1
00:00:01,480 --> 00:00:05,320
Speaker 1: We're from Mahart. We're Innovation, Money and Power Collie in

2
00:00:05,440 --> 00:00:10,560
Speaker 1: Silicon Vallet NBN. This is Bloomberg Technology with Caroline Hyde and.

3
00:00:10,720 --> 00:00:11,800
Speaker 2: Ed Ludlow.

4
00:00:25,960 --> 00:00:29,320
Speaker 3: Live from San Francisco to our TV and radio audiences

5
00:00:29,360 --> 00:00:33,320
Speaker 3: around the world. Welcome to a special edition of Bloomberg Technology.

6
00:00:33,320 --> 00:00:35,920
Speaker 3: I'm Ed Ludlow. In just a few moments in vidio,

7
00:00:36,000 --> 00:00:39,320
Speaker 3: CEO Jensen Wang will join us for a live interview

8
00:00:39,400 --> 00:00:43,040
Speaker 3: following their latest earnings report, the company posting a revenue

9
00:00:43,040 --> 00:00:46,280
Speaker 3: forecast that beat consensus that fell short to some of

10
00:00:46,320 --> 00:00:51,360
Speaker 3: the most optimistic estimates, stoking concern that the explosive growth

11
00:00:51,800 --> 00:00:55,200
Speaker 3: is waning. Let's get right to Bloomberg Semiconductor correspondent Ian King,

12
00:00:55,240 --> 00:00:57,840
Speaker 3: who joins me on set. Let's start with the basics,

13
00:00:57,880 --> 00:01:00,760
Speaker 3: the fiscal third quarter forecast and we learned through it.

14
00:01:00,920 --> 00:01:04,680
Speaker 4: Yeah, that forecast was fine. If you compare it to consensus,

15
00:01:04,680 --> 00:01:07,240
Speaker 4: it was there or thereabouts. And for most of the

16
00:01:07,319 --> 00:01:09,720
Speaker 4: companies in the world, that would be great and everybody

17
00:01:09,760 --> 00:01:11,920
Speaker 4: would be happy. But this is in video. This is

18
00:01:11,959 --> 00:01:15,840
Speaker 4: a company which you know, beats pretty much every quarter

19
00:01:16,160 --> 00:01:19,160
Speaker 4: and by an order of magnitude, and it didn't do that,

20
00:01:19,240 --> 00:01:21,200
Speaker 4: and it didn't indicate it would do that, and that

21
00:01:21,280 --> 00:01:23,119
Speaker 4: raised a lot of questions. As we heard on the.

22
00:01:23,040 --> 00:01:25,680
Speaker 3: Call, there are many storylines. I think the demand from

23
00:01:25,680 --> 00:01:29,039
Speaker 3: the hyperscalers is clearly intact, but Blackwell was everything. I

24
00:01:29,080 --> 00:01:32,039
Speaker 3: want to play a SoundBite of what Jensen Wong said

25
00:01:32,200 --> 00:01:33,480
Speaker 3: about Blackwell. Listened to this.

26
00:01:34,520 --> 00:01:38,640
Speaker 2: The change to the mask is complete. There were no

27
00:01:38,959 --> 00:01:46,880
Speaker 2: functional changes necessary, and so we're sampling functional samples of

28
00:01:47,640 --> 00:01:51,880
Speaker 2: Blackwell Grace Blackwell in a variety of system configurations as

29
00:01:51,920 --> 00:01:52,360
Speaker 2: we speak.

30
00:01:53,960 --> 00:01:56,400
Speaker 3: The main point here is that there was not a

31
00:01:56,480 --> 00:02:00,520
Speaker 3: design issue with Blackwell itself as had been reported, but

32
00:02:00,680 --> 00:02:03,800
Speaker 3: based on what Nvidia said, this was about the production mechanism.

33
00:02:04,000 --> 00:02:06,840
Speaker 3: Blue Bazine King, you did a very good job in

34
00:02:06,880 --> 00:02:10,320
Speaker 3: the top Life blog of explaining a GPU mask. Could

35
00:02:10,320 --> 00:02:12,720
Speaker 3: you just try to give a short version of that

36
00:02:12,760 --> 00:02:13,400
Speaker 3: to our audience.

37
00:02:13,440 --> 00:02:17,079
Speaker 4: Yeah, the mask is basically the blueprint, which basically that

38
00:02:17,560 --> 00:02:20,280
Speaker 4: is used to burn in the circuit onto the surface

39
00:02:20,320 --> 00:02:22,280
Speaker 4: of the chip to give it its function, right, and

40
00:02:22,360 --> 00:02:25,280
Speaker 4: it is a very important step. That's the fundamental blueprint.

41
00:02:25,639 --> 00:02:29,240
Speaker 4: And they were saying we didn't get it wrong, but

42
00:02:29,440 --> 00:02:31,920
Speaker 4: when it came to manufacturing, it didn't produce as many

43
00:02:31,960 --> 00:02:34,360
Speaker 4: good chips as we wanted, so we made some tweaks

44
00:02:34,360 --> 00:02:37,000
Speaker 4: to that. Didn't have to redesign it, but made some tweaks,

45
00:02:37,280 --> 00:02:39,120
Speaker 4: and that is helping us to get a better yield.

46
00:02:39,400 --> 00:02:42,480
Speaker 3: It's worth noting at this stage the stocks down almost

47
00:02:42,520 --> 00:02:44,640
Speaker 3: seven percent in after hours, have been down more at

48
00:02:44,680 --> 00:02:47,360
Speaker 3: the conclusion of the call. I think basically because we

49
00:02:47,360 --> 00:02:50,560
Speaker 3: didn't learn enough about Blackwell. What they said was in

50
00:02:50,600 --> 00:02:54,040
Speaker 3: the fiscal fourth quarter of fiscal year twenty five, there

51
00:02:54,040 --> 00:02:58,840
Speaker 3: will be several billion dollars of sales through Blackwell. Why

52
00:02:58,880 --> 00:03:00,960
Speaker 3: does the market want more and what does it want?

53
00:03:01,200 --> 00:03:04,800
Speaker 4: They wanted a reassurance from in videos management, and they

54
00:03:04,840 --> 00:03:09,080
Speaker 4: wanted reassurance in the form of details. They wanted gens

55
00:03:09,120 --> 00:03:11,520
Speaker 4: to PuTTY's arm around everybody and say, don't worry, It'll

56
00:03:11,520 --> 00:03:13,480
Speaker 4: be fine. This is how much I'm going to get.

