1
00:00:02,720 --> 00:00:13,960
Speaker 1: Bloomberg Audio Studios, Podcasts, radio News.

2
00:00:18,520 --> 00:00:22,040
Speaker 2: Hello and welcome to another episode of The Odd Laws podcast.

3
00:00:22,120 --> 00:00:24,360
Speaker 3: I'm Joe Wisenthal and I'm Tracy Alloway.

4
00:00:24,720 --> 00:00:27,760
Speaker 2: Tracy, I love doing the podcast, but I'm kind of

5
00:00:27,840 --> 00:00:32,479
Speaker 2: thinking about becoming a PhD level AI researcher if I

6
00:00:32,520 --> 00:00:33,519
Speaker 2: do a career pivot.

7
00:00:33,600 --> 00:00:37,000
Speaker 3: It seems something. Have you seen?

8
00:00:37,920 --> 00:00:41,080
Speaker 2: Have you seen like the salaries supposedly I still don't

9
00:00:41,080 --> 00:00:43,640
Speaker 2: actually believe them, But have you seen like the salaries

10
00:00:44,080 --> 00:00:47,200
Speaker 2: and comp packages that supposedly Meta and a few others

11
00:00:47,240 --> 00:00:47,839
Speaker 2: are paying for it.

12
00:00:47,960 --> 00:00:50,800
Speaker 4: I have seen some of the headlines. I have also

13
00:00:50,960 --> 00:00:54,320
Speaker 4: learned a bunch of new words like exploding offer and

14
00:00:54,440 --> 00:00:58,800
Speaker 4: acquahireh exploding offer sounds dangerous. What you want is an

15
00:00:58,800 --> 00:01:00,840
Speaker 4: explosive offer that you can't turn down.

16
00:01:01,000 --> 00:01:02,840
Speaker 2: I take it the exploding hire is like one that

17
00:01:02,960 --> 00:01:03,880
Speaker 2: lasts like five minutes.

18
00:01:04,000 --> 00:01:05,839
Speaker 4: Yeah, so you have to make a decision right away,

19
00:01:05,959 --> 00:01:09,319
Speaker 4: and the company that you're currently working for can't counteroffer.

20
00:01:09,640 --> 00:01:12,680
Speaker 2: So I read these headlines about some engineer something getting

21
00:01:12,680 --> 00:01:14,559
Speaker 2: paid one hundred million or two hundred and fifty million.

22
00:01:14,920 --> 00:01:17,320
Speaker 2: I actually literally don't believe them, Like I actually literally

23
00:01:17,360 --> 00:01:19,880
Speaker 2: think that's fake notes. Really, no but I.

24
00:01:19,840 --> 00:01:21,560
Speaker 4: Kind of just think it's probably made up of like

25
00:01:21,640 --> 00:01:23,399
Speaker 4: all this weird equity structure.

26
00:01:23,760 --> 00:01:26,559
Speaker 2: But I'm sure it's not all upfront anyway. We sort

27
00:01:26,600 --> 00:01:28,360
Speaker 2: of seem to be in a moment and it's actually

28
00:01:28,360 --> 00:01:33,000
Speaker 2: not just AI. I would say journalism, finance, et cetera,

29
00:01:33,400 --> 00:01:37,080
Speaker 2: where it's like the sportsification of a lot of different industries,

30
00:01:37,240 --> 00:01:42,880
Speaker 2: individual talent, the demand for individual superstars, and anything doing

31
00:01:43,040 --> 00:01:43,479
Speaker 2: very well.

32
00:01:44,000 --> 00:01:46,440
Speaker 4: I have a bunch of questions when it comes specifically

33
00:01:46,480 --> 00:01:50,520
Speaker 4: to AI, such as what makes an AI talent because

34
00:01:50,520 --> 00:01:53,160
Speaker 4: I assume if you're Mark Zuckerberg and you're trying to

35
00:01:53,200 --> 00:01:56,360
Speaker 4: assemble your AI dream team or something, Okay, sure you

36
00:01:56,400 --> 00:01:59,240
Speaker 4: have like a level of technical insight, and maybe you

37
00:01:59,280 --> 00:02:02,280
Speaker 4: hear people talk talking about, oh, this one guy is fantastic,

38
00:02:03,040 --> 00:02:04,840
Speaker 4: But I assume it's not like you can look at

39
00:02:04,880 --> 00:02:08,400
Speaker 4: their like individual levels of code. You can't see what

40
00:02:08,400 --> 00:02:10,680
Speaker 4: they're doing on like a daily basis. I don't know.

41
00:02:11,080 --> 00:02:13,000
Speaker 2: I don't know either. I don't know anything about this stuff,

42
00:02:13,040 --> 00:02:14,960
Speaker 2: but I'm really excited to say we do have the

43
00:02:15,000 --> 00:02:17,600
Speaker 2: perfect guests, A couple of guys who are right in

44
00:02:17,639 --> 00:02:20,280
Speaker 2: the middle of all of this and also who have

45
00:02:20,400 --> 00:02:23,000
Speaker 2: sort of been ahead of the curve in terms of this,

46
00:02:23,440 --> 00:02:26,880
Speaker 2: like I said, the sportification of the industry. We're going

47
00:02:26,919 --> 00:02:29,799
Speaker 2: to be speaking with John Coogan and Jordie Hayes. They

48
00:02:29,840 --> 00:02:33,799
Speaker 2: are the co hosts of TBPN. It's a live show podcast.

49
00:02:33,840 --> 00:02:38,640
Speaker 2: It's become one of my favorite new media properties. It exists,

50
00:02:38,639 --> 00:02:40,880
Speaker 2: and I'm really excited we have them both in the

51
00:02:40,960 --> 00:02:43,600
Speaker 2: studio here with us today. So John and Jordi, thank

52
00:02:43,639 --> 00:02:45,440
Speaker 2: you so much for coming on aut Loots.

53
00:02:45,560 --> 00:02:46,320
Speaker 5: Thanks for having us.

54
00:02:46,400 --> 00:02:47,519
Speaker 3: We're so excited to be here.

55
00:02:47,720 --> 00:02:49,160
Speaker 2: I love by the way you guys like make these

56
00:02:49,200 --> 00:02:51,960
Speaker 2: like baseball cards every time, So it's so clever for

57
00:02:51,960 --> 00:02:54,560
Speaker 2: those who are not paying attention. Why do you sort

58
00:02:54,600 --> 00:02:57,280
Speaker 2: of give the high level overview of like what you

59
00:02:57,320 --> 00:03:00,919
Speaker 2: see happening in this like crazy talent for people who

60
00:03:00,919 --> 00:03:03,600
Speaker 2: don't know about this, what is actually going on? This

61
00:03:04,040 --> 00:03:07,400
Speaker 2: talent war for people who know how to do AI

62
00:03:07,560 --> 00:03:09,360
Speaker 2: or train a model or whatever it is.

63
00:03:09,680 --> 00:03:12,760
Speaker 5: Yeah, the League or the mag seven the serious teams

64
00:03:12,800 --> 00:03:16,400
Speaker 5: are all worth over a trillion dollars. Tesla's in a

65
00:03:16,440 --> 00:03:19,040
Speaker 5: different kind of boat some days, but pretty much there's

66
00:03:19,080 --> 00:03:22,840
Speaker 5: multiply seven trillion plus dollar companies now, so there is

67
00:03:22,840 --> 00:03:25,760
Speaker 5: an inordinate amount of money to flow around. And if

68
00:03:25,800 --> 00:03:29,520
Speaker 5: you think about investing zero point one percent of your

69
00:03:29,520 --> 00:03:33,240
Speaker 5: market cap to make your business potentially five percent better,

70
00:03:33,280 --> 00:03:35,880
Speaker 5: that's a trade you take all day. The number is staggering,

71
00:03:36,200 --> 00:03:41,040
Speaker 5: but now you're seeing companies pay hundreds of millions of dollars.

72
00:03:41,040 --> 00:03:42,960
Speaker 5: I think you mentioned that. You know, there's a lot

73
00:03:42,960 --> 00:03:45,200
Speaker 5: of debate over whether that was fake news or not.

74
00:03:45,640 --> 00:03:47,920
Speaker 5: They seem like they are very real offers and they

75
00:03:47,960 --> 00:03:49,920
Speaker 5: are indeed happening, and there's a whole bunch of different

76
00:03:49,920 --> 00:03:52,080
Speaker 5: ways that you can kind of underwrite that. But we

77
00:03:52,080 --> 00:03:55,600
Speaker 5: could start with going through just a little bit of

78
00:03:55,640 --> 00:03:58,360
Speaker 5: the history here of how we got into this place

79
00:03:58,440 --> 00:04:01,000
Speaker 5: or what these AI researchers are actually doing. I'm happy

80
00:04:01,040 --> 00:04:02,520
Speaker 5: to kind of answer any question.

81
00:04:03,000 --> 00:04:05,560
Speaker 4: Okay, well, why don't we start with the baseball analogy?

82
00:04:05,840 --> 00:04:10,960
Speaker 4: Then if I was collecting AI engineer cards, who's the

83
00:04:10,960 --> 00:04:13,360
Speaker 4: most valuable? Like who do I actually want to have?

84
00:04:13,560 --> 00:04:15,840
Speaker 4: And then secondly, what are the stats that are.

85
00:04:17,040 --> 00:04:18,960
Speaker 6: So taking a little bit of a step back talking

86
00:04:18,960 --> 00:04:22,320
Speaker 6: about kind of the sportsification of tech and business, which

87
00:04:22,360 --> 00:04:25,640
Speaker 6: has been a huge catalyst for our show. I think

88
00:04:25,680 --> 00:04:29,640
Speaker 6: we realized early on people would call TVPN the Sports

89
00:04:29,640 --> 00:04:32,920
Speaker 6: Center for Tech or some analogy like that. We were

90
00:04:32,960 --> 00:04:34,919
Speaker 6: a little bit it sounded cool. We didn't really know

91
00:04:34,920 --> 00:04:37,720
Speaker 6: what that meant. John and I don't watch sports at all,

92
00:04:37,920 --> 00:04:43,400
Speaker 6: but for basically, for basically like two decades now, I'm

93
00:04:43,440 --> 00:04:47,279
Speaker 6: in my late twenties, John mid thirties, and we've followed

94
00:04:47,320 --> 00:04:50,279
Speaker 6: tech in business the way that our college friends follow sports.

95
00:04:50,360 --> 00:04:56,960
Speaker 6: So in sports you have players, personalities, coaches, managers, leagues, teams,

96
00:04:57,120 --> 00:05:01,000
Speaker 6: and people obsess over all the details. Fully understood that,

97
00:05:01,040 --> 00:05:03,560
Speaker 6: I mean, I understood kind of the draw, but it

98
00:05:03,640 --> 00:05:05,560
Speaker 6: just was never for me. Whereas John and I would

99
00:05:05,560 --> 00:05:08,680
Speaker 6: pick up the newspaper as teenagers and we were tracking

100
00:05:08,720 --> 00:05:13,960
Speaker 6: the talent, the CEOs, the companies, the markets, the industries,

101
00:05:14,520 --> 00:05:17,560
Speaker 6: and it's just something that we obsessed over. And so

102
00:05:18,360 --> 00:05:21,120
Speaker 6: this year has been amazing. As these AI researchers have

103
00:05:21,279 --> 00:05:24,560
Speaker 6: been getting these sort of superstar max contracts, which is

104
00:05:24,560 --> 00:05:27,039
Speaker 6: what we would call them, we would start joking on

105
00:05:27,080 --> 00:05:29,400
Speaker 6: the show and be like, look, this guy just went.

106
00:05:29,279 --> 00:05:29,840
Speaker 3: Over to Meta.

107
00:05:29,880 --> 00:05:33,040
Speaker 6: It's probably you know, four year contracts, you know, one year, Cliff.

108
00:05:33,240 --> 00:05:35,200
Speaker 6: And it just got more and more and more real

109
00:05:35,760 --> 00:05:39,960
Speaker 6: as the kind of demand for world class AI researchers

110
00:05:40,000 --> 00:05:43,680
Speaker 6: completely outstripped the supply, and we actually put out something

111
00:05:44,120 --> 00:05:46,200
Speaker 6: a couple of days ago called the Metas List, and

112
00:05:46,320 --> 00:05:48,440
Speaker 6: John can go a little bit more into into the

113
00:05:48,520 --> 00:05:51,039
Speaker 6: name behind that. It was kind of our take on

114
00:05:51,080 --> 00:05:53,600
Speaker 6: the Midas List, which is, we built a list of

115
00:05:54,160 --> 00:05:57,200
Speaker 6: top roughly one hundred AI researchers and rank them on

116
00:05:57,240 --> 00:05:58,960
Speaker 6: a bunch of different factors. And so when you talk

117
00:05:59,000 --> 00:06:03,640
Speaker 6: about what goes into what makes a great AI researcher,

118
00:06:03,920 --> 00:06:06,280
Speaker 6: there's a bunch of different factors. One that you can

119
00:06:06,320 --> 00:06:08,800
Speaker 6: get right into the numbers with is like citations. So

120
00:06:09,240 --> 00:06:12,479
Speaker 6: if somebody is a researcher, they've been publishing, you know, studies, papers,

121
00:06:12,520 --> 00:06:14,560
Speaker 6: et cetera for quite a while at this point, and

122
00:06:14,600 --> 00:06:18,400
Speaker 6: you can just see quantitatively what their contributions have been

123
00:06:18,440 --> 00:06:19,080
Speaker 6: to the industry.

124
00:06:19,080 --> 00:06:21,560
Speaker 3: And so that's like a good starting point to understand.

125
00:06:21,600 --> 00:06:26,599
Speaker 6: And historically Elon even said earlier this week, honestly, it

126
00:06:26,680 --> 00:06:28,600
Speaker 6: was like an hour after we posted the Metaalist.

127
00:06:28,680 --> 00:06:30,160
Speaker 3: Probably unrelated, but who knows.

