1
00:00:02,520 --> 00:00:15,040
Speaker 1: Bloomberg Audio Studios, podcasts, radio news. Bloomberg Tech is live

2
00:00:15,120 --> 00:00:18,599
Speaker 1: from the heart of Silicon Valley with Ed Lar though

3
00:00:18,760 --> 00:00:19,880
Speaker 1: in San Francisco.

4
00:00:23,280 --> 00:00:24,799
Speaker 2: This is Bloomberg Tech coming up.

5
00:00:24,880 --> 00:00:27,920
Speaker 3: Tech markets are mixed as US around jitters revived risk

6
00:00:27,960 --> 00:00:31,319
Speaker 3: off sentiment and send oil higher plus Open AI gets

7
00:00:31,320 --> 00:00:34,000
Speaker 3: a five hundred and twenty million dollar credit line from

8
00:00:34,080 --> 00:00:37,000
Speaker 3: Bank of America, while the chat GPT maker prepares to

9
00:00:37,040 --> 00:00:41,159
Speaker 3: roll up its most advanced model tomorrow. Eric Hippo from

10
00:00:41,240 --> 00:00:44,400
Speaker 3: Lera Hippo discusses the early stage investing landscape at a

11
00:00:44,440 --> 00:00:48,159
Speaker 3: time of ballooning AI valuations, New York's tech scene, and

12
00:00:48,320 --> 00:00:50,240
Speaker 3: a whole lot more.

13
00:00:50,600 --> 00:00:51,560
Speaker 2: Welcome to the show.

14
00:00:51,680 --> 00:00:53,640
Speaker 3: Let's get to check in on these markets, and it

15
00:00:53,800 --> 00:00:56,600
Speaker 3: is a mixed picture right now. It's difficult to say

16
00:00:56,840 --> 00:00:59,800
Speaker 3: what the main driver is. Earlier in the morning, headlines

17
00:00:59,800 --> 00:01:03,320
Speaker 3: from the President of the United States about the status

18
00:01:03,320 --> 00:01:06,400
Speaker 3: of relations with Iran did have an impact on sentiment.

19
00:01:06,520 --> 00:01:10,959
Speaker 3: Semiconductors are rebounding from a volatile period of selling, which

20
00:01:11,000 --> 00:01:14,360
Speaker 3: all stems from Samsung's prelim earnings. But there is sort

21
00:01:14,400 --> 00:01:19,280
Speaker 3: of a reassessment of the AI trade. I'd say it's

22
00:01:19,520 --> 00:01:22,000
Speaker 3: what's happening with the war in Iran as we await

23
00:01:22,000 --> 00:01:25,119
Speaker 3: a press conference from the President in Ankora to start

24
00:01:25,160 --> 00:01:26,920
Speaker 3: any moment now. By the way, I want to bring

25
00:01:26,959 --> 00:01:30,280
Speaker 3: in Bloomberg's Washington correspondent Tyler Kendall. We've heard from the

26
00:01:30,319 --> 00:01:34,640
Speaker 3: President a few times this morning already. How would we

27
00:01:34,720 --> 00:01:38,200
Speaker 3: summarize the United States position right now on Iran?

28
00:01:39,480 --> 00:01:42,120
Speaker 4: Well, and there were a lot of fast moving developments

29
00:01:42,120 --> 00:01:44,440
Speaker 4: and we're about to get more when President Trump takes

30
00:01:44,480 --> 00:01:47,160
Speaker 4: the stage there. But he is essentially threatening that we

31
00:01:47,200 --> 00:01:50,560
Speaker 4: are going to see further US strikes on Iran tonight

32
00:01:50,680 --> 00:01:53,760
Speaker 4: after a wave of attacks overnight in response to Iranian

33
00:01:53,800 --> 00:01:56,880
Speaker 4: aggression in the Strait of Removes, because over the last

34
00:01:56,920 --> 00:02:02,160
Speaker 4: few days we've seen at least three commercial tankers targeted

35
00:02:02,200 --> 00:02:05,440
Speaker 4: by Iran for transiting through the strait using that US

36
00:02:05,440 --> 00:02:08,040
Speaker 4: approved corridor that gets them closer to the coast of

37
00:02:08,040 --> 00:02:12,000
Speaker 4: Oman and further away from the Iranian coast. And we're

38
00:02:12,000 --> 00:02:15,040
Speaker 4: hearing from Iran within the last hour of reporting on

39
00:02:15,120 --> 00:02:18,320
Speaker 4: Iranian state TV, threatening that the country will close down

40
00:02:18,440 --> 00:02:21,720
Speaker 4: the strait completely if the US does follow through with

41
00:02:21,800 --> 00:02:24,720
Speaker 4: its threats for further strikes on the country, and it

42
00:02:24,760 --> 00:02:27,720
Speaker 4: hasn't just been escalation when it comes to kinetic action,

43
00:02:27,880 --> 00:02:30,919
Speaker 4: but we're also seeing just yesterday the US Treasury Department

44
00:02:31,000 --> 00:02:34,080
Speaker 4: moving to revoke that critical waiver when it comes to

45
00:02:34,160 --> 00:02:36,960
Speaker 4: allowing Iran to sell its crewed to market. So the

46
00:02:37,080 --> 00:02:39,880
Speaker 4: US is now taking that further stance when it comes

47
00:02:39,880 --> 00:02:43,480
Speaker 4: to squeezing Aran on the economic front, as it appears

48
00:02:43,639 --> 00:02:47,160
Speaker 4: that these negotiations at this point have really been stalled.

49
00:02:47,280 --> 00:02:50,120
Speaker 4: Now even with all of the said ed, President Trump

50
00:02:50,200 --> 00:02:52,480
Speaker 4: here is saying that he thinks the ceasefire is over,

51
00:02:52,560 --> 00:02:56,600
Speaker 4: but he did leave open that pathway that negotiations could continue.

52
00:02:56,639 --> 00:02:59,240
Speaker 4: He said that he wasn't going to stop his diplomatic

53
00:02:59,280 --> 00:03:02,000
Speaker 4: team from truck buying at least, but in his opinion

54
00:03:02,040 --> 00:03:04,119
Speaker 4: and in his words, he thinks at this point it's

55
00:03:04,160 --> 00:03:04,480
Speaker 4: just not.

56
00:03:04,480 --> 00:03:05,120
Speaker 5: Worth that effort.

57
00:03:05,160 --> 00:03:06,960
Speaker 2: Bloombos Tyler Kenda in Washington. Thank you.

58
00:03:07,040 --> 00:03:09,200
Speaker 3: We will hear from the President very soon again. To

59
00:03:09,280 --> 00:03:14,040
Speaker 3: summarize markets, tech generally lower semiconductors outperforming. Let's try and

60
00:03:14,040 --> 00:03:16,800
Speaker 3: get to our top tech story. Bank of America is

61
00:03:16,880 --> 00:03:21,080
Speaker 3: changing its tune on Open Ai. After initially passing on

62
00:03:21,240 --> 00:03:24,160
Speaker 3: lending to the AI company, it's now extending a five

63
00:03:24,240 --> 00:03:27,200
Speaker 3: hundred and twenty million dollar credit line. That's, according to sources,

64
00:03:27,240 --> 00:03:30,760
Speaker 3: joining US with the exclusive Bloombo Shreda Nadarajan. Okay, so

65
00:03:30,800 --> 00:03:32,800
Speaker 3: we're saying that this is a U turn. Give us

66
00:03:32,840 --> 00:03:33,519
Speaker 3: the backstory.

67
00:03:33,720 --> 00:03:35,000
Speaker 2: Well, it is an interesting clash.

68
00:03:35,040 --> 00:03:39,120
Speaker 6: The AI sector right now is represented by companies like

69
00:03:39,320 --> 00:03:43,360
Speaker 6: open Ai, high growth, massive cash burn companies that are

70
00:03:43,360 --> 00:03:45,600
Speaker 6: dominating the sector. And on the other hand you have

71
00:03:45,640 --> 00:03:48,520
Speaker 6: the old god Wall Street Bank of America, the second

72
00:03:48,640 --> 00:03:51,880
Speaker 6: largest lender in the United States, that is defined by

73
00:03:51,880 --> 00:03:55,120
Speaker 6: its responsible growth mantra. So when a company like that

74
00:03:55,240 --> 00:03:58,000
Speaker 6: comes out and us for a credit line, you can

75
00:03:58,040 --> 00:04:00,520
Speaker 6: see why that bank would be reticent and had said

76
00:04:00,520 --> 00:04:04,400
Speaker 6: no previously when pretty much the entire list of Bank

77
00:04:04,440 --> 00:04:08,000
Speaker 6: of America's rivals lined up to extend a credit line

78
00:04:08,000 --> 00:04:08,600
Speaker 6: to open Ai.

79
00:04:08,840 --> 00:04:09,800
Speaker 2: They said no again.

80
00:04:09,640 --> 00:04:12,600
Speaker 6: In March twenty twenty six, as recently as March, but

81
00:04:12,760 --> 00:04:16,520
Speaker 6: now in recent weeks, Bank of America finally is on board.

82
00:04:16,760 --> 00:04:19,600
Speaker 6: And the best way to explain this is it showcases

83
00:04:19,760 --> 00:04:24,559
Speaker 6: the internal struggle between maintaining this responsible growth philosophy, which

84
00:04:24,680 --> 00:04:29,520
Speaker 6: unfortunately translates to a more conservative approach, versus bankers wanting

85
00:04:29,560 --> 00:04:32,720
Speaker 6: to chase the hottest deals out there, and there's always

86
00:04:32,720 --> 00:04:33,800
Speaker 6: going to be some risk on them.

87
00:04:33,880 --> 00:04:36,920
Speaker 3: The hottest deal could be an open AI IPO, right,

88
00:04:37,000 --> 00:04:40,240
Speaker 3: We and others reported last week the week before that

89
00:04:40,839 --> 00:04:44,560
Speaker 3: an open AI ipo looking like twenty twenty seven. Right,

90
00:04:44,600 --> 00:04:46,680
Speaker 3: originally maybe this the end of this year, now twenty

91
00:04:46,720 --> 00:04:49,800
Speaker 3: twenty seven. Bank for America wants to get in on that, right, Yeah,

92
00:04:49,839 --> 00:04:50,160
Speaker 3: I mean.

93
00:04:50,080 --> 00:04:52,359
Speaker 6: The reality of modern day Silicon Valley is if you

94
00:04:52,480 --> 00:04:55,840
Speaker 6: want an admission ticket on the big ticket IPOs, it

95
00:04:55,920 --> 00:04:58,880
Speaker 6: is almost expected that you support the company in the

96
00:04:59,000 --> 00:05:02,040
Speaker 6: lead up to that, and extending a credit line. Participating

97
00:05:02,080 --> 00:05:05,240
Speaker 6: in a credit facility is one obvious way to do it.

98
00:05:05,640 --> 00:05:06,600
Speaker 2: Yes, we still.

