred megawats

21
00:01:12,400 --> 00:01:14,959
Speaker 1: of Stargate Abilene, a project that's been a work in

22
00:01:15,000 --> 00:01:18,399
Speaker 1: progress since twenty twenty four, appears to be complete, and

23
00:01:18,480 --> 00:01:20,319
Speaker 1: to my knowledge as well, there's only like two hundred

24
00:01:20,360 --> 00:01:22,640
Speaker 1: megawats of power out there. They need a lot more

25
00:01:22,680 --> 00:01:24,240
Speaker 1: than that. That's meant to be a one point two

26
00:01:24,280 --> 00:01:29,560
Speaker 1: gigawat center. Not great now. Several other data center projects

27
00:01:29,560 --> 00:01:33,320
Speaker 1: are allegedly in the works too, Stargate Shackleford, Texas and Wisconsin,

28
00:01:33,560 --> 00:01:38,240
Speaker 1: Stargate New Mexico, and theoretically at least Stargate Michigan through

29
00:01:38,240 --> 00:01:41,440
Speaker 1: Blue Owl, who invested in basically every Stargate project and

30
00:01:41,480 --> 00:01:45,800
Speaker 1: pulled out of Michigan due to financing conditions. Souring Oracle

31
00:01:45,840 --> 00:01:49,320
Speaker 1: had previously raised fifty six billion dollars in bonds and debt,

32
00:01:49,360 --> 00:01:52,920
Speaker 1: with thirty eight billion of that specifically yearmarked for Shackleford

33
00:01:52,920 --> 00:01:56,600
Speaker 1: and Wisconsin. That debt is payable over four years, with

34
00:01:56,720 --> 00:01:59,640
Speaker 1: two one year extensions and payments for interest only until

35
00:01:59,640 --> 00:02:04,200
Speaker 1: construc duction is complete. One little problem, I'm afraid. According

36
00:02:04,200 --> 00:02:07,320
Speaker 1: to Jerome Darling at TD Cohen, it's upwards of forty

37
00:02:07,320 --> 00:02:09,720
Speaker 1: two million dollars in megawatt to build a data center.

38
00:02:09,960 --> 00:02:12,000
Speaker 1: Completing these data centers is going to cost at least

39
00:02:12,000 --> 00:02:13,840
Speaker 1: one hundred and eighty nine billion dollars for the four

40
00:02:13,880 --> 00:02:17,040
Speaker 1: and a half gigawats an Oracle was and is short

41
00:02:17,080 --> 00:02:20,320
Speaker 1: by quite some measure, so it sold twenty five billion

42
00:02:20,360 --> 00:02:22,680
Speaker 1: dollars more in bonds and plans to sell another twenty

43
00:02:22,720 --> 00:02:25,959
Speaker 1: billion dollars of stock in an at the money share sale,

44
00:02:26,280 --> 00:02:28,920
Speaker 1: literally printing new shares and dumping them onto the market

45
00:02:29,080 --> 00:02:32,360
Speaker 1: in intervals set by the bankers behind the deal. Mathematicians

46
00:02:32,400 --> 00:02:34,280
Speaker 1: in the audience have likely run the numbers at this

47
00:02:34,360 --> 00:02:36,760
Speaker 1: point and worked out that this is only one hundred

48
00:02:36,760 --> 00:02:38,960
Speaker 1: and one billion dollars, which is less than one hundred

49
00:02:38,960 --> 00:02:41,480
Speaker 1: and eighty nine billion dollars. May seem obvious, but I

50
00:02:41,480 --> 00:02:44,560
Speaker 1: don't know. The fucking markets don't appear to be able

51
00:02:44,560 --> 00:02:47,919
Speaker 1: to count. Now. The other problem Oracle has is that

52
00:02:48,080 --> 00:02:50,560
Speaker 1: the only way it can pay these loans off is

53
00:02:50,600 --> 00:02:52,520
Speaker 1: if open Ai is able to pay them, and I