57
00:03:13,720 --> 00:03:17,040
Speaker 4: They asked consistently, We're asked questions of how many billions

58
00:03:17,160 --> 00:03:20,800
Speaker 4: and when will those billions exactly come in and collect. Kretz,

59
00:03:21,080 --> 00:03:25,200
Speaker 4: the CFO and Jensen Wang essentially avoided that question and

60
00:03:25,360 --> 00:03:27,480
Speaker 4: refuse to give that precise reassurance.

61
00:03:28,600 --> 00:03:30,800
Speaker 3: We're showing some of the after ours reaction, not just

62
00:03:30,840 --> 00:03:32,919
Speaker 3: in in video itself, but some of its peers, both

63
00:03:32,960 --> 00:03:36,760
Speaker 3: on the chip making side the server equipment providers, and

64
00:03:37,080 --> 00:03:39,960
Speaker 3: I think AMD and ARM in particular are very noteworthy.

65
00:03:40,280 --> 00:03:41,960
Speaker 3: Bloombogty and King stay with us. I want to go

66
00:03:42,000 --> 00:03:44,680
Speaker 3: out to Chicago and Bloomberg's Ryan for Lostelka on I

67
00:03:44,760 --> 00:03:47,840
Speaker 3: Equities team and Ryan. That's the broad summary from me

68
00:03:48,080 --> 00:03:50,680
Speaker 3: about the names moving and after hours there's probably a

69
00:03:50,680 --> 00:03:53,280
Speaker 3: bigger picture after ours movement in the markets as well,

70
00:03:53,480 --> 00:03:55,839
Speaker 3: start with in video and work outward from there.

71
00:03:56,640 --> 00:03:58,960
Speaker 5: Sure, well, one thing I would say about in videos

72
00:03:59,000 --> 00:04:01,440
Speaker 5: after hours decline is that it does come after a

73
00:04:01,600 --> 00:04:04,000
Speaker 5: very strong year to day performance. I think it closed

74
00:04:04,040 --> 00:04:05,960
Speaker 5: up more than one hundred and fifty percent this year,

75
00:04:06,280 --> 00:04:09,120
Speaker 5: So even though the re forecast was maybe a little

76
00:04:09,160 --> 00:04:11,800
Speaker 5: bit shy of some of the most optimistic expectations, it

77
00:04:11,800 --> 00:04:14,600
Speaker 5: did beat expectations. And it is, you know, coming off

78
00:04:14,640 --> 00:04:17,080
Speaker 5: such a huge gain. So it's not necessarily surprising to

79
00:04:17,120 --> 00:04:19,720
Speaker 5: see a little bit of a consolidation now. And I

80
00:04:19,760 --> 00:04:21,600
Speaker 5: think even the declient is a little bit less and

81
00:04:21,640 --> 00:04:24,960
Speaker 5: the options market was anticipating. So just some context there

82
00:04:24,960 --> 00:04:27,159
Speaker 5: for looking at the declient. But you're right that pretty

83
00:04:27,200 --> 00:04:30,960
Speaker 5: much we are seeing widespread weakness following this report, all

84
00:04:30,960 --> 00:04:34,160
Speaker 5: the megacaps are, you know, modestly lower. We are seeing

85
00:04:34,240 --> 00:04:37,080
Speaker 5: much more pronounced weakness in other chip makers and chip

86
00:04:37,120 --> 00:04:39,960
Speaker 5: design companies and so forth. So yeah, certainly it does

87
00:04:40,000 --> 00:04:42,320
Speaker 5: seem like the initial read through here is negative, but

88
00:04:42,360 --> 00:04:44,640
Speaker 5: again it does come after a very strong start to

89
00:04:44,680 --> 00:04:45,000
Speaker 5: the year.

90
00:04:46,080 --> 00:04:48,680
Speaker 3: We made a big deal about this earnings print for

91
00:04:48,800 --> 00:04:51,640
Speaker 3: quite a long time. I look at something like an

92
00:04:51,640 --> 00:04:55,800
Speaker 3: investco QQQ, the ETF that tracks the nast that one hundred,

93
00:04:56,120 --> 00:04:58,479
Speaker 3: and I think it's down around a percentage point, right,

94
00:04:58,560 --> 00:05:02,599
Speaker 3: But in true in the market's reaction, was this the

95
00:05:02,640 --> 00:05:05,039
Speaker 3: macro level event that we thought it was going to be.

96
00:05:07,200 --> 00:05:09,200
Speaker 5: That's a great question. I would say that it's kind

97
00:05:09,200 --> 00:05:12,039
Speaker 5: of close enough in line with expectations, even if it

98
00:05:12,120 --> 00:05:14,440
Speaker 5: is a little bit shy the most optimistic ones that

99
00:05:14,480 --> 00:05:16,440
Speaker 5: I don't think this is going to really cause people

100
00:05:16,440 --> 00:05:19,280
Speaker 5: to really change how they're allocating, change their opinions on

101
00:05:19,680 --> 00:05:22,760
Speaker 5: AI's as kind of fundamental secular driver. But maybe in

102
00:05:22,800 --> 00:05:24,920
Speaker 5: the near term we do see a little bit of weakness.

103
00:05:24,960 --> 00:05:27,400
Speaker 5: I mean, again, some of these stocks have been moving

104
00:05:27,480 --> 00:05:29,200
Speaker 5: up so much there have been sort of kind of

105
00:05:29,240 --> 00:05:32,440
Speaker 5: growing calls about their valuation. Some concerns about that. I

106
00:05:32,440 --> 00:05:35,080
Speaker 5: don't know if this is the kind of absolute blowout

107
00:05:35,120 --> 00:05:37,400
Speaker 5: that will just you know, cause people to continue piling

108
00:05:37,440 --> 00:05:40,200
Speaker 5: into the the way that they were doing earlier this year.

109
00:05:41,120 --> 00:05:43,120
Speaker 3: There is a lot in the news cycle inclusive of

110
00:05:43,200 --> 00:05:45,200
Speaker 3: and outside of video. I think one of the names

111
00:05:45,200 --> 00:05:47,880
Speaker 3: that you note in after ours is super Micro. It's

112
00:05:47,920 --> 00:05:52,240
Speaker 3: down significantly. There's an video relationship to that, and then

113
00:05:52,279 --> 00:05:54,720
Speaker 3: there's the news around that name in and of itself.