99
00:05:06,360 --> 00:05:08,719
Speaker 6: Don't have clarity on when Opening I could go public.

100
00:05:08,720 --> 00:05:10,599
Speaker 6: Could it be later this year, could it be next year?

101
00:05:10,800 --> 00:05:13,039
Speaker 6: But really the window for Bank of America to come

102
00:05:13,080 --> 00:05:17,160
Speaker 6: in would have been closing down. This doesn't necessarily guarantee

103
00:05:17,160 --> 00:05:19,279
Speaker 6: a seat at the table, but it certainly goes a

104
00:05:19,320 --> 00:05:21,200
Speaker 6: long way towards helping them get there.

105
00:05:21,320 --> 00:05:23,880
Speaker 3: Bloombashida, Nada Raja, and thank you very much. Staying with

106
00:05:24,000 --> 00:05:26,880
Speaker 3: Open Ai, the company is making news on the product.

107
00:05:26,480 --> 00:05:27,320
Speaker 2: Front as well.

108
00:05:27,440 --> 00:05:30,279
Speaker 3: It's now set to expand access to its most advanced

109
00:05:30,320 --> 00:05:34,880
Speaker 3: AI model GPT five point six tomorrow. Bloomberg's AI Edit

110
00:05:35,000 --> 00:05:38,320
Speaker 3: s Sephigman's with us with the latest. This is an

111
00:05:38,360 --> 00:05:42,320
Speaker 3: interesting case study because it is open AI more widely

112
00:05:42,360 --> 00:05:45,320
Speaker 3: releasing a model that the US government has had time

113
00:05:45,400 --> 00:05:48,280
Speaker 3: to review per the executive order of the President of

114
00:05:48,320 --> 00:05:50,600
Speaker 3: the United States last month. What do we need to know?

115
00:05:51,760 --> 00:05:54,440
Speaker 7: Yeah, I mean it has shades what happened with Anthropic

116
00:05:54,520 --> 00:05:56,800
Speaker 7: that there are some distinctions here. There was not really

117
00:05:56,800 --> 00:06:00,520
Speaker 7: any formal export controller ban on opening as models throughout

118
00:06:00,520 --> 00:06:03,240
Speaker 7: this felt a little bit more voluntary, but the government

119
00:06:03,320 --> 00:06:05,640
Speaker 7: had pressured open Ai to do a staggered release of

120
00:06:05,720 --> 00:06:08,000
Speaker 7: the model and approve the initial list of partners who

121
00:06:08,000 --> 00:06:10,480
Speaker 7: were able to access it. Now they're saying it'll be

122
00:06:10,520 --> 00:06:13,960
Speaker 7: available publicly to everyone starting tomorrow, But we don't really

123
00:06:14,000 --> 00:06:16,480
Speaker 7: have much clarity on what if anything has changed. You know,

124
00:06:16,520 --> 00:06:20,440
Speaker 7: with Anthropic there were additional cybersecurity safeguards that were agreed

125
00:06:20,480 --> 00:06:22,640
Speaker 7: to and implement it as part of the company bringing

126
00:06:22,720 --> 00:06:26,320
Speaker 7: back It's fable amit those models here, we don't really

127
00:06:26,560 --> 00:06:29,320
Speaker 7: know what meaningfully is shifted. What we do know is

128
00:06:29,360 --> 00:06:32,240
Speaker 7: that this does not feel like a workable long term

129
00:06:33,320 --> 00:06:35,960
Speaker 7: strategy at least for AI developers to have to go

130
00:06:36,040 --> 00:06:37,400
Speaker 7: through this kind of ad hoc reaction.

131
00:06:37,920 --> 00:06:41,600
Speaker 3: Let's talk about the model itself or themselves. There was

132
00:06:41,640 --> 00:06:44,320
Speaker 3: a limited preview, but basically open ai is doing this

133
00:06:44,480 --> 00:06:48,080
Speaker 3: in three tiers. What do we know about this generation

134
00:06:48,200 --> 00:06:49,440
Speaker 3: of model from open Ai.

135
00:06:50,320 --> 00:06:52,280
Speaker 7: Yeah, I mean more of what we've been seeing in

136
00:06:52,360 --> 00:06:54,960
Speaker 7: recent months from both openI and Anthropic that this model,

137
00:06:55,160 --> 00:06:58,000
Speaker 7: this family of models should be more capable when it

138
00:06:58,040 --> 00:07:02,960
Speaker 7: comes to cybersecurity tasks and both to defending cybersecurity as

139
00:07:03,000 --> 00:07:06,800
Speaker 7: well as AIA coding efforts. And ultimately, as we were

140
00:07:06,800 --> 00:07:09,960
Speaker 7: talking about earlier with the IBO coming Opening Anthropic or

141
00:07:10,000 --> 00:07:13,360
Speaker 7: buying against each other to cater to this market for

142
00:07:13,520 --> 00:07:16,800
Speaker 7: coding and cybersecurity tools and getting as many business customers

143
00:07:16,800 --> 00:07:18,840
Speaker 7: as they can. So it's really paramount to get these

144
00:07:18,880 --> 00:07:21,200
Speaker 7: new models out as fast as they can in the

145
00:07:21,240 --> 00:07:23,200
Speaker 7: face of increasing opposition from the government.

146
00:07:23,360 --> 00:07:25,360
Speaker 3: So some quick context what you said at the beginning,

147
00:07:25,600 --> 00:07:28,560
Speaker 3: you know, Sam Altman seemed to sort of be more

148
00:07:28,800 --> 00:07:31,320
Speaker 3: on a voluntary basis willing to do this with the

149
00:07:31,400 --> 00:07:34,080
Speaker 3: US government. Could you give some of the background please.

150
00:07:34,880 --> 00:07:37,040
Speaker 7: Yeah, I mean, again, this is not the Commerce Department

151
00:07:37,080 --> 00:07:39,760
Speaker 7: as far as we know, putting out a formal export

152
00:07:39,840 --> 00:07:42,080
Speaker 7: control ban on opening Eye's models in the way that

153
00:07:42,120 --> 00:07:42,480
Speaker 7: happened with.

154
00:07:42,520 --> 00:07:44,600
Speaker 8: Anthropic and throughout in recent months.

155
00:07:44,400 --> 00:07:46,240
Speaker 7: It's probably there's been a bit more of an adversarial

156
00:07:46,320 --> 00:07:49,760
Speaker 7: or tense relationship between Anthropic and the government than there

157
00:07:49,800 --> 00:07:52,600
Speaker 7: has been between Open AI and the government. That said,

158
00:07:52,640 --> 00:07:56,280
Speaker 7: if the government is pressuring, whether formally or informally, the

159
00:07:56,360 --> 00:07:59,280
Speaker 7: company to move a bit more slowly here, you know,

160
00:07:59,480 --> 00:08:02,520
Speaker 7: it's hard for a company to buck that pressure. So

161
00:08:02,600 --> 00:08:04,720
Speaker 7: I think we've seen this happen seemingly in a more

162
00:08:04,800 --> 00:08:06,440
Speaker 7: voluntary way, though under some dress.

163
00:08:06,480 --> 00:08:09,640
Speaker 3: Bloomberg's AI editor Seth Figeman, thank you very much. An

164
00:08:09,680 --> 00:08:12,760
Speaker 3: update on Amazon's twenty five billion dollar bond sale. Demand

165
00:08:13,080 --> 00:08:16,720
Speaker 3: peaked it's sixty two billion dollars before settling at roughly

166
00:08:16,840 --> 00:08:19,720
Speaker 3: forty one billion dollars in final orders, so about one

167
00:08:19,800 --> 00:08:22,680
Speaker 3: point six times the size of the deal. In terms

168
00:08:22,720 --> 00:08:26,080
Speaker 3: of over subscription, that's well above the average, below the average,

169
00:08:26,080 --> 00:08:29,480
Speaker 3: I should say, for US investment grade bond offerings this year,

170
00:08:29,560 --> 00:08:33,199
Speaker 3: which have been typically around four times oversubscribed according to

171
00:08:33,240 --> 00:08:37,160
Speaker 3: Bloomberg data. Even after the offering, investors had to higher

172
00:08:37,640 --> 00:08:40,880
Speaker 3: than usual premium opportunity. Amazon could not generate the kind

173
00:08:40,880 --> 00:08:44,160
Speaker 3: of blockbuster to demand that some had expected, suggesting that

174
00:08:44,240 --> 00:08:47,000
Speaker 3: there is a limit to investor's appetite even for the

175
00:08:47,080 --> 00:08:49,679
Speaker 3: debt of the biggest tech companies. The company, which is

176
00:08:49,760 --> 00:08:52,600
Speaker 3: expected to spend almost two hundred billion dollars this year,

177
00:08:52,920 --> 00:08:55,880
Speaker 3: has tapped different currencies to fund its planned and is

178
00:08:55,960 --> 00:08:58,920
Speaker 3: the biggest debt issuer among the firms that are leading

179
00:08:59,040 --> 00:08:59,800
Speaker 3: the AI boom.

180
00:09:00,280 --> 00:09:01,760
Speaker 2: If you've been wondering whether the.

181
00:09:01,840 --> 00:09:04,240
Speaker 3: Rally in tech has run its course, our next guest

182
00:09:04,320 --> 00:09:08,280
Speaker 3: says no. She sees recent market moves as a rotation,

183
00:09:08,760 --> 00:09:11,960
Speaker 3: with AI demand and tech fundamentals still looking strong. Margi

184
00:09:12,080 --> 00:09:14,599
Speaker 3: Vittel of All String Global Investments stoves us now it

185
00:09:14,679 --> 00:09:18,520
Speaker 3: has been so difficult to get to the root of

186
00:09:18,600 --> 00:09:20,960
Speaker 3: what's been driving the market in just the last forty

187
00:09:21,000 --> 00:09:24,800
Speaker 3: eight hours. A case study for you is Samsung incredible

188
00:09:24,880 --> 00:09:27,040
Speaker 3: numbers on the top and bottom line, and yet the

189
00:09:27,160 --> 00:09:30,600
Speaker 3: sell off in semiconductors that followed was difficult to gauge

190
00:09:31,320 --> 00:09:32,080
Speaker 3: what was happening.

191
00:09:33,400 --> 00:09:33,559
Speaker 5: Well.

