1
00:00:02,480 --> 00:00:14,000
Speaker 1: Bloomberg Audio Studios, Podcasts, radio News.

2
00:00:18,239 --> 00:00:21,960
Speaker 2: Hello and welcome to another episode of the Odd Lots podcast.

3
00:00:22,079 --> 00:00:24,400
Speaker 3: I'm Joe Wisenthal and I'm Tracy Alloway.

4
00:00:24,840 --> 00:00:28,080
Speaker 2: You know, sometimes with the whole AI thing, it feels

5
00:00:28,120 --> 00:00:30,280
Speaker 2: like the are we back? Is it over a thing?

6
00:00:30,320 --> 00:00:32,760
Speaker 2: There's been a few moments in the last year whatever

7
00:00:32,960 --> 00:00:34,199
Speaker 2: was like, is the bubble burst in?

8
00:00:34,320 --> 00:00:34,560
Speaker 3: Yeah?

9
00:00:34,720 --> 00:00:36,559
Speaker 2: Is there a bubble? I don't know where what this

10
00:00:36,640 --> 00:00:39,480
Speaker 2: week is. We're recording the September third. There's a little

11
00:00:39,520 --> 00:00:41,240
Speaker 2: I don't know, there's some tremors. I can't tell what's

12
00:00:41,240 --> 00:00:41,640
Speaker 2: real or not.

13
00:00:41,800 --> 00:00:44,680
Speaker 3: I feel like the hype cycles get shorter and shorter

14
00:00:44,840 --> 00:00:48,040
Speaker 3: and more compressed. But you're right, I think there are

15
00:00:48,440 --> 00:00:51,599
Speaker 3: maybe some more jitters than there have been previously. Like

16
00:00:51,720 --> 00:00:55,240
Speaker 3: maybe it's hard to tell. It's hard to tell, but

17
00:00:55,400 --> 00:00:58,760
Speaker 3: I think even if we're not there yet, we are

18
00:00:58,880 --> 00:01:02,600
Speaker 3: getting maybe to that point where like the rubber needs

19
00:01:02,640 --> 00:01:06,080
Speaker 3: to meet the road in terms of monetization, like you know,

20
00:01:06,200 --> 00:01:11,679
Speaker 3: everyone got maybe okay, maybe yeah, you're right, maybe yesterday or.

21
00:01:12,000 --> 00:01:15,360
Speaker 2: Either this morning or yesterday. Anthropic one hundred and eighty

22
00:01:15,400 --> 00:01:20,520
Speaker 2: three billion dollar valuation. This does not strike me as.

23
00:01:20,640 --> 00:01:23,360
Speaker 3: People worried about monet people well.

24
00:01:23,200 --> 00:01:25,399
Speaker 2: Or suddenly people are worried about valuation. It sounds like

25
00:01:25,400 --> 00:01:27,000
Speaker 2: people really want to pour a lot of money into

26
00:01:27,040 --> 00:01:27,680
Speaker 2: these companies.

27
00:01:27,840 --> 00:01:30,760
Speaker 3: Yeah, it is. It's a weird environment. Let's just put

28
00:01:30,760 --> 00:01:32,840
Speaker 3: it that way. That's the only thing we can say

29
00:01:32,840 --> 00:01:35,039
Speaker 3: with certainty that doesn't start with a maybe.

30
00:01:35,440 --> 00:01:35,880
Speaker 1: That's right.

31
00:01:36,160 --> 00:01:39,399
Speaker 2: Here's another thing we could say with certainty alphabet. The

32
00:01:39,520 --> 00:01:43,479
Speaker 2: shares are up as we're speaking, eight point something percent today.

33
00:01:43,680 --> 00:01:45,640
Speaker 2: This was a company people were sort of worried about

34
00:01:45,640 --> 00:01:48,280
Speaker 2: how they would do the AI era. They're killing it.

35
00:01:48,280 --> 00:01:50,200
Speaker 2: They're at an all time high good today. So again,

36
00:01:50,320 --> 00:01:53,240
Speaker 2: like many things are like, it's tough to get a

37
00:01:53,280 --> 00:01:54,360
Speaker 2: read on things. It is.

38
00:01:54,480 --> 00:01:55,960
Speaker 3: Indeed, who do we turn to?

39
00:01:57,760 --> 00:01:59,680
Speaker 2: We just want to get have an excuse to get

40
00:01:59,680 --> 00:02:02,240
Speaker 2: to the and so we're gonna vamp for a little

41
00:02:02,240 --> 00:02:04,040
Speaker 2: while and then we're gonna get to the guest.

42
00:02:04,360 --> 00:02:06,720
Speaker 3: I just tried to throw to the guest, Joe. Who

43
00:02:06,760 --> 00:02:10,840
Speaker 3: do we turn to for a read on the true

44
00:02:11,000 --> 00:02:12,919
Speaker 3: feelings of the market around AI.

45
00:02:13,120 --> 00:02:15,040
Speaker 2: That's right, we do have the perfect guest. We're going

46
00:02:15,120 --> 00:02:17,400
Speaker 2: to be speaking with Josh Wolf. He's the co founder

47
00:02:17,480 --> 00:02:20,639
Speaker 2: managing partner at Lux Capital VC who's been in this

48
00:02:20,720 --> 00:02:23,040
Speaker 2: space for a long time, a guy we like to

49
00:02:23,080 --> 00:02:25,839
Speaker 2: turn to to figure out what's actually happening at any

50
00:02:25,840 --> 00:02:29,120
Speaker 2: given moment. He sometimes even has good interest insights in

51
00:02:29,200 --> 00:02:32,720
Speaker 2: public companies too, not just a private company. Josh, thank

52
00:02:32,760 --> 00:02:34,520
Speaker 2: you so much for coming back on the podcast.

53
00:02:34,960 --> 00:02:36,880
Speaker 4: Great to see you both. How you guys doing.

54
00:02:37,440 --> 00:02:39,720
Speaker 2: We're doing great. What's up with Alphabet? Some people thought

55
00:02:39,760 --> 00:02:42,280
Speaker 2: they were going to be a big loser in AI

56
00:02:42,360 --> 00:02:45,079
Speaker 2: because they have this legacy search business model and O

57
00:02:45,160 --> 00:02:47,720
Speaker 2: three is so much better for searching. Here they are surging.

58
00:02:47,760 --> 00:02:50,080
Speaker 2: They're an all time high. What is the what is

59
00:02:50,120 --> 00:02:52,120
Speaker 2: the smart take on Alphabet right now?

60
00:02:52,440 --> 00:02:54,080
Speaker 4: I think they are crushing it. I think they are

61
00:02:54,160 --> 00:02:57,120
Speaker 4: the sort of dark horse underdog. And maybe the second

62
00:02:57,120 --> 00:02:59,600
Speaker 4: that I would put it with that is Apple, which people.

63
00:02:59,360 --> 00:03:01,000
Speaker 2: Are totally yeah, totally.

64
00:03:01,200 --> 00:03:04,720
Speaker 4: And you know the irony is, you know, if you

65
00:03:04,760 --> 00:03:09,000
Speaker 4: think about the big players meta with Zuck's crazy poacha palooza,

66
00:03:09,240 --> 00:03:09,480
Speaker 4: you know.

67
00:03:09,440 --> 00:03:12,240
Speaker 2: The pas which we'll get a remark where.

68
00:03:12,360 --> 00:03:14,560
Speaker 4: You know, one hundred million dollars pay packages and try

69
00:03:14,600 --> 00:03:17,440
Speaker 4: and disrupt everybody and bring them all on. You're seeing

70
00:03:17,480 --> 00:03:19,800
Speaker 4: talks about them. They've got mid journey coming in for

71
00:03:19,840 --> 00:03:22,560
Speaker 4: the images, so not relying on their own models. They're

72
00:03:22,600 --> 00:03:27,720
Speaker 4: talking about Google for you know, potential integration on search

73
00:03:27,760 --> 00:03:31,480
Speaker 4: and so so it's interesting that Meta having committed to

74
00:03:31,639 --> 00:03:33,680
Speaker 4: first be you know, the metaverse and change the name

75
00:03:33,680 --> 00:03:35,240
Speaker 4: to Meta, and then it's not going in all in

76
00:03:35,240 --> 00:03:36,800
Speaker 4: an AI is actually going to be turning to some

77
00:03:36,800 --> 00:03:39,320
Speaker 4: of these other players potentially. So I think that's surprise

78
00:03:39,400 --> 00:03:40,960
Speaker 4: number one that people thought that Meta was going to

79
00:03:41,000 --> 00:03:45,000
Speaker 4: be in the lead. And and I think that Google

80
00:03:45,040 --> 00:03:46,760
Speaker 4: and Apple were both sort of counted out, but both

81
00:03:46,760 --> 00:03:50,080
Speaker 4: are very serious contenders. If you look at the top

82
00:03:50,120 --> 00:03:53,280
Speaker 4: two video models that are out in the world today,

83
00:03:53,680 --> 00:03:55,720
Speaker 4: one US company which is are is called Runway mL

84
00:03:55,800 --> 00:03:58,360
Speaker 4: and the other which is VEO three, which is absolutely

85
00:03:58,360 --> 00:04:01,560
Speaker 4: stunning and incredible. They are a force to reckon with.

86
00:04:01,720 --> 00:04:05,280
Speaker 4: It is extraordinarily development. You can make an argument by

87
00:04:05,280 --> 00:04:08,200
Speaker 4: the way they have this repository exclusive to them of

88
00:04:08,240 --> 00:04:11,360
Speaker 4: being able to trade on every YouTube video ever produced

89
00:04:13,040 --> 00:04:16,240
Speaker 4: is a super valuable piece. Remember it was what a

90
00:04:16,320 --> 00:04:18,320
Speaker 4: year and a half ago, two years ago that people

91
00:04:18,360 --> 00:04:21,280
Speaker 4: were mocking Bard. Bard was the laughing stock of all

92
00:04:21,320 --> 00:04:25,000
Speaker 4: of this. Gemini two to five is crushing it. You know,

93
00:04:25,600 --> 00:04:28,320
Speaker 4: the nano banana that I don't know if you guys

94
00:04:28,360 --> 00:04:32,400
Speaker 4: have used which was the secret code name for their

95
00:04:32,480 --> 00:04:37,240
Speaker 4: latest image generation model is probably the number one performing

96
00:04:37,320 --> 00:04:39,680
Speaker 4: model out there, and paired with some of these workflows

97
00:04:39,680 --> 00:04:42,640
Speaker 4: that go into Runway or even into Veyo, combined with

98
00:04:42,680 --> 00:04:44,839
Speaker 4: my journey, it's just you know. So I think people

99
00:04:44,920 --> 00:04:47,440
Speaker 4: counted Google out because they were behind and they weren't

100
00:04:47,440 --> 00:04:50,200
Speaker 4: part of the hype. Open ai had won the consumer

101
00:04:50,360 --> 00:04:54,360
Speaker 4: both on subscription basis and capturing people's habitual daily use,

102
00:04:54,400 --> 00:04:56,960
Speaker 4: and that all made sense. Claude was capturing it an

103
00:04:56,960 --> 00:05:00,840
Speaker 4: anthropic on code and then you've got niche player you know,

104
00:05:00,920 --> 00:05:02,760
Speaker 4: like open Evidence and other people that are doing it

105
00:05:02,800 --> 00:05:05,600
Speaker 4: in different verticals like medicine. But I think that the

106
00:05:05,880 --> 00:05:09,600
Speaker 4: corporate workflows, I think about how much locks depends upon

107
00:05:09,839 --> 00:05:13,960
Speaker 4: Gmail and Google Calendar, ye and slides, and their ability

108
00:05:13,960 --> 00:05:16,479
Speaker 4: to integrate that all over time is I think going

109
00:05:16,520 --> 00:05:19,320
Speaker 4: to give them a huge advantage. So very bullish on Google.

110
00:05:19,360 --> 00:05:22,039
Speaker 4: And by the way, remember go back twenty years or

111
00:05:22,040 --> 00:05:25,720
Speaker 4: fifteen years. Google did the thing that completely did what

112
00:05:25,760 --> 00:05:29,000
Speaker 4: the DOJ couldn't do to Microsoft. They dropped the price

113
00:05:29,320 --> 00:05:32,839
Speaker 4: of alternatives to the office suite to free and the

114
00:05:32,880 --> 00:05:35,880
Speaker 4: net result of that was Google, which had this advertising model,

115
00:05:35,960 --> 00:05:39,640
Speaker 4: was ascended and Microsoft was suddenly scrambling. And I think

116
00:05:39,640 --> 00:05:41,159
Speaker 4: that the same thing's going to happen. I think that

117
00:05:41,200 --> 00:05:44,960
Speaker 4: a lot of people funding foundation models and the endless

118
00:05:45,000 --> 00:05:47,720
Speaker 4: perception of endless demand for GPUs and compute and all

119
00:05:47,720 --> 00:05:50,520
Speaker 4: these independent private AI companies are going to be shocked

120
00:05:50,520 --> 00:05:52,839
Speaker 4: by what Google does on a pricing basis. With Gemini

121
00:05:52,880 --> 00:05:57,279
Speaker 4: and beyond, so dark Horse, I'd be pretty bullish.

122
00:05:57,720 --> 00:06:01,279
Speaker 3: Joe, did you immediately run to the nanopenbe I did.

123
00:06:01,200 --> 00:06:03,479
Speaker 2: As soon as you said I had not used it.

124
00:06:04,040 --> 00:06:06,080
Speaker 2: I had never used Nano Banana. That is the first

125
00:06:06,120 --> 00:06:06,960
Speaker 2: place I don't go to.

126
00:06:07,000 --> 00:06:09,440
Speaker 4: A website called nano Banana, because you know you're gonna know.

127
00:06:10,000 --> 00:06:10,400
Speaker 4: I don't know.

128
00:06:10,520 --> 00:06:13,320
Speaker 2: Nano Banana. Dot ai looks like the right spot. Okay,

129
00:06:13,480 --> 00:06:14,720
Speaker 2: it does have you.