128
00:06:30,600 --> 00:06:34,320
Speaker 6: He's basically saying that AI researchers a and engineers, We're

129
00:06:34,360 --> 00:06:36,560
Speaker 6: just going to call them engineers now, and that makes

130
00:06:36,600 --> 00:06:39,279
Speaker 6: a lot of sense for someone like Elon to do

131
00:06:39,360 --> 00:06:43,680
Speaker 6: as somebody who's always been very engineering focused, less focused

132
00:06:43,720 --> 00:06:48,320
Speaker 6: on you know, entirely net new innovation. More so, how

133
00:06:48,320 --> 00:06:52,600
Speaker 6: do we make the best possible versions of products that exists.

134
00:06:52,640 --> 00:06:55,440
Speaker 6: And it's not to say that he hasn't innovated in

135
00:06:55,480 --> 00:06:58,839
Speaker 6: a bunch of different ways, but it is a really

136
00:06:58,880 --> 00:07:01,400
Speaker 6: wild moment in time, and it's been a fun moment

137
00:07:01,520 --> 00:07:05,360
Speaker 6: because historically you hear about this minas list investor, you know,

138
00:07:05,640 --> 00:07:08,240
Speaker 6: making two billion dollars of carry on some deal, or

139
00:07:08,279 --> 00:07:10,600
Speaker 6: this founder you know I p O ing. Today we

140
00:07:10,680 --> 00:07:13,120
Speaker 6: have you know, the Pigma I p O and uh,

141
00:07:13,360 --> 00:07:15,400
Speaker 6: you know there's a bunch of people that are going

142
00:07:15,480 --> 00:07:18,160
Speaker 6: to make billions of dollars there, But you don't hear

143
00:07:18,240 --> 00:07:22,160
Speaker 6: about the hundredth engineer that was hired, yeah, making you

144
00:07:22,240 --> 00:07:26,400
Speaker 6: start a one hundred million dollars signing bonus. And all

145
00:07:26,440 --> 00:07:28,400
Speaker 6: of this makes a ton of sense in the context

146
00:07:28,400 --> 00:07:30,960
Speaker 6: of what John said, because you look metas up I

147
00:07:31,200 --> 00:07:33,960
Speaker 6: checked this morning one hundred and ninety five billion dollars

148
00:07:34,280 --> 00:07:37,520
Speaker 6: new market cap and think about how many one hundred

149
00:07:37,520 --> 00:07:40,520
Speaker 6: million dollars signing bonuses you can make against that kind

150
00:07:40,520 --> 00:07:41,080
Speaker 6: of market cap.

151
00:07:41,120 --> 00:07:44,160
Speaker 3: Should we answer your question, yes, who the who's the bus?

152
00:07:44,240 --> 00:07:49,000
Speaker 5: Who's the whale? To have the who is the light whale?

153
00:07:51,240 --> 00:07:56,280
Speaker 4: Don't say anything about whales or whale products or hunting.

154
00:07:56,160 --> 00:07:59,840
Speaker 2: Large is the way.

155
00:08:00,240 --> 00:08:02,760
Speaker 5: Perhaps he is the Michael Jordan of AI researcher, but

156
00:08:03,320 --> 00:08:06,080
Speaker 5: Ilias Sitskiver is really He's at the top of our

157
00:08:06,160 --> 00:08:09,880
Speaker 5: list for a variety of reasons. And if you're not

158
00:08:09,920 --> 00:08:13,120
Speaker 5: familiar with Ilias Sutzkiver, he is an AI researcher who

159
00:08:13,320 --> 00:08:16,600
Speaker 5: was at open AI for a long time, and there's

160
00:08:16,640 --> 00:08:19,280
Speaker 5: a few different ways to characterize him. He is both

161
00:08:19,320 --> 00:08:23,640
Speaker 5: coming up with new ways to implement AI algorithms the

162
00:08:23,680 --> 00:08:26,720
Speaker 5: way you train the model, but he's also very good

163
00:08:26,840 --> 00:08:32,040
Speaker 5: at for a long time identifying which the shortest path

164
00:08:32,120 --> 00:08:35,160
Speaker 5: in the tech tree. So there are branches of choices

165
00:08:35,200 --> 00:08:38,000
Speaker 5: that you need to make as you develop the new

166
00:08:38,440 --> 00:08:42,440
Speaker 5: AI models, and he was very early at while he

167
00:08:42,520 --> 00:08:45,160
Speaker 5: was at open AI, he identified that the transformer paper

168
00:08:45,200 --> 00:08:48,000
Speaker 5: from Google. He didn't invent that it was at Google,

169
00:08:48,320 --> 00:08:51,920
Speaker 5: but the transformer technology was extremely important and that it

170
00:08:52,080 --> 00:08:55,679
Speaker 5: had the ability to do remarkable things once scaled up massively,

171
00:08:56,080 --> 00:08:59,120
Speaker 5: and so he was the driving force between kind of

172
00:08:59,200 --> 00:09:03,040
Speaker 5: identifying the transformers as the correct path. Now we look

173
00:09:03,120 --> 00:09:05,079
Speaker 5: back on Attention as All you Need, which is the

174
00:09:05,160 --> 00:09:07,959
Speaker 5: name of the paper that defines the architecture that is

175
00:09:08,080 --> 00:09:10,719
Speaker 5: used in these modern large language models that you use

176
00:09:10,720 --> 00:09:11,200
Speaker 5: when you're.

177
00:09:11,080 --> 00:09:11,880
Speaker 3: In chat GPT.

178
00:09:12,520 --> 00:09:15,880
Speaker 5: There were twenty five other potential paths that we could

179
00:09:15,880 --> 00:09:19,280
Speaker 5: have gone down. You know, the story goes is that

180
00:09:19,480 --> 00:09:22,240
Speaker 5: he really identified that and said, let's go really really

181
00:09:22,280 --> 00:09:22,760
Speaker 5: hard on that.

182
00:09:22,880 --> 00:09:26,120
Speaker 4: So he can sort of identify the innovation, what's new,

183
00:09:26,240 --> 00:09:29,560
Speaker 4: and then also identify the most efficient path to actually

184
00:09:29,559 --> 00:09:30,080
Speaker 4: execute on.

185
00:09:30,240 --> 00:09:32,640
Speaker 5: Yes, and if you're familiar with what the more the

186
00:09:32,640 --> 00:09:37,319
Speaker 5: more modern, So the first arc of LMS and these

187
00:09:37,440 --> 00:09:41,360
Speaker 5: chatbots scaling up where really this tack the transformer paper,

188
00:09:41,559 --> 00:09:44,160
Speaker 5: understand that that's the correct architecture, and then scale it

189
00:09:44,240 --> 00:09:45,880
Speaker 5: up really really big. So you need to be able

190
00:09:45,920 --> 00:09:48,600
Speaker 5: to not just write the code to implement that that

191
00:09:48,679 --> 00:09:52,160
Speaker 5: particular algorithm. You need to marshal the capital to say

192
00:09:52,320 --> 00:09:54,120
Speaker 5: we're going really really big and we're going to build

193
00:09:54,120 --> 00:09:55,400
Speaker 5: the big data center. We're going to spend a lot

194
00:09:55,400 --> 00:09:56,920
Speaker 5: of money, but it's going to be worth it because

195
00:09:56,920 --> 00:10:00,480
Speaker 5: we understand the trade offs here. Then the second kind

196
00:10:00,480 --> 00:10:04,160
Speaker 5: of innovation or correct call he had was during the

197
00:10:04,200 --> 00:10:07,600
Speaker 5: sam Altman, Ouster and return. He was working on a

198
00:10:07,600 --> 00:10:10,480
Speaker 5: project that was code named q Star, and there was

199
00:10:10,520 --> 00:10:13,960
Speaker 5: a lot of speculation about what q star was. Was

200
00:10:13,960 --> 00:10:15,719
Speaker 5: it the secret super intelligence thing?

201
00:10:15,760 --> 00:10:16,680
Speaker 3: What did Ilius see?

202
00:10:16,880 --> 00:10:19,440
Speaker 5: Yes, yeah, because he was going back and forth in

203
00:10:19,480 --> 00:10:22,120
Speaker 5: support of Sam Altman and then leaving and then and

204
00:10:22,160 --> 00:10:23,600
Speaker 5: then it was back and forth, and there's a lot

205
00:10:23,640 --> 00:10:25,959
Speaker 5: of drama there. Always keep them guessing, Always keep them

206
00:10:25,960 --> 00:10:28,599
Speaker 5: guessing for sure, and so there was there was a

207
00:10:28,600 --> 00:10:30,560
Speaker 5: lot of rumors, but what it wound up being was

208
00:10:30,679 --> 00:10:34,200
Speaker 5: just reasoning, which is now what is available in UH

209
00:10:34,280 --> 00:10:37,240
Speaker 5: if you use any of the three models in chat GPT,

210
00:10:37,679 --> 00:10:41,320
Speaker 5: or you use deep SEKR one and it's it goes

211
00:10:41,360 --> 00:10:43,920
Speaker 5: by a number of different names. The project was rebranded

212
00:10:44,040 --> 00:10:47,120
Speaker 5: Strawberry and then the O series, And you can think

213
00:10:47,160 --> 00:10:50,400
Speaker 5: about this as the test time. Inference is the other buzzword,

214
00:10:50,440 --> 00:10:54,400
Speaker 5: but basically the LM, you're using the same foundation model,

215
00:10:54,640 --> 00:10:56,880
Speaker 5: but you're running it a ton more to come up

216
00:10:56,920 --> 00:10:59,520
Speaker 5: with a bunch of potential answers to questions and then

217
00:10:59,679 --> 00:11:04,880
Speaker 5: narrowing that down and essentially applying what's called reinforcement learning

218
00:11:05,040 --> 00:11:08,559
Speaker 5: to the transformer architecture and the pre training that's happening,

219
00:11:08,559 --> 00:11:11,080
Speaker 5: and so he was very critical in leading that project,

220
00:11:11,240 --> 00:11:14,040
Speaker 5: and so that's allowed him to eventually, once he left

221
00:11:14,080 --> 00:11:17,760
Speaker 5: open ai, go start a new company called Safe Superintelligence SSI,

222
00:11:18,040 --> 00:11:20,480
Speaker 5: and he's gotten that company to what a thirty billion

223
00:11:20,520 --> 00:11:23,120
Speaker 5: dollar evaluation, thirty two billion dollar valuation.

224
00:11:22,880 --> 00:11:26,199
Speaker 6: Which ties into the talent war because I think it's

225
00:11:26,440 --> 00:11:28,480
Speaker 6: I don't know that it's been officially announced yet, but

226
00:11:28,720 --> 00:11:33,400
Speaker 6: people are expecting Daniel Gross to join Meta, who.

227
00:11:33,280 --> 00:11:37,040
Speaker 5: Was the CEO and co founder of that company. Side

228
00:11:37,040 --> 00:11:40,160
Speaker 5: and so back to your back to the white dynamic.

229
00:11:40,320 --> 00:11:42,640
Speaker 3: Yeah, yeah, if you guys get together, they start.

230
00:11:42,400 --> 00:11:45,360
Speaker 6: A company within a year, it's worth thirty two billion dollars,

231
00:11:46,040 --> 00:11:48,240
Speaker 6: and one of the people on that team decides to

232
00:11:48,280 --> 00:11:53,320
Speaker 6: actually leave and join Meta. It sounds insane, probably leaving

233
00:11:53,360 --> 00:11:56,240
Speaker 6: billions of dollars on the table, but you can kind

234
00:11:56,240 --> 00:11:59,480
Speaker 6: of trace the this talent war actually back to the

235
00:11:59,480 --> 00:12:02,680
Speaker 6: open ai founding team. When you think about all the

236
00:12:02,679 --> 00:12:05,960
Speaker 6: different players, right, you have Mira Murti who's now with

237
00:12:06,040 --> 00:12:08,680
Speaker 6: Thinking Machines, she was a CTO of open Ai. You

238
00:12:08,760 --> 00:12:12,640
Speaker 6: have Ilia Sutskiverer with SSI, you have Elon with XAI,

239
00:12:13,080 --> 00:12:15,160
Speaker 6: and not to mention, you have these sort of hyperscalers

240
00:12:15,160 --> 00:12:17,800
Speaker 6: who are also competing for the same talent. So yeah,

241
00:12:17,800 --> 00:12:20,959
Speaker 6: that's really kind of the origin story in the genesis,

242
00:12:21,040 --> 00:12:24,160
Speaker 6: And it's been amazing to see open AI's progress despite

243
00:12:24,559 --> 00:12:27,400
Speaker 6: you know, the recent reporting is around you know, losing

244
00:12:27,400 --> 00:12:30,439
Speaker 6: a bunch of top researchers, which is real, but it

245
00:12:30,600 --> 00:12:33,440
Speaker 6: wasn't that long ago that they lost like two very

246
00:12:33,640 --> 00:12:38,120
Speaker 6: very very key senior execs in Ilia and Mira.

247
00:12:53,720 --> 00:12:55,960
Speaker 2: So a few stats were recording this. By the way,

248
00:12:56,120 --> 00:12:59,280
Speaker 2: July thirty first, Meta came out with earnings last night.

249
00:12:59,320 --> 00:13:01,160
Speaker 2: The stock is up like ten percent on stuff like

250
00:13:01,200 --> 00:13:04,360
Speaker 2: one hundred and fifty billion more dollars. The other thing is,

251
00:13:04,760 --> 00:13:08,360
Speaker 2: and I think this gets ties into the logic behind

252
00:13:08,360 --> 00:13:10,320
Speaker 2: the comp so they're going to spend like something like

253
00:13:10,360 --> 00:13:13,520
Speaker 2: another seventy billion on capac some stuff like that. So

254
00:13:13,559 --> 00:13:17,640
Speaker 2: we know these are incredibly computationally intensive things. They're electricity

255
00:13:18,240 --> 00:13:23,920
Speaker 2: intensive things. You mentioned paper citations, but also actually having

256
00:13:23,960 --> 00:13:26,640
Speaker 2: the experience of one of these runs, and it does

257
00:13:26,679 --> 00:13:29,160
Speaker 2: seem very highly intuitive to me that if you've done

258
00:13:29,200 --> 00:13:32,120
Speaker 2: it before, and if you could even cut down the

259
00:13:32,160 --> 00:13:35,320
Speaker 2: cost of a training run or set up a big

260
00:13:35,600 --> 00:13:39,240
Speaker 2: computer rack for you can pay five very quickly you

261
00:13:39,320 --> 00:13:44,079
Speaker 2: have five percent less, right, exactly happened. Instantly pay then

262
00:13:44,080 --> 00:13:46,439
Speaker 2: you instantly pay for your salary, I could say, because

263
00:13:46,520 --> 00:13:49,079
Speaker 2: and this is different from the B to B SaaS yer, right,

264
00:13:49,400 --> 00:13:52,240
Speaker 2: because you can there's these huge kpax costs if you

265
00:13:52,240 --> 00:13:55,400
Speaker 2: could just marginally improve the efficiency that that pays that

266
00:13:55,440 --> 00:13:57,000
Speaker 2: one hundred million sellar salary.