192
00:09:33,600 --> 00:09:36,679
Speaker 9: It is remarkable that across the board in the tech sector,

193
00:09:36,760 --> 00:09:41,920
Speaker 9: the hyperscalers, Samsung, and memory stocks all had fantastic numbers

194
00:09:42,200 --> 00:09:45,479
Speaker 9: all had a looks as if it was even accelerating

195
00:09:45,600 --> 00:09:48,400
Speaker 9: from the current outstanding levels, and I think you really

196
00:09:48,440 --> 00:09:51,199
Speaker 9: have to look at short term volatility. We had the

197
00:09:51,280 --> 00:09:53,520
Speaker 9: Korean market which was very vald I think some of

198
00:09:53,559 --> 00:09:56,320
Speaker 9: best splashed over into our market. You had some new

199
00:09:56,640 --> 00:10:00,280
Speaker 9: very large new issues techs like SpaceX, for example, and

200
00:10:00,360 --> 00:10:02,360
Speaker 9: you had midyear where a lot of people are adjusting

201
00:10:02,400 --> 00:10:05,880
Speaker 9: their portfolios, particularly after some of the stocks have had

202
00:10:05,920 --> 00:10:09,040
Speaker 9: such an enormous run, others not so much, some of

203
00:10:09,080 --> 00:10:11,160
Speaker 9: the megastocks. So I think you're seeing a lot of

204
00:10:11,240 --> 00:10:15,200
Speaker 9: jostling short term trading really unrelated to the fundamentals, which

205
00:10:15,240 --> 00:10:17,480
Speaker 9: are as strong as they've ever been, if not stronger,

206
00:10:17,960 --> 00:10:22,239
Speaker 9: plus a backdrop of a very strong economy, capital expenders

207
00:10:22,360 --> 00:10:25,000
Speaker 9: very strong. Those are all sectors of the economy that

208
00:10:25,120 --> 00:10:28,640
Speaker 9: we'll be using tech products also in addition to AI

209
00:10:28,880 --> 00:10:31,679
Speaker 9: and so forth. So really it is a disconnect, but

210
00:10:31,760 --> 00:10:34,080
Speaker 9: I think it's a pretty good opportunity because a lot

211
00:10:34,160 --> 00:10:36,440
Speaker 9: of these stocks have cheapened up over the last four

212
00:10:36,520 --> 00:10:38,800
Speaker 9: days and really some over the last six months.

213
00:10:39,920 --> 00:10:40,839
Speaker 2: Margie, here's what I know.

214
00:10:41,840 --> 00:10:45,319
Speaker 3: The semiconductors are coming off their best quarter ever in

215
00:10:45,440 --> 00:10:49,280
Speaker 3: a long time. In videos, trading at eighteen times forward earnings,

216
00:10:49,520 --> 00:10:51,440
Speaker 3: cheaper than the S and P five hundred right now.

217
00:10:52,280 --> 00:10:56,040
Speaker 3: And from a memory perspective, nothing has changed. Supply is tight,

218
00:10:56,480 --> 00:10:59,560
Speaker 3: Deram pricing is going up. Nand pricing is going up.

219
00:11:00,080 --> 00:11:01,719
Speaker 3: Take all of that in aggregate and tell me what

220
00:11:01,840 --> 00:11:02,560
Speaker 3: it tells us.

221
00:11:04,480 --> 00:11:06,439
Speaker 9: Well, it tells you one thing about in Nvidia that

222
00:11:06,920 --> 00:11:09,719
Speaker 9: it is so widely owned that most people have an

223
00:11:09,760 --> 00:11:13,000
Speaker 9: overweight in Vidia, and that as a result, although they

224
00:11:13,000 --> 00:11:15,800
Speaker 9: again they've reported very very good numbers, a very good outlook,

225
00:11:16,200 --> 00:11:19,120
Speaker 9: the stock is really stagnated. If you look at the pe,

226
00:11:19,360 --> 00:11:22,040
Speaker 9: the dividendal cash flow yield, it really looks like a

227
00:11:22,400 --> 00:11:25,480
Speaker 9: consumer staple as far as its financial profile, even though

228
00:11:25,559 --> 00:11:28,000
Speaker 9: the growth and the profit margins are very high. So

229
00:11:28,080 --> 00:11:31,079
Speaker 9: I think it says the market is differentiating among these

230
00:11:31,120 --> 00:11:34,280
Speaker 9: companies number one and number two. There's just a reality

231
00:11:34,400 --> 00:11:38,320
Speaker 9: that great as Nvidia is, if most investors, which seems

232
00:11:38,360 --> 00:11:41,640
Speaker 9: to me are have an overweight position in Nvidia, there

233
00:11:41,720 --> 00:11:44,079
Speaker 9: is really no marginal buyer left to come in and

234
00:11:44,280 --> 00:11:46,520
Speaker 9: raise the stock, even in the face of good numbers.

235
00:11:46,800 --> 00:11:48,760
Speaker 9: And I think you're seeing that in contrast to say,

236
00:11:48,840 --> 00:11:51,520
Speaker 9: memory stocks, which I would say have been under owned

237
00:11:51,960 --> 00:11:56,080
Speaker 9: and had spectacular performances here, but again they're under owned,

238
00:11:56,120 --> 00:11:59,440
Speaker 9: so people really have an appetite to increase their holdings

239
00:11:59,480 --> 00:12:00,440
Speaker 9: in some of those stocks.

240
00:12:01,000 --> 00:12:03,960
Speaker 3: Margie, we have some breaking news in the markets. Oil

241
00:12:04,080 --> 00:12:06,880
Speaker 3: has really extended gains. We're now seeing Brent crewed rising

242
00:12:06,920 --> 00:12:10,080
Speaker 3: above eighty dollars a barrel, now just a touch below

243
00:12:10,120 --> 00:12:13,360
Speaker 3: seventy nine dollars eighty three cents. Of course, we're expecting

244
00:12:13,440 --> 00:12:15,679
Speaker 3: to hear from the President of the United States in

245
00:12:15,800 --> 00:12:20,079
Speaker 3: his NATO press conference any moment. Now, how is the

246
00:12:20,280 --> 00:12:24,600
Speaker 3: ongoing situation with Iran impacting the technology sector.

247
00:12:26,240 --> 00:12:30,120
Speaker 9: I think it has very, very very minor impact, particularly

248
00:12:30,160 --> 00:12:33,960
Speaker 9: in the US because we as much lower foreign oil

249
00:12:34,040 --> 00:12:38,959
Speaker 9: dependents were actually are exporters of energy products, and really

250
00:12:39,040 --> 00:12:41,120
Speaker 9: if you look at the energy sector itself, you could

251
00:12:41,160 --> 00:12:44,080
Speaker 9: say that this is going to really command those companies,

252
00:12:44,160 --> 00:12:49,719
Speaker 9: whether exploration, transport, to actually use more technology products in

253
00:12:49,920 --> 00:12:53,480
Speaker 9: order to cut out their vulnerability versus Iran. So I

254
00:12:53,520 --> 00:12:57,280
Speaker 9: would say net that indigestion in the middle is caused

255
00:12:57,280 --> 00:13:00,360
Speaker 9: by Iran is actually a positive for investments in the

256
00:13:00,440 --> 00:13:04,080
Speaker 9: tech sector, for the energy sector, So again it's certainly

257
00:13:04,160 --> 00:13:06,760
Speaker 9: not a negative, and I think really a positive because

258
00:13:07,120 --> 00:13:10,679
Speaker 9: the increase the demand for high chech products for exploration

259
00:13:10,840 --> 00:13:12,280
Speaker 9: and transport and so forth.

260
00:13:12,720 --> 00:13:16,480
Speaker 3: Weg here's one for you, the link between capital expenditures

261
00:13:16,920 --> 00:13:19,760
Speaker 3: and real world inflation of the world of the data center.

262
00:13:20,040 --> 00:13:23,439
Speaker 3: We talked about Amazon and its activity in the bomb market,

263
00:13:23,960 --> 00:13:26,319
Speaker 3: and the idea that Amazon is a case study. CAPEX

264
00:13:26,320 --> 00:13:29,720
Speaker 3: two hundred billion dollars this year, maybe three hundred billion

265
00:13:29,760 --> 00:13:32,400
Speaker 3: dollars next year. But what I see on the bloomberg

266
00:13:32,720 --> 00:13:36,280
Speaker 3: is construction and labor inflation, memory inflation.

267
00:13:36,559 --> 00:13:38,400
Speaker 2: How does that factor entire CAPEX?

268
00:13:40,240 --> 00:13:45,520
Speaker 9: Well, really, it says to me that inflation supports higher

269
00:13:45,600 --> 00:13:48,680
Speaker 9: profits across the board. And you've seen that actually looking

270
00:13:48,760 --> 00:13:53,480
Speaker 9: at results for the first quarter that companies are reporting

271
00:13:53,600 --> 00:13:56,600
Speaker 9: record high profit margins in chech of course, but also

272
00:13:56,679 --> 00:14:00,400
Speaker 9: in other sectors too. Why because they have the ability

273
00:14:00,520 --> 00:14:04,520
Speaker 9: to pass on input increases in prices and maintain those

274
00:14:04,640 --> 00:14:07,360
Speaker 9: very high profit margins. If you remember, I would say

275
00:14:07,360 --> 00:14:10,560
Speaker 9: about six months ago, people were worried that companies would

276
00:14:10,600 --> 00:14:12,880
Speaker 9: not be able to pass on price increases and their

277
00:14:12,960 --> 00:14:15,959
Speaker 9: margin to be squeezed. And it shows that actually this

278
00:14:16,120 --> 00:14:18,760
Speaker 9: little bit of inflation is helping companies, helping their profit

279
00:14:18,920 --> 00:14:22,120
Speaker 9: margins and I think supporting stock prices. And really, when

280
00:14:22,160 --> 00:14:23,880
Speaker 9: you look and you say, well, profits in the first

281
00:14:23,960 --> 00:14:27,080
Speaker 9: quarter were up twenty eight percent, revenues up eleven percent,

282
00:14:27,440 --> 00:14:31,240
Speaker 9: profit margins I think nineteen percent. You know, even if

283
00:14:31,280 --> 00:14:33,680
Speaker 9: the FED raised rates by a quarter of a point two,

284
00:14:34,080 --> 00:14:36,240
Speaker 9: you know, three and three quarter to four and a half,

285
00:14:37,200 --> 00:14:39,800
Speaker 9: that really doesn't It's such a small amount compared to

286
00:14:39,880 --> 00:14:42,680
Speaker 9: those other very very big numbers. So we think it's

287
00:14:42,840 --> 00:14:45,480
Speaker 9: I won't say your material, but pretty close to it. Certainly,

288
00:14:45,560 --> 00:14:48,600
Speaker 9: the fit is indicated it wants to be more passive,

289
00:14:49,000 --> 00:14:51,520
Speaker 9: not as aggressive as they've been in the previous years.

290
00:14:51,600 --> 00:14:55,000
Speaker 9: So that's really a positive to say, lower volatility in

291
00:14:55,000 --> 00:14:58,480
Speaker 9: the financial markets. All sectors benefit. Tech of course does too,

292
00:14:58,520 --> 00:14:59,920
Speaker 9: but really all the companies benefit.