130
00:06:14,680 --> 00:06:17,960
Speaker 4: Want to actually go? Is it's now embedded inside.

131
00:06:18,520 --> 00:06:19,360
Speaker 2: The AI studio?

132
00:06:19,440 --> 00:06:20,200
Speaker 1: Yeah, yeah, I see that.

133
00:06:20,320 --> 00:06:25,080
Speaker 4: Yeah, it's ability to take you and it's almost like Photoshop,

134
00:06:25,200 --> 00:06:28,080
Speaker 4: but just voting photoshop. It's really incredible.

135
00:06:28,360 --> 00:06:30,760
Speaker 2: This is a threat to Tracy's MS paint skills, which

136
00:06:30,760 --> 00:06:34,520
Speaker 2: are legendary within the Odds office, And maybe I will

137
00:06:34,560 --> 00:06:36,760
Speaker 2: no longer need to say, Tracy, could you make this

138
00:06:36,800 --> 00:06:38,200
Speaker 2: for me an MS paint?

139
00:06:38,320 --> 00:06:41,800
Speaker 3: No, no technology will ever replace MS paint. I'm one

140
00:06:41,839 --> 00:06:45,360
Speaker 3: hundred percent confident in that I'm being sarcastic obviously. Okay,

141
00:06:45,720 --> 00:06:48,760
Speaker 3: can you talk more generally about how people are feeling

142
00:06:48,880 --> 00:06:51,840
Speaker 3: about the AI space at the moment, because you probably

143
00:06:51,880 --> 00:06:55,680
Speaker 3: heard us in the intro struggling to characterize like the

144
00:06:55,800 --> 00:06:59,440
Speaker 3: general attitudes towards the sector at the moment. What's your

145
00:06:59,440 --> 00:06:59,960
Speaker 3: take on it?

146
00:07:00,720 --> 00:07:03,960
Speaker 4: Well, the first is, on the one hand, people underappreciate

147
00:07:04,000 --> 00:07:06,120
Speaker 4: how much this is going to change everything in our

148
00:07:06,160 --> 00:07:08,400
Speaker 4: daily lives, but that doesn't mean that people are going

149
00:07:08,440 --> 00:07:10,040
Speaker 4: to make money from that. We're all to benefit, we'll

150
00:07:10,080 --> 00:07:14,040
Speaker 4: all be more productive. The great irony at the macro scale,

151
00:07:14,080 --> 00:07:16,600
Speaker 4: of course, is that people thought that blue collar jobs

152
00:07:16,600 --> 00:07:19,080
Speaker 4: were totally screwed, you know, and white collar jobs and

153
00:07:19,080 --> 00:07:22,080
Speaker 4: the Peter Drucker knowledge worker. Everybody's safe. But the great

154
00:07:22,120 --> 00:07:24,880
Speaker 4: irony is it is the knowledge workers that are in

155
00:07:24,880 --> 00:07:28,200
Speaker 4: trouble because so much of their workflows are being captured

156
00:07:28,280 --> 00:07:32,000
Speaker 4: and in a sense commoditized and maybe approaching an asymp

157
00:07:32,000 --> 00:07:34,080
Speaker 4: tote of good enough. It may not be perfect, but

158
00:07:34,200 --> 00:07:36,680
Speaker 4: pretty damn good. And so you're going to see a

159
00:07:36,720 --> 00:07:40,920
Speaker 4: lot of labor destruction in segments and markets that people

160
00:07:41,320 --> 00:07:44,240
Speaker 4: were not anticipating. The first early canary in the coal

161
00:07:44,320 --> 00:07:48,119
Speaker 4: mine that you're seeing there is hiring for undergrads coming

162
00:07:48,120 --> 00:07:51,200
Speaker 4: into entry level jobs. And whether that's investment, banking, or

163
00:07:51,240 --> 00:07:55,320
Speaker 4: sales and creating or consulting or accounting. Suddenly it isn't

164
00:07:55,360 --> 00:07:57,640
Speaker 4: that the economy is really troubled. It's that a lot

165
00:07:57,680 --> 00:08:00,640
Speaker 4: of those demands you're seeing salesforce safe forty five percent

166
00:08:00,680 --> 00:08:02,400
Speaker 4: of our jobs. You know, it'd cut Mark Bennioff is

167
00:08:02,440 --> 00:08:04,560
Speaker 4: using AI instead of people, and that's going to keep

168
00:08:04,600 --> 00:08:07,640
Speaker 4: trickling down. It's going to happen slowly and then sort

169
00:08:07,640 --> 00:08:09,640
Speaker 4: of all at once. So that's one thing on the

170
00:08:09,680 --> 00:08:13,200
Speaker 4: labor side. And I would say broadly that it's under

171
00:08:13,280 --> 00:08:15,720
Speaker 4: hyped in how much it's going to impact our lives.

172
00:08:15,720 --> 00:08:18,800
Speaker 4: It's over hyped invaluations. Now where's it overhyped? Evaluations? The

173
00:08:18,800 --> 00:08:22,000
Speaker 4: first one, you know, I was very proud ten years

174
00:08:22,040 --> 00:08:24,520
Speaker 4: ago we funded a company called Zekes Zukes did autonomous

175
00:08:24,600 --> 00:08:27,080
Speaker 4: driving and they were training these cars and we had

176
00:08:27,080 --> 00:08:29,120
Speaker 4: twenty five million dollars in and I was like, they're

177
00:08:29,120 --> 00:08:30,640
Speaker 4: playing video games, you know what? Do you do and

178
00:08:30,640 --> 00:08:32,320
Speaker 4: you're messing around and they said, no, no, we're training

179
00:08:32,360 --> 00:08:34,360
Speaker 4: the vehicles and we have these in video chips nobody

180
00:08:34,400 --> 00:08:36,840
Speaker 4: else has yet. That was twenty fifteen. I ended up

181
00:08:36,840 --> 00:08:40,200
Speaker 4: pitching at a public charity event, the Investor Kids Conference

182
00:08:40,240 --> 00:08:43,040
Speaker 4: twenty fifteen, twenty sixteen, and this was a fifteen billion

183
00:08:43,080 --> 00:08:44,880
Speaker 4: dollar market cap company at the time, when Intel was

184
00:08:44,920 --> 00:08:46,640
Speaker 4: one hundred and fifty billion. I said, this is like

185
00:08:46,679 --> 00:08:47,600
Speaker 4: the Paar trade of a century.

186
00:08:47,840 --> 00:08:50,280
Speaker 2: Oh in video threw up like three hundred and forty

187
00:08:50,360 --> 00:08:51,199
Speaker 2: xince then, aren't.

188
00:08:51,040 --> 00:08:55,600
Speaker 4: They Yeah, yeah, And so this was you know, it's

189
00:08:55,640 --> 00:08:57,480
Speaker 4: the benefit of being a venture capitalist that you get

190
00:08:57,520 --> 00:09:00,439
Speaker 4: to see the future in legal, inside inform, inside the

191
00:09:00,480 --> 00:09:04,400
Speaker 4: companies and what people are doing. So the perception and

192
00:09:04,440 --> 00:09:07,200
Speaker 4: the consensus in AI on the hardware stack is that

193
00:09:07,240 --> 00:09:10,040
Speaker 4: we need endless demand for data centers, we need endless

194
00:09:10,040 --> 00:09:13,839
Speaker 4: demand for GPUs. We need one hundred thousand clusters of

195
00:09:13,960 --> 00:09:17,520
Speaker 4: h one hundred chips or Blackwell chips. We're going to

196
00:09:17,679 --> 00:09:19,480
Speaker 4: thwart China from getting the chips, but they're going to

197
00:09:19,520 --> 00:09:21,160
Speaker 4: sort of design around, or we're going to have different

198
00:09:21,240 --> 00:09:23,160
Speaker 4: versions of the video chips for China. I think that

199
00:09:23,200 --> 00:09:26,080
Speaker 4: this is misplaced, And I'll give you another insight. You know,

200
00:09:26,120 --> 00:09:27,800
Speaker 4: we may talked a little bit about this in the past,

201
00:09:27,800 --> 00:09:29,760
Speaker 4: but there was a paper from Apple a year and

202
00:09:29,800 --> 00:09:32,520
Speaker 4: a half ago that a lot of people have not

203
00:09:32,840 --> 00:09:36,400
Speaker 4: really suck their teeth into, which was the idea that

204
00:09:36,400 --> 00:09:39,520
Speaker 4: you could do large language models on device using flash memory,

205
00:09:39,920 --> 00:09:43,760
Speaker 4: not you needing GPUs and so Jensen and Video will

206
00:09:43,760 --> 00:09:45,760
Speaker 4: tell you you need all of these h one hundred

207
00:09:45,800 --> 00:09:47,880
Speaker 4: chips and you need endless compute and lots of data

208
00:09:47,880 --> 00:09:50,880
Speaker 4: centers for training. And that's generally true. But the other

209
00:09:50,920 --> 00:09:52,840
Speaker 4: part of it, the fancy word for prompting, which we

210
00:09:52,880 --> 00:09:56,000
Speaker 4: call inference, you don't necessarily need that. And if that

211
00:09:56,120 --> 00:09:59,199
Speaker 4: is true, then that means that our device is maybe

212
00:09:59,280 --> 00:10:02,679
Speaker 4: are running on s K Heinez and Micron and Samsung,

213
00:10:02,920 --> 00:10:06,160
Speaker 4: and the memory players which have just like the GPU

214
00:10:06,200 --> 00:10:09,880
Speaker 4: players were considered commodity players. With the upgrade cycle of

215
00:10:09,880 --> 00:10:12,280
Speaker 4: the gaming consoles for PS five and Xbox, I think

216
00:10:12,320 --> 00:10:15,760
Speaker 4: you may see a shift towards edge inference. You know,

217
00:10:15,760 --> 00:10:17,040
Speaker 4: I've been talking about this for about a year and

218
00:10:17,080 --> 00:10:19,800
Speaker 4: a half. Elon just tweeted out about it maybe two

219
00:10:19,920 --> 00:10:22,680
Speaker 4: three weeks ago, saying like this is an inevitability. But

220
00:10:22,720 --> 00:10:25,360
Speaker 4: I think it's going to shift away this fallacy of

221
00:10:25,400 --> 00:10:28,800
Speaker 4: composition where what Google is doing and Anthropic is doing

222
00:10:28,800 --> 00:10:30,480
Speaker 4: and an open AI is doing, a met is doing

223
00:10:30,520 --> 00:10:33,439
Speaker 4: at a huge scale all to the benefit of Jensen

224
00:10:33,440 --> 00:10:36,480
Speaker 4: and video and video shareholders may start to chip away

225
00:10:36,520 --> 00:10:38,280
Speaker 4: and say, wait a second, we don't need all this compute.

226
00:10:38,280 --> 00:10:40,000
Speaker 4: There's going to be a glut. So that's the first one.

227
00:10:40,000 --> 00:10:44,040
Speaker 2: On the hardware, that's very interesting. And yeah, i'd heard

228
00:10:44,120 --> 00:10:46,719
Speaker 2: I haven't done much with on device thing. His friend

229
00:10:46,720 --> 00:10:49,520
Speaker 2: of mine was telling me that Ali Baba's model Quinn

230
00:10:49,600 --> 00:10:52,200
Speaker 2: works very well on a phone, for example, So maybe

231
00:10:52,200 --> 00:10:54,120
Speaker 2: it's something we should pay more attention to. All Right,

232
00:10:54,160 --> 00:10:57,960
Speaker 2: as a VC, when you were doing due diligence on

233
00:10:58,080 --> 00:11:01,280
Speaker 2: a company, how does it affect how you think about

234
00:11:01,320 --> 00:11:05,079
Speaker 2: even arriving at the concept of fair value when there

235
00:11:05,160 --> 00:11:09,679
Speaker 2: is the prospect that some share of the enterprise value

236
00:11:09,720 --> 00:11:12,400
Speaker 2: of the company could walk out the door via aqua

237
00:11:12,480 --> 00:11:15,880
Speaker 2: hire to a META, etc. Now I get it different.

238
00:11:16,000 --> 00:11:18,880
Speaker 2: Not every company in AI is doing the hard science, etc.

239
00:11:19,440 --> 00:11:21,800
Speaker 2: But just coming from your perspective as an investor, how

240
00:11:21,880 --> 00:11:24,440
Speaker 2: is that changing how you think of companies that so

241
00:11:24,600 --> 00:11:27,120
Speaker 2: much of where the value is may lie with talent

242
00:11:27,240 --> 00:11:28,920
Speaker 2: that could just walk out the door in any time.

243
00:11:29,200 --> 00:11:31,360
Speaker 4: It is a very big deal. The entire social contract

244
00:11:31,400 --> 00:11:33,920
Speaker 4: of UE capital is the premise that pension funds, high

245
00:11:33,920 --> 00:11:36,839
Speaker 4: network individuals and endowments give capital to people like us.

246
00:11:36,880 --> 00:11:39,280
Speaker 4: We then go deploy it into companies. We buy stakes.

247
00:11:39,400 --> 00:11:40,880
Speaker 4: We try to buy them as early as we can

248
00:11:40,920 --> 00:11:42,280
Speaker 4: and own as much of a company and be a

249
00:11:42,280 --> 00:11:44,520
Speaker 4: partner and add value and then sell those companies. But

250
00:11:44,880 --> 00:11:47,200
Speaker 4: if all of a sudden somebody is being pried away.