54
00:02:52,600 --> 00:02:55,160
Speaker 1: need to be clear how difficult that will be. That

55
00:02:55,200 --> 00:02:58,800
Speaker 1: three hundred billion dollar, five year long compute deal guarantees

56
00:02:58,840 --> 00:03:01,239
Speaker 1: that at least one year will be over eighty billion

57
00:03:01,280 --> 00:03:04,560
Speaker 1: dollars in payments. For some context, the entire revenue of

58
00:03:04,600 --> 00:03:07,520
Speaker 1: MICROSOFTS as your cloud in twenty twenty five was seventy

59
00:03:07,560 --> 00:03:11,440
Speaker 1: five billion dollars. How does open ai pay that, especially

60
00:03:11,480 --> 00:03:14,919
Speaker 1: when Oracles' build capacity but they don't have the money.

61
00:03:14,960 --> 00:03:18,240
Speaker 1: They only allegedly made thirteen billion dollars last year and

62
00:03:18,280 --> 00:03:20,720
Speaker 1: lost nine billion dollars on top of that. How the

63
00:03:20,760 --> 00:03:22,960
Speaker 1: hell do they pay this? I don't know, but everyone

64
00:03:22,960 --> 00:03:25,880
Speaker 1: seems to think they will, or at least they did now.

65
00:03:25,960 --> 00:03:27,880
Speaker 1: In video, of course, went out to say that they

66
00:03:27,919 --> 00:03:30,200
Speaker 1: would absolutely be investing a lot of money in open

67
00:03:30,240 --> 00:03:34,320
Speaker 1: AI's supposed upcoming round, this one hundred billion dollar one.

68
00:03:34,520 --> 00:03:36,760
Speaker 1: Different to this one hundred billion dollar deal they were

69
00:03:36,760 --> 00:03:39,920
Speaker 1: talking about. Open ai is currently raising one hundred billion

70
00:03:39,960 --> 00:03:43,720
Speaker 1: dollars allegedly, but in video is only putting twenty billion

71
00:03:43,760 --> 00:03:47,320
Speaker 1: dollars in. And Jensen went on CNBC with Jim Kramer

72
00:03:47,360 --> 00:03:50,360
Speaker 1: and had the most bizarre conversation I'll link to it

73
00:03:50,400 --> 00:03:53,480
Speaker 1: in the episode notes. He was just saying, oh, yes,

74
00:03:53,680 --> 00:03:56,400
Speaker 1: open ai is wonderful, cutting over Jim Kramer's Jim Kramer

75
00:03:56,440 --> 00:03:58,800
Speaker 1: just said they going, well, you know, we love your deal.

76
00:03:58,920 --> 00:04:01,080
Speaker 1: Is to base to bash this deal ever, sounding like

77
00:04:01,120 --> 00:04:04,520
Speaker 1: Sylvester the Cat, and Jensen is just cutting over him

78
00:04:04,520 --> 00:04:07,520
Speaker 1: saying what open a ey is amazing. No drama, no

79
00:04:07,640 --> 00:04:10,400
Speaker 1: drama at all. And let me tell you something. The

80
00:04:10,480 --> 00:04:13,320
Speaker 1: reason he's saying no drama is because open ai also

81
00:04:13,480 --> 00:04:16,800
Speaker 1: leaked a thing saying that that Sam Altman and open

82
00:04:16,800 --> 00:04:20,080
Speaker 1: Ai were unhappy with in video chips, it's all so good.

83
00:04:20,279 --> 00:04:22,560
Speaker 1: This is this is the kind of stuff that happens

84
00:04:22,560 --> 00:04:27,039
Speaker 1: when things are going well and well. Oracles saw this,

85
00:04:27,080 --> 00:04:28,839
Speaker 1: and Oracles saw this back and forth, and they thought,

86
00:04:28,880 --> 00:04:30,799
Speaker 1: we need to help, we need to help the markets.