114
00:05:56,000 --> 00:05:58,960
Speaker 5: Yeah. Absolutely so. We saw yesterday a short report came

115
00:05:59,000 --> 00:06:02,760
Speaker 5: out Today's delaying the filing of the tin K. You know,

116
00:06:02,839 --> 00:06:04,800
Speaker 5: both of those cars does a weakness in the stock.

117
00:06:04,839 --> 00:06:06,680
Speaker 5: I think today it was down more than double digits,

118
00:06:06,760 --> 00:06:08,960
Speaker 5: So you know, a lot of reasons to be concerned

119
00:06:08,960 --> 00:06:12,120
Speaker 5: there in general. And it is one of Nvidia's biggest customers.

120
00:06:12,160 --> 00:06:14,320
Speaker 5: I think it's the third largest, So you know, this

121
00:06:14,400 --> 00:06:17,160
Speaker 5: is just another reason for people who might have been

122
00:06:17,240 --> 00:06:19,680
Speaker 5: kind of souring on super Micro to you know, maybe

123
00:06:19,720 --> 00:06:20,799
Speaker 5: be pulling the cell button.

124
00:06:20,640 --> 00:06:21,040
Speaker 2: A little bit.

125
00:06:22,279 --> 00:06:25,159
Speaker 3: Bloomdo's Ryan for Selica with the after hours action out

126
00:06:25,160 --> 00:06:27,719
Speaker 3: of Chicago. Thank you. Let's get the reaction from the

127
00:06:27,760 --> 00:06:31,279
Speaker 3: cell side with one of the stocks, relative bears d

128
00:06:31,360 --> 00:06:34,599
Speaker 3: A Davidson's Gil Luria had a neutral rating and a

129
00:06:34,640 --> 00:06:38,000
Speaker 3: street low ninety dollars price target on shares of Nvidia.

130
00:06:38,080 --> 00:06:41,920
Speaker 3: He joins US now and Gil, I guess does that

131
00:06:42,080 --> 00:06:44,839
Speaker 3: change your barished perspective on the stock and your main

132
00:06:44,920 --> 00:06:49,480
Speaker 3: takeaway from the analyst school, the quarter was.

133
00:06:49,400 --> 00:06:53,400
Speaker 1: A very good quarter there. Ability to continue to grow

134
00:06:53,920 --> 00:06:58,919
Speaker 1: delivering chips at this rate at this scale is fantastic

135
00:06:58,960 --> 00:07:02,080
Speaker 1: and unprecedented. I don't think there's much of a revenue

136
00:07:02,080 --> 00:07:04,960
Speaker 1: issue here, of a growth issue here. I think a

137
00:07:05,000 --> 00:07:08,160
Speaker 1: little bit of a pushback is probably more around the

138
00:07:08,200 --> 00:07:10,960
Speaker 1: margin situation. You had a good conversation with Ian about

139
00:07:10,960 --> 00:07:13,920
Speaker 1: Blackwell and some of the moving pieces part of what's

140
00:07:13,960 --> 00:07:17,800
Speaker 1: happening with Blackwell. It's a new product, it's got a

141
00:07:17,840 --> 00:07:21,160
Speaker 1: few hitches along the way that put pressure on gross margins,

142
00:07:21,440 --> 00:07:25,640
Speaker 1: and then operating margins grew more than anticipated, so less

143
00:07:25,680 --> 00:07:28,800
Speaker 1: of the top line upside flowed to the bottom line.

144
00:07:28,840 --> 00:07:31,840
Speaker 1: And that's probably where investors are getting a little bit

145
00:07:31,880 --> 00:07:35,000
Speaker 1: more cautious than they were before, because in the last

146
00:07:35,040 --> 00:07:39,240
Speaker 1: several quarters, the huge upside to revenue flowed all the

147
00:07:39,240 --> 00:07:42,720
Speaker 1: way to the bottom line, creating huge upside there to

148
00:07:43,160 --> 00:07:45,960
Speaker 1: believe in and Nvidia right now is to believe twenty

149
00:07:46,040 --> 00:07:48,400
Speaker 1: twenty six calendar is going to be about two hundred

150
00:07:48,440 --> 00:07:51,440
Speaker 1: billion of revenue and about four and a half to

151
00:07:51,560 --> 00:07:55,320
Speaker 1: five dollars of earnings. That's only possible if they continue

152
00:07:55,360 --> 00:07:58,160
Speaker 1: to get this type of growth on the top line

153
00:07:58,280 --> 00:08:01,800
Speaker 1: and more leverage. And the second part of that equation

154
00:08:01,880 --> 00:08:03,720
Speaker 1: looks a little less certain right now.

155
00:08:04,280 --> 00:08:06,480
Speaker 3: Gil. When you said that two hundred billion dollar figure,

156
00:08:06,520 --> 00:08:09,640
Speaker 3: Bloombergsy and King, who's covered semi conductors at this company

157
00:08:09,920 --> 00:08:12,200
Speaker 3: since nineteen ninety eight, sat next to me still, for

158
00:08:12,280 --> 00:08:15,480
Speaker 3: what it's worth, gave a little smile. I mean, it's

159
00:08:15,480 --> 00:08:18,520
Speaker 3: an astonishing figure. I think there's a lot of emphasis

160
00:08:18,560 --> 00:08:21,400
Speaker 3: here on Blackwell. How did you react to the news

161
00:08:21,440 --> 00:08:24,960
Speaker 3: of several billion dollars in fiscal fourth with Blackwell and

162
00:08:25,000 --> 00:08:29,360
Speaker 3: the explanation that this was a production issue impacting the ramp,

163
00:08:29,440 --> 00:08:32,920
Speaker 3: not a sort of core design issue around the product itself.

164
00:08:34,160 --> 00:08:35,960
Speaker 1: So, first of all, our estimate is far less than

165
00:08:36,000 --> 00:08:38,520
Speaker 1: two hundred billion for twenty twenty six.

166
00:08:38,440 --> 00:08:40,800
Speaker 3: But that's you're a relative bear, Gill and I got

167
00:08:40,800 --> 00:08:41,360
Speaker 3: that bit right.