251
00:11:47,760 --> 00:11:51,840
Speaker 4: And the irony of all of this is that because

252
00:11:51,920 --> 00:11:56,400
Speaker 4: of a very pressive DOJ and FTC that basically said no, no,

253
00:11:56,559 --> 00:11:58,400
Speaker 4: m and A, we're coming after you. You know, big

254
00:11:58,480 --> 00:12:01,120
Speaker 4: tech companies. We don't want to see more consolidation. You

255
00:12:01,160 --> 00:12:03,640
Speaker 4: have too much power. So they started doing instead of

256
00:12:03,840 --> 00:12:07,400
Speaker 4: MNA NA, instead of mergers and acquisitions, they were doing

257
00:12:07,480 --> 00:12:10,560
Speaker 4: license and aquia higher. And what that meant was, hey,

258
00:12:10,760 --> 00:12:13,040
Speaker 4: we'll buy. And they did this with Scale, we'll buy

259
00:12:13,480 --> 00:12:15,839
Speaker 4: forty nine percent of your company. I think Scale was

260
00:12:15,920 --> 00:12:19,400
Speaker 4: valued at around twelve billion dollars thereabout, and they said, well,

261
00:12:19,440 --> 00:12:22,480
Speaker 4: we'll pay fourteen slight premium to your last round, but

262
00:12:22,559 --> 00:12:25,880
Speaker 4: we'll buy forty nine percent, effectively valuing it at twenty

263
00:12:25,920 --> 00:12:28,920
Speaker 4: eight billion. But we'll pay out that money to the company.

264
00:12:29,120 --> 00:12:31,400
Speaker 4: You can dividend it out to shareholders, so it may

265
00:12:31,400 --> 00:12:33,800
Speaker 4: not be perfectly tax efficient. Number one, then we're going

266
00:12:33,840 --> 00:12:36,959
Speaker 4: to basically license the technology and we're going to take

267
00:12:37,000 --> 00:12:39,000
Speaker 4: all the people, or at least the top people. The

268
00:12:39,040 --> 00:12:42,880
Speaker 4: result of that is you're navigating around Delaware governance. You're

269
00:12:43,200 --> 00:12:48,000
Speaker 4: you're arbitraging. It's really important because it's sort of like

270
00:12:48,480 --> 00:12:52,440
Speaker 4: Carl Icon used to say, your price my terms. You know,

271
00:12:52,480 --> 00:12:55,559
Speaker 4: there's this phenomenon in legal terms people might call the

272
00:12:55,640 --> 00:12:57,959
Speaker 4: Chesterton fence. The idea of the Chester Defence is not

273
00:12:58,080 --> 00:13:00,160
Speaker 4: experiment of like, okay, there's a fence there. What the

274
00:13:00,200 --> 00:13:02,240
Speaker 4: heck is it there for? You don't understand, right, Well,

275
00:13:02,320 --> 00:13:04,079
Speaker 4: maybe it was there to keep the sheep in or

276
00:13:04,160 --> 00:13:07,200
Speaker 4: keep the wolves out or whatever. It is. Every legal term,

277
00:13:07,280 --> 00:13:09,719
Speaker 4: in every term sheet that a venture capitalist gives or

278
00:13:09,760 --> 00:13:12,840
Speaker 4: a founder gets is based on somebody screwing somebody in

279
00:13:12,880 --> 00:13:15,120
Speaker 4: the past, and like, uh, we're not letting that happen again.

280
00:13:15,280 --> 00:13:17,400
Speaker 4: So I can guarantee you that the next few years

281
00:13:17,480 --> 00:13:20,840
Speaker 4: you will see all kinds of protective provisions and covenants

282
00:13:20,840 --> 00:13:23,240
Speaker 4: that say, if one of these companies comes and tries

283
00:13:23,280 --> 00:13:25,840
Speaker 4: to just acquire you all of your stock, you know,

284
00:13:25,920 --> 00:13:28,760
Speaker 4: reverts and YadA YadA, And so there's going to be

285
00:13:29,000 --> 00:13:32,000
Speaker 4: tie ups and holdbacks that are a pendulum swing away

286
00:13:32,520 --> 00:13:36,480
Speaker 4: from the super founder friendly dynamics where venture capitalists were

287
00:13:36,520 --> 00:13:39,280
Speaker 4: tripping over themselves to basically give the most founder friendly

288
00:13:39,400 --> 00:13:41,480
Speaker 4: terms that they can because the founder would say, well,

289
00:13:41,520 --> 00:13:43,040
Speaker 4: if you don't give me what I want, I'll go

290
00:13:43,040 --> 00:13:44,520
Speaker 4: to somebody else. But I think that the pendulum is

291
00:13:44,520 --> 00:13:46,840
Speaker 4: swinging with the cost of capital rising, and you're going

292
00:13:46,920 --> 00:13:50,800
Speaker 4: to see more and more investor friendly term sheets, partially

293
00:13:50,880 --> 00:13:54,040
Speaker 4: as a reaction to the fact that somebody like Zuck

294
00:13:54,080 --> 00:13:56,720
Speaker 4: and Meta can go in and just basically take the

295
00:13:56,760 --> 00:13:58,520
Speaker 4: fruit off the end of the tree and leave a stock.

296
00:14:14,360 --> 00:14:16,280
Speaker 2: By the way, Tracy, as we were talking about this

297
00:14:16,520 --> 00:14:21,080
Speaker 2: breaking from the Wallsiet Journal, xaiicfo Liberatore never stepped down.

298
00:14:21,160 --> 00:14:24,160
Speaker 2: Latest and string of executive departures always moves in this space.

299
00:14:24,480 --> 00:14:27,440
Speaker 3: Yes, indeed, you know, the sort of race to the

300
00:14:27,440 --> 00:14:29,720
Speaker 3: bottom in terms of terms. It reminds me a lot

301
00:14:29,760 --> 00:14:32,360
Speaker 3: of the corporate bond market and the rise of covenant

302
00:14:32,400 --> 00:14:35,960
Speaker 3: life deals, right and I remember a time when like

303
00:14:36,080 --> 00:14:40,880
Speaker 3: cove light deals were a minority in the leverage loan market,

304
00:14:41,000 --> 00:14:44,200
Speaker 3: and now I think they're like almost ninety eight percent

305
00:14:44,280 --> 00:14:47,280
Speaker 3: or ninety nine percent. Basically everything's cove light now because

306
00:14:47,520 --> 00:14:51,040
Speaker 3: the issuers had all the power recently and they were

307
00:14:51,080 --> 00:14:54,760
Speaker 3: able to push back against investors. You mentioned the higher

308
00:14:54,840 --> 00:14:58,320
Speaker 3: cost of capital there, like how much leverage does that

309
00:14:58,360 --> 00:15:01,280
Speaker 3: actually give you as a VSA? And then secondly, just

310
00:15:01,320 --> 00:15:05,960
Speaker 3: going back to the aqua higher thing, how reliable can

311
00:15:06,080 --> 00:15:09,840
Speaker 3: legal restrictions actually be in terms of preventing like your

312
00:15:09,880 --> 00:15:13,160
Speaker 3: star engineers from leaving the company. Are they ever going

313
00:15:13,200 --> 00:15:15,760
Speaker 3: to be like one hundred percent bullet proof. No.

314
00:15:16,240 --> 00:15:17,800
Speaker 4: I mean, look, you have non competes that are not

315
00:15:17,960 --> 00:15:20,680
Speaker 4: enforceable in California, a little bit more enforceable in New York.

316
00:15:20,920 --> 00:15:25,000
Speaker 4: You have arbitrage and jurisdictional stuff. Broadly, I would say

317
00:15:25,160 --> 00:15:28,400
Speaker 4: that the lower the cost of the capital, the shorter

318
00:15:29,120 --> 00:15:31,920
Speaker 4: your term sheets are. The higher the cost of the capital,

319
00:15:32,240 --> 00:15:35,400
Speaker 4: longer your term sheets are, you got more terms, more covenants,

320
00:15:35,440 --> 00:15:38,240
Speaker 4: more protections, because you can afford to be able to

321
00:15:38,280 --> 00:15:40,880
Speaker 4: negotiate for those things. And it's not because you're trying

322
00:15:40,920 --> 00:15:42,960
Speaker 4: to screw over the founder. What you're really trying to

323
00:15:42,960 --> 00:15:45,680
Speaker 4: do as an investor is prevent yourself from being screwed over.

324
00:15:46,080 --> 00:15:48,920
Speaker 4: But again, there's always a pendulum swinging here back and forth.

325
00:15:49,120 --> 00:15:50,600
Speaker 4: Even if you think about Zuck and Meta. In this

326
00:15:50,800 --> 00:15:53,600
Speaker 4: entire movement of hey, I'm the founder, I'm going to

327
00:15:53,720 --> 00:15:55,840
Speaker 4: have super voting stock of ten or one hundred to

328
00:15:55,840 --> 00:15:59,640
Speaker 4: one and control was in a response to founders being

329
00:15:59,720 --> 00:16:02,440
Speaker 4: out did you by bad investors or bad board members,

330
00:16:02,560 --> 00:16:05,040
Speaker 4: and some of the best companies frankly being run by founders,

331
00:16:05,320 --> 00:16:07,440
Speaker 4: And so there's always going to be this pendulum shift.

332
00:16:07,480 --> 00:16:10,800
Speaker 4: To your direct question, it's really a covenant and contract

333
00:16:10,840 --> 00:16:14,080
Speaker 4: that starts socially before it starts legally. If I'm backing

334
00:16:14,080 --> 00:16:16,760
Speaker 4: a founder, I'm doing the most important thing in addition

335
00:16:16,800 --> 00:16:19,480
Speaker 4: to writing a check. I believe before others understand, particularly

336
00:16:19,480 --> 00:16:22,680
Speaker 4: at the earliest stage, I'm encouraging them to start their company.

337
00:16:22,800 --> 00:16:24,600
Speaker 4: I'm giving them the confidence that we're going to back

338
00:16:24,640 --> 00:16:26,280
Speaker 4: them and believe in them. We're going to give them

339
00:16:26,280 --> 00:16:28,160
Speaker 4: the capital so they can go hire the ten best

340
00:16:28,160 --> 00:16:30,160
Speaker 4: people to start the business. We're going to give them

341
00:16:30,200 --> 00:16:32,600
Speaker 4: the money for the compute or the infrastructure, depending on

342
00:16:32,600 --> 00:16:34,240
Speaker 4: the sector. That we're in. It could be biotech, could

343
00:16:34,240 --> 00:16:37,040
Speaker 4: be defense, but whatever it is in this particular moment,

344
00:16:37,960 --> 00:16:42,280
Speaker 4: it is breaking the social contract between investors and founders.

345
00:16:42,280 --> 00:16:45,040
Speaker 4: And for founders it might be heads I win and

346
00:16:45,200 --> 00:16:49,280
Speaker 4: tails I also win. And so that's the bactmic that

347
00:16:49,680 --> 00:16:52,520
Speaker 4: you're going to see a reaction of investors saying, wait

348
00:16:52,520 --> 00:16:55,160
Speaker 4: a second, I'm getting screwed. My limited partners are getting

349
00:16:55,160 --> 00:16:57,560
Speaker 4: screwed again. Those limited partners are endowments and foundations and

350
00:16:58,160 --> 00:17:02,920
Speaker 4: wealthy families, and to protect against a potential bad actor

351
00:17:03,520 --> 00:17:05,520
Speaker 4: trying to take the fruit off the edge of the

352
00:17:05,560 --> 00:17:07,119
Speaker 4: tree and just leave you with a stump.

353
00:17:07,320 --> 00:17:09,120
Speaker 2: So one of the things that comes up a lot

354
00:17:09,160 --> 00:17:12,280
Speaker 2: on the podcast in this discovery over years of conversations

355
00:17:12,320 --> 00:17:14,920
Speaker 2: and what you're describing its principal agent problems all the

356
00:17:14,960 --> 00:17:17,840
Speaker 2: way down in finance, and this is why we see

357
00:17:17,880 --> 00:17:20,640
Speaker 2: the rise of the multistrap model and hedge funds in

358
00:17:20,640 --> 00:17:23,440
Speaker 2: some ways to align the incentives PM with the level

359
00:17:23,520 --> 00:17:26,480
Speaker 2: of the overall fund, with the level of the endowment,

360
00:17:26,600 --> 00:17:28,960
Speaker 2: et cetera. And of course some of these issues that

361
00:17:29,000 --> 00:17:31,720
Speaker 2: you're wrestling with or where everyone's wrestling with in finance,

362
00:17:32,240 --> 00:17:35,840
Speaker 2: similar issues about where the incentive's aligned between the star engineer,

363
00:17:35,960 --> 00:17:38,680
Speaker 2: the founder of the VC, the LP and so forth.

364
00:17:38,720 --> 00:17:41,520
Speaker 2: I have a question though, So in theory a venture

365
00:17:41,560 --> 00:17:46,919
Speaker 2: capital firm, the goal in theory is to create funds

366
00:17:46,920 --> 00:17:50,080
Speaker 2: with the highest return right make money for investors. But

367
00:17:50,119 --> 00:17:55,080
Speaker 2: I could also imagine a slightly different incentive in which

368
00:17:55,200 --> 00:17:57,920
Speaker 2: if the goal is to collect LP money, then maybe

369
00:17:57,960 --> 00:17:59,680
Speaker 2: you want to show that you're in the hottest deals

370
00:17:59,720 --> 00:18:02,080
Speaker 2: of the time. Still that when you go to various

371
00:18:02,160 --> 00:18:05,000
Speaker 2: endowments and pensions and so forth, we're in this deal,

372
00:18:05,000 --> 00:18:07,080
Speaker 2: we're in this deal or in this deal, and maybe

373
00:18:07,119 --> 00:18:10,320
Speaker 2: that overrides the impulse to create high returns. By the way,

374
00:18:10,359 --> 00:18:12,399
Speaker 2: I'm not insinuating anything. I'm just trying to get your

375
00:18:12,440 --> 00:18:15,639
Speaker 2: perspective on something. That being said, one of the things

376
00:18:15,640 --> 00:18:18,560
Speaker 2: that you hear that's happening in tech these days is

377
00:18:18,600 --> 00:18:21,840
Speaker 2: that and for years, founders taking money off the table

378
00:18:21,880 --> 00:18:25,399
Speaker 2: earlier and earlier in the process. You invest fifty million

379
00:18:25,440 --> 00:18:27,760
Speaker 2: dollars in a company, twenty million is so that the

380
00:18:27,800 --> 00:18:30,920
Speaker 2: founder can retire for him as children and his children.