267
00:13:57,280 --> 00:14:00,680
Speaker 5: And so this this literally just happened right before Mark

268
00:14:00,760 --> 00:14:03,760
Speaker 5: Zuckerberg went on the talent acquisition spray that he's been

269
00:14:03,800 --> 00:14:07,920
Speaker 5: on for the last few months. Meta's main AI model

270
00:14:08,000 --> 00:14:11,800
Speaker 5: is called Lama and they've released a series of versions

271
00:14:11,920 --> 00:14:15,320
Speaker 5: of that model, and Lama four was the latest and greatest,

272
00:14:15,679 --> 00:14:18,880
Speaker 5: and it didn't go very well. And the rumors about

273
00:14:18,920 --> 00:14:23,320
Speaker 5: why it kind of failed to deliver on expectations was

274
00:14:23,360 --> 00:14:25,880
Speaker 5: that they kind of went down the wrong path in

275
00:14:25,920 --> 00:14:28,400
Speaker 5: the tech tree and they focused a little bit too

276
00:14:28,480 --> 00:14:34,120
Speaker 5: much on scale insanely costly. So they've released a part

277
00:14:34,240 --> 00:14:37,160
Speaker 5: of Lama four, but they have yet to release Malama

278
00:14:37,200 --> 00:14:40,000
Speaker 5: for Behemoth, which is their biggest model, the most the

279
00:14:40,040 --> 00:14:44,760
Speaker 5: most aggressive, and they had just little details in the implementation,

280
00:14:44,880 --> 00:14:48,120
Speaker 5: little choices of how you chunk the attention in the

281
00:14:48,160 --> 00:14:50,560
Speaker 5: transformer model, you know, we're operating in this level of

282
00:14:50,560 --> 00:14:54,480
Speaker 5: abstraction up here talking about you know, it's a prediction model,

283
00:14:54,520 --> 00:14:56,520
Speaker 5: and then we talk about transformers a little bit. There

284
00:14:56,520 --> 00:15:00,240
Speaker 5: are sub algorithms within these systems that you make the

285
00:15:00,240 --> 00:15:03,160
Speaker 5: wrong decision and you could get a vastly different.

286
00:15:03,160 --> 00:15:05,120
Speaker 2: Electricity bill goes up by one hundred.

287
00:15:04,920 --> 00:15:07,040
Speaker 5: Many exactly, so all of that money that was spent

288
00:15:07,080 --> 00:15:09,840
Speaker 5: on that electricity and then build out of course that

289
00:15:09,960 --> 00:15:12,360
Speaker 5: it can absorb that. But when you think about the

290
00:15:13,040 --> 00:15:14,640
Speaker 5: cost of getting correct.

291
00:15:14,400 --> 00:15:17,320
Speaker 6: One thing is is a lot of these projects will

292
00:15:17,320 --> 00:15:20,480
Speaker 6: be energy constrained too, So efficiency is going to matter

293
00:15:20,640 --> 00:15:23,080
Speaker 6: a lot. Oh, it hasn't been the core focus today.

294
00:15:23,320 --> 00:15:25,600
Speaker 6: If you talk to anybody at the big labs, they

295
00:15:25,640 --> 00:15:29,120
Speaker 6: are focused on maxing out intelligence at the cost of

296
00:15:29,160 --> 00:15:31,640
Speaker 6: efficiency because they know they can. You know, you want

297
00:15:31,680 --> 00:15:33,200
Speaker 6: to be on the bleeding edge. You want to have

298
00:15:33,240 --> 00:15:34,040
Speaker 6: the smartest model.

299
00:15:34,080 --> 00:15:34,560
Speaker 3: You want to be.

300
00:15:35,000 --> 00:15:38,400
Speaker 6: You know, there's debates around which benchmarks actually matter, how

301
00:15:38,440 --> 00:15:43,080
Speaker 6: important AI benchmarks really are. But energy is going to

302
00:15:43,400 --> 00:15:45,640
Speaker 6: you know, a lot of people are saying Zuck won't

303
00:15:45,640 --> 00:15:47,600
Speaker 6: even be able to spend as much as he wants

304
00:15:47,640 --> 00:15:50,840
Speaker 6: to spend on data center development purely because of the

305
00:15:51,000 --> 00:15:51,760
Speaker 6: energy constraint.

306
00:15:52,720 --> 00:15:55,560
Speaker 4: Can you talk a little bit about what happened with Windsurf,

307
00:15:55,640 --> 00:15:57,760
Speaker 4: because this is this is the one that seems to

308
00:15:57,760 --> 00:16:01,080
Speaker 4: have like captured everyone's attention, in part because there was

309
00:16:01,240 --> 00:16:03,720
Speaker 4: the drama. It almost seemed like it came out of

310
00:16:03,760 --> 00:16:06,880
Speaker 4: like a Silicon Valley the show script or something where

311
00:16:07,080 --> 00:16:09,960
Speaker 4: all the employees were gathering expecting to hear that they

312
00:16:09,960 --> 00:16:11,920
Speaker 4: were going to be bought by Open Ai, and then

313
00:16:11,920 --> 00:16:14,680
Speaker 4: they find out that actually their CTO has just been

314
00:16:14,680 --> 00:16:17,120
Speaker 4: bought by someone else completely different, and they're sort of

315
00:16:17,200 --> 00:16:18,360
Speaker 4: left in the lart ceo.

316
00:16:18,880 --> 00:16:21,560
Speaker 3: Oh is top CEO in top fifty?

317
00:16:21,640 --> 00:16:24,880
Speaker 5: Well, how about I give some prehistory on acquires and

318
00:16:24,880 --> 00:16:26,920
Speaker 5: you can take take us through that actual weekend. So

319
00:16:27,360 --> 00:16:30,560
Speaker 5: there has been a trend of sort of these zombie

320
00:16:30,560 --> 00:16:34,440
Speaker 5: aqua hires. Everyone has different names for this, but effectively,

321
00:16:34,640 --> 00:16:37,320
Speaker 5: when a very large company, usually a hyperscaler, wants to

322
00:16:37,360 --> 00:16:41,040
Speaker 5: go and acquire a company that has AI talent, they

323
00:16:41,120 --> 00:16:42,760
Speaker 5: used to just buy the whole company. And this was

324
00:16:42,800 --> 00:16:45,560
Speaker 5: part of the Silicon Valley social contract that even if

325
00:16:45,600 --> 00:16:50,120
Speaker 5: I am in operations or sales or finance or HR,

326
00:16:50,480 --> 00:16:52,800
Speaker 5: if I join a hot startup and it gets acquired

327
00:16:52,840 --> 00:16:54,920
Speaker 5: I'm coming along for the ride to exit and I

328
00:16:54,960 --> 00:16:56,720
Speaker 5: get the job at Google at least for a little bit.

329
00:16:56,760 --> 00:17:00,480
Speaker 5: And then yeah, if I underperform Google or Meta Amazon,

330
00:17:00,480 --> 00:17:03,280
Speaker 5: they might lay me off. But even if I'm somewhat redundant,

331
00:17:03,320 --> 00:17:05,240
Speaker 5: I'm coming along for the ride and my cashing out

332
00:17:05,240 --> 00:17:07,639
Speaker 5: my shares, and these the reason that you go and

333
00:17:07,680 --> 00:17:09,720
Speaker 5: take the risk and take the little bit of the

334
00:17:09,800 --> 00:17:12,159
Speaker 5: rougher ride that is a startup. You don't get as

335
00:17:12,160 --> 00:17:14,199
Speaker 5: many amenities, but you get the lottery tickets in the

336
00:17:14,240 --> 00:17:17,960
Speaker 5: form of stock options that hopefully pay out. But there's

337
00:17:18,000 --> 00:17:20,720
Speaker 5: a big question about how much of this is the

338
00:17:20,760 --> 00:17:23,760
Speaker 5: acquirer just not wanting to deal with the d duplication

339
00:17:23,920 --> 00:17:27,280
Speaker 5: of the back office. There's also the question of the FTC.

340
00:17:27,680 --> 00:17:30,280
Speaker 5: A lot of the FTC rulings have made it much

341
00:17:30,400 --> 00:17:34,560
Speaker 5: harder to get these these big acquisitions across the finish line.

342
00:17:34,680 --> 00:17:37,639
Speaker 5: And even if they even if the FDC does approve,

343
00:17:37,680 --> 00:17:39,280
Speaker 5: they can often hold it up for six months. And

344
00:17:39,320 --> 00:17:41,400
Speaker 5: we're in a race where if you deliver the best

345
00:17:41,440 --> 00:17:43,480
Speaker 5: model today, you're going to make more money. You're going

346
00:17:43,520 --> 00:17:45,280
Speaker 5: to be the hot company, you're going to acquire. It's

347
00:17:45,280 --> 00:17:48,720
Speaker 5: all this big snowball so companies. This started with the

348
00:17:48,800 --> 00:17:50,800
Speaker 5: very few, but character Ai was the big one with

349
00:17:50,840 --> 00:17:51,960
Speaker 5: Noam Shazir.

350
00:17:51,840 --> 00:17:55,399
Speaker 6: Who was in a very very interesting situation where so

351
00:17:55,560 --> 00:17:58,840
Speaker 6: Nome wanted to go back to Google, which made sense.

352
00:17:58,920 --> 00:18:02,600
Speaker 6: Google wanted him there as well researches of a yes,

353
00:18:02,640 --> 00:18:04,840
Speaker 6: and so having him go back there made a lot

354
00:18:04,840 --> 00:18:05,160
Speaker 6: of sense.

355
00:18:05,160 --> 00:18:06,480
Speaker 3: I think he had built this platform.

356
00:18:06,520 --> 00:18:10,080
Speaker 6: He built like the first at scale AI companionship platform.

357
00:18:10,119 --> 00:18:13,639
Speaker 6: If you look at character AI's site traffic today, I

358
00:18:13,680 --> 00:18:16,000
Speaker 6: mean it's it's one of the largest sites on the

359
00:18:16,040 --> 00:18:18,760
Speaker 6: internet still, which is wild. But I think he realized,

360
00:18:18,960 --> 00:18:20,800
Speaker 6: I want to work on AI research. I don't want

361
00:18:20,800 --> 00:18:24,280
Speaker 6: to work on AI girlfriends, boyfriends, that that kind of thing.

362
00:18:24,800 --> 00:18:26,359
Speaker 3: And as part of that.

363
00:18:26,320 --> 00:18:32,080
Speaker 6: Deal, character Ai effectively became entirely employee owned and they

364
00:18:32,080 --> 00:18:34,000
Speaker 6: had a really strong balance sheet. They had a crazy

365
00:18:34,119 --> 00:18:37,160
Speaker 6: amount of users. They weren't monetizing maybe that well yet.

366
00:18:37,160 --> 00:18:38,879
Speaker 6: But I think a lot of the employees in that

367
00:18:38,920 --> 00:18:41,160
Speaker 6: situation where like this is pretty cool. We're basically running

368
00:18:41,160 --> 00:18:43,000
Speaker 6: a co op where we all own a lot of

369
00:18:43,040 --> 00:18:43,520
Speaker 6: this company.

370
00:18:43,520 --> 00:18:43,960
Speaker 3: We don't have.

371
00:18:43,920 --> 00:18:47,040
Speaker 6: Investors, we don't have the same kind of pressure to perform,

372
00:18:47,359 --> 00:18:51,360
Speaker 6: and that space is competitive, right, even Elon is competing

373
00:18:51,400 --> 00:18:54,640
Speaker 6: there now Chat GPT gets used as a companion, but.

374
00:18:55,000 --> 00:18:56,600
Speaker 3: Not competitive like Cogen.

375
00:18:56,720 --> 00:18:58,800
Speaker 6: And so that was the dynamic here that was insane

376
00:18:59,160 --> 00:19:04,360
Speaker 6: because Google clearly cares about code generation. They see anthropic adding.

377
00:19:04,520 --> 00:19:07,400
Speaker 6: They went from one to four billion dollars in run

378
00:19:07,480 --> 00:19:11,480
Speaker 6: rate this year. They're pacing to be somewhere around ten

379
00:19:11,680 --> 00:19:13,199
Speaker 6: at the end of the year, and so that's a

380
00:19:13,200 --> 00:19:16,080
Speaker 6: Google size market, right. Google is going to ultimately care

381
00:19:16,080 --> 00:19:18,639
Speaker 6: about Cogen, So it made sense for them to say, hey,

382
00:19:18,720 --> 00:19:23,280
Speaker 6: let's get you know, fifty more hyper talented engineers. This

383
00:19:23,320 --> 00:19:25,840
Speaker 6: is winter we're talking about. The issue though, is if

384
00:19:25,880 --> 00:19:28,480
Speaker 6: you didn't get brought over to the Google ship, you

385
00:19:28,520 --> 00:19:30,600
Speaker 6: were on what I what I was calling a ghost ship.

386
00:19:31,240 --> 00:19:31,440
Speaker 3: John.

387
00:19:31,720 --> 00:19:34,520
Speaker 5: Everyone who was pro the strategy, they would call it

388
00:19:34,520 --> 00:19:36,720
Speaker 5: the remain co I was.

