1
00:00:03,200 --> 00:00:07,880
Speaker 1: Media Greetings, greetums, hello, and salutations. I'm ed Zytron and

2
00:00:07,920 --> 00:00:18,440
Speaker 1: this is your better offline monologue for the week. Earlier

3
00:00:18,520 --> 00:00:21,360
Speaker 1: this week, Bloomberg reported and I quote, that Meta is

4
00:00:21,400 --> 00:00:24,840
Speaker 1: developing plans for a cloud infrastructure business that will sell

5
00:00:24,880 --> 00:00:27,600
Speaker 1: access to its AI computing power and models. A few

6
00:00:27,600 --> 00:00:31,080
Speaker 1: weeks after, CEO Mark Zuckerberg said its annual shareholders meeting

7
00:00:31,080 --> 00:00:34,120
Speaker 1: that he thought that Meta had a use for its compute,

8
00:00:34,159 --> 00:00:36,959
Speaker 1: but it could sell it if it didn't. I genuinely

9
00:00:37,000 --> 00:00:38,840
Speaker 1: love that he said that. It's like, ah, I think

10
00:00:38,920 --> 00:00:41,280
Speaker 1: I spent one hundred billion dollars for a good reason.

11
00:00:41,760 --> 00:00:45,360
Speaker 1: Don't know yet though, And there's no other sensible way

12
00:00:45,400 --> 00:00:48,280
Speaker 1: at least to interpret this news other than that Meta

13
00:00:48,360 --> 00:00:52,240
Speaker 1: is admitting that it overbuilt its GPU capacity after five

14
00:00:52,280 --> 00:00:55,520
Speaker 1: different reorganizations of its AI team, a fourteen billion dollar

15
00:00:55,600 --> 00:00:59,360
Speaker 1: acquisition of Scalai for no imaginable reason, multiple versions of

16
00:00:59,400 --> 00:01:03,240
Speaker 1: their loath some pervert glasses and celebrity faced AI chatbots

17
00:01:03,240 --> 00:01:05,480
Speaker 1: that have led an old man with dementia to his death,

18
00:01:05,680 --> 00:01:07,280
Speaker 1: and I'm not kidding. I'll link you to that great

19
00:01:07,360 --> 00:01:11,680
Speaker 1: Jeff Howitz piece from Reuters. In reality, Meta never had

20
00:01:11,720 --> 00:01:14,560
Speaker 1: any reason to get into AI other than the fact

21
00:01:14,600 --> 00:01:17,000
Speaker 1: that Zuckerberg is a directionless oath that can never be

22
00:01:17,040 --> 00:01:20,039
Speaker 1: fired from the company surrounded by Loadso yes, men like

23
00:01:20,120 --> 00:01:23,200
Speaker 1: bos Adam Mursky and Havier Olivan, who cater to his

24
00:01:23,280 --> 00:01:25,840
Speaker 1: every women, say yes to any idea, no matter how

25
00:01:25,880 --> 00:01:29,680
Speaker 1: goddamn stupid they are. Every time I hear from somebody

26
00:01:29,720 --> 00:01:32,560
Speaker 1: at Meta, they vibrate with sadness and despair, lost in

27
00:01:32,600 --> 00:01:35,399
Speaker 1: a company that exists as a large social experiment to

28
00:01:35,400 --> 00:01:37,800
Speaker 1: try and conjure up mora ad revenue for a platform

29
00:01:37,800 --> 00:01:41,520
Speaker 1: designed to engage far more than build any meaningful connections.

30
00:01:42,319 --> 00:01:44,720
Speaker 1: AI balls and the recently can cast argue that this

31
00:01:44,800 --> 00:01:47,720
Speaker 1: is actually a smart move because now Meta can monetize

32
00:01:47,720 --> 00:01:50,560
Speaker 1: its AI capex, the kind of ridiculous thing you say

33
00:01:50,560 --> 00:01:53,200
Speaker 1: when you like big tech are fundamentally out of ideas

34
00:01:53,200 --> 00:01:56,840
Speaker 1: and horribly desperate. Meta is doing this because Meta has

35
00:01:56,920 --> 00:02:00,440
Speaker 1: no idea what to do other than to try something anything,

36
00:02:00,520 --> 00:02:04,400
Speaker 1: to make some money out of these goddamn GPUs. The

37
00:02:04,440 --> 00:02:06,560
Speaker 1: real question is who Meta will sell to and how

38
00:02:06,640 --> 00:02:09,600
Speaker 1: much it will sell. Epoch estimates that it has around

39
00:02:09,600 --> 00:02:12,440
Speaker 1: one point sixty one gigawatts of total capacity, as is