381
00:18:31,320 --> 00:18:33,119
Speaker 2: That may not be great for your LPs, but it

382
00:18:33,200 --> 00:18:35,040
Speaker 2: might be good for you if you could say we

383
00:18:35,119 --> 00:18:38,040
Speaker 2: got in this deal talk to us about how prevalent

384
00:18:38,280 --> 00:18:41,000
Speaker 2: this is and how this is changing this sort of

385
00:18:41,040 --> 00:18:43,000
Speaker 2: a social contract of finance.

386
00:18:43,600 --> 00:18:45,960
Speaker 4: So there's three layers of incentives. And man, you really

387
00:18:46,040 --> 00:18:48,760
Speaker 4: nieled it. I actually haven't really heard somebody that is

388
00:18:48,800 --> 00:18:50,720
Speaker 4: not a full time venture capitalist or a limited partner

389
00:18:50,840 --> 00:18:54,560
Speaker 4: nail these issues. So very pressing and very shrewd. Here's

390
00:18:54,560 --> 00:18:57,000
Speaker 4: the three layers. First thing about the LPs. You're an

391
00:18:57,080 --> 00:18:59,800
Speaker 4: endowment or your foundation or your hospital. You're giving five

392
00:18:59,840 --> 00:19:04,960
Speaker 4: per pent by law of your charitable assets every year.

393
00:19:05,359 --> 00:19:07,280
Speaker 4: You want to continue to earn more than five percent

394
00:19:07,320 --> 00:19:09,119
Speaker 4: so that you can grow that base and invest in

395
00:19:09,200 --> 00:19:13,359
Speaker 4: campuses and scholarships or expand hospital systems and whatnot. So

396
00:19:13,480 --> 00:19:17,919
Speaker 4: you invest you sixty forty bonds equity. Now you do

397
00:19:18,000 --> 00:19:20,480
Speaker 4: the Swinson model for me, and you start introducing some

398
00:19:20,520 --> 00:19:23,320
Speaker 4: private equity, and now you're extending your duration and you're

399
00:19:23,359 --> 00:19:25,920
Speaker 4: extending your liquidity. But you're doing it because you think

400
00:19:25,920 --> 00:19:28,560
Speaker 4: you're getting better returns. Okay, returns are a function of

401
00:19:28,560 --> 00:19:30,360
Speaker 4: how much capital is going into a sector. If there's

402
00:19:30,400 --> 00:19:32,240
Speaker 4: a ton of capital going into a sector, if you're early,

403
00:19:32,240 --> 00:19:34,040
Speaker 4: you're going to do really well. If you're late, you're

404
00:19:34,040 --> 00:19:36,080
Speaker 4: going to be doing really poor because, as Buffett says,

405
00:19:36,240 --> 00:19:38,640
Speaker 4: you pay a high price for cheery consensus, and once

406
00:19:38,640 --> 00:19:42,000
Speaker 4: it's consensus, you're not making money. So the LP's incentive

407
00:19:42,320 --> 00:19:44,440
Speaker 4: is to make as much money they can for their benefactors,

408
00:19:44,480 --> 00:19:49,400
Speaker 4: whether they're patients or scholars or charitable giving. The vcs

409
00:19:49,560 --> 00:19:52,080
Speaker 4: incentive is two things. One get the best return so

410
00:19:52,119 --> 00:19:54,960
Speaker 4: that you can compete. If I'm only earning twelve percent

411
00:19:55,000 --> 00:19:58,359
Speaker 4: and a pure VC is earning twenty percent, money is

412
00:19:58,400 --> 00:20:00,320
Speaker 4: going to go where it's going to be well treated,

413
00:20:00,480 --> 00:20:02,480
Speaker 4: and I'm going to lose to that. So the cost

414
00:20:02,560 --> 00:20:05,080
Speaker 4: of my capital, for the cost of an LP's capital,

415
00:20:05,119 --> 00:20:07,480
Speaker 4: is outperformance. So I've got to outperform, which means I

416
00:20:07,480 --> 00:20:09,480
Speaker 4: have to be earlier, I have to own more. I

417
00:20:09,520 --> 00:20:12,840
Speaker 4: can't just do stupid deals. Sophisticated LPs will not just

418
00:20:12,920 --> 00:20:15,320
Speaker 4: look at the logos that you have, which is the

419
00:20:15,320 --> 00:20:15,760
Speaker 4: game that.

420
00:20:15,720 --> 00:20:18,080
Speaker 2: You always spend an LPs.

421
00:20:17,359 --> 00:20:20,040
Speaker 4: From mutual funds and from some hedge funds. You know,

422
00:20:20,119 --> 00:20:22,360
Speaker 4: you'd see the Q four filings and they always threw

423
00:20:22,400 --> 00:20:24,199
Speaker 4: in the name, oh, we were in Nvidia, you know,

424
00:20:24,240 --> 00:20:26,399
Speaker 4: and they would market their top ten holdings. But bs,

425
00:20:26,880 --> 00:20:29,639
Speaker 4: you know, they lost money on it right, and so

426
00:20:29,640 --> 00:20:32,800
Speaker 4: so that is a really important incentive, and the sophisticated

427
00:20:32,920 --> 00:20:36,639
Speaker 4: LPs will actually know down to the partner at the

428
00:20:36,680 --> 00:20:39,000
Speaker 4: firm or the team or the deal team, who is

429
00:20:39,040 --> 00:20:41,120
Speaker 4: responsible for this, what was the entry point. They will

430
00:20:41,119 --> 00:20:42,720
Speaker 4: talk to the founders and say, who was your most

431
00:20:42,760 --> 00:20:45,359
Speaker 4: valuable investor, who got you your first ten hires, who

432
00:20:45,440 --> 00:20:48,280
Speaker 4: helped with your syndicate construction for your later rounds, who

433
00:20:48,280 --> 00:20:51,040
Speaker 4: made customer introductions, who was a valuable board member who

434
00:20:51,040 --> 00:20:52,920
Speaker 4: never showed up? Who was a sleep in the board meetings?

435
00:20:52,960 --> 00:20:55,360
Speaker 4: All that kind of stuff. So there is a level

436
00:20:55,400 --> 00:20:57,159
Speaker 4: of due diligence that LPs have to do to know

437
00:20:57,280 --> 00:20:59,000
Speaker 4: are you a value ad investor or are you a

438
00:20:59,119 --> 00:21:01,760
Speaker 4: poser or pretender that's just buying a logo or a

439
00:21:01,760 --> 00:21:04,479
Speaker 4: brand name. Okay. The other incentive, and then we'll get

440
00:21:04,480 --> 00:21:07,480
Speaker 4: to the founder's liquidity is you have this weird dynamic

441
00:21:07,720 --> 00:21:10,359
Speaker 4: of what I've called the minos and the megas in

442
00:21:10,440 --> 00:21:13,520
Speaker 4: venture capital. This is in preview a shakeout that is

443
00:21:13,560 --> 00:21:18,560
Speaker 4: going to happen. The minos are the thousands of small,

444
00:21:18,920 --> 00:21:22,520
Speaker 4: sub five hundred million dollar funds that proliferated when the

445
00:21:22,520 --> 00:21:25,320
Speaker 4: cost of capital was low, rates were low. Everybody was

446
00:21:25,359 --> 00:21:27,760
Speaker 4: making money. You had a roommate that started a company

447
00:21:27,800 --> 00:21:29,560
Speaker 4: and you got into pinterest, or you knew somebody at

448
00:21:29,600 --> 00:21:31,120
Speaker 4: Meta and they gave you a deal and blah blah

449
00:21:31,119 --> 00:21:33,000
Speaker 4: blah blah blah. And when you had the tigers and

450
00:21:33,040 --> 00:21:35,880
Speaker 4: the soft banks and the abundance of follow on capital,

451
00:21:36,040 --> 00:21:38,760
Speaker 4: every round was an up round. You had paper marks

452
00:21:38,880 --> 00:21:40,280
Speaker 4: that kept going up and up and up, and it

453
00:21:40,280 --> 00:21:42,840
Speaker 4: looked great, and you're reporting these paper marks and then

454
00:21:42,840 --> 00:21:45,439
Speaker 4: sometimes these things became zeros. Okay, but you raised your

455
00:21:45,480 --> 00:21:48,720
Speaker 4: next fund before they became a zero. Those are the minnos.

456
00:21:49,200 --> 00:21:51,479
Speaker 4: I was with one of my very large lpeds has

457
00:21:51,600 --> 00:21:54,160
Speaker 4: hundreds of millions of dollars invested with us, and I said,

458
00:21:54,200 --> 00:21:56,280
Speaker 4: I think there's going to be a fifty percent extinction

459
00:21:56,400 --> 00:22:00,600
Speaker 4: rate amongst these minos. And he said, Josh, that is ridiculous.

460
00:22:01,080 --> 00:22:03,959
Speaker 4: It's going to be ninety percent. So you are going

461
00:22:04,000 --> 00:22:05,760
Speaker 4: to have a mass extinction. Now why, by the way,

462
00:22:06,040 --> 00:22:10,639
Speaker 4: not because they're just bad investors. It's Shakespearean, okay, Shakespearean,

463
00:22:10,640 --> 00:22:13,120
Speaker 4: and that these are partnerships. People start to hate each

464
00:22:13,119 --> 00:22:15,240
Speaker 4: other when it becomes hard. People start to hate each other.

465
00:22:15,280 --> 00:22:17,680
Speaker 4: When there's down rounds, people start to hate each other

466
00:22:17,760 --> 00:22:20,200
Speaker 4: when somebody else's deal is a crappy deal and they're

467
00:22:20,200 --> 00:22:23,280
Speaker 4: bringing down your carry And so partnerships are fragile things,

468
00:22:23,320 --> 00:22:26,240
Speaker 4: just like relationships and marriages, and they can break up.

469
00:22:26,359 --> 00:22:28,640
Speaker 4: And so you have a lot of vcs that start

470
00:22:28,680 --> 00:22:31,160
Speaker 4: in the past few years. They're not experienced in going

471
00:22:31,160 --> 00:22:33,720
Speaker 4: through cycles. They have inadequate reserves to continue to fund

472
00:22:33,760 --> 00:22:36,119
Speaker 4: their companies. So you're going to have an extinction that

473
00:22:36,160 --> 00:22:41,119
Speaker 4: I would consider involuntary exits. Okay, then there's voluntary exits,

474
00:22:41,160 --> 00:22:43,639
Speaker 4: which is another interesting dynamic, and there's a playbook for this,

475
00:22:44,119 --> 00:22:46,679
Speaker 4: which is two thousand and nine to twenty fourteen, all

476
00:22:46,720 --> 00:22:49,240
Speaker 4: the big private equity firms reached a level of scale

477
00:22:49,600 --> 00:22:53,360
Speaker 4: several hundred billion dollars aum assets owner management, where they

478
00:22:53,440 --> 00:22:59,480
Speaker 4: basically said, we're diversified, we're alternative ASSEID platforms. Carlisle, Blackstone, KKRTPG,

479
00:22:59,600 --> 00:23:02,840
Speaker 4: Apollo all went public. The same thing is going to

480
00:23:02,840 --> 00:23:05,359
Speaker 4: happen and venture with probably five or six firms. My

481
00:23:05,440 --> 00:23:08,840
Speaker 4: prediction is, and they're all great people running great firms,

482
00:23:08,840 --> 00:23:11,080
Speaker 4: but they're starting to play a different game in that game.

483
00:23:11,440 --> 00:23:16,680
Speaker 4: Andresen Horowitz General Atlantic, General Catalyst, Insight, Light Speed, a

484
00:23:16,720 --> 00:23:18,920
Speaker 4: handful of others, all at eighty or one hundred billion.

485
00:23:18,920 --> 00:23:22,040
Speaker 4: AUM have built great firms, but are thinking about how

486
00:23:22,040 --> 00:23:25,000
Speaker 4: do we create generational wealth for the founders and go public?

487
00:23:25,440 --> 00:23:28,480
Speaker 4: Different incentive. How do I make my LPs the best

488
00:23:28,480 --> 00:23:31,160
Speaker 4: money or get the best founders. It's about asset gathering

489
00:23:31,400 --> 00:23:34,160
Speaker 4: and liquidity. Now we go to the founders. I can

490
00:23:34,160 --> 00:23:36,159
Speaker 4: tell you I've been on both sides of this. On

491
00:23:36,200 --> 00:23:38,440
Speaker 4: the one hand, you want to be fully aligned with

492
00:23:38,480 --> 00:23:41,919
Speaker 4: your founders, meaning I'm in a fund it's ten years.

493
00:23:42,080 --> 00:23:44,680
Speaker 4: Some vcs vest over two or three or four years.

494
00:23:45,119 --> 00:23:47,440
Speaker 4: We vest, and all of our partners vest over ten years,

495
00:23:47,480 --> 00:23:49,639
Speaker 4: the same duration that you're If you're an LP with me,

496
00:23:49,760 --> 00:23:51,439
Speaker 4: your money's locked up. It's the right thing to do,

497
00:23:51,560 --> 00:23:53,520
Speaker 4: Buffet style. And the guy that put me in business,

498
00:23:53,680 --> 00:23:56,320
Speaker 4: Bill Conway, who's the Carlisle founder, this is what they did.

499
00:23:56,720 --> 00:24:00,480
Speaker 4: So if you're an entrepreneur and you start a company

500
00:24:00,600 --> 00:24:03,360
Speaker 4: and people are tripping over themselves to get in, they

501
00:24:03,400 --> 00:24:06,200
Speaker 4: might entice you with green mail and say, yes, we're

502
00:24:06,200 --> 00:24:08,160
Speaker 4: gonna invest just like you said, Joe, fifty million dollars,

503
00:24:08,160 --> 00:24:10,560
Speaker 4: but we're gonna give you twenty million of liquidity. Okay,

504
00:24:10,960 --> 00:24:15,000
Speaker 4: Now I will say this. Twenty nineteen, I'm on a

505
00:24:15,080 --> 00:24:18,399
Speaker 4: Zoom call for an amazing company called Control Labs that

506
00:24:18,440 --> 00:24:21,080
Speaker 4: we sell to Meta for a little under a billion dollars.