168
00:08:41,400 --> 00:08:45,800
Speaker 1: I think in terms of Blackwell, what happened to in

169
00:08:45,840 --> 00:08:47,319
Speaker 1: n videos they used to be on a two year

170
00:08:47,360 --> 00:08:49,719
Speaker 1: cycle for the data center product, and they decided to

171
00:08:49,760 --> 00:08:53,160
Speaker 1: aggressively move to a one year cycle and then put

172
00:08:53,200 --> 00:08:56,120
Speaker 1: a lot of new feature functionality and bundle into the

173
00:08:56,120 --> 00:09:00,200
Speaker 1: Blackwell platform. So that was a very aggressive agenda, and

174
00:09:00,240 --> 00:09:03,520
Speaker 1: it's not that surprising that there's some challenges that they

175
00:09:03,559 --> 00:09:07,480
Speaker 1: have to encounter. If they can still ship in their

176
00:09:07,559 --> 00:09:11,439
Speaker 1: fiscal fourth quarter, that's a very good sign. And by

177
00:09:11,480 --> 00:09:14,800
Speaker 1: the way, their customers don't really care that much. Their

178
00:09:14,840 --> 00:09:17,200
Speaker 1: customers are just going to continue to buy the latest

179
00:09:17,240 --> 00:09:19,640
Speaker 1: and greatest, and if that's the Age two hundred, they'll

180
00:09:19,679 --> 00:09:22,120
Speaker 1: just continue to buy h two hundred. They have made

181
00:09:22,120 --> 00:09:27,160
Speaker 1: that abundantly clear on their conference calls, Microsoft, Meta, Amazon, Google, Tesla.

182
00:09:27,440 --> 00:09:30,400
Speaker 1: Elon Musk made it abundantly clear they'll buy as much

183
00:09:30,440 --> 00:09:33,640
Speaker 1: GPUs as and Video can produce, at least this year,

184
00:09:34,120 --> 00:09:36,679
Speaker 1: because they're saying the demand signals are there. As long

185
00:09:36,679 --> 00:09:39,000
Speaker 1: as the demand signals are there, they'll buy whatever and

186
00:09:39,120 --> 00:09:41,480
Speaker 1: Video is selling, and towards the end of the year,

187
00:09:41,760 --> 00:09:43,280
Speaker 1: a piece of that will be Blackwell.

188
00:09:44,520 --> 00:09:47,560
Speaker 3: Gil Jensen Wong was it pains and gave a detailed

189
00:09:47,600 --> 00:09:51,440
Speaker 3: answer that basically summarizing video as a systems vendor the

190
00:09:51,480 --> 00:09:55,920
Speaker 3: discussion around Envy link, CPUs, everything beyond GPU. How much

191
00:09:56,000 --> 00:09:58,720
Speaker 3: credit do you assign the company for that, the sort

192
00:09:58,760 --> 00:10:00,760
Speaker 3: of complete systems that they sell.

193
00:10:01,880 --> 00:10:05,199
Speaker 1: A tremendous amount of credit. Let's start with the beginning.

194
00:10:05,840 --> 00:10:10,040
Speaker 1: When history is written, we'll talk about how Jensen Wangen

195
00:10:10,160 --> 00:10:14,199
Speaker 1: Nvidia created the capabilities that make generative AI possible today

196
00:10:14,200 --> 00:10:16,640
Speaker 1: in a similar way that Elon Musk is responsible for

197
00:10:17,000 --> 00:10:20,240
Speaker 1: the electric vehicle. And part of the genius wasn't just

198
00:10:20,320 --> 00:10:23,600
Speaker 1: seeing that this is coming, that accelerate computers coming, AI

199
00:10:23,720 --> 00:10:27,360
Speaker 1: capabilities is coming well before anybody else did. It's also

200
00:10:27,480 --> 00:10:30,920
Speaker 1: realizing that in order to create a moat around the chip,

201
00:10:30,960 --> 00:10:33,679
Speaker 1: they need to do a couple of things. One is

202
00:10:34,080 --> 00:10:37,120
Speaker 1: to own the proprietary software around that KUDA, but the

203
00:10:37,200 --> 00:10:40,920
Speaker 1: other thing is to enhance the bundle as much as possible.

204
00:10:41,320 --> 00:10:45,280
Speaker 1: These GPUs work in a raise when they work together,

205
00:10:45,520 --> 00:10:48,679
Speaker 1: ten thousands or sometimes tens of thousands of GPUs together.

206
00:10:49,200 --> 00:10:55,439
Speaker 1: When you box those up in a box with multiple GPUs, CPUs, wiring,

207
00:10:55,600 --> 00:10:59,600
Speaker 1: cabling and firmware, you make it harder for your competitors

208
00:10:59,640 --> 00:11:02,040
Speaker 1: to catch. And they've done a remarkable job of that.

209
00:11:03,120 --> 00:11:05,520
Speaker 3: I also wonder about all the rest The story thus

210
00:11:05,559 --> 00:11:09,040
Speaker 3: far has been about the hyperscale is a meta five names,

211
00:11:09,160 --> 00:11:12,720
Speaker 3: essentially being the core customer of Nvidia. We got some

212
00:11:12,920 --> 00:11:16,120
Speaker 3: detail on low double digit billions sales this year in

213
00:11:16,200 --> 00:11:19,560
Speaker 3: sovereign AI. Do you see evidence that the market is

214
00:11:19,640 --> 00:11:22,119
Speaker 3: broadening out beyond just the cloud providers?

215
00:11:23,320 --> 00:11:26,000
Speaker 1: Not really. It seems like those five are still a

216
00:11:26,040 --> 00:11:30,640
Speaker 1: substantial amount of the revenue. And those five have in

217
00:11:30,679 --> 00:11:33,680
Speaker 1: their most recent calls said some interesting things. One of

218
00:11:33,720 --> 00:11:36,719
Speaker 1: them is that they consistently said they are over investing.