293
00:15:00,480 --> 00:15:03,520
Speaker 3: Margie Pteroville, Spring Global Investments, thank you so much, dynamic

294
00:15:03,560 --> 00:15:05,960
Speaker 3: conversation on tech and the markets. I'm taking a look

295
00:15:05,960 --> 00:15:08,880
Speaker 3: at Ali Barba. These are the US listed shares, and

296
00:15:09,200 --> 00:15:11,640
Speaker 3: you know, it's a serious gain eleven percent. What happened

297
00:15:12,120 --> 00:15:15,040
Speaker 3: was a pre earnings briefing for the analysts, and it

298
00:15:15,120 --> 00:15:18,080
Speaker 3: looks like the market, in part based on that, has

299
00:15:18,160 --> 00:15:23,160
Speaker 3: become very optimistic about the potential earnings potential value Barber,

300
00:15:23,240 --> 00:15:26,440
Speaker 3: but also more broadly, this kind of shift of capital

301
00:15:26,760 --> 00:15:29,400
Speaker 3: that we have been seeing in the short term into

302
00:15:29,520 --> 00:15:32,520
Speaker 3: major Chinese technology and internet companies.

303
00:15:32,560 --> 00:15:34,320
Speaker 2: It's an interesting move. We'll keep tracking it.

304
00:15:34,600 --> 00:15:39,240
Speaker 3: Coming up, Apple's chip partnership with Broadcom actually gets even bigger.

305
00:15:39,280 --> 00:15:43,320
Speaker 3: The expanded deal is expected to exceed thirty billion dollars.

306
00:15:43,360 --> 00:15:56,000
Speaker 3: We have the details. Next, this is Bloomberg Tech. Apple's

307
00:15:56,000 --> 00:15:58,240
Speaker 3: writing a bigger check for its US supply chain. The

308
00:15:58,280 --> 00:16:02,400
Speaker 3: company says it's expands broad Com partnership well top thirty

309
00:16:02,480 --> 00:16:06,080
Speaker 3: billion US dollars. Bloombos Consumer Tech and Apple Managing editor

310
00:16:06,080 --> 00:16:07,560
Speaker 3: Mark Gummons with us here in New York City.

311
00:16:08,640 --> 00:16:09,120
Speaker 2: Very quickly.

312
00:16:09,120 --> 00:16:11,400
Speaker 3: We're getting more details of this, right, it's not just

313
00:16:11,480 --> 00:16:14,720
Speaker 3: the headline figure, it's the sort of structure of it. Again,

314
00:16:15,000 --> 00:16:17,240
Speaker 3: very us focus well of the new details.

315
00:16:17,440 --> 00:16:19,440
Speaker 10: Yeah, I know Apple loves to put out these announcements

316
00:16:19,480 --> 00:16:24,000
Speaker 10: every so often. The President Donald Trump obviously loves these

317
00:16:24,040 --> 00:16:28,120
Speaker 10: types of announcements. Tim Cook, Apple CEO, had a six

318
00:16:28,280 --> 00:16:32,120
Speaker 10: hundred billion dollar commitment announcement in the Oval Office over

319
00:16:32,200 --> 00:16:34,880
Speaker 10: a year ago. This is part of that thirty billion

320
00:16:35,560 --> 00:16:38,400
Speaker 10: in expenses for these chips. These are developed by Broadcom

321
00:16:38,520 --> 00:16:41,120
Speaker 10: produced in the United States as part of that capital

322
00:16:41,160 --> 00:16:43,880
Speaker 10: expendature investment of one and a half billion from Apple

323
00:16:44,000 --> 00:16:48,400
Speaker 10: into broadcoms Fort Collins, Colorado facility. The Trump administration, obviously,

324
00:16:48,520 --> 00:16:51,120
Speaker 10: like the other announcements, is probably eating this one up.

325
00:16:51,280 --> 00:16:53,000
Speaker 3: Let me go back to what we talked about in

326
00:16:53,080 --> 00:16:55,120
Speaker 3: the last couple of days, because it was really picked up.

327
00:16:55,560 --> 00:16:58,640
Speaker 3: What are these broad coom design chips going to actually

328
00:16:58,640 --> 00:16:59,560
Speaker 3: be used for with Apple?

329
00:16:59,720 --> 00:17:02,240
Speaker 10: You know, Apples working with Broadcom on a range of things.

330
00:17:02,440 --> 00:17:04,840
Speaker 10: In terms of this announcement, they're saying this has to

331
00:17:04,880 --> 00:17:08,000
Speaker 10: do with wireless components. The key here is that Broadcom

332
00:17:08,480 --> 00:17:12,520
Speaker 10: has long been Apple's supplier of Bluetooth plus Wi Fi chips.

333
00:17:12,560 --> 00:17:14,000
Speaker 10: In order to get your phone to connect to the

334
00:17:14,040 --> 00:17:16,879
Speaker 10: internet on your home network or your office or what

335
00:17:17,040 --> 00:17:19,439
Speaker 10: have you, and connect to other devices over Bluetooth, they

336
00:17:19,480 --> 00:17:21,840
Speaker 10: relied on Broadcom, but they designed them out and they

337
00:17:21,920 --> 00:17:24,639
Speaker 10: build their own end one chips. Those are expanding rapidly

338
00:17:24,760 --> 00:17:27,920
Speaker 10: into iPhones, iPads, and macs. You'll see a range of

339
00:17:27,960 --> 00:17:30,560
Speaker 10: new smart home devices with this Apple wireless chip in

340
00:17:30,640 --> 00:17:33,359
Speaker 10: the coming months, new HomePod, many new Apple TV. But

341
00:17:33,480 --> 00:17:35,960
Speaker 10: they are making what are called RF filters it's a

342
00:17:36,040 --> 00:17:39,840
Speaker 10: component that works with the new again in house cellular modem,

343
00:17:40,080 --> 00:17:41,840
Speaker 10: the C one, the C two, the C three that

344
00:17:41,880 --> 00:17:45,800
Speaker 10: Apple's rolling out over the next few years. Broadcom's also

345
00:17:45,880 --> 00:17:49,600
Speaker 10: working with Apple on a six chips, a component related

346
00:17:49,920 --> 00:17:53,240
Speaker 10: to upcoming Apple Intelligence servers that are going to be

347
00:17:53,320 --> 00:17:56,760
Speaker 10: deployed over the next year and so forth. So a

348
00:17:56,840 --> 00:17:59,040
Speaker 10: lot of working on with Broadcom, even though they've been

349
00:17:59,080 --> 00:18:01,480
Speaker 10: designed out of their primary component.

350
00:18:01,400 --> 00:18:04,280
Speaker 3: Of Bloomberg's Mark Gumman, who leads our coverage of consumer technology,

351
00:18:04,359 --> 00:18:07,200
Speaker 3: thank you very much. Billions of dollars in planned investments

352
00:18:07,280 --> 00:18:10,399
Speaker 3: to develop the US chip sector could be at risk

353
00:18:10,760 --> 00:18:14,000
Speaker 3: from a lack of workers. According to analysis from McKinsey,

354
00:18:14,160 --> 00:18:17,320
Speaker 3: the Chip Industry Group, SEMI, and the National Science Foundation,

355
00:18:17,800 --> 00:18:20,879
Speaker 3: the US could face a deficit of one hundred and

356
00:18:21,040 --> 00:18:25,240
Speaker 3: fifty seven thousand full time skilled workers by twenty thirty.

357
00:18:25,240 --> 00:18:28,440
Speaker 3: Bloombergs Maggie Eastland joins us with the details and that chart.

358
00:18:29,000 --> 00:18:32,800
Speaker 3: I mean this is a drastic deficit of skilled labor,

359
00:18:33,119 --> 00:18:36,520
Speaker 3: the exact type of labor that is necessary to get

360
00:18:36,600 --> 00:18:37,159
Speaker 3: this built.

361
00:18:37,600 --> 00:18:40,760
Speaker 2: Take us inside it, absolutely so.

362
00:18:41,040 --> 00:18:45,919
Speaker 11: The US has been trying to create a manufacturing renaissance,

363
00:18:46,080 --> 00:18:49,240
Speaker 11: especially in the semiconductor industry for many years now. The

364
00:18:49,359 --> 00:18:51,920
Speaker 11: Chips Act of twenty twenty two tried to address some

365
00:18:52,040 --> 00:18:55,640
Speaker 11: of those financial shortfalls, those extra costs that companies face

366
00:18:55,680 --> 00:18:57,600
Speaker 11: to build in the US. But a key part of

367
00:18:57,640 --> 00:18:59,920
Speaker 11: this that still isn't figured out is all the work

368
00:19:00,280 --> 00:19:02,760
Speaker 11: that will be needed. One thing I found really interesting

369
00:19:02,880 --> 00:19:06,120
Speaker 11: in this reporting is that, you know, most US engineering

370
00:19:06,200 --> 00:19:09,440
Speaker 11: students don't want to work in the chip industry. In fact,

371
00:19:09,560 --> 00:19:12,720
Speaker 11: only three percent of those students ultimately go on to

372
00:19:12,800 --> 00:19:14,760
Speaker 11: work in the chip industry, when there are so many

373
00:19:14,840 --> 00:19:19,000
Speaker 11: other exciting careers in tech in AI that are you know,

374
00:19:19,200 --> 00:19:22,560
Speaker 11: viewed as a bit more vanguard and exciting maybe than hardware.

375
00:19:22,760 --> 00:19:24,840
Speaker 3: Right, Maggie, let's just go back to basics, like what

376
00:19:25,000 --> 00:19:27,159
Speaker 3: kind of jobs are these? What kind of skills are

377
00:19:27,160 --> 00:19:28,200
Speaker 3: we literally talking about?

378
00:19:29,160 --> 00:19:34,879
Speaker 11: Absolutely so the biggest gap is in manufacturing engineers. You know,

379
00:19:34,960 --> 00:19:36,760
Speaker 11: there are a lot of engineering students in the US,

380
00:19:36,840 --> 00:19:40,280
Speaker 11: but most are not going into manufacturing. There are also technicians,

381
00:19:41,080 --> 00:19:44,119
Speaker 11: so these are high skilled jobs, but they don't require as.

382
00:19:43,960 --> 00:19:45,159
Speaker 2: Advanced of education.

383
00:19:46,119 --> 00:19:48,040
Speaker 11: The Chips Act did address some of this, and there

384
00:19:48,080 --> 00:19:51,399
Speaker 11: are a lot of employer funded programs to increase the

385
00:19:51,480 --> 00:19:54,280
Speaker 11: number of technicians. The harder problem is actually going to

386
00:19:54,359 --> 00:19:55,480
Speaker 11: be the engineers.

387
00:19:55,960 --> 00:19:58,359
Speaker 3: If you are working in that field, Maggie and I

388
00:19:58,520 --> 00:20:00,320
Speaker 3: easy to find reach out. I'd love to hear what

389
00:20:00,400 --> 00:20:03,400
Speaker 3: your experiences in industry, Maggie very quickly. You just touched

390
00:20:03,440 --> 00:20:05,760
Speaker 3: on it there. What are the companies doing right? Are

391
00:20:05,800 --> 00:20:09,040
Speaker 3: they having to hire competitively with big pay packages. What's

392
00:20:09,080 --> 00:20:11,320
Speaker 3: the solution to bridge the gap in the short term.