389
00:19:38,119 --> 00:19:40,439
Speaker 6: But imagine you're at a company and the dynamic with

390
00:19:40,480 --> 00:19:43,120
Speaker 6: Windsurf was fascinating because the company, if you had joined

391
00:19:43,160 --> 00:19:46,199
Speaker 6: in August of last year Windsurf, you could have the

392
00:19:46,240 --> 00:19:49,160
Speaker 6: product hadn't launched it, so you could have worked and

393
00:19:49,520 --> 00:19:52,159
Speaker 6: worked up to the product launch launched it. Seeing this

394
00:19:52,280 --> 00:19:54,880
Speaker 6: meteoric growth, the last reported number they had with some

395
00:19:54,880 --> 00:19:57,240
Speaker 6: something like eighty million of air R. You have a

396
00:19:57,440 --> 00:20:00,360
Speaker 6: term sheet to get acquired from open Ai for three

397
00:20:00,400 --> 00:20:03,960
Speaker 6: billion dollars and that falls through, and then all these

398
00:20:03,960 --> 00:20:05,800
Speaker 6: employees are looking around and they're like, wait, I didn't

399
00:20:05,800 --> 00:20:08,280
Speaker 6: even hit my I haven't even hit my one year cliff.

400
00:20:08,359 --> 00:20:12,000
Speaker 6: Yet I don't actually have a right to. I mean,

401
00:20:12,720 --> 00:20:16,520
Speaker 6: they might technically own shares or options depending on how

402
00:20:16,560 --> 00:20:20,000
Speaker 6: it was structured, but they as I'm sure everyone here

403
00:20:20,240 --> 00:20:25,280
Speaker 6: knows you, oftentimes, like sometimes founders would would try to

404
00:20:25,320 --> 00:20:28,720
Speaker 6: accelerate their employees, get them compensated as part of a

405
00:20:28,760 --> 00:20:31,560
Speaker 6: transaction like that, but there's not necessarily a contractual right

406
00:20:31,600 --> 00:20:33,840
Speaker 6: to do that, and it's part of a negotiation. And

407
00:20:33,880 --> 00:20:36,800
Speaker 6: so in this case, you have this team that's effectively

408
00:20:36,880 --> 00:20:39,920
Speaker 6: split up. And we were covering the whole thing live

409
00:20:39,960 --> 00:20:42,520
Speaker 6: because we were hearing that. You know, employees that Friday

410
00:20:42,520 --> 00:20:45,480
Speaker 6: were like crying, and there was a ton of confusion

411
00:20:45,520 --> 00:20:48,440
Speaker 6: and chaos and everybody was learning facts kind of over

412
00:20:48,480 --> 00:20:52,120
Speaker 6: that forty eight hour period. Luckily, our friend Scott Wu

413
00:20:52,320 --> 00:20:55,639
Speaker 6: at Cognition, the company that ultimately bought the remain co,

414
00:20:56,359 --> 00:20:59,520
Speaker 6: flew to meet the Windsurf team, the new CEO Windsurf

415
00:20:59,560 --> 00:21:02,520
Speaker 6: and Basic. They spent the weekend doing this insane deal

416
00:21:02,520 --> 00:21:04,680
Speaker 6: and it ended up being a great outcome, I think

417
00:21:04,680 --> 00:21:05,760
Speaker 6: for everybody.

418
00:21:05,560 --> 00:21:09,560
Speaker 4: Didn't involve Windsurf have their like marketing department in the

419
00:21:09,640 --> 00:21:13,440
Speaker 4: room or something. When they made the announcement, the initial announcement,

420
00:21:13,480 --> 00:21:15,560
Speaker 4: and everyone was like because they were expecting to be

421
00:21:15,600 --> 00:21:17,760
Speaker 4: bought by open Ai, right, They're like, we're going to

422
00:21:17,840 --> 00:21:20,800
Speaker 4: film this for posterity, and then it turns out to

423
00:21:20,800 --> 00:21:21,160
Speaker 4: be something.

424
00:21:21,920 --> 00:21:24,000
Speaker 5: Yeah. Yeah, I believe the timeline is that, you know,

425
00:21:24,080 --> 00:21:28,320
Speaker 5: Windsurf launches a year ago, is in a knockout, drag

426
00:21:28,440 --> 00:21:32,200
Speaker 5: out fight with Cursor. Cursor's doing very well. They're also

427
00:21:32,240 --> 00:21:34,720
Speaker 5: going to eighty million or so opening.

428
00:21:34,560 --> 00:21:37,000
Speaker 3: Cursors in the hundreds of millions of revenue.

429
00:21:37,119 --> 00:21:39,800
Speaker 6: Yeah, Windsurf was very clearly the number two player in

430
00:21:39,840 --> 00:21:41,160
Speaker 6: the AI id E mark.

431
00:21:41,280 --> 00:21:45,240
Speaker 5: Yeah, and so Cursor is staying independent, but open Ai

432
00:21:45,359 --> 00:21:48,399
Speaker 5: wants to continue to get a foothold in this space,

433
00:21:48,440 --> 00:21:52,679
Speaker 5: so they make an offer that falls through. The rumor

434
00:21:52,840 --> 00:21:57,280
Speaker 5: was because Microsoft would have had ip look through exposure

435
00:21:57,800 --> 00:22:00,760
Speaker 5: via the more complex open ai structure.

436
00:22:00,400 --> 00:22:02,760
Speaker 3: Which is ongoing, which is continuous.

437
00:22:02,760 --> 00:22:06,200
Speaker 5: Be ongoing, and then when when Google came through, they

438
00:22:06,560 --> 00:22:08,920
Speaker 5: said that they wanted just to buy the team leave

439
00:22:08,960 --> 00:22:12,879
Speaker 5: the remain co because it's potentially cleaner from an FTC perspective.

440
00:22:13,160 --> 00:22:14,639
Speaker 5: But there's a whole bunch of different reasons and no

441
00:22:14,680 --> 00:22:18,520
Speaker 5: one really comments on exactly what happens. But when these

442
00:22:18,560 --> 00:22:22,160
Speaker 5: deals happen. This just happened with SCALEAI and Meta. The CEO,

443
00:22:22,400 --> 00:22:26,240
Speaker 5: even though there's some sort of amorphous FTC risk, the

444
00:22:26,600 --> 00:22:28,960
Speaker 5: CEO if they're going through one of these zombie acquisitions,

445
00:22:29,200 --> 00:22:33,000
Speaker 5: does have the ability to kind of set the team

446
00:22:33,200 --> 00:22:36,639
Speaker 5: up with certain expectations and say, hey, we're going to

447
00:22:36,720 --> 00:22:39,000
Speaker 5: take care of you. Trust me. You're going to get

448
00:22:39,000 --> 00:22:41,760
Speaker 5: a check that you would expect based on your ownership.

449
00:22:41,800 --> 00:22:44,200
Speaker 5: So if you own point zero one percent of the company,

450
00:22:44,240 --> 00:22:46,520
Speaker 5: you would expect that you get x of the headline number,

451
00:22:46,560 --> 00:22:49,080
Speaker 5: and yes, it's coming to you. The weird thing about

452
00:22:49,119 --> 00:22:52,840
Speaker 5: the Windsurf deal was that that was not messaged and

453
00:22:52,880 --> 00:22:55,680
Speaker 5: so there was like this z the zombie ship was

454
00:22:55,720 --> 00:22:56,560
Speaker 5: more zombie five.

455
00:22:57,119 --> 00:22:59,760
Speaker 6: So like yeah, and and imagine, imagine you're a you're

456
00:22:59,760 --> 00:23:02,840
Speaker 6: an hyper competitive market where you're competing with Google and

457
00:23:02,920 --> 00:23:06,640
Speaker 6: Anthropic and Cursor and all these different players, and then

458
00:23:06,760 --> 00:23:09,960
Speaker 6: you lose your CEO and your top fifty engineers, and

459
00:23:09,960 --> 00:23:12,960
Speaker 6: they're like, and you guys are gonna do against don't worry.

460
00:23:12,960 --> 00:23:15,359
Speaker 6: You guys have a strong balance sheet. You guys are

461
00:23:15,359 --> 00:23:17,200
Speaker 6: going to do great. And we're also gonna be We're

462
00:23:17,200 --> 00:23:20,639
Speaker 6: also gonna be competing with you at at at at Google,

463
00:23:20,760 --> 00:23:22,560
Speaker 6: But you guys are gonna be fod luck guys and

464
00:23:22,600 --> 00:23:23,240
Speaker 6: so and.

465
00:23:23,400 --> 00:23:26,399
Speaker 2: Thanks, thanks for the help. But this break but like, okay,

466
00:23:26,800 --> 00:23:29,600
Speaker 2: cognition came in. It sounds like all the employees will

467
00:23:29,600 --> 00:23:31,679
Speaker 2: get something. It's something in the world because there was

468
00:23:31,720 --> 00:23:34,520
Speaker 2: cash on the balance but it didn't have to be

469
00:23:34,720 --> 00:23:37,840
Speaker 2: that outcome. And so there's two things here that strike me.

470
00:23:38,040 --> 00:23:41,359
Speaker 2: One is, again, this does not seem like the B

471
00:23:41,440 --> 00:23:44,000
Speaker 2: to B SaaS era, where it's like, if you created

472
00:23:44,040 --> 00:23:47,639
Speaker 2: the hottest building product for dentist offices and that was

473
00:23:47,680 --> 00:23:50,159
Speaker 2: great gathering traction, no one would like just buy the

474
00:23:50,200 --> 00:23:52,920
Speaker 2: talent that built that. So that's really different, right, because

475
00:23:53,119 --> 00:23:56,119
Speaker 2: so the the ability to take value out via the

476
00:23:56,119 --> 00:23:59,320
Speaker 2: talent channel rather than buying the product itself. But then

477
00:23:59,359 --> 00:24:01,920
Speaker 2: also and I'm you know, the knock on effects. Okay, fine,

478
00:24:02,080 --> 00:24:05,280
Speaker 2: the Windsurf employees did fine, but going forward, there's no

479
00:24:05,320 --> 00:24:07,680
Speaker 2: guarantee that they could have. And so I'm wondering, from

480
00:24:07,800 --> 00:24:11,320
Speaker 2: either a VC perspective or a future employees perspective, how

481
00:24:11,440 --> 00:24:13,879
Speaker 2: this is going to change the sort of calculation that

482
00:24:13,960 --> 00:24:17,399
Speaker 2: anyone makes when participating in a new AI startup, the

483
00:24:17,440 --> 00:24:20,200
Speaker 2: fact that the enterprise may not be where the value

484
00:24:20,200 --> 00:24:20,679
Speaker 2: actually is.

485
00:24:20,800 --> 00:24:24,000
Speaker 5: So the first question is, like, this kind of is

486
00:24:24,119 --> 00:24:26,919
Speaker 5: the example you gave of, like buying the team that

487
00:24:26,960 --> 00:24:30,240
Speaker 5: built the dentist B to B SaaS because Windsurf does

488
00:24:30,280 --> 00:24:34,280
Speaker 5: not train foundation models and Google is exceptional in training

489
00:24:34,320 --> 00:24:37,480
Speaker 5: foundation models. With Gemini and deep Mind, the deep Mind

490
00:24:37,480 --> 00:24:42,400
Speaker 5: team is extremely well staffed on AI researchers and continually

491
00:24:42,880 --> 00:24:46,960
Speaker 5: seems to push the frontier both the qualitative and quantitative metrics.

492
00:24:47,160 --> 00:24:49,760
Speaker 5: And if they were weak anywhere, it was maybe products exactly.

493
00:24:49,840 --> 00:24:52,000
Speaker 5: And so this is the like, I don't want to

494
00:24:52,080 --> 00:24:54,960
Speaker 5: be like they're just product people. Sure, Windsurf has some

495
00:24:55,000 --> 00:24:59,320
Speaker 5: amazing AI researchers, some amazing AI engineers, but what Windsurf

496
00:24:59,520 --> 00:25:02,040
Speaker 5: really did. They did not train a frontier model that

497
00:25:02,440 --> 00:25:07,160
Speaker 5: was about to disrupt Gemini go deep minds foundation model

498
00:25:07,359 --> 00:25:10,280
Speaker 5: they were going after. You know, would you use a

499
00:25:10,320 --> 00:25:13,679
Speaker 5: Google product, No, you would use Windsurf on top of

500
00:25:13,720 --> 00:25:16,000
Speaker 5: anthropy person or another foundation model.

501
00:25:15,800 --> 00:25:16,960
Speaker 3: And ultimately don't.

502
00:25:17,000 --> 00:25:20,880
Speaker 6: I don't think it's a systemic risk to the social

503
00:25:21,000 --> 00:25:24,639
Speaker 6: contract of our industry. And the reason for that is

504
00:25:24,680 --> 00:25:27,800
Speaker 6: that when you see a deal like the Google Windsurf

505
00:25:27,880 --> 00:25:29,840
Speaker 6: deal get done, it was very concerning. A lot of

506
00:25:29,840 --> 00:25:32,960
Speaker 6: people were extremely angry. It ended up being a good outcome.

507
00:25:33,119 --> 00:25:36,919
Speaker 6: A lot of employees were looking around, probably concerned about

508
00:25:37,200 --> 00:25:39,840
Speaker 6: you know, is the million dollars of stock that I've

509
00:25:39,880 --> 00:25:42,159
Speaker 6: been working for for years worth anything at all? I

510
00:25:42,160 --> 00:25:44,520
Speaker 6: think that was a good question to ask. But I

511
00:25:44,520 --> 00:25:48,399
Speaker 6: think there's two things. One is is AI is a category,

512
00:25:48,440 --> 00:25:51,240
Speaker 6: and of course it's touching kind of every category of

513
00:25:51,320 --> 00:25:54,560
Speaker 6: venture in the private markets. But we're not seeing You're

514
00:25:54,560 --> 00:25:58,040
Speaker 6: not seeing a defense tech founding team or engineering group

515
00:25:58,440 --> 00:25:59,960
Speaker 6: get There's no real.

516
00:25:59,760 --> 00:26:01,080
Speaker 3: Act hire value there.