40
00:02:12,440 --> 00:02:14,440
Speaker 1: the fourth quarter of twenty twenty five. And if that's

41
00:02:14,480 --> 00:02:17,040
Speaker 1: the case, it's invariable that it gets snapped up by

42
00:02:17,080 --> 00:02:19,480
Speaker 1: that Anthropic or open AI. And by the way, when

43
00:02:19,480 --> 00:02:21,840
Speaker 1: that happens, we're at the top. We're done. Wrap this

44
00:02:21,919 --> 00:02:25,160
Speaker 1: bad boy up. And if it doesn't do that, I'm

45
00:02:25,200 --> 00:02:27,400
Speaker 1: not sure what the demand looks like outside of those

46
00:02:27,400 --> 00:02:30,200
Speaker 1: two companies, who I estimate take up anywhere from seventy

47
00:02:30,240 --> 00:02:34,400
Speaker 1: percent to eighty five percent of all global AI compute. Meanwhile,

48
00:02:34,400 --> 00:02:37,120
Speaker 1: the information reports that Invidia is offering and I quote,

49
00:02:37,280 --> 00:02:40,560
Speaker 1: financial backstops to young cloud providers that renteds GPUs in

50
00:02:40,600 --> 00:02:43,440
Speaker 1: exchange for a share of their revenues, a euphemism for

51
00:02:43,520 --> 00:02:47,320
Speaker 1: literally paying their customers to buy their stuff. This is

52
00:02:47,360 --> 00:02:50,760
Speaker 1: our direct admission that Innvidia doesn't see there's a divers

53
00:02:50,840 --> 00:02:53,440
Speaker 1: or rigorous demand for AI compute. Not. The only way

54
00:02:53,480 --> 00:02:55,840
Speaker 1: that these horrible little neoclouds can ever run a business

55
00:02:55,880 --> 00:02:58,359
Speaker 1: renting out GPUs is by letting them latch onto the

56
00:02:58,360 --> 00:03:02,640
Speaker 1: side of Nvidia like a parasite. It's blatantly obvious the

57
00:03:02,680 --> 00:03:05,800
Speaker 1: demand doesn't exist at scale for AI compute, that the

58
00:03:05,800 --> 00:03:09,320
Speaker 1: only real customers are two bulbous fail suns Anthropic and

59
00:03:09,320 --> 00:03:12,000
Speaker 1: open AI that lose tens of billions of dollars a

60
00:03:12,080 --> 00:03:14,920
Speaker 1: year with no path to profitability. This is not a

61
00:03:14,960 --> 00:03:18,639
Speaker 1: real industry, but a reflection of the desperation and directionlessness

62
00:03:18,800 --> 00:03:21,160
Speaker 1: of a tech industry that's run out of hypergrowth ideas

63
00:03:21,280 --> 00:03:23,639
Speaker 1: after giving up on investing at the seed stage and

64
00:03:23,680 --> 00:03:27,520
Speaker 1: chasing out those who actually build real things. This industry

65
00:03:27,560 --> 00:03:30,760
Speaker 1: deserves the punishment that's coming for it. The debt fueled

66
00:03:30,800 --> 00:03:34,840
Speaker 1: hyperscaler capex, no it loads, refused cash dump has pumped

67
00:03:34,840 --> 00:03:37,600
Speaker 1: a global stock market and left the world invested in

68
00:03:37,640 --> 00:03:41,400
Speaker 1: an impossible payoff of trillions of dollars of revenue well

69
00:03:41,800 --> 00:03:45,600
Speaker 1: that will never exist. Everybody involved in inflating this bubble

70
00:03:45,600 --> 00:03:48,960
Speaker 1: should be ashamed of themselves. Everybody who refused to heed

71
00:03:49,000 --> 00:03:51,320
Speaker 1: my warnings in favor of the mythologies of Amazon Web

72
00:03:51,400 --> 00:03:53,680
Speaker 1: services in Uber should know that they are partly to

73
00:03:53,720 --> 00:03:56,560
Speaker 1: blame for the chaos that will follow. This was blatantly

74
00:03:56,680 --> 00:04:00,200
Speaker 1: obvious years ago. It was obvious before I started this. Shh,

75
00:04:00,480 --> 00:04:03,680
Speaker 1: it's been so obvious for so long. I've got better

76
00:04:03,720 --> 00:04:06,320
Speaker 1: at this, I've got more nuanced in my arguments, but

77
00:04:06,360 --> 00:04:09,280
Speaker 1: my arguments have stayed the same. There is no business

78
00:04:09,280 --> 00:04:12,520
Speaker 1: model in generative AI, there is no way of serving

79
00:04:12,560 --> 00:04:16,000
Speaker 1: inference at scale in a way that is consistently profitable.