219
00:11:37,120 --> 00:11:39,520
Speaker 1: So they said they definitely want to invest more. We're

220
00:11:39,520 --> 00:11:41,520
Speaker 1: seeing that in results today. But they also use the

221
00:11:41,600 --> 00:11:45,120
Speaker 1: term over investing. Over investing is something you do before

222
00:11:45,520 --> 00:11:48,240
Speaker 1: you do the other thing, and they made it clear

223
00:11:48,280 --> 00:11:50,400
Speaker 1: that they're only going to do so when they have

224
00:11:50,480 --> 00:11:55,520
Speaker 1: the demand signals. If for whatever reason, demand for AI

225
00:11:55,679 --> 00:12:01,160
Speaker 1: capacity goes down, for instance, because compute doesn't continue to

226
00:12:01,280 --> 00:12:05,800
Speaker 1: drive performance of foundational models, you're going to see those

227
00:12:05,840 --> 00:12:10,120
Speaker 1: same five actors pull back and say, you know, maybe

228
00:12:10,120 --> 00:12:12,600
Speaker 1: we have enough data center capacity, or maybe what we

229
00:12:12,679 --> 00:12:15,760
Speaker 1: have in the pipeline is enough. And that's the concern

230
00:12:15,840 --> 00:12:19,079
Speaker 1: for twenty twenty five and twenty six is that those

231
00:12:19,679 --> 00:12:23,480
Speaker 1: very companies could do that. Beyond that, these are the

232
00:12:23,559 --> 00:12:27,160
Speaker 1: same companies that are developing their own chips. They're at

233
00:12:27,240 --> 00:12:30,320
Speaker 1: various stages of this, but Google's TPUs are probably as

234
00:12:30,360 --> 00:12:33,040
Speaker 1: good as in videos, so in terms of their internal

235
00:12:33,120 --> 00:12:36,680
Speaker 1: use and selling to customers such as Apple, Google's already

236
00:12:36,679 --> 00:12:41,560
Speaker 1: caught up, Amazon's catching up, Meta and Microsoft are earlier stages.

237
00:12:42,120 --> 00:12:45,880
Speaker 1: But these same companies that are buying every GPU from

238
00:12:46,040 --> 00:12:48,160
Speaker 1: video they can gil will be the ones that will

239
00:12:48,200 --> 00:12:50,480
Speaker 1: buy possibly less next year in the year after that.

240
00:12:51,240 --> 00:12:53,560
Speaker 3: I just want to reflect for our TV and radio audience,

241
00:12:53,600 --> 00:12:55,960
Speaker 3: there are people tuned in around the world. My understanding

242
00:12:56,040 --> 00:12:58,400
Speaker 3: is there are watch parties in cities from New York

243
00:12:58,440 --> 00:13:01,920
Speaker 3: to London. For this earnings. We've hyped it up for weeks.

244
00:13:02,360 --> 00:13:05,240
Speaker 3: Probably this is the most unique earnings print that I've

245
00:13:05,280 --> 00:13:07,840
Speaker 3: covered in my career. What about for you in the

246
00:13:07,880 --> 00:13:10,720
Speaker 3: semiconductor space, Why is this so different?

247
00:13:11,720 --> 00:13:14,600
Speaker 1: Well, so, first of all, I tried to participate by

248
00:13:14,640 --> 00:13:19,439
Speaker 1: wearing a video green today. But importantly and video is

249
00:13:19,480 --> 00:13:22,719
Speaker 1: important because of how much they've added to the cumulative

250
00:13:22,720 --> 00:13:25,880
Speaker 1: earnings of this in P. Five hundred and the cumulative performance.

251
00:13:26,360 --> 00:13:31,559
Speaker 1: But I wouldn't necessarily extrapolate their success or sometimes setbacks

252
00:13:31,800 --> 00:13:34,320
Speaker 1: to the rest of the market because what they do

253
00:13:34,440 --> 00:13:38,480
Speaker 1: is very specific. The growth and data center construction and

254
00:13:38,880 --> 00:13:45,200
Speaker 1: data center a GPU fulfillment into those data centers is

255
00:13:45,240 --> 00:13:48,720
Speaker 1: a very small set of companies. I'd argue, and Video

256
00:13:48,800 --> 00:13:51,280
Speaker 1: is sucking the oxygen for the room from a lot

257
00:13:51,320 --> 00:13:54,880
Speaker 1: of other technology companies. So we celebrate within Video because

258
00:13:54,880 --> 00:13:57,800
Speaker 1: they've added so much value I wouldn't necessarily read through

259
00:13:57,880 --> 00:13:59,000
Speaker 1: to the rest of technology.

260
00:14:00,080 --> 00:14:04,880
Speaker 3: Guillerya of DA Davidson, the relatively most bearish or perhaps

261
00:14:04,960 --> 00:14:07,800
Speaker 3: the least bullish name on the street. Thank you so

262
00:14:07,880 --> 00:14:11,600
Speaker 3: welcome to this special edition of Bloomberg Technology for our

263
00:14:11,600 --> 00:14:14,360
Speaker 3: TV and radio audience around the world. Joining me now

264
00:14:14,840 --> 00:14:18,920
Speaker 3: is Jensen One, CEO of Nvidia, straight from the analyst's call,

265
00:14:18,960 --> 00:14:22,200
Speaker 3: and Jensen, good evening, good afternoon to you. I think

266
00:14:22,280 --> 00:14:26,040
Speaker 3: the market wanted more on Blackwell, they wanted more specifics,

267
00:14:26,080 --> 00:14:27,840
Speaker 3: and I'm trying to go through all of the call

268
00:14:27,880 --> 00:14:31,560
Speaker 3: and the transcript. It seems like a very clearly this

269
00:14:31,680 --> 00:14:34,480
Speaker 3: was a production issue and not a fundamental design issue

270
00:14:34,480 --> 00:14:38,080
Speaker 3: with Blackwell, but the deployment in the real world, what

271
00:14:38,160 --> 00:14:40,440
Speaker 3: does that look like tangibly and is there a sort

272
00:14:40,440 --> 00:14:43,560
Speaker 3: of delay in the timeline of that deployment and thus

273
00:14:43,640 --> 00:14:44,840
Speaker 3: revenue from that product.

274
00:14:47,320 --> 00:14:53,320
Speaker 6: I let's see, that's just the fact that I was

275
00:14:53,360 --> 00:14:56,840
Speaker 6: so clear and it wasn't clear enough kind of tripped

276
00:14:56,880 --> 00:15:00,120
Speaker 6: me up there right away. And so let's see. We

277
00:15:00,120 --> 00:15:03,320
Speaker 6: we made a mass change to improve the yield. Functionality

278
00:15:03,320 --> 00:15:06,520
Speaker 6: of Blackwell is wonderful. We're sampling Blackwell all over the

279
00:15:06,520 --> 00:15:12,800
Speaker 6: world today. We show people giving tours to people of

280
00:15:12,840 --> 00:15:15,920
Speaker 6: the Blackwall systems that we have up and running. You

281
00:15:15,960 --> 00:15:19,240
Speaker 6: could find pictures of Blackwall systems all over the web.