87
00:04:31,080 --> 00:04:32,919
Speaker 1: So they put out a tweet and I'm just going

88
00:04:33,000 --> 00:04:35,960
Speaker 1: to read it to you. The in video open Ai

89
00:04:36,120 --> 00:04:39,120
Speaker 1: deal has zero impact on our financial relationship with open Ai.

90
00:04:39,480 --> 00:04:42,520
Speaker 1: We remain highly confident in open aiy's ability to raise

91
00:04:42,560 --> 00:04:46,400
Speaker 1: funds and meet its commitments. Sadly, Oracles, the in video

92
00:04:46,440 --> 00:04:49,080
Speaker 1: open Ai deal has zero impact on our financial relationship.

93
00:04:49,120 --> 00:04:51,120
Speaker 1: T shirt appears to have people asking them a lot

94
00:04:51,160 --> 00:04:55,560
Speaker 1: of questions already answered by the shirt jokes aside, that's

95
00:04:55,600 --> 00:04:58,240
Speaker 1: the kind of thing that Alex Wilhelm tech reporter said.

96
00:04:58,240 --> 00:05:01,360
Speaker 1: It's bank run language. You don't that will always have

97
00:05:01,480 --> 00:05:04,479
Speaker 1: the opposite effect saying, well, it's I said this on

98
00:05:04,520 --> 00:05:06,200
Speaker 1: an interview recently. It's like walking into the room and

99
00:05:06,240 --> 00:05:09,960
Speaker 1: saying I didn't fucking kill anyone. People are gonna people

100
00:05:10,000 --> 00:05:12,320
Speaker 1: are probably gonna ask why you need to bring that

101
00:05:12,400 --> 00:05:15,440
Speaker 1: up at this time. The other part of that statement

102
00:05:15,480 --> 00:05:18,400
Speaker 1: that got me was we're highly confident in open AI's

103
00:05:18,400 --> 00:05:22,200
Speaker 1: ability to raise funds. They're not paying this out of

104
00:05:22,200 --> 00:05:26,120
Speaker 1: cash flow, so Oracle is just depending their entire existence

105
00:05:26,160 --> 00:05:28,920
Speaker 1: on open AI's ability to raise some venture capitalists, it's

106
00:05:28,960 --> 00:05:32,320
Speaker 1: all very good, it's all very very good, and it

107
00:05:32,400 --> 00:05:35,560
Speaker 1: all times very very poorly, with a TD Cohen analyst

108
00:05:35,680 --> 00:05:38,360
Speaker 1: letter that suggested Oracle would have to fire upwards of

109
00:05:38,400 --> 00:05:41,760
Speaker 1: twenty five thousand people and potentially sell its valuable set

110
00:05:42,080 --> 00:05:45,600
Speaker 1: SNA healthcare IT infrastructure just to make it through. And

111
00:05:45,640 --> 00:05:48,760
Speaker 1: again they paid twenty eight billion dollars with Serna, and

112
00:05:48,800 --> 00:05:50,799
Speaker 1: TD Cohen said that would be eight to ten billion

113
00:05:50,800 --> 00:05:53,360
Speaker 1: dollars worth of cash flow, again nowhere near one hundred

114
00:05:53,360 --> 00:05:58,080
Speaker 1: and eighty nine billion dollars. Look, look, I need people

115
00:05:58,120 --> 00:06:01,279
Speaker 1: to realize that Oracle cannot afford this out. It's committed

116
00:06:01,279 --> 00:06:04,080
Speaker 1: to building ten gigawats of capacity for open Ai, meaning

117
00:06:04,080 --> 00:06:05,719
Speaker 1: that on top of the one hundred and eighty nine

118
00:06:05,720 --> 00:06:09,160
Speaker 1: billion dollars it needs already. Oracle needs another two hundred

119
00:06:09,240 --> 00:06:12,839
Speaker 1: and thirty one billion dollars just to build those data centers.

120
00:06:13,120 --> 00:06:15,520
Speaker 1: And on its last earnings, Oracle had a cash flow

121
00:06:15,560 --> 00:06:22,120
Speaker 1: of negative thirteen billion dollars. It's it's not looking very positive.