393
00:20:12,400 --> 00:20:16,200
Speaker 11: Well, companies told McKinsey and Semi and NSF in this

394
00:20:16,280 --> 00:20:20,960
Speaker 11: survey that they are already facing difficulties hiring engineers. You know,

395
00:20:21,160 --> 00:20:23,480
Speaker 11: immigration may fill some of this gap. There's been some

396
00:20:23,600 --> 00:20:27,840
Speaker 11: challenges there, there are you know, this survey is recommending

397
00:20:28,000 --> 00:20:31,600
Speaker 11: that the companies work with universities a bit more to

398
00:20:31,720 --> 00:20:36,119
Speaker 11: increase interest in semiconductor careers. But really what this analysis

399
00:20:36,200 --> 00:20:38,440
Speaker 11: is showing is that the companies need to be doing

400
00:20:38,520 --> 00:20:40,680
Speaker 11: more work in order to fill these engineering gaps.

401
00:20:40,960 --> 00:20:42,639
Speaker 2: Makes Maggie Eastland thank you very much.

402
00:20:42,720 --> 00:20:45,720
Speaker 3: Indeed, coming up on the program, Chinese aifirm GPU was

403
00:20:45,760 --> 00:20:49,240
Speaker 3: sell four billion dollars of shares after rallying nearly fifteen

404
00:20:49,440 --> 00:20:53,760
Speaker 3: hundred percent from earlier this year. Look, we're still awaiting

405
00:20:53,840 --> 00:20:57,280
Speaker 3: a press conference from President Trump in Uncora, Turkey, set

406
00:20:57,359 --> 00:21:00,320
Speaker 3: to start any moment. A lot of headlines this morning

407
00:21:00,400 --> 00:21:03,680
Speaker 3: relating to the two Iran The President had said that

408
00:21:03,720 --> 00:21:06,399
Speaker 3: the US will probably strike Iran again tonight. Then he

409
00:21:06,520 --> 00:21:10,359
Speaker 3: said that he will continue to deliberate. We'll get the

410
00:21:10,440 --> 00:21:13,080
Speaker 3: latest headlines on that very shortly. This is Bloomberg Tech.

411
00:21:21,640 --> 00:21:23,960
Speaker 12: It's time now for talking tech. I'm you hira Ana.

412
00:21:24,119 --> 00:21:24,560
Speaker 2: First up.

413
00:21:24,880 --> 00:21:28,000
Speaker 12: Chinese AI firm Jipu is slated to sell four billion

414
00:21:28,040 --> 00:21:32,000
Speaker 12: dollars of shares after its stock stored nearly fifteen hundred

415
00:21:32,119 --> 00:21:35,160
Speaker 12: percent following a January listing in Hong Kong. The firm,

416
00:21:35,280 --> 00:21:38,440
Speaker 12: which trades as Knowledge at List Technology, will use the

417
00:21:38,520 --> 00:21:42,960
Speaker 12: money for research and development, expansion, and other investments. Plus

418
00:21:43,040 --> 00:21:46,280
Speaker 12: Samsung says it has started mass production of its most

419
00:21:46,440 --> 00:21:50,639
Speaker 12: advanced AI data center storage drive, designed for Nvidia's upcoming

420
00:21:50,840 --> 00:21:54,159
Speaker 12: VERA Rubin platform. The company says the new drive is

421
00:21:54,240 --> 00:21:56,800
Speaker 12: more than twice as fast as its predecessor and is

422
00:21:56,880 --> 00:22:00,360
Speaker 12: built to handle the massive data demands of AI. Works

423
00:22:00,960 --> 00:22:04,520
Speaker 12: and Meta debuted a new AI image generation system called

424
00:22:04,720 --> 00:22:08,560
Speaker 12: news Image, its first major release since chief AI officer

425
00:22:08,640 --> 00:22:12,680
Speaker 12: Alexander Wing took over the company's AI efforts. News will

426
00:22:12,720 --> 00:22:16,880
Speaker 12: be integrated into metas apps including Instagram and WhatsApp, allowing

427
00:22:17,480 --> 00:22:21,640
Speaker 12: users to generate new images or edit existing photos.

428
00:22:21,880 --> 00:22:23,080
Speaker 2: Ed okay, thank you, Hira.

429
00:22:23,200 --> 00:22:25,879
Speaker 3: Coming up, It's day two of the annual Allen and Co.

430
00:22:26,160 --> 00:22:30,440
Speaker 3: Sun Valley Conference, also known as the Billionaires Summer Camp.

431
00:22:30,520 --> 00:22:31,080
Speaker 2: Stay with us.

432
00:22:31,080 --> 00:22:32,600
Speaker 3: We're going to go to the ground. We're going to

433
00:22:32,640 --> 00:22:35,720
Speaker 3: get the latest details and some of the news. We're

434
00:22:35,760 --> 00:22:38,920
Speaker 3: still awaiting a press conference from the President of the

435
00:22:39,040 --> 00:22:42,080
Speaker 3: United States in Ankara, set to start any moment.

436
00:22:42,240 --> 00:22:42,400
Speaker 8: Now.

437
00:22:42,480 --> 00:22:45,080
Speaker 3: We will bring you the latest headlines with a focus

438
00:22:45,200 --> 00:22:47,280
Speaker 3: on this situation with Iran.

439
00:22:47,600 --> 00:23:02,359
Speaker 2: That's next. Stay with us. This is Bloomberg Tech. Welcome

440
00:23:02,359 --> 00:23:04,800
Speaker 2: back to Bloomberg Tech and video is cheap.

441
00:23:05,000 --> 00:23:07,720
Speaker 3: The stock's down about fifteen percent from a mayp send

442
00:23:07,760 --> 00:23:11,040
Speaker 3: me the company's valuation to pre AI boom levels. His

443
00:23:11,119 --> 00:23:15,520
Speaker 3: investors turned towards semiconductor manufacturers. Bloomberg's calm and Ryanicky, she's

444
00:23:15,560 --> 00:23:16,320
Speaker 3: got the charts high.

445
00:23:17,440 --> 00:23:18,480
Speaker 2: All right, thanks, Ed.

446
00:23:18,600 --> 00:23:21,080
Speaker 12: Yeah, So let's take a look at Nvidia's valuation here.

447
00:23:21,359 --> 00:23:24,880
Speaker 4: So that fifteen percent decline in stock price has raised.

448
00:23:24,720 --> 00:23:28,560
Speaker 5: Upon using AI to execute tasks and sharing use cases

449
00:23:28,600 --> 00:23:30,880
Speaker 5: with each other. In terms of who we've seen so far.

450
00:23:31,160 --> 00:23:33,200
Speaker 5: I mean, it's been a lot of the usual suspects.

451
00:23:33,480 --> 00:23:38,080
Speaker 5: David Zaslov of course, Bob Iiger, Disney's new CEO, Josh doyamorrow.

452
00:23:38,359 --> 00:23:42,119
Speaker 5: We saw Sam Altman yesterday as well this morning, Lachlan Murdoch.

453
00:23:42,480 --> 00:23:44,480
Speaker 5: One thing that's been a little bit different this year, though,

454
00:23:44,640 --> 00:23:46,679
Speaker 5: is that it seems like folks are a little bit

455
00:23:46,840 --> 00:23:49,640
Speaker 5: more reluctant to come and talk to us in yours past.

456
00:23:49,960 --> 00:23:52,600
Speaker 5: You know, David Zaslov has always been very eager to

457
00:23:52,640 --> 00:23:55,080
Speaker 5: get his time with the press, but he, you know,

458
00:23:55,280 --> 00:23:58,520
Speaker 5: not even he gave us time. We saw Comcats executives

459
00:23:58,720 --> 00:24:02,040
Speaker 5: Brian Roberts and Mike Cavin yesterday. They of course, you know,

460
00:24:02,119 --> 00:24:04,320
Speaker 5: smiled at the cameras, but wouldn't come and engage. And

461
00:24:04,680 --> 00:24:06,680
Speaker 5: one theory for why that's happening is a lot of

462
00:24:06,720 --> 00:24:11,400
Speaker 5: these companies have deals in the pipeline or on their

463
00:24:11,480 --> 00:24:13,720
Speaker 5: wish lists that are really going to hinge on getting

464
00:24:13,760 --> 00:24:16,000
Speaker 5: regulatory approval, and it could be that in this very

465
00:24:16,040 --> 00:24:19,520
Speaker 5: politicized environment, they're reluctant to speak out and say anything

466
00:24:19,560 --> 00:24:21,560
Speaker 5: that's going to muck up their ability to get these

467
00:24:21,640 --> 00:24:23,440
Speaker 5: deals across the finish line.

468
00:24:23,320 --> 00:24:26,879
Speaker 3: And remind us the deal is Paramount and Warner Brothers,

469
00:24:27,000 --> 00:24:27,840
Speaker 3: what's the latest there?

470
00:24:31,040 --> 00:24:33,480
Speaker 5: So, the latest on the Warner Brothers Paramount deal is

471
00:24:33,600 --> 00:24:37,480
Speaker 5: that they are still awaiting regulatory approval in the EU

472
00:24:37,680 --> 00:24:38,160
Speaker 5: and the UK.

473
00:24:39,320 --> 00:24:40,280
Speaker 2: The EU has set a.

474
00:24:40,280 --> 00:24:43,200
Speaker 5: Deadline of July twenty second to decide whether or not

475
00:24:43,320 --> 00:24:48,040
Speaker 5: they want to clear the deal or investigate further. We

476
00:24:48,320 --> 00:24:52,080
Speaker 5: of course asked Warner Brothers CEO David Saslov about their

477
00:24:52,400 --> 00:24:55,200
Speaker 5: regulatory strategy and whether they plan to offer any concessions

478
00:24:55,320 --> 00:24:59,440
Speaker 5: to the EU, but he declined to comment on that.

479
00:24:59,720 --> 00:25:02,480
Speaker 5: And Comcast, of course, has also announced that it's planning

480
00:25:02,560 --> 00:25:06,800
Speaker 5: to separate its NBC Universal Media division from the cable unit,

481
00:25:06,920 --> 00:25:10,840
Speaker 5: and if those do result in sales on either side,

482
00:25:11,040 --> 00:25:15,600
Speaker 5: that will also require regulatory approval here and in other places.

483
00:25:15,880 --> 00:25:18,159
Speaker 3: Bloom makes Michelle Davis in Son Valley, Idaho, thank you

484
00:25:18,320 --> 00:25:21,040
Speaker 3: very much. INDEENA coming up. Eric Hippo from Lera Hippo

485
00:25:21,480 --> 00:25:24,880
Speaker 3: discusses how his firm's choosing early stage startups to back

486
00:25:24,920 --> 00:25:27,840
Speaker 3: in a landscape where late stage companies are kind of

487
00:25:27,840 --> 00:25:30,320
Speaker 3: all the rage and valuations right now, kind of through

488
00:25:30,359 --> 00:25:33,080
Speaker 3: the roof. We are still awaiting a press conference from

489
00:25:33,160 --> 00:25:35,800
Speaker 3: President Trump in un Chris set to start any moment

490
00:25:35,880 --> 00:25:37,919
Speaker 3: now and we will bring you that as it happens.