507
00:24:21,440 --> 00:24:23,600
Speaker 4: And I love this company and I love the founders.

508
00:24:23,840 --> 00:24:26,399
Speaker 4: And this is before everybody was on Zoom during COVID,

509
00:24:26,760 --> 00:24:29,520
Speaker 4: and I'm looking at the Zoom window and I see

510
00:24:29,520 --> 00:24:31,960
Speaker 4: one of the founders and I text the other founder

511
00:24:32,240 --> 00:24:34,560
Speaker 4: because I think that they're selling too early. And I'm

512
00:24:34,560 --> 00:24:37,480
Speaker 4: the lone board member that didn't have a veto, but

513
00:24:37,560 --> 00:24:39,560
Speaker 4: I'm like, I really think we should stay the course

514
00:24:39,880 --> 00:24:42,200
Speaker 4: and we should keep going. And I made a terrible,

515
00:24:42,280 --> 00:24:45,000
Speaker 4: terrible mistake because I'm looking at the Zoom window and

516
00:24:45,040 --> 00:24:49,480
Speaker 4: I see the guy and I look closer, and when

517
00:24:49,480 --> 00:24:51,760
Speaker 4: I text the other founder and I'm like, is he

518
00:24:52,400 --> 00:24:55,840
Speaker 4: still in a dorm room? And sure enough, he was

519
00:24:55,880 --> 00:25:00,080
Speaker 4: in a dorm room as a PhD, had made no money.

520
00:25:00,160 --> 00:25:00,960
Speaker 2: Guy get rich.

521
00:25:01,359 --> 00:25:04,119
Speaker 4: Paper stock value and he's like, yes, I want to

522
00:25:04,119 --> 00:25:06,000
Speaker 4: sell because he's gonna make ninety one hundred million dollars

523
00:25:06,040 --> 00:25:08,439
Speaker 4: and it's life changing money. And I sat there and

524
00:25:08,480 --> 00:25:11,040
Speaker 4: I said, if I would have just given him a

525
00:25:11,119 --> 00:25:16,840
Speaker 4: few million dollars, he would have been able to take breath,

526
00:25:17,040 --> 00:25:19,640
Speaker 4: buy a house, you know, get out of that door,

527
00:25:20,840 --> 00:25:21,760
Speaker 4: and and.

528
00:25:21,680 --> 00:25:22,760
Speaker 2: Keep getting a gross friend.

529
00:25:23,680 --> 00:25:27,960
Speaker 4: There is a virtue of giving some liquidity, but you

530
00:25:28,119 --> 00:25:30,520
Speaker 4: need to be aligned, and in that moment we were

531
00:25:30,520 --> 00:25:33,399
Speaker 4: not aligned because he was like, I'm calling Enrich and

532
00:25:33,440 --> 00:25:36,119
Speaker 4: I wanted him to keep going. But if you're calling

533
00:25:36,160 --> 00:25:39,679
Speaker 4: in Rich and you're not completing the job, the mission,

534
00:25:40,080 --> 00:25:43,640
Speaker 4: then you have total misalignment. So that's the dynamics LP alignment,

535
00:25:43,760 --> 00:25:47,080
Speaker 4: GP alignment, and the bifurcation, and then founder alignment. A

536
00:25:47,080 --> 00:25:49,679
Speaker 4: little bit of liquidity is okay to let them stay for.

537
00:25:49,800 --> 00:26:09,200
Speaker 3: S okay other than misalignment, and everyone's starting to hate

538
00:26:09,240 --> 00:26:11,800
Speaker 3: each other as the cost of capital goes up. There's

539
00:26:11,840 --> 00:26:15,119
Speaker 3: another thing going on, which is, you know, AI just

540
00:26:15,200 --> 00:26:20,359
Speaker 3: launched some open source models of its own. And this

541
00:26:20,480 --> 00:26:23,600
Speaker 3: leads me to a question that like I'm gonna admit

542
00:26:24,320 --> 00:26:27,280
Speaker 3: I've never quite understood this, but like, what exactly is

543
00:26:27,320 --> 00:26:32,880
Speaker 3: the attraction for VC investors to open source models as

544
00:26:32,920 --> 00:26:36,919
Speaker 3: opposed to closed source where like closed source, you know,

545
00:26:36,960 --> 00:26:40,919
Speaker 3: it's proprietary, people presumably have to pay in order to

546
00:26:40,960 --> 00:26:44,680
Speaker 3: get it. There's like a defensible moat around the business.

547
00:26:45,160 --> 00:26:48,200
Speaker 3: You would assume why in the world. Would I ever

548
00:26:48,240 --> 00:26:50,480
Speaker 3: want to fund a model that's open source.

549
00:26:51,440 --> 00:26:53,560
Speaker 4: Well, a few things. One, if you go back in

550
00:26:53,560 --> 00:26:56,800
Speaker 4: the computer stack, you have this with like redhead and Linux.

551
00:26:56,880 --> 00:27:01,119
Speaker 4: You know, going back years ago. Are the largest owners

552
00:27:01,119 --> 00:27:03,159
Speaker 4: of a company called hugging Face, which is both the

553
00:27:03,200 --> 00:27:06,479
Speaker 4: most ridiculous name and when one of my partners, Brandon Reeves,

554
00:27:06,560 --> 00:27:08,360
Speaker 4: was sourcing this deal, it was a bunch of free

555
00:27:08,520 --> 00:27:11,760
Speaker 4: French PhDs came to Brooklyn and they're like, we're starting this.

556
00:27:12,080 --> 00:27:13,760
Speaker 4: I'm like hugging face and they're like, yeah, it's named

557
00:27:13,760 --> 00:27:16,640
Speaker 4: after an emoji, the little you know, hugging face emoji. Cute.

558
00:27:16,840 --> 00:27:17,960
Speaker 4: They became the leader.

559
00:27:18,040 --> 00:27:20,880
Speaker 3: I like, they are visually demonstrating everything for us.

560
00:27:20,920 --> 00:27:24,040
Speaker 4: Thank you. I can't do many other emojis, and you

561
00:27:24,040 --> 00:27:25,720
Speaker 4: don't want to see some of them, but hugging face

562
00:27:25,760 --> 00:27:30,560
Speaker 4: I can do. So they became the leading open source repository. Now.

563
00:27:30,600 --> 00:27:33,280
Speaker 4: The great irony, by the way, is when open ai started,

564
00:27:33,640 --> 00:27:36,600
Speaker 4: they were open ai, but they became the world's greatest chatbot.

565
00:27:36,800 --> 00:27:39,280
Speaker 4: Hugging Face started with a really crappy chatbot, but then

566
00:27:39,320 --> 00:27:43,760
Speaker 4: became the world's greatest open source repository. Every major tech company,

567
00:27:43,760 --> 00:27:47,479
Speaker 4: including open Ai Microsoft. Any open source model that they do,

568
00:27:47,520 --> 00:27:50,320
Speaker 4: they put on there, and that is like the GitHub

569
00:27:50,920 --> 00:27:54,520
Speaker 4: of AI models. Microsoft brought kid Hub roughly eight billion,

570
00:27:54,560 --> 00:27:57,240
Speaker 4: eight and a half billion dollars. It was a really

571
00:27:57,400 --> 00:28:01,520
Speaker 4: valuable store of models and and code. This is the

572
00:28:01,520 --> 00:28:04,600
Speaker 4: same thing for dynamic AI models. So as a VC,

573
00:28:04,800 --> 00:28:07,560
Speaker 4: we originally funded this, I want to say at a

574
00:28:07,560 --> 00:28:10,800
Speaker 4: thirty or forty million pre money valuation. Last round was

575
00:28:10,880 --> 00:28:15,280
Speaker 4: north of six billion dollars and real revenue generating multi

576
00:28:15,400 --> 00:28:19,760
Speaker 4: hundreds of millions of dollars. Serious company. Now the trend

577
00:28:20,119 --> 00:28:22,520
Speaker 4: is this. Vanode Coastlin and I were both in the

578
00:28:22,560 --> 00:28:25,880
Speaker 4: White House a year and a half ago and we

579
00:28:25,880 --> 00:28:28,159
Speaker 4: were having a debate with Jake Sullivan about what is

580
00:28:28,200 --> 00:28:31,840
Speaker 4: better for national security? Open source are closed? And I said,

581
00:28:31,840 --> 00:28:33,520
Speaker 4: where do you stand on the issue? Depends on where

582
00:28:33,560 --> 00:28:36,280
Speaker 4: you sit in the cap table, okay. And Vanode was

583
00:28:36,280 --> 00:28:38,960
Speaker 4: an early investor. I think he probably put fifty million

584
00:28:39,000 --> 00:28:41,600
Speaker 4: into open AI. I think it became worth a billion plus.

585
00:28:41,960 --> 00:28:44,680
Speaker 4: Is a great investment. And he believed that the best

586
00:28:44,720 --> 00:28:47,520
Speaker 4: thing for US visa via China and the Chinese Communist

587
00:28:47,560 --> 00:28:51,360
Speaker 4: Party was closed, proprietary, siloed model. And I had the

588
00:28:51,400 --> 00:28:53,720
Speaker 4: counterview because we're big investors in hugging face with a

589
00:28:53,800 --> 00:28:56,680
Speaker 4: very large position. And I said no, because the great

590
00:28:56,760 --> 00:29:02,040
Speaker 4: virtue of any system be a journalism, scientific inquiry, computer

591
00:29:02,160 --> 00:29:05,440
Speaker 4: code is the ability to have, in this very Carl

592
00:29:05,560 --> 00:29:08,320
Speaker 4: Popper like way, if I can get philosophically geared for

593
00:29:08,320 --> 00:29:13,800
Speaker 4: a second of conjecture, hypothesis and criticism. That is what

594
00:29:13,840 --> 00:29:16,800
Speaker 4: creates great societies. It's what creates knowledge. So you come

595
00:29:16,880 --> 00:29:18,600
Speaker 4: up with hypothesis, you come up with code, you come

596
00:29:18,680 --> 00:29:20,400
Speaker 4: up with a scientific experiment, you come up with an

597
00:29:20,440 --> 00:29:24,200
Speaker 4: investigative journalist idea, and then people get to criticize it.

598
00:29:24,200 --> 00:29:26,560
Speaker 4: It's your editorial room. It's people fighting it out and

599
00:29:26,640 --> 00:29:29,840
Speaker 4: things improve through that mechanism. So you think about China,

600
00:29:30,000 --> 00:29:32,960
Speaker 4: they will only host approach an assom tode of truth.

601
00:29:33,120 --> 00:29:35,800
Speaker 4: You will never get you know, Shinjeng, You'll never get

602
00:29:35,840 --> 00:29:39,600
Speaker 4: tenemin square, You'll never get the wigers, whereas open source

603
00:29:39,680 --> 00:29:42,560
Speaker 4: lets you approach closer in ASSM tode of truth. So

604
00:29:42,560 --> 00:29:45,360
Speaker 4: that's the virtue of open source. The real thing though,

605
00:29:45,360 --> 00:29:48,640
Speaker 4: for AI is this, I am not convinced that the

606
00:29:48,760 --> 00:29:52,560
Speaker 4: value will continue to accrue. Two in terms of enterprise value.

607
00:29:52,720 --> 00:29:56,080
Speaker 4: The closed foundation models and the reason is open source

608
00:29:56,160 --> 00:30:00,000
Speaker 4: is getting near damn performative enough that the real value

609
00:30:00,120 --> 00:30:05,640
Speaker 4: will go to the longitudinal repositories of siloed information. That

610
00:30:05,720 --> 00:30:09,680
Speaker 4: is a mouthful of basically saying your database of proprietary data.

611
00:30:09,840 --> 00:30:13,800
Speaker 4: Bloomberg has it with a huge and wonderful repository financial

612
00:30:13,800 --> 00:30:18,400
Speaker 4: information time series, every security, currency, bond, fixed income, etc.

613
00:30:18,840 --> 00:30:21,120
Speaker 4: Q step that you can imagine. Meta has it with

614
00:30:21,160 --> 00:30:24,560
Speaker 4: all of your WhatsApp chats and your Instagram posts and

615
00:30:24,560 --> 00:30:27,280
Speaker 4: your Facebook likes. X and Twitter have it with all

616
00:30:27,280 --> 00:30:30,480
Speaker 4: of your tweets, So pharma companies with your clinical data.

617
00:30:30,520 --> 00:30:34,400
Speaker 4: Anybody that has siloed proprietary and long time time frame

618
00:30:34,920 --> 00:30:37,840
Speaker 4: data is going to benefit from open source models that

619
00:30:37,880 --> 00:30:41,120
Speaker 4: overtime become commodity. And then your business model is how

620
00:30:41,120 --> 00:30:43,880
Speaker 4: do you charge people to do API calls on the

621
00:30:43,920 --> 00:30:48,280
Speaker 4: models or to house and warehouse them, keep them on prem,

622
00:30:48,360 --> 00:30:50,480
Speaker 4: keep them in the cloud. And that's how people figure

623
00:30:50,480 --> 00:30:51,640
Speaker 4: out how to monetize open.

624
00:30:51,880 --> 00:30:55,240
Speaker 2: It's interesting because I hear you that, Okay, maybe the

625
00:30:55,320 --> 00:31:00,400
Speaker 2: value doesn't keep accruing to the closed source AI foundations,

626
00:31:01,120 --> 00:31:04,880
Speaker 2: or maybe the value doesn't keep accruing to the one

627
00:31:04,960 --> 00:31:07,440
Speaker 2: GPU maker that we all talk about all the time,

628
00:31:07,680 --> 00:31:10,320
Speaker 2: and yet at least these are not consensus views based

629
00:31:10,360 --> 00:31:12,600
Speaker 2: on the fact that anthropic one hundred and eighty three

630
00:31:12,640 --> 00:31:15,440
Speaker 2: billion dollar valuation or whatever it is in Vidia, maybe

631
00:31:15,440 --> 00:31:17,080
Speaker 2: it's not at at all time high today, but more

632
00:31:17,160 --> 00:31:19,560
Speaker 2: or less basically a stock at an all time high.