517
00:26:01,119 --> 00:26:04,400
Speaker 6: The acquihires that are getting done in hard tech are hey,

518
00:26:04,440 --> 00:26:08,119
Speaker 6: you clearly have good engineering capabilities. We're happy to have

519
00:26:08,119 --> 00:26:11,400
Speaker 6: you join the team, but there's no real premium being

520
00:26:11,440 --> 00:26:13,879
Speaker 6: placed on that kind of talent. Might be able to

521
00:26:13,920 --> 00:26:17,080
Speaker 6: get a great comp packages, but they're certainly not getting

522
00:26:17,080 --> 00:26:19,840
Speaker 6: these sort of multi hundred million dollar premiums. The other

523
00:26:19,880 --> 00:26:23,280
Speaker 6: factor here is, like we have a set like right now,

524
00:26:23,440 --> 00:26:25,760
Speaker 6: if you are one of the top one hundred AI researchers,

525
00:26:25,800 --> 00:26:28,959
Speaker 6: you could probably get one hundred million dollar comp package

526
00:26:29,080 --> 00:26:31,040
Speaker 6: like within a week if you really wanted it, right,

527
00:26:31,080 --> 00:26:33,440
Speaker 6: And there's even people that are at Thinking you know,

528
00:26:33,520 --> 00:26:35,679
Speaker 6: the reporting from this week was that and it was

529
00:26:35,760 --> 00:26:38,360
Speaker 6: kind of hotly debated, but that people at Thinking Machines,

530
00:26:38,440 --> 00:26:43,080
Speaker 6: Miramrati's company, former CTO of Open AI, we're turning down

531
00:26:43,080 --> 00:26:45,440
Speaker 6: these sort of one hundred million, multi hundred million dollar

532
00:26:45,840 --> 00:26:47,639
Speaker 6: There was a rumor that somebody had turned down a

533
00:26:47,680 --> 00:26:53,040
Speaker 6: billion dollar five year contract. And I don't believe that

534
00:26:53,280 --> 00:26:56,719
Speaker 6: those deals will be getting done in three years now.

535
00:26:56,760 --> 00:26:58,520
Speaker 6: I might be wrong, and there's going to be like

536
00:26:58,600 --> 00:27:03,520
Speaker 6: probably some exceptions, but the idea that the thirtieth AI

537
00:27:03,600 --> 00:27:05,720
Speaker 6: researcher on your team is going to make more than

538
00:27:05,800 --> 00:27:09,560
Speaker 6: a mags Heaven CEO, like that doesn't feel hyper.

539
00:27:09,800 --> 00:27:10,800
Speaker 3: Something has to change either.

540
00:27:10,840 --> 00:27:12,560
Speaker 5: Tim Cook has to make more money.

541
00:27:13,600 --> 00:27:14,920
Speaker 3: Tim Cook is looking.

542
00:27:15,000 --> 00:27:17,479
Speaker 2: For assistances on Wall Street were like a top trader

543
00:27:17,600 --> 00:27:20,880
Speaker 2: to consistently make more than or will often make more

544
00:27:20,920 --> 00:27:23,200
Speaker 2: than the CEO, absolutely because.

545
00:27:22,960 --> 00:27:39,520
Speaker 4: They're the ones that got the money in the door. Yeah,

546
00:27:40,000 --> 00:27:42,240
Speaker 4: can you talk a little bit more about I guess

547
00:27:42,240 --> 00:27:46,040
Speaker 4: the fungibility of the IP here in both the practical

548
00:27:46,080 --> 00:27:49,239
Speaker 4: and legal sense. So if I'm a VC or some

549
00:27:49,280 --> 00:27:53,040
Speaker 4: sort of investor, I invest in an AI company because

550
00:27:53,080 --> 00:27:56,480
Speaker 4: I get to own that technology, that intellectual property. But

551
00:27:56,600 --> 00:27:59,240
Speaker 4: then let's say the main guy walks out the door,

552
00:27:59,320 --> 00:28:02,840
Speaker 4: gets hired by Google or whoever, Like, how much of

553
00:28:02,880 --> 00:28:06,560
Speaker 4: what he learned at his original firm or developed is

554
00:28:06,600 --> 00:28:08,720
Speaker 4: immediately going to be replicated at Google?

555
00:28:08,760 --> 00:28:09,840
Speaker 3: I bite almost all of it.

556
00:28:10,000 --> 00:28:12,400
Speaker 5: Yeah, I guess this is like, well, that's exactly why

557
00:28:12,400 --> 00:28:13,439
Speaker 5: they're paying that much money.

558
00:28:13,480 --> 00:28:15,880
Speaker 4: This is a lot more way of saying how many

559
00:28:15,960 --> 00:28:18,840
Speaker 4: lawsuits are we going to get after all these aquhiers?

560
00:28:19,080 --> 00:28:21,479
Speaker 6: So I think one way you can look at some

561
00:28:21,560 --> 00:28:24,679
Speaker 6: of Zuck's deals are like unauthorized aquhiers. It's like the

562
00:28:24,720 --> 00:28:26,919
Speaker 6: other CEO is not even participating in the deal. But

563
00:28:27,200 --> 00:28:28,840
Speaker 6: you know, Zuck is able to come in and be like,

564
00:28:29,040 --> 00:28:32,119
Speaker 6: would I pay if this, if these ten people were

565
00:28:32,400 --> 00:28:34,960
Speaker 6: working on a company independently, would I pay a billion

566
00:28:35,000 --> 00:28:37,840
Speaker 6: dollars to bring them onto my team? One hundred percent? Okay,

567
00:28:37,920 --> 00:28:39,560
Speaker 6: let's see the deal. It doesn't matter that you're kind

568
00:28:39,560 --> 00:28:42,520
Speaker 6: of piecing it all up from a. I think going

569
00:28:42,560 --> 00:28:44,720
Speaker 6: back to like the kind of Wall Street example of

570
00:28:44,760 --> 00:28:47,120
Speaker 6: like a top trader who maybe is like putting up

571
00:28:47,280 --> 00:28:51,320
Speaker 6: consistently incredible returns with some sort of like differentiated approach,

572
00:28:51,560 --> 00:28:53,240
Speaker 6: that person can go and as long as they have

573
00:28:53,280 --> 00:28:55,160
Speaker 6: capital at the new company, they're going to be able

574
00:28:55,200 --> 00:28:58,720
Speaker 6: to imagine to continue to just follow that same type

575
00:28:58,720 --> 00:29:02,080
Speaker 6: of strategy. I do think that AI and large language

576
00:29:02,080 --> 00:29:04,560
Speaker 6: models are somewhat different in that you have to look

577
00:29:04,600 --> 00:29:07,560
Speaker 6: at what is the AI product that these companies are

578
00:29:07,600 --> 00:29:09,360
Speaker 6: going to sell. How much of an impact is that

579
00:29:09,440 --> 00:29:12,920
Speaker 6: individual engineer researcher are going to have. And you know,

580
00:29:12,960 --> 00:29:15,720
Speaker 6: it's fairly obvious with open Ai their consumer tech company,

581
00:29:15,800 --> 00:29:18,880
Speaker 6: right they sell subscriptions to chat GBT, they have some

582
00:29:18,960 --> 00:29:23,080
Speaker 6: other use cases. It's obvious that companies like Anthropic, where

583
00:29:23,160 --> 00:29:25,640
Speaker 6: they're in the cogeneration business, they don't really have a

584
00:29:25,720 --> 00:29:28,520
Speaker 6: consumer business. And at Meta it's a little bit less

585
00:29:28,520 --> 00:29:31,400
Speaker 6: clear right now because the sort of ongoing product strategy

586
00:29:31,480 --> 00:29:35,200
Speaker 6: is not entirely clear yet. The thing that is obvious

587
00:29:35,320 --> 00:29:38,440
Speaker 6: is that if you can make a twenty billion dollars

588
00:29:38,440 --> 00:29:41,640
Speaker 6: training around more efficient than you pay for yourself, pretty quickly,

589
00:29:41,760 --> 00:29:42,440
Speaker 6: we should.

590
00:29:42,160 --> 00:29:44,560
Speaker 4: Just point out the Wall Street analogy is not perfect, right,

591
00:29:44,600 --> 00:29:47,120
Speaker 4: because if you're a high flyer at a Wall Street firm,

592
00:29:47,160 --> 00:29:50,000
Speaker 4: either client facing or if you're trading, if you join

593
00:29:50,080 --> 00:29:53,400
Speaker 4: someone else, you usually have a non compete agreement in

594
00:29:53,440 --> 00:29:56,080
Speaker 4: one form or another. You can't take all your existing

595
00:29:56,120 --> 00:29:59,200
Speaker 4: clients with you and then be you're often on an

596
00:29:59,320 --> 00:30:02,920
Speaker 4: enforced garden for a really long time. Yeah, I imagine

597
00:30:02,960 --> 00:30:05,400
Speaker 4: in the massive race that is AI at the moment,

598
00:30:05,840 --> 00:30:06,640
Speaker 4: there's no gardening.

599
00:30:06,760 --> 00:30:09,800
Speaker 5: Also, the nature of the intellectual property I believe is

600
00:30:09,840 --> 00:30:12,360
Speaker 5: a little bit different still. I think of this. I

601
00:30:12,400 --> 00:30:15,920
Speaker 5: think it's a twenty twelve example from Citadel in Chicago

602
00:30:15,960 --> 00:30:20,120
Speaker 5: where a hygh ferquency trader stole some code and they

603
00:30:20,200 --> 00:30:22,920
Speaker 5: sent and he had it on his hard drive. You

604
00:30:22,960 --> 00:30:24,720
Speaker 5: know this story, Yes, it was really.

605
00:30:24,600 --> 00:30:26,600
Speaker 4: Famous when it happened. He like sent it to himself

606
00:30:26,640 --> 00:30:27,120
Speaker 4: for yeah.

607
00:30:27,200 --> 00:30:30,800
Speaker 5: Yeah, So he expltraded some code, some very you know,

608
00:30:30,920 --> 00:30:33,440
Speaker 5: definitive code of how to make money in the market

609
00:30:33,520 --> 00:30:36,520
Speaker 5: and get an edge, and they he figured out that

610
00:30:36,560 --> 00:30:38,800
Speaker 5: they were on his trail, and he threw his hard

611
00:30:38,880 --> 00:30:41,080
Speaker 5: drive into the river and they sent scuba divers into

612
00:30:41,080 --> 00:30:42,000
Speaker 5: the river and got it back.

613
00:30:42,080 --> 00:30:42,240
Speaker 3: Here.

614
00:30:43,400 --> 00:30:47,120
Speaker 6: The other high profile SV example was the guy going

615
00:30:47,200 --> 00:30:50,080
Speaker 6: from Google self driving to yes.

616
00:30:50,360 --> 00:30:52,920
Speaker 5: Yeah and so and so taking specific code.

617
00:30:53,040 --> 00:30:53,880
Speaker 3: That's not happening.

618
00:30:53,920 --> 00:30:56,200
Speaker 5: I mean, there's probably going to be like one example

619
00:30:56,200 --> 00:30:58,840
Speaker 5: of this that pops up, but mostly it's just you

620
00:30:58,880 --> 00:31:02,640
Speaker 5: get someone who says, I understand that at my previous job,

621
00:31:02,840 --> 00:31:05,400
Speaker 5: we scaled up the transformer a little too far and

622
00:31:05,440 --> 00:31:07,960
Speaker 5: we didn't focus on reasoning enough, and so we need

623
00:31:07,960 --> 00:31:11,360
Speaker 5: to shift this spend to test time inference. And it's

624
00:31:11,400 --> 00:31:14,560
Speaker 5: almost like you're leaving you're leaving a company and you're

625
00:31:14,640 --> 00:31:16,760
Speaker 5: and you're just you're you're leaving with like a mindset

626
00:31:16,800 --> 00:31:19,840
Speaker 5: of like the right balance of capex toopics or something.

627
00:31:19,920 --> 00:31:22,480
Speaker 6: What was the recent Jane Street example that came out

628
00:31:22,480 --> 00:31:23,720
Speaker 6: in that lawsuit around it?

629
00:31:23,760 --> 00:31:25,400
Speaker 3: Was it the strategy?

630
00:31:25,800 --> 00:31:26,000
Speaker 4: Yeah?

631
00:31:26,120 --> 00:31:28,040
Speaker 3: It only we we only really all.

632
00:31:27,960 --> 00:31:31,320
Speaker 6: Learned what was happening because there was this lawsuit over

633
00:31:31,880 --> 00:31:35,720
Speaker 6: over basically the team bringing over some type of IP

634
00:31:36,200 --> 00:31:37,080
Speaker 6: or methodology.

635
00:31:37,360 --> 00:31:38,560
Speaker 5: Yeah, can we talk.

636
00:31:38,440 --> 00:31:42,440
Speaker 2: About later the broader sort of economic world right now,

637
00:31:42,440 --> 00:31:44,480
Speaker 2: because there's a bunch of other sort of things going

638
00:31:44,520 --> 00:31:46,280
Speaker 2: on and things that you talk about.

639
00:31:46,320 --> 00:31:49,320
Speaker 6: Touch on them last thing, because I think it's pretty interesting.

640
00:31:49,360 --> 00:31:51,960
Speaker 6: So we put out the metaslist I think it was Monday,

641
00:31:52,480 --> 00:31:55,080
Speaker 6: and we immediately had a ton of inbound from a

642
00:31:55,080 --> 00:31:59,400
Speaker 6: lot of these researchers basically like critiquing like the ranking right,

643
00:31:59,440 --> 00:32:02,160
Speaker 6: and the ranking was like we got we got people that.

644
00:32:02,240 --> 00:32:04,120
Speaker 3: Work in AI to kind of like rank them. We

645
00:32:04,520 --> 00:32:10,680
Speaker 3: the beauty of ranks? Do you get so much the beauty?

646
00:32:11,040 --> 00:32:13,680
Speaker 6: And there's one uh, there was one example that was

647
00:32:13,680 --> 00:32:17,000
Speaker 6: interesting where somebody that was ranked fairly high that I

648
00:32:17,040 --> 00:32:19,920
Speaker 6: know has gotten one of these nine figure comp packages

649
00:32:20,520 --> 00:32:24,479
Speaker 6: and somebody that worked with him basically said this person

650
00:32:24,680 --> 00:32:27,239
Speaker 6: was kicked off of every team they worked on and

651
00:32:27,360 --> 00:32:32,160
Speaker 6: just like effectively like over multi year period, like consistently.