491
00:25:37,960 --> 00:25:39,720
Speaker 2: Stay with us. This is Bloomberg Tech.

492
00:25:50,240 --> 00:25:53,520
Speaker 3: Blue Origin is seizing on investor enthusiasm for space. It's

493
00:25:53,600 --> 00:25:56,200
Speaker 3: raising ten billion dollars at a one hundred and thirty

494
00:25:56,240 --> 00:26:00,840
Speaker 3: billion dollar valuation in its first external funding round auded sources.

495
00:26:01,040 --> 00:26:03,760
Speaker 3: The rocket company, which has until now been financed by

496
00:26:03,800 --> 00:26:07,000
Speaker 3: its founder Jeff Bezos, blues stepping up efforts to take

497
00:26:07,040 --> 00:26:10,480
Speaker 3: on rivals space X, though it's new Glen rocket, which

498
00:26:10,520 --> 00:26:14,720
Speaker 3: can take tayloads to space, is temporarily grounded following a

499
00:26:14,840 --> 00:26:19,399
Speaker 3: launch pad explosion in May. Money's flowing in Swentch capital,

500
00:26:19,520 --> 00:26:23,440
Speaker 3: but it's increasingly concentrated. Pitchbook says that AI accounted for

501
00:26:23,520 --> 00:26:26,600
Speaker 3: nearly seventy seven percent of global VC deal value in

502
00:26:26,640 --> 00:26:28,879
Speaker 3: the first half of this year, with more than forty

503
00:26:28,960 --> 00:26:32,520
Speaker 3: two percent going to just three companies, Open Ai, Anthropic

504
00:26:32,720 --> 00:26:36,240
Speaker 3: and Xai. But beneath those mega rounds, investors are becoming

505
00:26:36,280 --> 00:26:40,000
Speaker 3: more selective, especially early on joining us to discuss that

506
00:26:40,160 --> 00:26:43,199
Speaker 3: environment is Eric Hippo, managing partner at Lera. Hippo, one

507
00:26:43,240 --> 00:26:46,520
Speaker 3: of New York's most active early stage venture capital firms.

508
00:26:46,640 --> 00:26:48,680
Speaker 3: I think there's a lot of consensus right now about

509
00:26:48,680 --> 00:26:49,200
Speaker 3: what's happening.

510
00:26:49,520 --> 00:26:49,680
Speaker 10: Eric.

511
00:26:50,040 --> 00:26:53,320
Speaker 3: Seed stage is very hot. Valuations in some cases are

512
00:26:53,440 --> 00:26:56,920
Speaker 3: extremely high. We can debate how high it's too high,

513
00:26:57,280 --> 00:26:59,840
Speaker 3: but it's the Series A spot I want to start.

514
00:27:00,320 --> 00:27:02,160
Speaker 2: You have this belief that it's it's tough.

515
00:27:02,800 --> 00:27:07,119
Speaker 13: Why is it tough, Well, it's it's first of all,

516
00:27:07,160 --> 00:27:10,600
Speaker 13: good morning, ed, it's going to be here. It's it's

517
00:27:10,680 --> 00:27:13,040
Speaker 13: kind of related to what you said, which is that

518
00:27:13,119 --> 00:27:17,479
Speaker 13: the valuations are very high. So you know, valuations are

519
00:27:18,480 --> 00:27:21,359
Speaker 13: fifty one hundred percent more than they were in the

520
00:27:21,400 --> 00:27:24,280
Speaker 13: past couple of years. So that's the starting point for us,

521
00:27:24,359 --> 00:27:28,560
Speaker 13: right we set first investors, and so companies can raise

522
00:27:28,640 --> 00:27:31,640
Speaker 13: these rounds at these crazy evaluations.

523
00:27:31,960 --> 00:27:37,919
Speaker 8: But then if they don't get into a high growth, fast.

524
00:27:37,680 --> 00:27:40,000
Speaker 13: Go to market mode, they can't raise the Series A

525
00:27:40,119 --> 00:27:42,560
Speaker 13: because the evaluation that they started with was too high.

526
00:27:43,080 --> 00:27:45,760
Speaker 8: So you can't you know, you can't have it both ways.

527
00:27:46,520 --> 00:27:50,080
Speaker 13: Either you start in the modest saluation you grow into

528
00:27:50,160 --> 00:27:54,000
Speaker 13: it for your Series A, or if you don't, then

529
00:27:54,680 --> 00:27:56,240
Speaker 13: the Series A investors will pass on you.

530
00:27:57,119 --> 00:27:59,800
Speaker 3: Valuations are extremely high, as you wrote to me, and

531
00:28:00,040 --> 00:28:05,520
Speaker 3: some cases completely nonsensical. So where does the power lay, Like, like,

532
00:28:05,680 --> 00:28:09,800
Speaker 3: who's allowing evaluation at the seed stage to a series

533
00:28:09,880 --> 00:28:12,480
Speaker 3: A to get to that level that the founders that

534
00:28:12,600 --> 00:28:16,280
Speaker 3: are raising or or those that drove and led the

535
00:28:16,359 --> 00:28:17,080
Speaker 3: seed itself.

536
00:28:18,440 --> 00:28:20,800
Speaker 8: I think it's it's those, it's the investors themselves.

537
00:28:20,880 --> 00:28:24,320
Speaker 13: The investors are trying to find those big winners, and

538
00:28:24,520 --> 00:28:27,680
Speaker 13: there's a there's a there's also particularly on the West Coast,

539
00:28:27,800 --> 00:28:30,120
Speaker 13: there's some sort of a consensus building. You know, let's

540
00:28:30,359 --> 00:28:33,399
Speaker 13: let's pick company A to be the winner in category B,

541
00:28:34,920 --> 00:28:38,560
Speaker 13: and so everybody piles on, and you know, it's it's

542
00:28:38,760 --> 00:28:41,080
Speaker 13: whether the company has the right mode or doesn't have

543
00:28:41,160 --> 00:28:43,640
Speaker 13: the right mode. And I would argue, you know, if

544
00:28:43,680 --> 00:28:46,960
Speaker 13: you're just a wrapper around the foundational models, you don't

545
00:28:47,000 --> 00:28:50,520
Speaker 13: have much of a mode. But nevertheless, you've been anointed

546
00:28:50,640 --> 00:28:53,400
Speaker 13: as the winner, and so you you raise more and

547
00:28:53,440 --> 00:28:57,320
Speaker 13: more money at higher and higher valuations, and then everybody

548
00:28:57,400 --> 00:29:00,520
Speaker 13: else is who is not anointed the winner is kind

549
00:29:00,520 --> 00:29:01,520
Speaker 13: of left on the wayside.

550
00:29:02,840 --> 00:29:05,160
Speaker 3: Eric, you've been at this a long time with respect.

551
00:29:05,400 --> 00:29:07,800
Speaker 3: I wondered if you could reflect on how the sort

552
00:29:07,840 --> 00:29:12,200
Speaker 3: of price discovery and vetting of a founder present day

553
00:29:12,320 --> 00:29:15,160
Speaker 3: is different to when you started out in your career,

554
00:29:15,320 --> 00:29:16,560
Speaker 3: even just five years ago.

555
00:29:17,960 --> 00:29:20,920
Speaker 13: I would say that there's way more importance that is

556
00:29:21,120 --> 00:29:26,080
Speaker 13: coming from us in the quality of the founders, particularly

557
00:29:26,120 --> 00:29:29,360
Speaker 13: at the early stage. You know, often there is it's

558
00:29:29,360 --> 00:29:32,080
Speaker 13: an idea, or there's a model, or there's a prototype,

559
00:29:32,120 --> 00:29:34,280
Speaker 13: but it hasn't really been battle tested.

560
00:29:34,080 --> 00:29:37,760
Speaker 8: Yet, and so the quality of the founder is really essential.

561
00:29:38,800 --> 00:29:40,520
Speaker 13: And it's not just the quality of the person in

562
00:29:40,640 --> 00:29:43,040
Speaker 13: terms of these people in terms of thinking and their

563
00:29:43,080 --> 00:29:47,160
Speaker 13: ability to execute, but it's also their personality and their resiliency.

564
00:29:47,520 --> 00:29:51,120
Speaker 13: You know, building a business is really really difficult, and

565
00:29:51,400 --> 00:29:53,080
Speaker 13: how you know, we're going to be married to them

566
00:29:53,200 --> 00:29:55,479
Speaker 13: for a long long time and we want to make

567
00:29:55,480 --> 00:29:57,480
Speaker 13: sure that they're the right people. So that really is

568
00:29:57,840 --> 00:30:01,280
Speaker 13: our main focus now. Obviously we're interested in what they're

569
00:30:01,320 --> 00:30:06,120
Speaker 13: doing in the domains. You know, we we avoid investing

570
00:30:06,320 --> 00:30:12,400
Speaker 13: in foundation models, anything that has heavy computing layers, a

571
00:30:12,440 --> 00:30:13,240
Speaker 13: lot of CAPEX.

572
00:30:13,320 --> 00:30:16,920
Speaker 8: That's really kind of for you know, the big megafunds

573
00:30:16,960 --> 00:30:17,760
Speaker 8: on the West Coast.

574
00:30:18,040 --> 00:30:21,960
Speaker 13: We're looking for people who have domain expertise, who have

575
00:30:22,120 --> 00:30:27,240
Speaker 13: a passionate idear and that and most most people obviously

576
00:30:27,280 --> 00:30:29,800
Speaker 13: these days are going to be using AI to launch

577
00:30:29,840 --> 00:30:32,800
Speaker 13: their products, and so it's it's you know, we we've

578
00:30:32,880 --> 00:30:35,880
Speaker 13: done this now for seventeen years. We're looking for that

579
00:30:36,120 --> 00:30:39,240
Speaker 13: needle in a haystack, that exceptional funding team.

580
00:30:39,680 --> 00:30:43,400
Speaker 3: It's so interesting because if you think the phrase seed

581
00:30:43,560 --> 00:30:45,880
Speaker 3: round I've written this year, like what's the point? You

582
00:30:45,920 --> 00:30:48,640
Speaker 3: can call it a coconut seed or a mango seed round.

583
00:30:49,200 --> 00:30:51,680
Speaker 3: But the sort of interesting part is that a small

584
00:30:51,760 --> 00:30:56,400
Speaker 3: group of largely researchers can raise such an incredibly large

585
00:30:56,480 --> 00:30:59,920
Speaker 3: seed for compute. So if you're looking for the profile

586
00:31:00,080 --> 00:31:03,440
Speaker 3: set up that doesn't need the compute, are they able

587
00:31:03,560 --> 00:31:07,560
Speaker 3: to command like that debut check size, you know, what

588
00:31:07,600 --> 00:31:08,080
Speaker 3: would they.

589
00:31:08,040 --> 00:31:08,440
Speaker 2: Need it for?