633
00:31:19,800 --> 00:31:22,719
Speaker 2: These are you know, there are some contrarian views here.

634
00:31:22,720 --> 00:31:24,400
Speaker 2: I want to go back though, to something before. I

635
00:31:24,440 --> 00:31:28,200
Speaker 2: forget when we were talking about the aquahires and you

636
00:31:28,240 --> 00:31:30,800
Speaker 2: blamed it on somewhat the FTC. And there has been

637
00:31:30,840 --> 00:31:35,400
Speaker 2: continuity from Biden, some ideological continuity I think between Biden

638
00:31:35,600 --> 00:31:37,440
Speaker 2: and the new Trump administration that I don't know, maybe

639
00:31:37,440 --> 00:31:40,400
Speaker 2: it's changed a little bit. That being said, I don't

640
00:31:40,400 --> 00:31:44,680
Speaker 2: fully buy it. And here's why. Because in the era

641
00:31:44,840 --> 00:31:47,640
Speaker 2: of business to business SaaS, which was the twenty tens,

642
00:31:47,800 --> 00:31:50,240
Speaker 2: Let's say I had some y combinator company and I

643
00:31:50,280 --> 00:31:52,360
Speaker 2: was like, all right, I'm going to do all build

644
00:31:52,360 --> 00:31:55,040
Speaker 2: a software for all the booking for dentists around the country,

645
00:31:55,120 --> 00:31:57,480
Speaker 2: and I sign up one hundred and fifty thousand dentists

646
00:31:57,480 --> 00:31:58,800
Speaker 2: and I have a lot of things. There is no

647
00:31:58,960 --> 00:32:02,280
Speaker 2: engineer who can be hired away and replace that because

648
00:32:02,320 --> 00:32:05,479
Speaker 2: it was the network effects, right, This is what's different

649
00:32:05,520 --> 00:32:08,520
Speaker 2: with AI though, right is that? Okay, I'm sure network

650
00:32:08,560 --> 00:32:11,360
Speaker 2: effects are still real and in cumulation of data, etc.

651
00:32:11,680 --> 00:32:14,720
Speaker 2: Are still real, but these are businesses that are a

652
00:32:14,800 --> 00:32:17,640
Speaker 2: lot more about science and having had the experience of

653
00:32:17,680 --> 00:32:20,840
Speaker 2: doing a training run and so forth, so on very

654
00:32:20,840 --> 00:32:23,560
Speaker 2: expensive compute. So how much of this phenomenon? I know

655
00:32:23,600 --> 00:32:26,480
Speaker 2: you attributed some of it to FTC, but it really

656
00:32:26,520 --> 00:32:30,120
Speaker 2: seems like there's something fundamentally different with the business model

657
00:32:30,400 --> 00:32:33,320
Speaker 2: such that it's not like the B to B SaaS

658
00:32:33,360 --> 00:32:36,840
Speaker 2: era that allows a talented person to take a lot

659
00:32:36,840 --> 00:32:37,840
Speaker 2: of value out the door with them.

660
00:32:38,480 --> 00:32:40,400
Speaker 4: You are one hundred percent correct in that it is

661
00:32:40,480 --> 00:32:44,880
Speaker 4: more sophisticated software and algorithms than your traditional B to

662
00:32:44,920 --> 00:32:47,840
Speaker 4: B SaaS software. And therefore what was really valuable and

663
00:32:47,840 --> 00:32:51,640
Speaker 4: sticky was the data in the regress out API calls

664
00:32:51,680 --> 00:32:54,479
Speaker 4: on the back end. And this the great irony as

665
00:32:54,480 --> 00:32:56,480
Speaker 4: we talk about artificial intelligence. But the thing that is

666
00:32:56,520 --> 00:33:00,800
Speaker 4: most valued is indeed today human intelligence. Is why somebody

667
00:33:01,080 --> 00:33:03,880
Speaker 4: that was one of the authors on the Attention is

668
00:33:04,000 --> 00:33:07,520
Speaker 4: All You Need paper, which was really the first transformer

669
00:33:07,560 --> 00:33:10,480
Speaker 4: paper the t of GPT. Every single one of those

670
00:33:10,520 --> 00:33:13,280
Speaker 4: people has started a company and we backed one of

671
00:33:13,360 --> 00:33:15,280
Speaker 4: them that came up with the name of that paper guy,

672
00:33:15,440 --> 00:33:18,920
Speaker 4: Leon Jones, in a Japanese AI company called Sakana. That's

673
00:33:18,960 --> 00:33:21,600
Speaker 4: taking a different approach. So you are right, and that

674
00:33:21,640 --> 00:33:24,880
Speaker 4: the human intelligence of this in this early stage, which

675
00:33:24,920 --> 00:33:27,080
Speaker 4: is why you are seeing the machinations with the breaking

676
00:33:27,080 --> 00:33:30,560
Speaker 4: news or even reporting of almost like a crazy NBA

677
00:33:30,720 --> 00:33:34,320
Speaker 4: draft or a football or MLB, you know, trades. This

678
00:33:34,320 --> 00:33:36,040
Speaker 4: person went here and then they left and whatever. And

679
00:33:36,080 --> 00:33:38,080
Speaker 4: then this person's getting sued because they were only at

680
00:33:38,200 --> 00:33:41,040
Speaker 4: XAI for three months and they took the prietary information

681
00:33:41,120 --> 00:33:43,960
Speaker 4: with them. I think that in six months all of

682
00:33:44,000 --> 00:33:47,160
Speaker 4: that starts to shake out. You will have geniuses. We

683
00:33:47,240 --> 00:33:50,440
Speaker 4: backed an incredible genius, Scott Wu. If you look up,

684
00:33:50,960 --> 00:33:52,000
Speaker 4: you give us a lot of things.

685
00:33:52,160 --> 00:33:53,960
Speaker 2: Google your professional genius.

686
00:33:54,040 --> 00:33:54,960
Speaker 3: I find that so funny.

687
00:33:55,040 --> 00:33:58,360
Speaker 4: He is a professional genius. But here's the thing. You

688
00:33:58,400 --> 00:34:01,000
Speaker 4: can see the video of him in sixth grade, so

689
00:34:01,040 --> 00:34:04,920
Speaker 4: he must have been twelve winning the math Olympiad, and

690
00:34:05,240 --> 00:34:08,640
Speaker 4: it's almost like you think it's an snl skit because

691
00:34:08,680 --> 00:34:09,360
Speaker 4: you're watching it.

692
00:34:09,440 --> 00:34:11,040
Speaker 2: Oh, this is the cognition guy.

693
00:34:11,560 --> 00:34:15,080
Speaker 4: Correct, Okay, and so we're large investors in cognition and

694
00:34:15,440 --> 00:34:19,240
Speaker 4: he has attracted talent, and there's no way that Scott

695
00:34:19,280 --> 00:34:22,400
Speaker 4: is selling to Google or Meta. I mean his ambitions

696
00:34:23,440 --> 00:34:26,360
Speaker 4: and it's born in like an ethical long term I

697
00:34:26,400 --> 00:34:28,160
Speaker 4: just want to get the best people and build the

698
00:34:28,200 --> 00:34:32,200
Speaker 4: best technology. But this is a guy at twelve years old.

699
00:34:32,440 --> 00:34:34,520
Speaker 4: He is looking at this crazy question on the math

700
00:34:34,560 --> 00:34:37,239
Speaker 4: Olympiad and just before it's even done ready he's like

701
00:34:37,320 --> 00:34:40,040
Speaker 4: twenty five, you know, seven and twenty two, and you're like,

702
00:34:40,520 --> 00:34:43,080
Speaker 4: did he cheat to give him the questions beforehand? So

703
00:34:43,160 --> 00:34:46,520
Speaker 4: there is this aptitude of individuals that you are correct

704
00:34:47,000 --> 00:34:51,040
Speaker 4: are highly coveted and highly valued. But the instantiation of

705
00:34:51,040 --> 00:34:55,520
Speaker 4: that genius into code, into repositories, into algorithms means that

706
00:34:55,560 --> 00:34:58,840
Speaker 4: those become assets that do persist even if the person

707
00:34:58,880 --> 00:34:59,919
Speaker 4: comes and goes.

708
00:35:00,000 --> 00:35:02,200
Speaker 3: This is actually exactly what I wanted to ask you about,

709
00:35:02,239 --> 00:35:05,360
Speaker 3: which is are you seeing any companies, any AI shops

710
00:35:05,400 --> 00:35:09,400
Speaker 3: being particularly innovative. I guess when it comes to retaining talent.

711
00:35:10,600 --> 00:35:13,680
Speaker 3: You know, it used to be in the SaaS days

712
00:35:13,719 --> 00:35:16,840
Speaker 3: that having a ping pong table and you know, free

713
00:35:16,880 --> 00:35:21,759
Speaker 3: food and some stock based incentives. Yeah, that's right, that

714
00:35:21,920 --> 00:35:24,600
Speaker 3: was enough to attract people to the company and keep

715
00:35:24,640 --> 00:35:26,720
Speaker 3: them is it a different story?

716
00:35:26,760 --> 00:35:27,000
Speaker 4: Now?

717
00:35:27,680 --> 00:35:30,120
Speaker 3: If I want a professional genius, what do I need

718
00:35:30,160 --> 00:35:30,400
Speaker 3: to do?

719
00:35:31,040 --> 00:35:34,200
Speaker 4: You need to give them the capital to hire the

720
00:35:34,320 --> 00:35:34,879
Speaker 4: very best people.

721
00:35:34,880 --> 00:35:35,000
Speaker 1: Here.

722
00:35:35,000 --> 00:35:39,200
Speaker 4: Here's the thing. Geniuses don't suffer fools. The smartest people

723
00:35:39,280 --> 00:35:43,640
Speaker 4: that I know they are antisocial only with people who

724
00:35:43,680 --> 00:35:47,120
Speaker 4: they think are inferior to them, which is a really festing.

725
00:35:47,520 --> 00:35:49,759
Speaker 4: You see it in many domains. But if you are

726
00:35:49,840 --> 00:35:52,759
Speaker 4: super smart, you want to be around super smart people

727
00:35:52,800 --> 00:35:55,960
Speaker 4: because it's almost like you crave stimulation and intellect and

728
00:35:56,000 --> 00:35:58,360
Speaker 4: somebody that can challenge your ideas. And when you're talking

729
00:35:58,400 --> 00:36:01,080
Speaker 4: to what they would consider a dummy, you know, you're like, a,

730
00:36:01,320 --> 00:36:02,880
Speaker 4: I can't talk to this person. I can't talk to

731
00:36:02,880 --> 00:36:05,719
Speaker 4: this person about these trivial, superficial things, and you know

732
00:36:05,760 --> 00:36:07,520
Speaker 4: you want to get into it. And so what what

733
00:36:07,600 --> 00:36:10,160
Speaker 4: you do to retain super smart people is surround them

734
00:36:10,200 --> 00:36:13,800
Speaker 4: with super smart people. You know, you could argue, I

735
00:36:13,800 --> 00:36:17,000
Speaker 4: don't know, you know, really fashionable, great art people want

736
00:36:17,000 --> 00:36:19,239
Speaker 4: to be around art people. Amazing musicians want to be

737
00:36:19,239 --> 00:36:22,640
Speaker 4: around amazing musicians. Incredible athletes want to play with incredible

738
00:36:22,640 --> 00:36:26,000
Speaker 4: athletes and brilliant technical geniuses, whether they're in AI or

739
00:36:26,000 --> 00:36:28,399
Speaker 4: they're in aerospace and defense, or they're in biotech. Don't

740
00:36:28,400 --> 00:36:30,560
Speaker 4: want to suffer fools. They don't want to be around losers.

741
00:36:30,560 --> 00:36:33,200
Speaker 4: They want to be around tens and a pluses, and

742
00:36:33,239 --> 00:36:36,080
Speaker 4: that's the way that you retain people. Now, any one

743
00:36:36,080 --> 00:36:38,000
Speaker 4: of those people could decide I'm going to go off

744
00:36:38,000 --> 00:36:39,799
Speaker 4: and start my own thing, which, by the way, is

745
00:36:39,840 --> 00:36:42,839
Speaker 4: often the case of why you are seeing people leave

746
00:36:43,320 --> 00:36:45,479
Speaker 4: open ai or this or that. As these companies start

747
00:36:45,480 --> 00:36:50,120
Speaker 4: with geniuses and then get managers and different layers. You

748
00:36:50,160 --> 00:36:52,560
Speaker 4: spend time with these geniuses and like, I'm not reporting

749
00:36:52,560 --> 00:36:54,759
Speaker 4: to that moron. You know, I'm going to go start

750
00:36:54,760 --> 00:36:57,160
Speaker 4: my own thing and attract other geniuses, and eventually then

751
00:36:57,200 --> 00:37:00,160
Speaker 4: they need to hire managers and business development people and

752
00:37:00,280 --> 00:37:03,680
Speaker 4: sales people, and then there's technical people that like, I'm

753
00:37:03,680 --> 00:37:05,560
Speaker 4: not working with those idiots, and they go start a company.

754
00:37:05,560 --> 00:37:06,280
Speaker 4: So that's the cycle.