282
00:15:20,240 --> 00:15:24,680
Speaker 6: We have started volume production. Volume production will ship in

283
00:15:25,240 --> 00:15:29,240
Speaker 6: Q four. Q four, we will have billions of dollars

284
00:15:29,280 --> 00:15:35,200
Speaker 6: of Blackwell revenues and we will ramp from there. We

285
00:15:35,240 --> 00:15:40,120
Speaker 6: will ramp from there. The demand for Blackwell far exceeds

286
00:15:40,280 --> 00:15:44,360
Speaker 6: its supply, of course in the beginning, because the demand

287
00:15:44,400 --> 00:15:47,040
Speaker 6: is so great, But we're going to have lots and

288
00:15:47,080 --> 00:15:50,480
Speaker 6: lots of supply and we will be able to ramp.

289
00:15:51,040 --> 00:15:54,280
Speaker 6: Starting in Q four, we have billions of dollars of

290
00:15:54,320 --> 00:15:57,040
Speaker 6: revenues and we will ramp from there into Q one

291
00:15:57,120 --> 00:15:59,240
Speaker 6: into Q two and two next year. We're going to

292
00:15:59,280 --> 00:16:00,000
Speaker 6: have a great next year.

293
00:16:00,120 --> 00:16:04,760
Speaker 3: Well, Jensen, what is the demand for accelerated computing beyond

294
00:16:04,760 --> 00:16:09,360
Speaker 3: the hyperscalers and meta.

295
00:16:08,200 --> 00:16:13,000
Speaker 6: Hyperscalers represent about forty five percent of our total data

296
00:16:13,000 --> 00:16:18,400
Speaker 6: center business. We're relatively diversified today. We have hyperscalers, we

297
00:16:18,440 --> 00:16:24,720
Speaker 6: have Internet service providers, we have sovereign AIS, we have

298
00:16:26,040 --> 00:16:32,680
Speaker 6: industries enterprises, so it's fairly fairly diversified. A site outside

299
00:16:32,720 --> 00:16:38,760
Speaker 6: of hyperscalers is the other fifty five percent. Now, the

300
00:16:38,840 --> 00:16:42,680
Speaker 6: application use across all of that, all of that data

301
00:16:42,680 --> 00:16:48,560
Speaker 6: center starts with accelerated computing. Accelerated computing does everything, of course,

302
00:16:49,040 --> 00:16:53,320
Speaker 6: from well the models the things that we know about,

303
00:16:53,320 --> 00:16:57,400
Speaker 6: which is generative AI, and that gets most of the attention.

304
00:16:58,120 --> 00:17:02,920
Speaker 6: But at the core we also do database processing, pre

305
00:17:03,080 --> 00:17:08,119
Speaker 6: and post processing of data before you use it for

306
00:17:08,280 --> 00:17:14,560
Speaker 6: generative AI, trans coding, scientific simulations, computer graphics of course,

307
00:17:14,680 --> 00:17:18,760
Speaker 6: image processing of course. And so there's tons of applications

308
00:17:18,800 --> 00:17:23,199
Speaker 6: that people use are accelerated computing for, and one of

309
00:17:23,200 --> 00:17:26,880
Speaker 6: them is generative AI. And so let's see what else

310
00:17:26,880 --> 00:17:29,480
Speaker 6: can I say? I think that's the.

311
00:17:29,640 --> 00:17:32,639
Speaker 3: Lever jump in Jensen please on Sovereign AI. You and

312
00:17:32,640 --> 00:17:35,400
Speaker 3: I've talked about that before, and it was so interesting

313
00:17:35,440 --> 00:17:38,800
Speaker 3: to hear something behind it that in this fiscal year

314
00:17:39,200 --> 00:17:41,000
Speaker 3: there will be low double digit I think you said

315
00:17:41,040 --> 00:17:44,119
Speaker 3: billions of dollars in sovereign AI sales. But to the

316
00:17:44,200 --> 00:17:46,960
Speaker 3: lay person, what does that mean. It means deals with

317
00:17:47,040 --> 00:17:48,880
Speaker 3: specific governments, if so, where.

318
00:17:50,480 --> 00:17:56,680
Speaker 6: It's not necessarily Sometimes it deals with a particular regional

319
00:17:57,240 --> 00:18:01,119
Speaker 6: service provider that's been funded by the government. And oftentimes

320
00:18:01,160 --> 00:18:03,520
Speaker 6: that's the case in a case of in the case

321
00:18:03,520 --> 00:18:07,760
Speaker 6: of Japan, for example, the the Japanese government came out

322
00:18:08,040 --> 00:18:14,320
Speaker 6: and UH offered UH subsidies of a couple of billion dollars.

323
00:18:14,400 --> 00:18:19,480
Speaker 6: I think for several different internet companies and telcos to

324
00:18:19,560 --> 00:18:24,320
Speaker 6: be able to fund their AI infrastructure. UH India has

325
00:18:24,440 --> 00:18:30,880
Speaker 6: a sovereign AI initiative going and they're building their AI infrastructure. Canada,

326
00:18:32,119 --> 00:18:40,920
Speaker 6: the UK, France, Italy, missing somebody, Singapore, Malaysia, UH. You know,

327
00:18:41,040 --> 00:18:46,200
Speaker 6: a large number of countries are subsidizing their regional data

328
00:18:46,200 --> 00:18:50,160
Speaker 6: centers so that they could become able to build out

329
00:18:50,160 --> 00:18:56,640
Speaker 6: their AI infrastructure. They recognize that their countries UH knowledge,

330
00:18:56,680 --> 00:19:00,760
Speaker 6: their countries data digital data is also their natural resource,

331
00:19:01,800 --> 00:19:03,800
Speaker 6: not just the land they're sitting on, not just the

332
00:19:03,840 --> 00:19:07,960
Speaker 6: air above them. But they realize now that their digital

333
00:19:08,040 --> 00:19:12,760
Speaker 6: knowledge is part of their natural and national resource, and

334
00:19:12,800 --> 00:19:16,720
Speaker 6: they are to harvest that and process that and transform

335
00:19:16,760 --> 00:19:22,159
Speaker 6: it into their national digital intelligence. And so this is

336
00:19:22,200 --> 00:19:24,480
Speaker 6: what we call sovereign AI. You could imagine almost every

337
00:19:24,520 --> 00:19:29,320
Speaker 6: single country in the world will eventually recognize this and

338
00:19:29,320 --> 00:19:30,840
Speaker 6: build out their AI infrastructure.