122
00:06:22,440 --> 00:06:25,359
Speaker 1: And yes, Oracle also just bought TikTok, a business that

123
00:06:25,600 --> 00:06:28,960
Speaker 1: also but it loses billions and billions of dollars. And yes,

124
00:06:29,279 --> 00:06:32,599
Speaker 1: Larry Ellison's wealth is almost entirely tied up in Oracle stock,

125
00:06:32,760 --> 00:06:35,520
Speaker 1: which is currently hurtling towards hell with a velocity reserve

126
00:06:35,560 --> 00:06:39,360
Speaker 1: for Thatcher and Reagan, I know you're gonna listen to this,

127
00:06:39,400 --> 00:06:40,720
Speaker 1: and I hear from a lot of you. No matter

128
00:06:40,760 --> 00:06:43,400
Speaker 1: how many times I say this, everyone's like, oh, the

129
00:06:43,440 --> 00:06:46,440
Speaker 1: two deals are not about it's not about making money.

130
00:06:46,680 --> 00:06:50,400
Speaker 1: It's about political influence. Wow, you're the first fucking person

131
00:06:50,480 --> 00:06:52,240
Speaker 1: to tell me that. No one has ever told me

132
00:06:52,279 --> 00:06:55,200
Speaker 1: that other than the first one hundred and fifty fucking people.

133
00:06:55,440 --> 00:06:59,040
Speaker 1: Not an original thought and also not a particularly effective one.

134
00:06:59,120 --> 00:07:01,919
Speaker 1: The algorithm on ticketok is already shitting its pants because

135
00:07:02,279 --> 00:07:04,760
Speaker 1: Oracle can't stop messing with it because they want to

136
00:07:04,760 --> 00:07:08,200
Speaker 1: do the propaganda thing. And guess how TikTok makes money

137
00:07:08,800 --> 00:07:11,520
Speaker 1: ads and how does TikTok show ads? Well, they keep

138
00:07:11,560 --> 00:07:14,080
Speaker 1: you on the app with the algorithm. How do you

139
00:07:14,120 --> 00:07:17,360
Speaker 1: think this is gonna work? I know it all feels

140
00:07:17,360 --> 00:07:19,640
Speaker 1: a little ridiculous, but there really is a future in

141
00:07:19,640 --> 00:07:24,200
Speaker 1: which Oracle collapses. Larry Ellison's stock is entirely tied up

142
00:07:24,200 --> 00:07:26,200
Speaker 1: in Oracle, and I just did a story called the

143
00:07:26,320 --> 00:07:28,680
Speaker 1: Hater's God to Oracle in my premium newsletter that goes

144
00:07:28,720 --> 00:07:31,040
Speaker 1: through this. But this is a company in a company,

145
00:07:31,240 --> 00:07:34,480
Speaker 1: I'm just keeping it a company in decline, and they

146
00:07:34,520 --> 00:07:37,840
Speaker 1: are really looking quite shaky. And even if they weren't

147
00:07:37,840 --> 00:07:40,240
Speaker 1: in decline, the amount of money they're trying to spend

148
00:07:40,280 --> 00:07:44,400
Speaker 1: here is genuinely impossible for them to afford. They are

149
00:07:44,400 --> 00:07:46,520
Speaker 1: going to have to stress the debt markets the levels

150
00:07:46,560 --> 00:07:49,400
Speaker 1: they've never seen, and even then they cannot afford the

151
00:07:49,400 --> 00:07:51,720
Speaker 1: payments unless open Ai can pay them more than anyone

152
00:07:51,800 --> 00:07:56,960
Speaker 1: has ever paid for Cloud ever. Ah nah, I'm excited.

153
00:07:57,040 --> 00:08:00,000
Speaker 1: I own to hold stock, and until we know more,

154
00:08:00,280 --> 00:08:10,320
Speaker 1: I really only have one question. Is that good? H