80
00:04:16,279 --> 00:04:18,279
Speaker 1: And people will argue about this, and people will have

81
00:04:18,400 --> 00:04:21,120
Speaker 1: the little gnatterings on their various blogs. I won't read them.

82
00:04:21,120 --> 00:04:24,760
Speaker 1: I've got stuff to do. Sorry, But the truth is

83
00:04:24,800 --> 00:04:28,240
Speaker 1: that everything I'm saying has been coming true. I've been

84
00:04:28,279 --> 00:04:30,320
Speaker 1: trying to warn people, all of you. I know I'm

85
00:04:30,360 --> 00:04:32,440
Speaker 1: speaking to an audience that kind of gets it already.

86
00:04:32,720 --> 00:04:34,440
Speaker 1: All of you have been sticking with me, and I

87
00:04:34,480 --> 00:04:36,640
Speaker 1: love it. I was just on CNBC, did a great

88
00:04:36,720 --> 00:04:38,839
Speaker 1: hit there or linked to it in the notes. Hate

89
00:04:38,880 --> 00:04:40,400
Speaker 1: to link to Twitter, but it's one of the only

90
00:04:40,400 --> 00:04:43,640
Speaker 1: places I can actually upload it with again, dmcaid, but

91
00:04:43,680 --> 00:04:45,240
Speaker 1: I think you're really gonna like that interview. And the

92
00:04:45,240 --> 00:04:48,320
Speaker 1: fact that the mainstream is finally opening up to these

93
00:04:48,360 --> 00:04:51,039
Speaker 1: stories is a sign that we're wrapping this up, and

94
00:04:51,080 --> 00:04:54,680
Speaker 1: they're trying to kind of layer on the news and

95
00:04:54,760 --> 00:04:57,719
Speaker 1: layer on the stories that say, hey, we didn't fall

96
00:04:57,760 --> 00:04:59,680
Speaker 1: behind this, were on this, and good on them for

97
00:04:59,760 --> 00:05:03,520
Speaker 1: doing again, albeit a little too late. In fact, I

98
00:05:03,520 --> 00:05:06,119
Speaker 1: think all of the kind of bubble discussion is too little,

99
00:05:06,120 --> 00:05:11,599
Speaker 1: too late, And now Clammy Scammy, Clamate, Klamatha, Clamuel Clemantha.

100
00:05:11,760 --> 00:05:14,560
Speaker 1: We love Clamia, Weird, don't like Clamantha. He wants to

101
00:05:14,560 --> 00:05:16,840
Speaker 1: give five percent of Open AI to the US government,

102
00:05:16,880 --> 00:05:20,360
Speaker 1: I assume in exchange for money. And I know some

103
00:05:20,400 --> 00:05:21,960
Speaker 1: of you are going to say that's a line up

104
00:05:22,000 --> 00:05:24,679
Speaker 1: to a bailout. No, it isn't. It will require congressional approval.

105
00:05:24,760 --> 00:05:27,400
Speaker 1: And this is the time when AI is one of

106
00:05:27,440 --> 00:05:30,240
Speaker 1: the least popular things in the world with voters. And

107
00:05:30,279 --> 00:05:33,760
Speaker 1: also no one wants to give forty billion dollars to

108
00:05:33,800 --> 00:05:38,440
Speaker 1: goddamn sam Altman. All signs to this, All signs point

109
00:05:38,480 --> 00:05:41,120
Speaker 1: to this. I mean blowing up sometime, I think in

110
00:05:41,120 --> 00:05:42,720
Speaker 1: the next six months. And I feel like I've been

111
00:05:42,720 --> 00:05:45,880
Speaker 1: saying that for about six months. But we're seeing the

112
00:05:45,920 --> 00:05:50,280
Speaker 1: desperation start and we're hearing these conversations happen. I think

113
00:05:50,279 --> 00:05:52,360
Speaker 1: things are going to get weird. Things are going to

114
00:05:52,400 --> 00:05:55,039
Speaker 1: get very weird. And next week we've got a great

115
00:05:55,080 --> 00:05:57,880
Speaker 1: joint episode with the Trash Future crew talking about Massiyoshi's

116
00:05:57,880 --> 00:06:00,840
Speaker 1: son's Goosebild shareholder speech and the collapse of the AI

117
00:06:01,000 --> 00:06:04,640
Speaker 1: hype cycle. You're gonna love it, a promise. Thanks for listening.

118
00:06:05,000 --> 00:06:05,640
Speaker 1: Zitron out