339
00:19:31,640 --> 00:19:34,359
Speaker 3: Jenson, you use the word resource, and that makes me

340
00:19:34,400 --> 00:19:37,320
Speaker 3: think about the energy requirements here. I think on the

341
00:19:37,320 --> 00:19:40,200
Speaker 3: cool you talk about how the next generation models will

342
00:19:40,240 --> 00:19:43,639
Speaker 3: have many orders of magnitude greater compute needs. But how

343
00:19:43,640 --> 00:19:46,359
Speaker 3: will the energy needs increase and what is the advantage

344
00:19:46,400 --> 00:19:51,159
Speaker 3: you feel in Vidia has in that sense, Well.

345
00:19:51,040 --> 00:19:53,879
Speaker 6: The most important thing that we do is increase the

346
00:19:54,040 --> 00:19:58,080
Speaker 6: performance of and increase the performance and efficiency of our

347
00:19:58,119 --> 00:20:03,320
Speaker 6: next generation. So Blackwell is many times more performance than

348
00:20:03,480 --> 00:20:07,120
Speaker 6: Hopper at the same level of power used. And so

349
00:20:07,200 --> 00:20:11,280
Speaker 6: that's energy efficiency, more performance with the same amount of

350
00:20:11,320 --> 00:20:14,640
Speaker 6: power or same performance at a lower power And that's

351
00:20:14,720 --> 00:20:18,919
Speaker 6: number one. And the second is using luca cooling. We

352
00:20:18,960 --> 00:20:21,880
Speaker 6: support air cool we support air cooling, we support liqual cooling,

353
00:20:21,960 --> 00:20:24,240
Speaker 6: but liqual cooling is a lot more energy efficient, and

354
00:20:24,320 --> 00:20:27,520
Speaker 6: so the combination of all of that you're going to

355
00:20:27,520 --> 00:20:31,119
Speaker 6: get a pretty large, pretty large step up. But the

356
00:20:31,160 --> 00:20:34,080
Speaker 6: important thing to also realize is that AI doesn't really

357
00:20:34,119 --> 00:20:37,120
Speaker 6: care where it goes to school, and so increasingly we're

358
00:20:37,160 --> 00:20:41,040
Speaker 6: going to see AI be trained somewhere else, have that

359
00:20:41,080 --> 00:20:44,399
Speaker 6: model come back and be used near the population, or

360
00:20:44,680 --> 00:20:47,720
Speaker 6: even running on your PC or your phone. And so

361
00:20:47,720 --> 00:20:50,600
Speaker 6: we're going to train large models. But the goal is

362
00:20:50,600 --> 00:20:53,040
Speaker 6: not to run the large models necessarily all the time.

363
00:20:53,200 --> 00:20:55,000
Speaker 6: You can, you can surely do that for some of

364
00:20:55,040 --> 00:21:00,280
Speaker 6: the premium services and the very high value AIS, but

365
00:21:00,480 --> 00:21:02,960
Speaker 6: it's very likely that these large models would then help

366
00:21:03,080 --> 00:21:06,320
Speaker 6: to train and teach smaller models, and what we'll end

367
00:21:06,400 --> 00:21:10,879
Speaker 6: up doing is have one large, few large models that

368
00:21:10,960 --> 00:21:13,200
Speaker 6: are able to train a whole bunch of small models,

369
00:21:13,200 --> 00:21:14,200
Speaker 6: and they run everywhere.

370
00:21:15,840 --> 00:21:20,800
Speaker 3: Jensen, you explain clearly that demand to build generative AI

371
00:21:20,960 --> 00:21:24,000
Speaker 3: product on models or even at the GPU level is

372
00:21:24,080 --> 00:21:28,080
Speaker 3: greater than current supply. In black Wealth's case in particular,

373
00:21:28,320 --> 00:21:31,159
Speaker 3: explain the supply dynamics to me for your products and

374
00:21:31,200 --> 00:21:34,440
Speaker 3: whether you see an improvement sequentially quarter on quarter or

375
00:21:34,480 --> 00:21:36,640
Speaker 3: at some point by the end of fiscal year into

376
00:21:36,680 --> 00:21:37,160
Speaker 3: next year.

377
00:21:38,440 --> 00:21:41,720
Speaker 6: Well, the fact that we're growing would suggest that our

378
00:21:41,840 --> 00:21:46,600
Speaker 6: supply is improving and our supply chain is quite large.

379
00:21:46,960 --> 00:21:49,359
Speaker 6: One of the largest supply chains in the world. We

380
00:21:49,440 --> 00:21:53,400
Speaker 6: have incredible partners and they're doing a great job supporting

381
00:21:53,480 --> 00:21:56,320
Speaker 6: us in our growth. As you know, we're one of

382
00:21:56,320 --> 00:22:00,199
Speaker 6: the fastest growing technology companies in history. None of that

383
00:22:00,200 --> 00:22:03,600
Speaker 6: would have been possible without very strong demand but also

384
00:22:03,720 --> 00:22:06,520
Speaker 6: very strong supply. We're expecting Q three to have more

385
00:22:06,520 --> 00:22:08,840
Speaker 6: supply than Q two, We're expecting Q four to have

386
00:22:08,880 --> 00:22:10,919
Speaker 6: more supply than Q three, and we're expecting Q one

387
00:22:10,960 --> 00:22:12,919
Speaker 6: to have more supply than Q four, And so I

388
00:22:12,920 --> 00:22:16,000
Speaker 6: think our supply, our supply condition going into next year

389
00:22:16,080 --> 00:22:18,800
Speaker 6: will be in will be a large improvement over this

390
00:22:19,000 --> 00:22:24,960
Speaker 6: last year with respect to demand. Blackwell is just such

391
00:22:24,960 --> 00:22:28,639
Speaker 6: a leap and there's several things that are happening, you know,

392
00:22:28,720 --> 00:22:31,919
Speaker 6: just the Foundation model makers themselves. The size of the

393
00:22:31,920 --> 00:22:35,879
Speaker 6: Foundation models are growing from hundreds of billions parameters to

394
00:22:36,080 --> 00:22:40,399
Speaker 6: trillions of parameters. They're also learning more languages. Instead of

395
00:22:40,600 --> 00:22:43,560
Speaker 6: just learning human language, they're learning the language of images

396
00:22:43,720 --> 00:22:49,080
Speaker 6: and sounds and videos, and they're even learning the language

397
00:22:49,080 --> 00:22:52,119
Speaker 6: of three D graphics. And whenever they are able to

398
00:22:52,160 --> 00:22:54,879
Speaker 6: learn these languages, they can understand what they see, but

399
00:22:54,920 --> 00:22:58,119
Speaker 6: they can also generate what they're asked to generate and

400
00:22:58,200 --> 00:23:02,400
Speaker 6: so they're learning the language of routines and chemicals and physics.