652
00:32:31,760 --> 00:32:33,720
Speaker 3: Demoted over and over and over.

653
00:32:34,200 --> 00:32:38,080
Speaker 6: And now now like got you know this incredible.

654
00:32:37,680 --> 00:32:40,719
Speaker 4: And soft skills in ranking?

655
00:32:40,920 --> 00:32:42,920
Speaker 3: Well yeah, and it was it was more of like

656
00:32:42,960 --> 00:32:44,200
Speaker 3: a technical ability thing.

657
00:32:44,280 --> 00:32:47,760
Speaker 6: It was not even because I think skills are getting

658
00:32:47,760 --> 00:32:51,280
Speaker 6: a little bit uh you know, they don't they don't.

659
00:32:51,480 --> 00:32:53,440
Speaker 6: You don't care that Lebron James like might get a

660
00:32:53,480 --> 00:32:56,960
Speaker 6: little angry at somebody if they underperform, right, No.

661
00:32:56,800 --> 00:33:00,120
Speaker 2: No, no, no, no, there's so this phenomenon of like superstars,

662
00:33:00,160 --> 00:33:02,440
Speaker 2: like it's not just it's not just in tech and

663
00:33:02,480 --> 00:33:05,720
Speaker 2: like we see it. I mentioned in journalism, where you know,

664
00:33:05,800 --> 00:33:09,120
Speaker 2: the sort of like median reporter job in a newsroom.

665
00:33:09,120 --> 00:33:11,800
Speaker 2: A lot of that's hollowing out, but you have some

666
00:33:11,840 --> 00:33:15,480
Speaker 2: people who become insanely well compensated, either because they're really

667
00:33:15,520 --> 00:33:17,880
Speaker 2: smart and they write a great newsletter, or they're really

668
00:33:17,920 --> 00:33:21,120
Speaker 2: good looking and they can do front facing video, et cetera,

669
00:33:21,400 --> 00:33:23,680
Speaker 2: or you know, like the four of us, you know,

670
00:33:23,880 --> 00:33:26,560
Speaker 2: like can do something on video or something like that.

671
00:33:26,760 --> 00:33:29,000
Speaker 2: You see it in the range of areas. And then

672
00:33:29,080 --> 00:33:31,360
Speaker 2: like the amount of betting that's going on, which of

673
00:33:31,400 --> 00:33:33,560
Speaker 2: course adds to this, and so the fact that all

674
00:33:33,600 --> 00:33:36,200
Speaker 2: of these events there's some market out there that you

675
00:33:36,240 --> 00:33:38,880
Speaker 2: can bet on that. And I'm curious, like from your coast,

676
00:33:39,040 --> 00:33:40,920
Speaker 2: you know, you're here in New York, but from your coast,

677
00:33:41,160 --> 00:33:43,520
Speaker 2: what does the world look like in terms of just

678
00:33:43,560 --> 00:33:45,360
Speaker 2: like the state of this economy.

679
00:33:45,520 --> 00:33:48,520
Speaker 5: Yeah, there's something interesting going on in tech where the

680
00:33:48,600 --> 00:33:50,760
Speaker 5: idea of a company going from one hundred billion to

681
00:33:50,800 --> 00:33:54,360
Speaker 5: a trillion seemed unfathomable. Yeah, and there's this take that

682
00:33:54,440 --> 00:33:56,680
Speaker 5: in fact, that was the easiest ten X of all,

683
00:33:57,360 --> 00:34:01,120
Speaker 5: where the hardest ten X was going from zero to one.

684
00:34:01,320 --> 00:34:04,160
Speaker 5: From going to zero to a million dollars and getting

685
00:34:04,200 --> 00:34:07,680
Speaker 5: these systems, inventing PageRank, and then once you had Google

686
00:34:07,760 --> 00:34:10,200
Speaker 5: humming at one hundred billion dollar market cap getting to

687
00:34:10,239 --> 00:34:12,279
Speaker 5: a trillion, not to you knock of the work that

688
00:34:12,280 --> 00:34:15,400
Speaker 5: they did to get that, but these the numbers have

689
00:34:15,480 --> 00:34:18,480
Speaker 5: just kept growing and growing at Internet scale, at Internet speed,

690
00:34:18,800 --> 00:34:22,680
Speaker 5: and so the leverage that you're getting from an AI

691
00:34:22,719 --> 00:34:26,680
Speaker 5: researcher is ever more increasing, and it just adds ten

692
00:34:26,920 --> 00:34:29,120
Speaker 5: x every couple of years. And so ye seeing the

693
00:34:29,200 --> 00:34:30,920
Speaker 5: compackages and increase that way, have.

694
00:34:31,080 --> 00:34:33,080
Speaker 6: A couple of things like the Internet is the greatest

695
00:34:33,160 --> 00:34:37,600
Speaker 6: distribution engine for information and digital products and apps and

696
00:34:37,640 --> 00:34:41,520
Speaker 6: services ever and so at least in venture, that just

697
00:34:41,600 --> 00:34:43,439
Speaker 6: means that everything is faster now.

698
00:34:43,520 --> 00:34:44,440
Speaker 3: Even in media too.

699
00:34:44,480 --> 00:34:47,480
Speaker 6: If you're if you are somebody's working at a at

700
00:34:47,520 --> 00:34:50,040
Speaker 6: a legacy media company and they set up a sub stack,

701
00:34:50,239 --> 00:34:53,520
Speaker 6: they can get a million dollars of ARR on the

702
00:34:53,520 --> 00:34:56,120
Speaker 6: first day that they launch. If somebody is working at

703
00:34:56,120 --> 00:34:58,239
Speaker 6: a media company and really good at a certain type

704
00:34:58,239 --> 00:35:02,080
Speaker 6: of YouTube video, they can launch and immediately the YouTube

705
00:35:02,120 --> 00:35:05,920
Speaker 6: algorithm will serve that video to all of their fans

706
00:35:06,080 --> 00:35:09,000
Speaker 6: within the first week. And that what that does is

707
00:35:09,000 --> 00:35:12,279
Speaker 6: it just changes. You know, the power dynamic between media

708
00:35:12,320 --> 00:35:13,520
Speaker 6: companies you see is.

709
00:35:13,600 --> 00:35:17,000
Speaker 5: Leverage come in through financial markets. If you can lever

710
00:35:17,200 --> 00:35:20,320
Speaker 5: up or just marshal more capital one idea can generate

711
00:35:20,360 --> 00:35:23,120
Speaker 5: billions of dollars in value. Same thing in technology, but

712
00:35:23,200 --> 00:35:26,320
Speaker 5: we aren't seeing it everywhere. I don't think we're seeing

713
00:35:26,360 --> 00:35:29,839
Speaker 5: it in hard skills would working like unless you get

714
00:35:29,840 --> 00:35:32,560
Speaker 5: into true like art territory. But there is this interesting

715
00:35:32,640 --> 00:35:34,839
Speaker 5: question about like what's up next, and we were kind

716
00:35:34,840 --> 00:35:38,280
Speaker 5: of noodling on like there might be a law firm

717
00:35:38,400 --> 00:35:41,279
Speaker 5: in the future that's extremely high leverage in the same

718
00:35:41,320 --> 00:35:44,240
Speaker 5: sense that you might have a lawyer who's so good

719
00:35:44,680 --> 00:35:47,279
Speaker 5: at what they do and they are so good at

720
00:35:47,280 --> 00:35:50,279
Speaker 5: resolving these and the and the connections and everything that

721
00:35:50,320 --> 00:35:53,600
Speaker 5: they're doing. They're making billions, but they have a very

722
00:35:53,680 --> 00:35:55,799
Speaker 5: very lean team. And so you see a much a

723
00:35:55,880 --> 00:35:59,160
Speaker 5: much steeper power law in law. And there's a there's

724
00:35:59,200 --> 00:36:01,839
Speaker 5: a number of other professional services that are going through

725
00:36:02,000 --> 00:36:07,160
Speaker 5: like a transformation with technology bringing increased leverage to the

726
00:36:07,200 --> 00:36:09,960
Speaker 5: profession and that could drive more of that power li

727
00:36:10,000 --> 00:36:11,000
Speaker 5: outcome in terms of earning.

728
00:36:11,080 --> 00:36:13,279
Speaker 6: Yeah, the thing you know in the private markets on

729
00:36:13,600 --> 00:36:16,560
Speaker 6: the West Coast, which is the dynamic right now that's

730
00:36:16,600 --> 00:36:18,440
Speaker 6: fascinating is like, I feel like a lot of people

731
00:36:18,480 --> 00:36:20,960
Speaker 6: have like a little bit of PTSD from the twenty

732
00:36:21,280 --> 00:36:25,239
Speaker 6: twenty to twenty twenty two era, where many of the

733
00:36:25,280 --> 00:36:28,799
Speaker 6: things that in hindsight were incredible top signals are like

734
00:36:28,880 --> 00:36:31,480
Speaker 6: all pop popping up again right now, and like we like,

735
00:36:31,760 --> 00:36:34,319
Speaker 6: we like trap, we track them just for fun. We

736
00:36:34,400 --> 00:36:36,520
Speaker 6: are not in the business of like calling the top

737
00:36:36,600 --> 00:36:38,840
Speaker 6: or anything like that. And in many ways it feels like,

738
00:36:39,400 --> 00:36:43,120
Speaker 6: you know, there's so many positive indicators. But the interesting

739
00:36:43,200 --> 00:36:46,240
Speaker 6: dynamic and in venture is like, as an angel investor,

740
00:36:46,239 --> 00:36:50,000
Speaker 6: I've invested in probably sixty five or so different companies

741
00:36:50,040 --> 00:36:53,120
Speaker 6: at the pre seed to Series A stage, and there

742
00:36:53,160 --> 00:36:54,640
Speaker 6: was a bunch of deals that I did in twenty

743
00:36:54,680 --> 00:36:57,400
Speaker 6: twenty one twenty twenty two that ultimately like I paid

744
00:36:57,440 --> 00:37:01,359
Speaker 6: too much or just like the team wasn't good enough

745
00:37:01,400 --> 00:37:03,719
Speaker 6: to execute against the vision they had. But there was

746
00:37:03,760 --> 00:37:05,560
Speaker 6: also just like a handful of deals that I did

747
00:37:05,560 --> 00:37:08,520
Speaker 6: that that have been you know, performed so well that

748
00:37:08,560 --> 00:37:10,600
Speaker 6: it doesn't matter that I did a bunch of silly deals.

749
00:37:10,640 --> 00:37:13,239
Speaker 6: And so venture right now is this interesting kind of

750
00:37:13,320 --> 00:37:16,440
Speaker 6: dynamic where everybody knows that it's crazy, Right, you have

751
00:37:16,640 --> 00:37:19,759
Speaker 6: hundreds of billions of dollars of value created in the

752
00:37:19,760 --> 00:37:22,520
Speaker 6: private markets that there's a lot of real revenue growth,

753
00:37:22,520 --> 00:37:24,440
Speaker 6: but there's also hundreds of billions of dollars of value

754
00:37:24,480 --> 00:37:27,200
Speaker 6: tied to zero revenue, right. And I think that there's

755
00:37:27,280 --> 00:37:30,520
Speaker 6: something that we've been tracking is like this underlying kind

756
00:37:30,560 --> 00:37:34,080
Speaker 6: of feeling from people that are maybe under thirty of

757
00:37:34,320 --> 00:37:36,880
Speaker 6: there's like this meme that's become very prevalent, and I

758
00:37:36,880 --> 00:37:40,879
Speaker 6: think it explains a lot of economic activity today, which

759
00:37:40,960 --> 00:37:44,200
Speaker 6: is young people feel that they have two years to

760
00:37:44,320 --> 00:37:48,239
Speaker 6: accumulate capital to escape the permanent underclass. And so I

761
00:37:48,239 --> 00:37:52,680
Speaker 6: think that drives a lot of investing activity today in

762
00:37:52,719 --> 00:37:55,520
Speaker 6: that you know, you'll see young people on the timeline

763
00:37:55,560 --> 00:37:58,560
Speaker 6: saying that like you'd be stupid not to use leverage,

764
00:37:58,640 --> 00:38:00,480
Speaker 6: you know, and they're and they're just like, you know,

765
00:38:00,560 --> 00:38:01,880
Speaker 6: they're not a professional.

766
00:38:01,520 --> 00:38:04,200
Speaker 3: Investor in any capacity. Don't do it.

767
00:38:05,880 --> 00:38:08,440
Speaker 6: And you know, any anytime you have people in venture

768
00:38:08,640 --> 00:38:13,160
Speaker 6: making public market stock predictions, we get a little concerned.

769
00:38:13,719 --> 00:38:16,960
Speaker 4: Are we going to get AI talent agents? Not AI agents,

770
00:38:16,960 --> 00:38:19,640
Speaker 4: We have those already, but AI talent agents because like.

771
00:38:19,560 --> 00:38:22,400
Speaker 5: We do so we do they're called venture capitalists.

772
00:38:22,560 --> 00:38:26,680
Speaker 4: Oh well, but yes, this was going to be a sentence.

773
00:38:26,880 --> 00:38:29,719
Speaker 4: So if the money is no longer in, you know,

774
00:38:29,760 --> 00:38:32,240
Speaker 4: I invest in a startup and eventually they get the exit,

775
00:38:32,360 --> 00:38:34,520
Speaker 4: they get acquired, or they list or whatever. If the

776
00:38:34,560 --> 00:38:38,400
Speaker 4: money is in, eventually the guy that started the startup

777
00:38:38,520 --> 00:38:42,520
Speaker 4: gets bought by someone like Meta or whoever. Wouldn't I

778
00:38:42,520 --> 00:38:45,200
Speaker 4: invest my money in that person versus in the startup

779
00:38:45,280 --> 00:38:46,080
Speaker 4: inventure service?