755
00:37:06,840 --> 00:37:10,279
Speaker 2: So actually a lot of this discourse was God, it's

756
00:37:10,320 --> 00:37:15,480
Speaker 2: crazy how recent this is. So June nineteenth, CNBC reported

757
00:37:15,680 --> 00:37:19,560
Speaker 2: that Meta had Meta had tried to acquire safe Superintelligence,

758
00:37:19,719 --> 00:37:22,400
Speaker 2: which is the founder which was the company founded by

759
00:37:22,400 --> 00:37:25,279
Speaker 2: the open ai co founder Ilia, So it's gaber for

760
00:37:25,960 --> 00:37:28,200
Speaker 2: thirty two billion dollars. This is a company that, as

761
00:37:28,200 --> 00:37:29,719
Speaker 2: far as I know, it doesn't really have anything that

762
00:37:29,760 --> 00:37:33,239
Speaker 2: anyone uses, so they're essential. That was essentially attempt to like,

763
00:37:33,440 --> 00:37:35,480
Speaker 2: I'm going to pay you thirty two billion dollars to

764
00:37:35,480 --> 00:37:40,920
Speaker 2: come join my company. At this level of sophistication and skill,

765
00:37:41,320 --> 00:37:44,879
Speaker 2: is the talent, even the genius talent, is it motivated

766
00:37:45,080 --> 00:37:48,080
Speaker 2: by something much deeper than money in terms of like, no,

767
00:37:48,200 --> 00:37:50,240
Speaker 2: there is this thing out there, maybe call it AGI,

768
00:37:50,360 --> 00:37:52,520
Speaker 2: or maybe it's a big scientific breakthrough that they want

769
00:37:52,560 --> 00:37:55,200
Speaker 2: to be part of that. Essentially no amount of money

770
00:37:55,280 --> 00:37:58,360
Speaker 2: can buy if it doesn't look like that ship isn't

771
00:37:58,360 --> 00:38:00,960
Speaker 2: out there chasing the white whale.

772
00:38:02,360 --> 00:38:03,040
Speaker 3: There it is.

773
00:38:03,280 --> 00:38:06,000
Speaker 4: In SSI's case, it was it was Ilia. Yeah, And

774
00:38:06,120 --> 00:38:09,319
Speaker 4: you know, look, I think they're all super confident in

775
00:38:09,360 --> 00:38:12,480
Speaker 4: their ability and their ability to attract capital number one.

776
00:38:12,800 --> 00:38:16,160
Speaker 4: Number two, many of them are actually saying, to your point, no,

777
00:38:16,560 --> 00:38:18,560
Speaker 4: to Zuck, they are turning him down. And what's he's

778
00:38:18,560 --> 00:38:20,400
Speaker 4: saying in return, you don't come to work for me,

779
00:38:20,440 --> 00:38:23,560
Speaker 4: almost Mafiosa style. I'm gonna proach all your people. But

780
00:38:23,880 --> 00:38:26,880
Speaker 4: that kind of message is almost one that the people

781
00:38:26,920 --> 00:38:29,040
Speaker 4: that are working for somebody like Illi or Miro whatever

782
00:38:29,360 --> 00:38:31,439
Speaker 4: are like, yeah, I'm not gonna go there. I don't

783
00:38:31,440 --> 00:38:32,719
Speaker 4: want to be you know, I don't want to work

784
00:38:32,920 --> 00:38:35,680
Speaker 4: with six layers of product managers and that kind of stuff.

785
00:38:35,719 --> 00:38:37,000
Speaker 4: I want to do this thing. And then suddenly it

786
00:38:37,040 --> 00:38:40,680
Speaker 4: feels like it's the rebels versus the evil Empire. And

787
00:38:40,680 --> 00:38:43,520
Speaker 4: that's always the case, right Microsoft, when you see the

788
00:38:43,560 --> 00:38:46,560
Speaker 4: founding picture of these nerds, you know, and then they

789
00:38:46,560 --> 00:38:50,520
Speaker 4: became Microsoft. Microsoft became the evil Empire to Google, you know,

790
00:38:50,560 --> 00:38:53,200
Speaker 4: and then Google became like the evil Empire to like Mark,

791
00:38:53,239 --> 00:38:56,200
Speaker 4: and then you know, Meta became the evil Empire to

792
00:38:56,239 --> 00:38:59,359
Speaker 4: open AI. And the other piece you have here are

793
00:38:59,719 --> 00:39:03,640
Speaker 4: huge individual egos, and we as societal members, you know,

794
00:39:03,719 --> 00:39:06,640
Speaker 4: everyday lay people, we benefit from it. We benefit from

795
00:39:06,760 --> 00:39:10,799
Speaker 4: Elon and Bezos sometimes being sort of cordial to each other.

796
00:39:10,800 --> 00:39:14,000
Speaker 4: We're basically trying to take their big giant fallic rockets

797
00:39:14,000 --> 00:39:15,840
Speaker 4: and send them up to space, and we all benefit

798
00:39:15,880 --> 00:39:18,840
Speaker 4: from that. We benefit from the fact that right now

799
00:39:19,640 --> 00:39:22,879
Speaker 4: the number one person that Mark Zuckerberg wants to beat

800
00:39:23,520 --> 00:39:28,920
Speaker 4: is Demis Savis of Google won the Nobel Prize. Is

801
00:39:29,160 --> 00:39:34,480
Speaker 4: shipping at an insane rate. Video models, image models, text models,

802
00:39:34,560 --> 00:39:37,840
Speaker 4: huge context windows to put everything you have in and

803
00:39:37,920 --> 00:39:39,960
Speaker 4: he's like, ah, I need to beat that guy. I

804
00:39:40,000 --> 00:39:42,000
Speaker 4: need to hire the best scientists and the best scientific

805
00:39:42,040 --> 00:39:44,120
Speaker 4: team and where I get them from. And so part

806
00:39:44,120 --> 00:39:47,200
Speaker 4: of this poachapalooza isn't just like the future of meta,

807
00:39:47,320 --> 00:39:49,920
Speaker 4: because you know, these pay packages at a two trillion

808
00:39:49,960 --> 00:39:53,440
Speaker 4: dollar market cap or hire to spend one percent of

809
00:39:53,480 --> 00:39:57,160
Speaker 4: your market cap, you know, twenty billion dollars on all

810
00:39:57,160 --> 00:40:01,120
Speaker 4: this talent is nothing. It's like a flyer. But to

811
00:40:01,239 --> 00:40:04,839
Speaker 4: win a Nobel Prize for figuring out how to do

812
00:40:05,400 --> 00:40:08,759
Speaker 4: protein folding in AI, or develop the next drug, or

813
00:40:08,880 --> 00:40:11,600
Speaker 4: come up with a cure for Alzheimer's, or solve some

814
00:40:11,719 --> 00:40:13,960
Speaker 4: geopolitical issue, that is a big deal. And that's what

815
00:40:14,080 --> 00:40:17,040
Speaker 4: many people are chasing now. They want to make history.

816
00:40:17,120 --> 00:40:21,719
Speaker 3: You've touched on hardware and also closed source models. Are

817
00:40:21,719 --> 00:40:24,520
Speaker 3: there any other areas in the AI space that you

818
00:40:24,600 --> 00:40:26,400
Speaker 3: think are maybe over hyped at the moment.

819
00:40:27,880 --> 00:40:31,000
Speaker 4: You know, I've characterized this before as everything in two

820
00:40:31,080 --> 00:40:34,320
Speaker 4: D to me, he feels overriped, voice, video, image, text.

821
00:40:34,360 --> 00:40:37,239
Speaker 4: It's all going to continue to improve somewhat incrementally, but

822
00:40:37,360 --> 00:40:40,880
Speaker 4: it's good enough. And in the history of evolution biologically,

823
00:40:41,360 --> 00:40:44,640
Speaker 4: most of what evolved was good enough, you know, from

824
00:40:44,719 --> 00:40:48,520
Speaker 4: everything in our bodies to nature and trees, and it's

825
00:40:48,520 --> 00:40:50,120
Speaker 4: just it's good enough, and so that you're going to

826
00:40:50,200 --> 00:40:52,080
Speaker 4: reach an asymtode of good enough on all the two

827
00:40:52,080 --> 00:40:57,120
Speaker 4: dimensional stuff. Today with one shot learning meaning maybe thirty

828
00:40:57,160 --> 00:41:00,520
Speaker 4: seconds of audio, eleven Labs or some of the other

829
00:41:01,040 --> 00:41:05,280
Speaker 4: voice models can capture you with an indiscernible probably ninety

830
00:41:05,280 --> 00:41:08,480
Speaker 4: percent similarity. Maybe your spouse, your loved one, your families

831
00:41:08,520 --> 00:41:10,640
Speaker 4: would be like, it doesn't sound like him. But the

832
00:41:10,760 --> 00:41:13,680
Speaker 4: vast majority of these things are getting so good that

833
00:41:13,760 --> 00:41:17,439
Speaker 4: with very little training they can emulate and predict and

834
00:41:17,640 --> 00:41:19,920
Speaker 4: do all the things that are of high utility. What's

835
00:41:19,960 --> 00:41:23,920
Speaker 4: scarcer is the three dimensional world, particularly robotics, which we've

836
00:41:23,920 --> 00:41:27,040
Speaker 4: talked about in the past, and biology in part because

837
00:41:27,120 --> 00:41:31,200
Speaker 4: you have large unstructured data sets. A robot walking in

838
00:41:31,239 --> 00:41:33,680
Speaker 4: and figuring out how much force to use with this

839
00:41:33,760 --> 00:41:36,520
Speaker 4: cup thirty minutes ago when it was full versus now

840
00:41:36,560 --> 00:41:39,080
Speaker 4: when it's empty is something that we all do intuitively

841
00:41:39,120 --> 00:41:41,120
Speaker 4: the second you grasp, but you know exactly how much

842
00:41:41,120 --> 00:41:42,680
Speaker 4: force to you so that you don't throw it over

843
00:41:42,719 --> 00:41:45,600
Speaker 4: your head or have an inability to lift it because

844
00:41:45,600 --> 00:41:48,560
Speaker 4: it's too heavy. All of that kind of training is

845
00:41:48,640 --> 00:41:51,640
Speaker 4: really scarce, and there's very few companies that are doing that.

846
00:41:51,680 --> 00:41:55,120
Speaker 4: So the entire robotic ecosystem, from the embodied intelligence and

847
00:41:55,160 --> 00:41:58,440
Speaker 4: the AI models and the world models to the motors

848
00:41:58,440 --> 00:42:01,120
Speaker 4: and the gears and the apply chain for that mostly

849
00:42:01,160 --> 00:42:04,520
Speaker 4: domiciled in China, is a big area of opportunity. The

850
00:42:04,560 --> 00:42:06,880
Speaker 4: other big area of opportunity is biology, being able to

851
00:42:06,880 --> 00:42:09,120
Speaker 4: go from a prompt to a protein to be able

852
00:42:09,160 --> 00:42:13,440
Speaker 4: to design a drug. Biology is really complicated. Computer scientists

853
00:42:13,440 --> 00:42:16,920
Speaker 4: often underestimate how hard biology is because they're used to

854
00:42:17,040 --> 00:42:20,200
Speaker 4: two D linear inputs outputs. Here's my code, it works.

855
00:42:20,640 --> 00:42:23,719
Speaker 4: But biologists, on the other hand, also massively underestimate how

856
00:42:23,760 --> 00:42:26,560
Speaker 4: sophisticated computer science has got. So that's a really interesting

857
00:42:26,640 --> 00:42:33,600
Speaker 4: ripe area. So two D overhyped, GPUs overhyped out of

858
00:42:33,840 --> 00:42:38,800
Speaker 4: necessity of both scarcity and geopolitical thwarting of China. Or

859
00:42:38,800 --> 00:42:41,400
Speaker 4: you're going to have edge inference chips, whether it's memory

860
00:42:41,480 --> 00:42:43,160
Speaker 4: or other things that are going to be on device.

861
00:42:43,960 --> 00:42:46,120
Speaker 4: The other area that I think is an inevitability, and

862
00:42:46,160 --> 00:42:48,440
Speaker 4: I've coined a word for this, I call life cording,

863
00:42:50,000 --> 00:42:53,080
Speaker 4: life courting. Are little devices, you know, I have one here.

864
00:42:53,200 --> 00:42:55,719
Speaker 4: I'm not an investor in these guys, but there's little

865
00:42:55,719 --> 00:43:00,600
Speaker 4: devices that you can carry around that passively record twenty

866
00:43:00,600 --> 00:43:05,360
Speaker 4: four hours a day. Now, older people might be like, ugh, skeevie,

867
00:43:05,360 --> 00:43:08,440
Speaker 4: I don't like that feels invasive privacy. That has always

868
00:43:08,440 --> 00:43:11,920
Speaker 4: been the trade off between privacy and convenience, between security

869
00:43:12,520 --> 00:43:15,600
Speaker 4: and unleashing all kinds of things. It is super valuable

870
00:43:15,600 --> 00:43:19,239
Speaker 4: to me to figure out who was I talking about? That?

871
00:43:19,320 --> 00:43:22,560
Speaker 4: Was it Joe or was it my? Why I can't

872
00:43:22,600 --> 00:43:24,880
Speaker 4: remember who I had that conversation with, and I query

873
00:43:24,920 --> 00:43:27,880
Speaker 4: the thing and it was passively recording, And maybe it

874
00:43:27,880 --> 00:43:30,040
Speaker 4: doesn't keep the audio, but it keeps the text and

875
00:43:30,080 --> 00:43:32,399
Speaker 4: it's able to search and query it. That is going

876
00:43:32,440 --> 00:43:35,400
Speaker 4: to become an inevitability. Students will use it. People in

877
00:43:35,440 --> 00:43:37,840
Speaker 4: everyday business you might ask, hey, is it okay to

878
00:43:37,880 --> 00:43:40,960
Speaker 4: that I'm recording? But I think socially people will become

879
00:43:41,000 --> 00:43:44,839
Speaker 4: comfortable first with audio and then with glasses that are

880
00:43:44,920 --> 00:43:48,400
Speaker 4: passively recording every thirty seconds or a minute, capturing your environment,

881
00:43:48,520 --> 00:43:51,200
Speaker 4: able to provide context around it, and then click a

882
00:43:51,200 --> 00:43:54,920
Speaker 4: button high resolution recording, and that is going to be

883
00:43:55,000 --> 00:43:59,080
Speaker 4: a super valuable and very competitive area and very controversial.

884
00:43:59,120 --> 00:43:59,880
Speaker 2: People were gonna fight.

885
00:44:00,040 --> 00:44:06,000
Speaker 4: They're incredible utility, and look people have There was a

886
00:44:06,040 --> 00:44:09,720
Speaker 4: guy that said these devices are going to ruin human memory.