590
00:31:09,800 --> 00:31:09,960
Speaker 4: Well?

591
00:31:10,120 --> 00:31:12,600
Speaker 13: No, they look everyone's going to use compute at some

592
00:31:12,680 --> 00:31:15,520
Speaker 13: point or another. I'm looking for companies that don't have

593
00:31:15,720 --> 00:31:19,000
Speaker 13: to be building the compute themselves, I see, but they'll

594
00:31:19,040 --> 00:31:20,960
Speaker 13: be able to rent the compute, they'll be able to

595
00:31:21,600 --> 00:31:25,680
Speaker 13: you know, offer tokens, which are really a measurement of compute.

596
00:31:26,880 --> 00:31:29,520
Speaker 13: So but we're looking for we're looking for people who

597
00:31:29,640 --> 00:31:34,480
Speaker 13: have original ideas that are you know, not just again

598
00:31:34,560 --> 00:31:38,520
Speaker 13: as I mentioned, wrappers around around the l l MS,

599
00:31:38,560 --> 00:31:41,440
Speaker 13: around the foundational models, and we're getting to see that

600
00:31:41,640 --> 00:31:45,840
Speaker 13: second generation of AI companies. And we saw this in mobile,

601
00:31:45,960 --> 00:31:50,600
Speaker 13: we saw this, you know previously, previous technological cycles. You know,

602
00:31:50,920 --> 00:31:54,000
Speaker 13: the first cycle is typically kind of low hanging fruit.

603
00:31:54,080 --> 00:31:57,040
Speaker 13: I'm going to automate this, I'm going to go into

604
00:31:57,080 --> 00:31:59,360
Speaker 13: customer service, I'm going to go into law and this

605
00:31:59,520 --> 00:32:02,920
Speaker 13: kind of thing. The second generation of people who invent

606
00:32:03,040 --> 00:32:07,000
Speaker 13: products that could not be made possible without the technology

607
00:32:07,080 --> 00:32:07,840
Speaker 13: of today, and the.

608
00:32:07,960 --> 00:32:10,600
Speaker 8: Technology of today obviously easier. Those are the people that

609
00:32:10,680 --> 00:32:11,160
Speaker 8: we're looking for.

610
00:32:11,520 --> 00:32:14,880
Speaker 3: You're deploying your ninth seed fund, and what you told

611
00:32:14,920 --> 00:32:17,840
Speaker 3: me was you were doing it with extreme discipline. So

612
00:32:17,920 --> 00:32:20,600
Speaker 3: I wonder how often it's more about resisting the urge

613
00:32:20,680 --> 00:32:25,640
Speaker 3: to participate in around potentially rather than go actively chasing around.

614
00:32:27,720 --> 00:32:27,920
Speaker 1: Yeah.

615
00:32:28,120 --> 00:32:31,360
Speaker 8: Look, that's where there's art coming into the science.

616
00:32:31,520 --> 00:32:31,640
Speaker 9: Right.

617
00:32:31,800 --> 00:32:34,280
Speaker 13: So the science is, you know, can I live with

618
00:32:34,440 --> 00:32:38,160
Speaker 13: a high valuation? And what will become of the company

619
00:32:38,240 --> 00:32:41,800
Speaker 13: if it starts at such a high valuation. The art is,

620
00:32:42,400 --> 00:32:47,320
Speaker 13: you know, do these people have taste? Are they do

621
00:32:47,440 --> 00:32:50,440
Speaker 13: they have an idea that it might might not command

622
00:32:50,840 --> 00:32:54,320
Speaker 13: a super high valuation today because it's brand new, it's novel,

623
00:32:54,440 --> 00:32:58,880
Speaker 13: it hasn't been proven and so again, you know, you

624
00:32:59,240 --> 00:33:01,560
Speaker 13: weave and you bob and you go this way, and

625
00:33:01,680 --> 00:33:06,080
Speaker 13: you're you know, our deal flow hasn't been as the

626
00:33:06,240 --> 00:33:07,480
Speaker 13: highest level it's ever been.

627
00:33:07,640 --> 00:33:10,680
Speaker 8: You know, we literally are talking to thousands of companies.

628
00:33:11,040 --> 00:33:13,160
Speaker 8: Our entire team talks two thousands.

629
00:33:12,840 --> 00:33:15,520
Speaker 13: Companies every year, and so we've become more and more

630
00:33:15,640 --> 00:33:20,280
Speaker 13: selective as a result of that. And and you know,

631
00:33:20,400 --> 00:33:24,360
Speaker 13: we use our experience, We use our kind of you know,

632
00:33:25,240 --> 00:33:28,880
Speaker 13: accumulated knowledge over the years to try to pick the

633
00:33:28,960 --> 00:33:29,440
Speaker 13: best teams.

634
00:33:29,920 --> 00:33:32,640
Speaker 3: Error Hippo, managing partner at Lera Hippo. It's been great

635
00:33:32,640 --> 00:33:34,680
Speaker 3: to have you back on Bloomberg Tech and thank you

636
00:33:34,760 --> 00:33:37,840
Speaker 3: for sharing that accrued knowledge with us. Coming up on

637
00:33:37,880 --> 00:33:41,160
Speaker 3: the program, San Bernova SEO Rodrigue Liang joins to discuss

638
00:33:41,160 --> 00:33:43,840
Speaker 3: the company's latest funding. It's a billion dollars and what's

639
00:33:43,960 --> 00:33:48,320
Speaker 3: driving the next wave of AI infrastructure investment but also innovation.

640
00:33:48,480 --> 00:33:51,600
Speaker 3: And we're still awaiting a press conference from President Trump

641
00:33:51,720 --> 00:33:53,720
Speaker 3: in Ankara, set to start any moment.

642
00:33:53,840 --> 00:33:55,800
Speaker 2: We will bring it to you when it happens. This

643
00:33:55,920 --> 00:33:56,680
Speaker 2: is Bloomberg Tech.

644
00:34:09,840 --> 00:34:12,880
Speaker 3: Chip startup sa Manova has completed the first close of

645
00:34:12,960 --> 00:34:15,880
Speaker 3: its Series F raising a billion dollars at an eleven

646
00:34:15,960 --> 00:34:18,359
Speaker 3: billion dollar buck evaluation. The companies, among a growing group

647
00:34:18,400 --> 00:34:22,359
Speaker 3: of startups, aimans challenge in video and supply AI infrastructure,

648
00:34:22,480 --> 00:34:26,880
Speaker 3: specifically for inference. Samanova CEO Rodrigu Liang is with us

649
00:34:27,200 --> 00:34:29,200
Speaker 3: with more. Let's start with the basics of what that

650
00:34:29,440 --> 00:34:32,560
Speaker 3: capital allows you to do in scaling the platform.

651
00:34:33,719 --> 00:34:37,040
Speaker 14: Well, look, you know, it's uh, the inference market has

652
00:34:37,160 --> 00:34:39,880
Speaker 14: broken everything open and that's what we focus. And so

653
00:34:40,040 --> 00:34:43,600
Speaker 14: we're seeing this incredible demand in the market for inferencing

654
00:34:43,680 --> 00:34:47,719
Speaker 14: and influencing at really high token speeds, and so the

655
00:34:47,840 --> 00:34:51,680
Speaker 14: capital allows us to really accelerate the supply chain, accelerate.

656
00:34:51,280 --> 00:34:54,000
Speaker 2: The build out of these rights, and deliver the rags.

657
00:34:53,800 --> 00:34:57,400
Speaker 3: To a broad wing of customs inferences split in two

658
00:34:57,440 --> 00:35:00,200
Speaker 3: phases pre filled decode and so, you know, to the

659
00:35:00,280 --> 00:35:02,080
Speaker 3: audience that might not be familiar with that, take the

660
00:35:02,120 --> 00:35:06,320
Speaker 3: opportunity Rodrigo to explain why a specific platform for the

661
00:35:06,440 --> 00:35:09,960
Speaker 3: d COODE phase is important, why it works better for inference.

662
00:35:11,040 --> 00:35:13,000
Speaker 14: Yeah, I mean the analogy I make is like, you know,

663
00:35:13,040 --> 00:35:15,000
Speaker 14: if you're the Disney Land ride and you have the

664
00:35:15,160 --> 00:35:17,719
Speaker 14: rights that can really busy and you want to organize

665
00:35:17,719 --> 00:35:20,120
Speaker 14: the queues ahead of time, and so in prefill, what

666
00:35:20,239 --> 00:35:22,800
Speaker 14: you're really able to do is organize this traffic to

667
00:35:23,000 --> 00:35:25,960
Speaker 14: maximize what you need to do when the decode phase

668
00:35:26,080 --> 00:35:28,520
Speaker 14: of inference shows up. And so what you're seeing is

669
00:35:28,600 --> 00:35:31,719
Speaker 14: that for use in video for the prefill portion and

670
00:35:31,800 --> 00:35:34,239
Speaker 14: then replace that in video RAC using someone over for

671
00:35:34,400 --> 00:35:37,080
Speaker 14: d code, you're getting two to three x through put

672
00:35:37,120 --> 00:35:40,800
Speaker 14: advantage on the same infrastructure the same cost, So driving

673
00:35:40,880 --> 00:35:44,760
Speaker 14: your output to a much higher level without increasing your investment.

674
00:35:45,000 --> 00:35:46,799
Speaker 2: So this is the bit that I find fascinating.

675
00:35:47,200 --> 00:35:50,960
Speaker 3: SM forty s M fifty, your proprietary tech, the way

676
00:35:51,040 --> 00:35:54,319
Speaker 3: that you put it out there, five to ten times

677
00:35:54,360 --> 00:35:57,800
Speaker 3: faster on inference on the dcode phase when compared to

678
00:35:57,880 --> 00:36:01,120
Speaker 3: in video GPU. But the reality is it is working

679
00:36:01,200 --> 00:36:05,399
Speaker 3: alongside other accelerators on the platform, right Explain.

680
00:36:05,120 --> 00:36:06,880
Speaker 8: That, Yeah, that's right.

681
00:36:07,160 --> 00:36:09,440
Speaker 14: So if you look at somewhat of a SEM forty,

682
00:36:09,640 --> 00:36:12,480
Speaker 14: it was originally designed for the enterprise, for on prem

683
00:36:12,640 --> 00:36:14,239
Speaker 14: use cases, so you could do the pre few, you

684
00:36:14,280 --> 00:36:17,200
Speaker 14: could do decode altogether, and you can actually run these

685
00:36:17,320 --> 00:36:20,839
Speaker 14: models at a really really high performance, very low power

686
00:36:20,920 --> 00:36:23,480
Speaker 14: ten kilo watch for rack versus one hundred killowats.

687
00:36:23,160 --> 00:36:24,759
Speaker 2: On a typical GP rack.