780
00:38:46,160 --> 00:38:48,480
Speaker 5: Yes, yeah, you can't quite do that, but there are

781
00:38:48,520 --> 00:38:51,200
Speaker 5: a ton of roles that are indexed to these high

782
00:38:51,200 --> 00:38:54,120
Speaker 5: performance packages. And yes, there are people right now in

783
00:38:54,160 --> 00:38:57,360
Speaker 5: Silicon Valley who are effectively talent agents who get a

784
00:38:57,440 --> 00:39:00,000
Speaker 5: cut of those big packages and they help me get

785
00:39:00,160 --> 00:39:01,160
Speaker 5: siate just.

786
00:39:01,440 --> 00:39:04,799
Speaker 3: Called just talent, called the venture capitalists. No, no, so so

787
00:39:05,000 --> 00:39:05,359
Speaker 3: you can.

788
00:39:05,440 --> 00:39:10,800
Speaker 5: You can you find a great researcher and you think that, okay,

789
00:39:11,120 --> 00:39:13,120
Speaker 5: maybe they will build a business, but I'm sure that

790
00:39:13,160 --> 00:39:15,040
Speaker 5: they are going to be worth hundreds of millions of

791
00:39:15,040 --> 00:39:17,560
Speaker 5: dollars a year, and you can just invest in their company.

792
00:39:17,800 --> 00:39:20,840
Speaker 5: You will almost certainly see a good return on that investment,

793
00:39:20,880 --> 00:39:23,640
Speaker 5: even if the product never gets to scale, because when

794
00:39:23,640 --> 00:39:26,000
Speaker 5: they get an ACQUA higher, you will get a.

795
00:39:25,920 --> 00:39:31,560
Speaker 2: Payout, not going there unlike the.

796
00:39:33,560 --> 00:39:36,360
Speaker 6: Are like, oh absolutely no, it's now it's now so

797
00:39:36,560 --> 00:39:40,160
Speaker 6: normalized to invest in college dropouts that you actually see

798
00:39:40,640 --> 00:39:42,799
Speaker 6: people like investing in high school.

799
00:39:43,400 --> 00:39:45,239
Speaker 5: I'm not kidding about this. So the Iomo Gold Medal

800
00:39:45,320 --> 00:39:49,080
Speaker 5: is the math Olympiad, and there are venture capitalists who

801
00:39:49,120 --> 00:39:52,120
Speaker 5: will give calls to every single student that performs well

802
00:39:52,120 --> 00:39:53,960
Speaker 5: on them on the math Olympiad. And these are high

803
00:39:53,960 --> 00:39:54,560
Speaker 5: school students.

804
00:39:54,600 --> 00:39:57,560
Speaker 4: So do they what are these investments? Exactly like, are

805
00:39:57,640 --> 00:39:59,359
Speaker 4: you're funding their tuition or no?

806
00:39:59,360 --> 00:40:02,240
Speaker 5: No, no, it's it's it's I am taking ten or twenty

807
00:40:02,239 --> 00:40:05,080
Speaker 5: percent of a company of a Delaware c core. Most

808
00:40:05,160 --> 00:40:08,719
Speaker 5: likely that you will build something in and who knows

809
00:40:08,760 --> 00:40:11,360
Speaker 5: where it goes. Maybe it turns into a great business,

810
00:40:11,440 --> 00:40:15,160
Speaker 5: maybe it turns into aquhire. But the downside is extremely

811
00:40:15,239 --> 00:40:18,560
Speaker 5: limited because there's always this at least right now, there's

812
00:40:18,600 --> 00:40:21,920
Speaker 5: this aqui hire, there's this AQUHIRER on the table where

813
00:40:22,440 --> 00:40:23,040
Speaker 5: it used to be.

814
00:40:23,080 --> 00:40:25,520
Speaker 6: That to be clear that it is only an AI

815
00:40:25,920 --> 00:40:29,960
Speaker 6: yes only for the people that you would consider to

816
00:40:29,960 --> 00:40:32,480
Speaker 6: be the top five hundred yes in there industry.

817
00:40:32,480 --> 00:40:35,239
Speaker 5: So to go back to look at Cursor windsurf thing

818
00:40:35,520 --> 00:40:38,040
Speaker 5: like Instagram was the power law winner got the billion

819
00:40:38,080 --> 00:40:42,640
Speaker 5: dollar acquisition from Meta Hipstomatic was the second largest photo

820
00:40:42,680 --> 00:40:47,000
Speaker 5: filtering app, did not get a billion dollar aquhier from Google. Right,

821
00:40:47,239 --> 00:40:49,560
Speaker 5: But now we're in the market where if there's a

822
00:40:49,680 --> 00:40:51,839
Speaker 5: leading product with a bunch of AI researchers over here,

823
00:40:51,840 --> 00:40:54,280
Speaker 5: and they get a multi billion dollar acquisition for the product,

824
00:40:54,280 --> 00:40:56,000
Speaker 5: and the product's working and it's growing and it is

825
00:40:56,000 --> 00:40:57,960
Speaker 5: a great business, and you buy it for the value

826
00:40:57,960 --> 00:41:00,960
Speaker 5: of the business, then the second best team might get

827
00:41:00,960 --> 00:41:04,000
Speaker 5: acquired just for talent, which is a completely different downside protection.

828
00:41:04,280 --> 00:41:08,319
Speaker 6: Yeah, and there's a lot of talent acquisitions traditional acquires

829
00:41:08,440 --> 00:41:11,799
Speaker 6: where it's only the talent that benefit, right in the

830
00:41:11,840 --> 00:41:14,120
Speaker 6: sense that they get basically a job offer at the

831
00:41:14,160 --> 00:41:17,520
Speaker 6: new company, and vcs get some capital back or in

832
00:41:17,560 --> 00:41:19,280
Speaker 6: some cases a small amount of money.

833
00:41:19,920 --> 00:41:23,960
Speaker 2: Can I ask a TVPN question This should be like

834
00:41:24,000 --> 00:41:27,000
Speaker 2: the ultimate compliment, which is that I got a d

835
00:41:27,120 --> 00:41:29,319
Speaker 2: M from some random person I've never seen them the

836
00:41:29,360 --> 00:41:34,319
Speaker 2: other day and he said I can build tvpns for X,

837
00:41:34,640 --> 00:41:37,319
Speaker 2: So like that's that's deck. That's sort of which means

838
00:41:37,320 --> 00:41:39,920
Speaker 2: it's sort of like it's become like Kleenex or one

839
00:41:39,960 --> 00:41:43,080
Speaker 2: of these things where it's like a category and where

840
00:41:43,120 --> 00:41:44,680
Speaker 2: are you going with it? What are your playing?

841
00:41:44,960 --> 00:41:48,000
Speaker 6: It's so funny that people are calling it TVPN for

842
00:41:48,280 --> 00:41:51,560
Speaker 6: X because our show, while it's unique in a variety

843
00:41:51,600 --> 00:41:56,000
Speaker 6: of ways, looks very much like traditional is interesting, and so.

844
00:41:56,520 --> 00:41:59,840
Speaker 3: We can't we can't take credit for we held it.

845
00:42:00,400 --> 00:42:03,560
Speaker 5: We invented media, we invented live streaming.

846
00:42:03,640 --> 00:42:06,600
Speaker 2: Of course, course normally it doesn't really I don't know,

847
00:42:06,880 --> 00:42:08,560
Speaker 2: why would I watch cable news on.

848
00:42:09,280 --> 00:42:11,840
Speaker 6: I think thing that's exciting is that in the private

849
00:42:11,880 --> 00:42:15,560
Speaker 6: markets and venture and tech, if you had a podcast,

850
00:42:15,640 --> 00:42:18,080
Speaker 6: there was one format which was a once a week

851
00:42:18,440 --> 00:42:22,600
Speaker 6: interview show, and it was a great strategy to do

852
00:42:22,640 --> 00:42:25,120
Speaker 6: that ten years ago, around the time that you guys.

853
00:42:26,560 --> 00:42:30,640
Speaker 3: It locked established.

854
00:42:30,960 --> 00:42:33,000
Speaker 5: But if we if we tried to clone this, yeah,

855
00:42:33,080 --> 00:42:34,839
Speaker 5: everyone would be like, why is that a knockoff? There's

856
00:42:34,840 --> 00:42:37,759
Speaker 5: nothing new about this, nothing fresh about this, Like why

857
00:42:37,840 --> 00:42:39,440
Speaker 5: do I want to go on your knockoff? I just

858
00:42:39,480 --> 00:42:40,120
Speaker 5: go on the real thing.

859
00:42:40,280 --> 00:42:40,560
Speaker 3: Yeah.

860
00:42:40,760 --> 00:42:43,920
Speaker 6: And so our edge in launching the show at the

861
00:42:43,960 --> 00:42:47,839
Speaker 6: beginning of the year is that media is not zero sum.

862
00:42:48,280 --> 00:42:49,480
Speaker 3: Content is not zero s.

863
00:42:49,719 --> 00:42:52,000
Speaker 6: We have a friend that jokes that he's so competitive

864
00:42:52,000 --> 00:42:55,160
Speaker 6: he wishes he wishes media zero sum it was truly

865
00:42:55,239 --> 00:42:58,560
Speaker 6: zero s. But our edge early was that we just

866
00:42:58,600 --> 00:43:01,359
Speaker 6: took it ten times more seriously than anyone else. So

867
00:43:01,680 --> 00:43:04,239
Speaker 6: everybody that was creating content for the private markets was

868
00:43:04,280 --> 00:43:07,160
Speaker 6: doing it as a part time gig, and we were

869
00:43:07,400 --> 00:43:09,399
Speaker 6: happy to compete with a bunch of people that were

870
00:43:09,480 --> 00:43:09,959
Speaker 6: part time.

871
00:43:10,560 --> 00:43:12,920
Speaker 2: You know what, Actually, I was wondering in twenty twenty

872
00:43:12,960 --> 00:43:17,400
Speaker 2: one or twenty twenty two that craziness. Someone once reached

873
00:43:17,440 --> 00:43:20,239
Speaker 2: out to me and they're like, you know what, you

874
00:43:20,280 --> 00:43:26,640
Speaker 2: should like, no direct, it's probably the no, but this

875
00:43:26,760 --> 00:43:28,680
Speaker 2: was this was the interesting thing. He said, you should

876
00:43:28,680 --> 00:43:30,920
Speaker 2: go independent and what you should do is attach a

877
00:43:31,000 --> 00:43:33,680
Speaker 2: VC armed to it, and because of the quality of

878
00:43:33,680 --> 00:43:35,760
Speaker 2: the guests that you could get, you would get really

879
00:43:36,200 --> 00:43:38,720
Speaker 2: good access to deal flow. But it seems like basically

880
00:43:39,520 --> 00:43:41,360
Speaker 2: it's very uncomfortable to me because I'm not going to

881
00:43:41,719 --> 00:43:45,800
Speaker 2: highlight startup founders and say, oh, you're the perfect guest.

882
00:43:46,040 --> 00:43:47,880
Speaker 2: And it's like because like I have like five percent.

883
00:43:48,640 --> 00:43:51,600
Speaker 2: But I'm curious about because you mentioned doing Angel the

884
00:43:51,680 --> 00:43:54,959
Speaker 2: sort of link of the TVPN at some point, like, yeah,

885
00:43:55,080 --> 00:43:57,600
Speaker 2: so an investing.

886
00:43:57,160 --> 00:44:01,000
Speaker 6: Platform Angel investing. I look at it as a hobby.

887
00:44:01,200 --> 00:44:04,000
Speaker 6: It's not a great financial activity, right, locking up, I've

888
00:44:04,040 --> 00:44:06,720
Speaker 6: heard locking out. I mean, like you can generate great returns,

889
00:44:06,719 --> 00:44:09,959
Speaker 6: but it's not the most logical way to invest your time.

890
00:44:10,000 --> 00:44:13,480
Speaker 6: But it's really fun. Like we just enjoy supporting founders

891
00:44:13,719 --> 00:44:15,839
Speaker 6: early when it's an idea and a team and it's

892
00:44:16,200 --> 00:44:20,080
Speaker 6: just it is genuinely addicting. I always joke about people

893
00:44:20,120 --> 00:44:23,560
Speaker 6: in San Francisco with Angel. You know, people across the country,

894
00:44:23,560 --> 00:44:28,160
Speaker 6: sports betting addictions san Francisco, you know, Angel investing addictions

895
00:44:28,160 --> 00:44:32,080
Speaker 6: are I think real. But for us, for us, that was,

896
00:44:32,160 --> 00:44:35,040
Speaker 6: you know, as we started having some success a lot

897
00:44:35,040 --> 00:44:38,040
Speaker 6: of people, they would probably get a message today, when's

898
00:44:38,040 --> 00:44:40,560
Speaker 6: the fun coming, when's the fund coming? And that's been

899
00:44:40,600 --> 00:44:44,239
Speaker 6: a way to monetize an audience within tech. If you

900
00:44:44,320 --> 00:44:47,080
Speaker 6: have an audience, go raise one hundred million dollar fund,

901
00:44:47,360 --> 00:44:49,719
Speaker 6: you get the fee streams do that, you get a

902
00:44:49,719 --> 00:44:54,160
Speaker 6: bunch of upside. And we joke because we have the shows.

903
00:44:54,160 --> 00:44:56,719
Speaker 6: Every single day we go live at eleven. We have

904
00:44:56,840 --> 00:45:00,440
Speaker 6: to eat, we have to work out, we have to

905
00:45:00,440 --> 00:45:02,880
Speaker 6: prep the show, we have to talk with partners, we

906
00:45:02,960 --> 00:45:05,600
Speaker 6: have to manage our team. There's all these different things

907
00:45:06,120 --> 00:45:09,640
Speaker 6: and so somebody might immediately think, Okay, these guys talk

908
00:45:09,719 --> 00:45:13,319
Speaker 6: to six founders and investors a day. They have this

909
00:45:13,440 --> 00:45:17,080
Speaker 6: like media property that like, I think everybody in ventured

910
00:45:17,160 --> 00:45:19,160
Speaker 6: now is going to see our content like once a

911
00:45:19,160 --> 00:45:22,000
Speaker 6: week in some form or another if they're if they're online.