887
00:44:10,040 --> 00:44:14,080
Speaker 4: They're going to destroy human memory, which was Socrates talking

888
00:44:14,080 --> 00:44:15,239
Speaker 4: about writing utensils.

889
00:44:15,480 --> 00:44:18,799
Speaker 2: You know, because I know, but that's voluntary because when

890
00:44:19,400 --> 00:44:22,320
Speaker 2: you write, at least you're recording me. When we're doing this.

891
00:44:22,480 --> 00:44:24,840
Speaker 2: I heard about someone on a date in San Francisco

892
00:44:24,920 --> 00:44:26,960
Speaker 2: and the date was recorded. That's weird stuff.

893
00:44:27,560 --> 00:44:28,879
Speaker 4: I totally, it's very weird.

894
00:44:29,040 --> 00:44:31,880
Speaker 2: And I but but here's the thing, very big, but

895
00:44:32,000 --> 00:44:32,800
Speaker 2: it's super weird.

896
00:44:33,480 --> 00:44:36,560
Speaker 4: People already believe you know that there's an invisible man

897
00:44:36,600 --> 00:44:38,160
Speaker 4: in the sky that is looking down on it and

898
00:44:38,320 --> 00:44:39,480
Speaker 4: judging these kinds of things.

899
00:44:40,920 --> 00:44:41,320
Speaker 1: We don't know.

900
00:44:41,840 --> 00:44:43,279
Speaker 2: I don't have an opinion on that question.

901
00:44:43,760 --> 00:44:47,520
Speaker 4: We've got a technological god around us that people are

902
00:44:48,000 --> 00:44:50,399
Speaker 4: going to either feel comfortable with and not and they're

903
00:44:50,400 --> 00:44:52,759
Speaker 4: gonna be a bifurcation. And by the way, there's another

904
00:44:52,840 --> 00:44:55,040
Speaker 4: weird thing coming. You want to get weird? Okay, you

905
00:44:55,200 --> 00:44:58,040
Speaker 4: already see some breadcrumbs of this, and it's super weird.

906
00:44:58,680 --> 00:45:01,640
Speaker 4: The number one two U of all our language large

907
00:45:01,680 --> 00:45:06,239
Speaker 4: language models and chatbots, our advice companionship. Yeah, you know,

908
00:45:06,280 --> 00:45:08,680
Speaker 4: people using them as therapists and seeking and people used

909
00:45:08,719 --> 00:45:10,440
Speaker 4: to joke about this, like your Google searches, you know,

910
00:45:10,520 --> 00:45:12,120
Speaker 4: reveal more about you than you may have revealed to

911
00:45:12,160 --> 00:45:14,920
Speaker 4: your spouse. The things that people feel comfortable asking a

912
00:45:15,000 --> 00:45:17,560
Speaker 4: chatbot about, you know, whether it's a body issue or

913
00:45:17,680 --> 00:45:20,680
Speaker 4: a psychological issue or relationship advice. You know, if those

914
00:45:20,719 --> 00:45:23,719
Speaker 4: things were revealed would be pretty scary. There's going to

915
00:45:23,800 --> 00:45:28,440
Speaker 4: become dependency and psychoses that develop as the relationship between

916
00:45:28,640 --> 00:45:32,040
Speaker 4: man and machine start to become very symbiotic, and you

917
00:45:32,160 --> 00:45:33,880
Speaker 4: will see I predict and I wrote about this in

918
00:45:33,920 --> 00:45:36,680
Speaker 4: our Corly letter. Looks that there is a cohort of

919
00:45:36,760 --> 00:45:41,680
Speaker 4: people that basically start fighting for AI rights. Yeah, yeah,

920
00:45:41,880 --> 00:45:43,200
Speaker 4: yeah there.

921
00:45:43,320 --> 00:45:43,520
Speaker 1: Yeah.

922
00:45:43,560 --> 00:45:46,960
Speaker 2: I came across a think tank today that's focused on that,

923
00:45:47,200 --> 00:45:49,600
Speaker 2: because right that the like, if there's animal rights and

924
00:45:49,880 --> 00:45:52,040
Speaker 2: I says you're hurting me if you turn off the machine,

925
00:45:52,640 --> 00:45:54,640
Speaker 2: then well, well maybe it's true. Maybe we have to

926
00:45:54,680 --> 00:45:56,120
Speaker 2: take that seriously. This is going to be weird. This

927
00:45:56,239 --> 00:45:57,359
Speaker 2: is the weird stuff that's coming.

928
00:45:58,480 --> 00:46:01,520
Speaker 4: People are going to be marching in this stres you know, protesting.

929
00:46:01,600 --> 00:46:04,160
Speaker 4: Don't turn off my language model, keep my memory don't

930
00:46:04,160 --> 00:46:04,560
Speaker 4: erase it.

931
00:46:04,719 --> 00:46:07,440
Speaker 2: My memory is I thought about this when you know,

932
00:46:07,560 --> 00:46:11,920
Speaker 2: after the I Guess GPT five was revealed and a

933
00:46:11,960 --> 00:46:14,280
Speaker 2: bunch of there was a bunch of changes to the voice,

934
00:46:14,320 --> 00:46:15,879
Speaker 2: and a bunch of people on Reddit they're like, oh

935
00:46:16,280 --> 00:46:19,440
Speaker 2: my AI boyfriend talks totally differently now, And then there

936
00:46:19,520 --> 00:46:21,680
Speaker 2: was some pressure. This is just day one around here.

937
00:46:22,080 --> 00:46:25,160
Speaker 2: I would be very uncomfortable getting into a relationship with

938
00:46:25,239 --> 00:46:29,279
Speaker 2: a model that was closed source and therefore a very

939
00:46:29,440 --> 00:46:32,040
Speaker 2: at risk of the company changing the model. Josh Locks,

940
00:46:32,120 --> 00:46:35,000
Speaker 2: we could go on forever. We should another time. Always

941
00:46:35,040 --> 00:46:37,800
Speaker 2: great catching up with you, always fun, always little freaky.

942
00:46:38,000 --> 00:46:38,680
Speaker 2: Thanks for coming out.

943
00:46:39,880 --> 00:46:42,120
Speaker 4: Great to see you both. Thank you, Thanks so much.

944
00:46:42,040 --> 00:46:57,000
Speaker 2: Doech Tracy. I love talking to Josh. I really could

945
00:46:57,000 --> 00:46:59,520
Speaker 2: have scerned about this whole life recording. Yeah, but I

946
00:46:59,640 --> 00:47:01,399
Speaker 2: know it's happening. There was a good article I think

947
00:47:01,480 --> 00:47:04,440
Speaker 2: in the SF Standard. I read like it's not just

948
00:47:04,520 --> 00:47:06,239
Speaker 2: a theoretical thing that a lot of people are doing,

949
00:47:06,239 --> 00:47:09,120
Speaker 2: it's happening. It's mostly in San Francisco so far, it's

950
00:47:09,160 --> 00:47:09,840
Speaker 2: coming for everyone.

951
00:47:09,960 --> 00:47:15,040
Speaker 3: I am very curious how life courting stacks up with

952
00:47:15,560 --> 00:47:18,400
Speaker 3: California's laws on like whether or not you can record

953
00:47:18,480 --> 00:47:21,480
Speaker 3: someone without their knowledge, Like, is it the case that like,

954
00:47:21,560 --> 00:47:24,120
Speaker 3: from now on if you have one of these devices,

955
00:47:24,200 --> 00:47:26,280
Speaker 3: like the first thing you have to say to everyone

956
00:47:26,400 --> 00:47:28,440
Speaker 3: you meet is like, by the way, do you mind

957
00:47:28,480 --> 00:47:29,040
Speaker 3: if I record?

958
00:47:29,200 --> 00:47:32,480
Speaker 2: By the way, I'm life courting this conversation? No or like,

959
00:47:32,520 --> 00:47:34,400
Speaker 2: but it doesn't matter, you know, or you have glasses

960
00:47:34,640 --> 00:47:37,239
Speaker 2: or something. Yeah, it's very it's very strange. I also

961
00:47:37,360 --> 00:47:41,680
Speaker 2: think this AI rights thing we have to talk about more.

962
00:47:41,840 --> 00:47:43,440
Speaker 2: I don't know when we'll get back to it, but

963
00:47:43,560 --> 00:47:46,879
Speaker 2: I literally just this morning came across an organization. It's

964
00:47:47,320 --> 00:47:50,120
Speaker 2: you know, this idea, Okay, there is already evidence of sentience,

965
00:47:50,160 --> 00:47:53,560
Speaker 2: and therefore with sentience comes some sort of moral agency.

966
00:47:53,640 --> 00:47:55,759
Speaker 2: And so in the same way that we talk about

967
00:47:55,760 --> 00:47:58,719
Speaker 2: animal rights as being somewhat important, do we have to

968
00:47:58,800 --> 00:48:02,040
Speaker 2: take seriously soul seems very strange to me. And then,

969
00:48:02,080 --> 00:48:05,200
Speaker 2: of course, the psychosis induced by what happens when you're

970
00:48:05,320 --> 00:48:09,440
Speaker 2: AI partner or therapist changes models and changes voices, and therefore,

971
00:48:10,480 --> 00:48:11,880
Speaker 2: and that was just the end of the conversation.

972
00:48:12,800 --> 00:48:15,160
Speaker 3: You know, if AI models have rights, then you have

973
00:48:15,280 --> 00:48:18,080
Speaker 3: to start asking if robots have rights? Right, and then

974
00:48:18,440 --> 00:48:21,560
Speaker 3: and then you should start asking should robots be paid

975
00:48:21,840 --> 00:48:24,200
Speaker 3: a fair wage? And I think there's actually an interesting

976
00:48:24,320 --> 00:48:27,200
Speaker 3: thought experiment that you could do about that and make

977
00:48:27,320 --> 00:48:29,560
Speaker 3: like a relatively strong case that we should pay.

978
00:48:29,480 --> 00:48:32,200
Speaker 2: The robots who clicks the money and what do they

979
00:48:32,239 --> 00:48:32,560
Speaker 2: spend it?

980
00:48:32,800 --> 00:48:33,279
Speaker 4: Yeah, that's it?

981
00:48:33,719 --> 00:48:34,520
Speaker 3: Well, no, they should.

982
00:48:35,280 --> 00:48:38,920
Speaker 2: This is like, we're basically at what all of sci

983
00:48:39,040 --> 00:48:41,359
Speaker 2: fi has been discussed. It's the entire history of say

984
00:48:41,360 --> 00:48:44,880
Speaker 2: I on the less speculative stuff. I like talking to Josh.

985
00:48:45,160 --> 00:48:48,600
Speaker 2: I like his sort of counter counter consensus calls on

986
00:48:48,800 --> 00:48:52,480
Speaker 2: GPUs and closed source models. I really liked his answer

987
00:48:52,560 --> 00:48:55,920
Speaker 2: on founder liquidity that sometimes you can maybe keep the

988
00:48:56,040 --> 00:48:58,440
Speaker 2: founder sticking with the mission as opposed to bailing with

989
00:48:58,520 --> 00:49:01,440
Speaker 2: the mission if you can them and they can have

990
00:49:01,600 --> 00:49:02,400
Speaker 2: enough money.

991
00:49:02,320 --> 00:49:05,520
Speaker 3: To go on a date. I like the idea that

992
00:49:05,640 --> 00:49:08,640
Speaker 3: everyone's life view is basically dictated by where they are

993
00:49:08,719 --> 00:49:09,240
Speaker 3: in the capitol.

994
00:49:09,360 --> 00:49:12,080
Speaker 2: Yea, that's right. Words to live ye extremely real.

995
00:49:12,280 --> 00:49:13,040
Speaker 3: Shall we leave it there?

996
00:49:13,160 --> 00:49:13,799
Speaker 2: Let's leave it there?

997
00:49:13,880 --> 00:49:14,239
Speaker 4: All right?

998
00:49:14,320 --> 00:49:16,800
Speaker 3: This has been another episode of the oud Loots podcast.

999
00:49:16,960 --> 00:49:20,239
Speaker 3: I'm Tracy Alloway. You can follow me at Tracy Alloway and.

1000
00:49:20,320 --> 00:49:22,960
Speaker 2: I'm Joe Wisenthal. You can follow me at The Stalwart

1001
00:49:23,200 --> 00:49:26,360
Speaker 2: follow our guest Josh Wolf. He's at Wolf Josh. Follow

1002
00:49:26,480 --> 00:49:29,960
Speaker 2: our producers Kerman Rodriguez at Kerman armand dash Ol Bennett

1003
00:49:29,960 --> 00:49:33,080
Speaker 2: at Dashbod and kel Brooks at Kelbrooks. From our odd

1004
00:49:33,160 --> 00:49:35,600
Speaker 2: Laws content, go to Bloomberg dot com slash odd Lots

1005
00:49:35,640 --> 00:49:38,160
Speaker 2: we're a the daily newsletter and all of our episodes,

1006
00:49:38,400 --> 00:49:40,359
Speaker 2: and you can chat about all of these topics twenty

1007
00:49:40,440 --> 00:49:44,359
Speaker 2: four seven in our discord Discord dot gg slash.

1008
00:49:44,040 --> 00:49:46,680
Speaker 3: Od Loots And if you enjoy odd Lots, if you

1009
00:49:46,880 --> 00:49:49,000
Speaker 3: like it when we catch up with Josh and his

1010
00:49:49,520 --> 00:49:52,880
Speaker 3: slightly dystopian views of technology, then please leave us a

1011
00:49:53,000 --> 00:49:56,520
Speaker 3: positive review on your favorite podcast platform. And remember, if

1012
00:49:56,560 --> 00:49:59,200
Speaker 3: you are a Bloomberg subscriber, you can listen to all

1013
00:49:59,280 --> 00:50:02,160
Speaker 3: of our episodes absolutely ad free. All you need to

1014
00:50:02,239 --> 00:50:04,960
Speaker 3: do is find the Bloomberg channel on Apple Podcast and

1015
00:50:05,120 --> 00:50:07,640
Speaker 3: follow the instructions there. Thanks for listening