688
00:36:24,960 --> 00:36:27,440
Speaker 14: Now, if you actually go at scale, when you look

689
00:36:27,480 --> 00:36:30,600
Speaker 14: at these cloud players, these very large frontier labs, what

690
00:36:30,680 --> 00:36:33,800
Speaker 14: they're doing is actually running lots and lots of traffic

691
00:36:33,880 --> 00:36:35,719
Speaker 14: behind it. And so now what you want to do

692
00:36:35,920 --> 00:36:38,520
Speaker 14: is actually take and video chips where are very good

693
00:36:38,600 --> 00:36:42,120
Speaker 14: for certain things, marry them together with somewhatov racks which

694
00:36:42,160 --> 00:36:46,040
Speaker 14: are made for very fast decode of inference, and then

695
00:36:46,080 --> 00:36:50,920
Speaker 14: getting the maximum throughput and maximizing the output of whatever

696
00:36:51,080 --> 00:36:53,160
Speaker 14: power you're able to secure from the data center.

697
00:36:54,360 --> 00:36:56,960
Speaker 3: Let's give some some size and scope, please, Rodrigo. How

698
00:36:57,040 --> 00:37:01,200
Speaker 3: many customers are actually running production workdes on SM forty

699
00:37:01,320 --> 00:37:04,360
Speaker 3: or SM fifty today And I guess you'd measure that

700
00:37:04,600 --> 00:37:07,680
Speaker 3: as tokens processed today or what would the unit of

701
00:37:07,760 --> 00:37:08,279
Speaker 3: measurement be?

702
00:37:09,600 --> 00:37:11,560
Speaker 14: Well today, look, you know, somewhat over we're in the

703
00:37:11,640 --> 00:37:15,440
Speaker 14: business of moving racks. Our customers are service providers, cloud

704
00:37:15,480 --> 00:37:18,160
Speaker 14: players in your model builders, So those are our customers.

705
00:37:18,360 --> 00:37:20,400
Speaker 2: We don't sell tokens. Who really sell racks.

706
00:37:20,440 --> 00:37:21,719
Speaker 14: And so if I you know, if I look at

707
00:37:21,800 --> 00:37:23,920
Speaker 14: kind of what we did the Summatova, we're ready in

708
00:37:24,080 --> 00:37:27,200
Speaker 14: the dozens of customers and we'll teach touch triple digit

709
00:37:27,280 --> 00:37:29,080
Speaker 14: by the end of the year, and so you were

710
00:37:29,120 --> 00:37:32,000
Speaker 14: in the model of the point racks, which then those

711
00:37:32,080 --> 00:37:35,400
Speaker 14: customers of ours would turned into token services that allow

712
00:37:35,480 --> 00:37:39,360
Speaker 14: them to offer these premium tokens, very fast tokens on

713
00:37:39,440 --> 00:37:42,760
Speaker 14: the largest models. How to their developers, how to their users.

714
00:37:43,239 --> 00:37:45,600
Speaker 3: JP Morgan is the big one, right, it's a really

715
00:37:45,719 --> 00:37:49,400
Speaker 3: interesting case study. Explain what it is that JP Morgan

716
00:37:49,480 --> 00:37:51,239
Speaker 3: is able to achieve with your technology.

717
00:37:52,200 --> 00:37:54,680
Speaker 14: Yeah, I'm really excited about this announcement. JP Morgan has

718
00:37:54,719 --> 00:37:57,640
Speaker 14: selected someatover to be the infants provider for the bank.

719
00:37:58,000 --> 00:38:02,040
Speaker 14: And what you're seeing now is we're the attention continues

720
00:38:02,080 --> 00:38:05,600
Speaker 14: to be on the frontier labs, on the hyper skilled clouds.

721
00:38:06,239 --> 00:38:10,200
Speaker 14: One big segment of inference or AI is waking up,

722
00:38:10,239 --> 00:38:11,200
Speaker 14: and that's enterprise.

723
00:38:11,600 --> 00:38:12,360
Speaker 2: The enterprise.

724
00:38:12,680 --> 00:38:15,800
Speaker 14: You know, there's Frankly, you know, JP Morgan has always

725
00:38:15,880 --> 00:38:18,880
Speaker 14: been the leader in actually using technology to drive their business,

726
00:38:19,120 --> 00:38:21,040
Speaker 14: and so they're using someone of them because they can

727
00:38:21,160 --> 00:38:25,919
Speaker 14: bring this technology onto their premise, private data, put into

728
00:38:25,960 --> 00:38:30,040
Speaker 14: these racks, secure data completely within their own firewalls, and

729
00:38:30,160 --> 00:38:33,200
Speaker 14: you can do all of this highly regulated business in

730
00:38:33,320 --> 00:38:39,040
Speaker 14: production within the fine confines of a nicely protected environment.

731
00:38:39,680 --> 00:38:40,080
Speaker 2: Ridrigo.

732
00:38:40,320 --> 00:38:42,920
Speaker 3: I think you'd admit right, there are many inferant specific

733
00:38:43,000 --> 00:38:45,680
Speaker 3: platforms out there in the world at different stages in

734
00:38:45,760 --> 00:38:48,160
Speaker 3: their life cycles. But you know, I want to go

735
00:38:48,239 --> 00:38:50,360
Speaker 3: back to this idea that the the SM forty and

736
00:38:50,440 --> 00:38:53,040
Speaker 3: s M fifty work in conjunction with an Agate, with

737
00:38:53,160 --> 00:38:57,520
Speaker 3: other accelerators, but in particular the GPU. The GPU architecture

738
00:38:57,680 --> 00:39:01,360
Speaker 3: relies on high bandwidth memory, for example, has talked a

739
00:39:01,440 --> 00:39:03,960
Speaker 3: lot about its use of s RAM, so it's not

740
00:39:04,120 --> 00:39:07,680
Speaker 3: affected by the bottleneck that is HBM. How does that

741
00:39:07,840 --> 00:39:10,799
Speaker 3: play out for san Manova, Well, some of.

742
00:39:11,080 --> 00:39:13,640
Speaker 14: You know if it's a data flow architecture which is

743
00:39:13,800 --> 00:39:16,360
Speaker 14: very s RAND based, and so what it does it

744
00:39:16,480 --> 00:39:19,360
Speaker 14: actually is able to actually take all the benefits of

745
00:39:19,440 --> 00:39:22,239
Speaker 14: the s RAM, the very very fast s RAM and

746
00:39:22,320 --> 00:39:25,960
Speaker 14: generate the speeds. One of the benefits of having HPM

747
00:39:26,440 --> 00:39:28,480
Speaker 14: is now you're able to run the big models where

748
00:39:28,560 --> 00:39:34,400
Speaker 14: the STRAM, traditional strand based architectures that are don't have HPM,

749
00:39:34,840 --> 00:39:37,800
Speaker 14: they can't run the trillion parameter models. They have to

750
00:39:37,880 --> 00:39:40,440
Speaker 14: quantize those models. And so we want to tackle the

751
00:39:40,520 --> 00:39:42,960
Speaker 14: hard models, and so we use HPM. But here's what

752
00:39:43,080 --> 00:39:46,320
Speaker 14: we did on Samarova. We use HBM that was n

753
00:39:46,400 --> 00:39:50,800
Speaker 14: Mino one technology HBM that was already in mature production,

754
00:39:51,120 --> 00:39:55,040
Speaker 14: which allows us to actually generate significantly more supply versus

755
00:39:55,160 --> 00:39:58,040
Speaker 14: competing with them video on say the latest and greatest,

756
00:39:58,160 --> 00:40:02,440
Speaker 14: newest HPM. And so we're delivering high performance using mature

757
00:40:02,600 --> 00:40:04,919
Speaker 14: HBM technology with the supply significantly better.

758
00:40:05,560 --> 00:40:08,760
Speaker 3: Let's end on going back to the round. Very interesting.

759
00:40:08,840 --> 00:40:11,680
Speaker 3: Who's now on the cap table or increased position? What

760
00:40:11,840 --> 00:40:14,520
Speaker 3: happens next for you guys like capital intensive on the

761
00:40:14,680 --> 00:40:17,239
Speaker 3: R and D side, on the scaling side, how are

762
00:40:17,239 --> 00:40:19,280
Speaker 3: you going to manage your finances going forward?

763
00:40:20,440 --> 00:40:22,120
Speaker 14: Look, I mean this is something that you know, we're

764
00:40:22,400 --> 00:40:24,880
Speaker 14: heading to scale, and you see the investors coming in,

765
00:40:25,040 --> 00:40:27,360
Speaker 14: you know, from General Atlantic to a capital group of

766
00:40:27,440 --> 00:40:31,960
Speaker 14: Tiro Price, Columbia Seligman, you know, just incredible investors that

767
00:40:32,080 --> 00:40:34,560
Speaker 14: have great track record going to the public markets. And

768
00:40:34,680 --> 00:40:38,680
Speaker 14: so we're actively driving the scale. We see incredible ramp

769
00:40:38,760 --> 00:40:41,440
Speaker 14: up on our demand. So we're just focused on delivery

770
00:40:41,480 --> 00:40:44,279
Speaker 14: and driving the revenue to the point that allows us

771
00:40:44,320 --> 00:40:46,840
Speaker 14: the option of actually going into the public markets.

772
00:40:46,960 --> 00:40:50,879
Speaker 3: Hey, Bloomberg tech technology, but also the money behind it. Rodrigong,

773
00:40:51,040 --> 00:40:53,680
Speaker 3: CEO of San Binova, thank you, very much for your time.

774
00:40:54,120 --> 00:40:55,120
Speaker 2: We're still awaiting a.

775
00:40:55,080 --> 00:40:58,680
Speaker 3: Press conference from the President of the United States in Turkey.

776
00:40:59,360 --> 00:41:02,800
Speaker 3: We will bring you that press conference and the president

777
00:41:03,120 --> 00:41:06,400
Speaker 3: as soon as it starts in Ankra. There have been

778
00:41:06,520 --> 00:41:10,120
Speaker 3: many headlines already this morning relating to Iran, and that

779
00:41:10,280 --> 00:41:13,040
Speaker 3: is the expectation of where that will focus, as well

780
00:41:13,080 --> 00:41:15,600
Speaker 3: as the discussion that has happened over a number of

781
00:41:15,719 --> 00:41:19,160
Speaker 3: days relating to NATO. That is why they are there.

782
00:41:19,480 --> 00:41:23,320
Speaker 3: That does it for this edition of Bloomberg Tech. The

783
00:41:23,480 --> 00:41:26,319
Speaker 3: markets right now are pretty mixed and pretty all over

784
00:41:26,400 --> 00:41:29,160
Speaker 3: the place. We had some really interesting insight from those

785
00:41:29,239 --> 00:41:31,440
Speaker 3: in the markets, both private and public.

786
00:41:31,760 --> 00:41:33,840
Speaker 2: Recap all of that on the podcast.

787
00:41:33,880 --> 00:41:35,160
Speaker 3: You know where to find it on all of the

788
00:41:35,200 --> 00:41:39,560
Speaker 3: Bloomberg platforms, as well as online on Apple, Spotify, and

789
00:41:39,800 --> 00:41:40,240
Speaker 3: iHeart