912
00:45:22,520 --> 00:45:25,000
Speaker 6: But I joke with them, I'm like, so we're live

913
00:45:25,160 --> 00:45:27,920
Speaker 6: for three hours every single day during the middle of

914
00:45:28,000 --> 00:45:31,360
Speaker 6: the day. And so if you think that we with

915
00:45:31,440 --> 00:45:35,640
Speaker 6: a show would would effectively compete at VC is about

916
00:45:35,640 --> 00:45:39,640
Speaker 6: winning allocation, right. You can be cool and connected and

917
00:45:39,640 --> 00:45:41,760
Speaker 6: have an audience that might get you one hundred K allocation,

918
00:45:42,040 --> 00:45:43,680
Speaker 6: but if you have one hundred million dollar fund, you

919
00:45:43,719 --> 00:45:47,160
Speaker 6: need to be putting size into deals. And so I

920
00:45:47,239 --> 00:45:49,560
Speaker 6: know that if John and Jordi have a VC fund

921
00:45:49,640 --> 00:45:51,640
Speaker 6: and we're competing with these other guys that have a.

922
00:45:51,600 --> 00:45:53,520
Speaker 3: Live show that they're live for three hours a day

923
00:45:53,520 --> 00:45:53,960
Speaker 3: and they have.

924
00:45:53,920 --> 00:45:56,800
Speaker 6: A VC fund, we'd smoke them because we'd be like, Okay,

925
00:45:56,840 --> 00:45:58,520
Speaker 6: well they're live. I'm going to fly to the founder,

926
00:45:58,520 --> 00:46:00,480
Speaker 6: I'm gonna meet with them, and I'm we're gonna win

927
00:46:00,480 --> 00:46:01,920
Speaker 6: this deal, and we're gonna be like, yeah, why why

928
00:46:01,920 --> 00:46:03,640
Speaker 6: don't you let them put in like two hundred k.

929
00:46:03,880 --> 00:46:06,640
Speaker 4: So the thing to watch out for when the VC

930
00:46:06,760 --> 00:46:09,480
Speaker 4: fund is coming, is when you start reducing your live hours,

931
00:46:09,680 --> 00:46:10,359
Speaker 4: that'll be the same.

932
00:46:11,440 --> 00:46:11,640
Speaker 1: Yeah.

933
00:46:11,760 --> 00:46:13,799
Speaker 6: I think I think we we didn't get into this

934
00:46:13,960 --> 00:46:15,520
Speaker 6: to start a fund, and I think there's a lot

935
00:46:15,520 --> 00:46:17,680
Speaker 6: of people that get into content because they see it

936
00:46:17,680 --> 00:46:20,359
Speaker 6: as a way to do that and we just love

937
00:46:20,480 --> 00:46:21,239
Speaker 6: talking about tech.

938
00:46:21,440 --> 00:46:24,520
Speaker 5: It actually was the key insight was that not enough people.

939
00:46:24,680 --> 00:46:27,360
Speaker 5: Yeah it's very fun, but not but very few people

940
00:46:27,360 --> 00:46:30,359
Speaker 5: in tech were actually taking media seriously. Yeah, it was

941
00:46:30,560 --> 00:46:33,200
Speaker 5: everyone had a fund, everyone that was the high status thing,

942
00:46:33,520 --> 00:46:36,439
Speaker 5: and doing media was like lower status or people didn't

943
00:46:36,480 --> 00:46:38,239
Speaker 5: think we could have as much of a power all

944
00:46:38,360 --> 00:46:41,520
Speaker 5: outcome as it very clearly can. And so this idea

945
00:46:41,640 --> 00:46:45,400
Speaker 5: of of just what if you actually took it completely

946
00:46:45,440 --> 00:46:47,320
Speaker 5: seriously and just made it the main.

947
00:46:47,200 --> 00:46:49,160
Speaker 6: Thing, and when you were We've had a bunch of

948
00:46:49,200 --> 00:46:54,040
Speaker 6: people copy our format, you know, major legacy media companies

949
00:46:54,160 --> 00:46:57,839
Speaker 6: all the way through friends of ours. It doesn't really

950
00:46:57,880 --> 00:47:00,080
Speaker 6: bother I mean it bothers me more than John. But

951
00:47:00,120 --> 00:47:02,760
Speaker 6: at the end of the day, if you want to

952
00:47:02,800 --> 00:47:07,200
Speaker 6: spend the tracy, yeah, I totally got it. And so

953
00:47:07,239 --> 00:47:09,600
Speaker 6: it's like it doesn't basically know so so John and

954
00:47:09,680 --> 00:47:13,000
Speaker 6: I John and I basically hang out for twelve hours

955
00:47:13,040 --> 00:47:17,080
Speaker 6: a day, and the entire time, the entire time we're thinking,

956
00:47:17,120 --> 00:47:20,320
Speaker 6: we're thinking about the show, we're just talking about the

957
00:47:20,360 --> 00:47:22,920
Speaker 6: show doesn't look very different than when we're hanging out offline.

958
00:47:23,320 --> 00:47:25,319
Speaker 6: And so I just joke, I'm like, Okay, if you

959
00:47:25,320 --> 00:47:27,480
Speaker 6: want to compete with us and you're willing to put

960
00:47:27,480 --> 00:47:30,800
Speaker 6: it one hundred hours a week in god by all means,

961
00:47:31,080 --> 00:47:33,399
Speaker 6: go for it like this probably your life's work.

962
00:47:33,600 --> 00:47:35,879
Speaker 3: But if it's not good luck, it doesn't matter.

963
00:47:36,239 --> 00:47:39,120
Speaker 2: John Cougan, Jordie Hayes, thank you so much for coming

964
00:47:39,120 --> 00:47:43,239
Speaker 2: on Odd Loss that's in good luck and looking forward

965
00:47:43,280 --> 00:47:45,520
Speaker 2: to continue watching TVPM.

966
00:47:45,640 --> 00:47:52,080
Speaker 7: We'll be live from Nicey later Odd Lots in the

967
00:47:52,120 --> 00:47:57,279
Speaker 7: same day and in the wake of meadow winch mech

968
00:47:57,360 --> 00:47:58,040
Speaker 7: me a.

969
00:47:58,400 --> 00:48:13,480
Speaker 3: Thank you for having us. Thank you so much, Tracy.

970
00:48:13,560 --> 00:48:14,279
Speaker 3: That was a lot of fun.

971
00:48:14,640 --> 00:48:17,040
Speaker 4: It was a little bit media naval gazing, a little

972
00:48:17,080 --> 00:48:17,719
Speaker 4: bit fun.

973
00:48:17,719 --> 00:48:20,239
Speaker 2: A little bit media naval gazing. I mean, there is

974
00:48:20,280 --> 00:48:23,160
Speaker 2: this thing that's happening. It's been happening in media for

975
00:48:23,200 --> 00:48:26,880
Speaker 2: a while, and of course it happens in Wall Street

976
00:48:27,800 --> 00:48:30,600
Speaker 2: and now happening in AI where you just have a

977
00:48:30,640 --> 00:48:34,840
Speaker 2: lot of talented people, and they wonder about the degree

978
00:48:35,000 --> 00:48:38,239
Speaker 2: to which they need their existing platform there. You know,

979
00:48:38,400 --> 00:48:40,960
Speaker 2: like a star banker can take a book of business,

980
00:48:41,040 --> 00:48:43,320
Speaker 2: or a star lawyer can take a book of business.

981
00:48:43,800 --> 00:48:45,640
Speaker 2: And in the case of a it's not taking a

982
00:48:45,640 --> 00:48:47,719
Speaker 2: book of business, right because they don't have like their

983
00:48:47,760 --> 00:48:52,200
Speaker 2: individual clients. But it's this knowledge and transport it and

984
00:48:52,239 --> 00:48:54,839
Speaker 2: it's instantly worth a lot of money for someone somewhere else.

985
00:48:54,920 --> 00:48:57,200
Speaker 4: The thing I thought was really interesting about that discussion

986
00:48:57,280 --> 00:49:01,080
Speaker 4: was the emphasis on how capital intent all of AI

987
00:49:01,320 --> 00:49:04,360
Speaker 4: is and so that kind of changes the economics of

988
00:49:04,400 --> 00:49:07,719
Speaker 4: why you're paying such a massive money for someone who's

989
00:49:07,800 --> 00:49:10,319
Speaker 4: able to eke out even a slight efficiency.

990
00:49:10,560 --> 00:49:12,760
Speaker 3: This is huge.

991
00:49:13,000 --> 00:49:16,040
Speaker 2: This was like this It suddenly is what made it

992
00:49:16,080 --> 00:49:19,000
Speaker 2: all made sense to me, right, because we know about

993
00:49:19,040 --> 00:49:22,879
Speaker 2: like how costly one training run is, right, and we

994
00:49:22,960 --> 00:49:26,200
Speaker 2: know just the insane numbers for data center set up,

995
00:49:26,239 --> 00:49:28,640
Speaker 2: et cetera. And I'm sure there's progress being made on

996
00:49:28,680 --> 00:49:33,920
Speaker 2: the literal design of like how you string together in video, GPS,

997
00:49:33,960 --> 00:49:36,080
Speaker 2: et cetera. And so if you could get some margin,

998
00:49:36,239 --> 00:49:37,960
Speaker 2: if you know have that know how to get some

999
00:49:37,960 --> 00:49:40,879
Speaker 2: marginal improvement out of it, And this is fundamentally what was.

1000
00:49:40,960 --> 00:49:42,720
Speaker 4: One trillion dollars for you?

1001
00:49:42,760 --> 00:49:45,680
Speaker 2: Maybe not a trill No, not a trilliont but like

1002
00:49:45,920 --> 00:49:49,359
Speaker 2: this was not the case when tech was not so

1003
00:49:49,440 --> 00:49:53,160
Speaker 2: capital intensive in the twenty tens. Where yeah, I'm sure

1004
00:49:53,200 --> 00:49:55,279
Speaker 2: talented people always made a lot of money and there's

1005
00:49:55,280 --> 00:49:59,040
Speaker 2: always improvements, but where it's so that link between some

1006
00:49:59,080 --> 00:50:02,160
Speaker 2: sort of efficiency gain and instant cost savings is so

1007
00:50:02,320 --> 00:50:04,080
Speaker 2: linear and so straightforward. Yeah.

1008
00:50:04,200 --> 00:50:06,840
Speaker 4: And I take their point that, Okay, vcs exist and

1009
00:50:06,840 --> 00:50:09,680
Speaker 4: they're already investing in some ways and specific talent. But

1010
00:50:09,719 --> 00:50:11,239
Speaker 4: I do kind of wonder if you're going to get

1011
00:50:11,239 --> 00:50:14,920
Speaker 4: some sort of like specialized headhunters at the very least,

1012
00:50:14,960 --> 00:50:17,040
Speaker 4: who are going to like seek out these big AI

1013
00:50:17,160 --> 00:50:19,640
Speaker 4: talents and try to like graft themselves onto them.

1014
00:50:19,719 --> 00:50:25,880
Speaker 2: Yeah, just people whose expertise is in reading through under cited,

1015
00:50:26,520 --> 00:50:28,719
Speaker 2: under cited AI research papers.

1016
00:50:29,160 --> 00:50:30,080
Speaker 3: Yeah, AI research.

1017
00:50:30,280 --> 00:50:32,480
Speaker 2: Well, we can't give you Sam Altman, but what if

1018
00:50:32,480 --> 00:50:34,400
Speaker 2: we could replace Sam Altman in the egg.

1019
00:50:34,280 --> 00:50:38,120
Speaker 4: Guy has fifty citations in the following papers? Should we

1020
00:50:38,160 --> 00:50:38,520
Speaker 4: leave it there?

1021
00:50:38,560 --> 00:50:39,239
Speaker 3: Let's leave it there.

1022
00:50:39,400 --> 00:50:41,640
Speaker 4: This has been another episode of the All Thoughts podcast.

1023
00:50:41,719 --> 00:50:44,959
Speaker 4: I'm Tracy Alloway. You can follow me at Tracy Alloway and.

1024
00:50:44,880 --> 00:50:47,560
Speaker 2: I'm Jill Wisenthal. You can follow me at the Stalwart,

1025
00:50:47,719 --> 00:50:50,520
Speaker 2: follow our guest Jordi Hayes He's at Jordi Hayes, and

1026
00:50:50,640 --> 00:50:54,760
Speaker 2: John Cougan at John Coogan, and check out TBPN at TBPN.

1027
00:50:55,080 --> 00:50:58,040
Speaker 2: Follow our producers Carmen Rodriguez at Carmen Arman, dash Ol

1028
00:50:58,040 --> 00:51:00,880
Speaker 2: Bennett at dashbod and kel Brooks at Kale Brooks. And

1029
00:51:00,920 --> 00:51:03,520
Speaker 2: from our Odlots content good to Bloomberg dot com slash

1030
00:51:03,520 --> 00:51:06,239
Speaker 2: odd lots were the daily newsletter and all of our episodes,

1031
00:51:06,440 --> 00:51:08,319
Speaker 2: and you can chat about all of these topics twenty

1032
00:51:08,360 --> 00:51:11,720
Speaker 2: four to seven in our discord Discord dot gg slash

1033
00:51:11,760 --> 00:51:12,440
Speaker 2: od loots.

1034
00:51:12,600 --> 00:51:14,759
Speaker 4: And if you enjoy odd Lots, if you like it

1035
00:51:14,800 --> 00:51:18,359
Speaker 4: when we talk about the sportification of AI talent, then

1036
00:51:18,360 --> 00:51:21,520
Speaker 4: please leave us a positive review on your favorite podcast platform.

1037
00:51:21,680 --> 00:51:24,520
Speaker 4: And remember, if you are a Bloomberg subscriber, you can

1038
00:51:24,560 --> 00:51:27,879
Speaker 4: listen to all of our episodes absolutely ad free. All

1039
00:51:27,920 --> 00:51:29,680
Speaker 4: you need to do is find the Bloomberg channel on

1040
00:51:29,719 --> 00:51:33,200
Speaker 4: Apple Podcasts and follow the instructions there. Thanks for listening.

1041
00:52:04,200 --> 00:52:04,239
Speaker 4: H