401
00:23:02,960 --> 00:23:04,840
Speaker 6: You know, it could be fluids and it could be

402
00:23:05,160 --> 00:23:08,760
Speaker 6: particle physics, and so they're learning all kinds of different languages,

403
00:23:09,320 --> 00:23:12,400
Speaker 6: learned the meaning of what we call modalities, but basically

404
00:23:12,440 --> 00:23:17,240
Speaker 6: learning languages. And so these models are growing in size,

405
00:23:17,240 --> 00:23:20,639
Speaker 6: they're learning from more data, and there are more model

406
00:23:20,640 --> 00:23:23,800
Speaker 6: makers then there was a year ago. And so the

407
00:23:23,880 --> 00:23:26,680
Speaker 6: number of model makers have grown substantially because of all

408
00:23:26,680 --> 00:23:29,520
Speaker 6: these different modalities. And so that's just one, just the

409
00:23:29,520 --> 00:23:34,400
Speaker 6: frontier model. The foundation model makers themselves haven't really grown tremendously.

410
00:23:34,840 --> 00:23:39,159
Speaker 6: And then the generative AI market has really diversified, you know,

411
00:23:39,280 --> 00:23:43,719
Speaker 6: beyond the Internet service makers to startups and now enterprises

412
00:23:43,760 --> 00:23:47,879
Speaker 6: are jumping in and different countries are jumping in, so

413
00:23:48,240 --> 00:23:49,359
Speaker 6: the demand is really growing.

414
00:23:50,000 --> 00:23:52,359
Speaker 3: Jensen, I'm sorry to cut you off. I will lose

415
00:23:52,600 --> 00:23:56,840
Speaker 3: your time soon. You've also diversified. And when I said

416
00:23:56,880 --> 00:23:58,639
Speaker 3: to our audience you were coming on, I got so

417
00:23:58,720 --> 00:24:02,320
Speaker 3: many questions. Probably the most common one is what is

418
00:24:02,359 --> 00:24:05,359
Speaker 3: in Vidia. We talked about you as a systems vendor,

419
00:24:05,600 --> 00:24:09,000
Speaker 3: but so many points on in Vidia Gpu Cloud, and

420
00:24:09,040 --> 00:24:12,280
Speaker 3: I want to ask finally, do you have plans to

421
00:24:12,280 --> 00:24:14,600
Speaker 3: become literally a cloud compute provider.

422
00:24:16,640 --> 00:24:22,119
Speaker 6: No. Our GPU cloud was designed to be the best

423
00:24:22,359 --> 00:24:27,800
Speaker 6: version of Nvidia Cloud that's built within each cloud. Nvidio

424
00:24:27,840 --> 00:24:34,760
Speaker 6: DGX cloud is built inside GCP, inside Azure, inside AWS,

425
00:24:34,760 --> 00:24:38,880
Speaker 6: inside OCI, and so we build our clouds within THEIRS

426
00:24:39,240 --> 00:24:42,119
Speaker 6: so that we can implement our best version of our cloud,

427
00:24:42,800 --> 00:24:46,880
Speaker 6: work with them to make that cloud, that that infrastructure,

428
00:24:46,920 --> 00:24:51,359
Speaker 6: that AI infrastructure in Nvidia infrastructure, as performance, as great

429
00:24:51,400 --> 00:24:57,760
Speaker 6: TCO as possible, and so that strategy has worked incredibly well.

430
00:24:58,320 --> 00:25:01,760
Speaker 6: And of course we are large consumers of it because

431
00:25:01,800 --> 00:25:04,200
Speaker 6: we create a lot of AI ourselves, because our chips

432
00:25:04,240 --> 00:25:07,000
Speaker 6: aren't possible to design without AI, our software is not

433
00:25:07,040 --> 00:25:09,520
Speaker 6: possible to write without AI, and so we use it

434
00:25:09,560 --> 00:25:12,919
Speaker 6: ourselves tremendous, you know, tremendous amount of it self driving cars,

435
00:25:12,920 --> 00:25:15,480
Speaker 6: the general robotics work that we're doing, the omniverse work

436
00:25:15,480 --> 00:25:18,639
Speaker 6: that we're doing. So we're using the DGX cloud for ourselves.

437
00:25:19,520 --> 00:25:23,360
Speaker 6: We also use it for an AI foundry. We make

438
00:25:23,440 --> 00:25:27,639
Speaker 6: AI models for companies that would like to have expertise

439
00:25:27,640 --> 00:25:30,320
Speaker 6: in doing so, and so we are AI. We're an AI,

440
00:25:30,440 --> 00:25:33,840
Speaker 6: we're a foundry for AI. Like tsmcs a foundry for

441
00:25:33,840 --> 00:25:37,399
Speaker 6: our chips and so so there are three fundamental reasons

442
00:25:37,400 --> 00:25:39,960
Speaker 6: why we do it. One is to have the best

443
00:25:40,160 --> 00:25:43,200
Speaker 6: version of Nvidia inside all the clouds. Two because we're

444
00:25:43,200 --> 00:25:46,119
Speaker 6: a large consumer to ourselves, and third because we use

445
00:25:46,160 --> 00:25:48,520
Speaker 6: it for AI foundry for to help every other company.

446
00:25:49,880 --> 00:25:52,239
Speaker 3: Jensen one, CEO and video. I want to thank you

447
00:25:52,280 --> 00:25:55,160
Speaker 3: for your time the extended conversation straight off the earnings

448
00:25:55,200 --> 00:25:59,160
Speaker 3: called thank you in earnings fiscal it's good to see

449
00:25:59,160 --> 00:26:02,000
Speaker 3: you too called second quarter. Then I point out that

450
00:26:02,080 --> 00:26:04,800
Speaker 3: in after hours the stock is down still almost seven percent.

451
00:26:05,040 --> 00:26:07,720
Speaker 3: There'll be a lot of analysis to be done overnight

452
00:26:07,760 --> 00:26:09,960
Speaker 3: by the cell side, by the byside, and we'll bring

453
00:26:10,000 --> 00:26:13,280
Speaker 3: you the best from Bloomberg's reporters and editors from San Francisco.

454
00:26:13,600 --> 00:26:26,399
Speaker 3: This is Bloomberg Technology