1
00:00:00,040 --> 00:00:07,000
Speaker 1: Bloomberg Audio Studios, Podcasts, Radio News.

2
00:00:07,840 --> 00:00:09,920
Speaker 2: One of the biggest stories that we have been following

3
00:00:10,000 --> 00:00:12,000
Speaker 2: in the last twenty four hours, though, of course, has

4
00:00:12,000 --> 00:00:15,080
Speaker 2: been the important keynote address from Jensen Wong at the

5
00:00:15,120 --> 00:00:18,560
Speaker 2: Consumer Electronics Show. It has been feeding a number of

6
00:00:18,560 --> 00:00:21,480
Speaker 2: storylines and moving the stock ever since that took place

7
00:00:21,520 --> 00:00:24,639
Speaker 2: in Las Vegas yesterday. And joining us right now for

8
00:00:24,760 --> 00:00:27,320
Speaker 2: a very special conversation, We've got good news for you.

9
00:00:27,400 --> 00:00:31,600
Speaker 2: It's Bloomberg's Ed Ludlow with the CEO of Nvidia, Jensen Wong,

10
00:00:31,600 --> 00:00:34,200
Speaker 2: as well as the president's CEO of Siemens, Roland Bush.

11
00:00:34,200 --> 00:00:37,320
Speaker 2: They're at the cees right now and we want to

12
00:00:37,360 --> 00:00:39,600
Speaker 2: hand things out to Las Vegas. Ed Ludlow, you have

13
00:00:39,680 --> 00:00:40,360
Speaker 2: the con.

14
00:00:42,840 --> 00:00:43,520
Speaker 3: Thank you, Joe.

15
00:00:43,560 --> 00:00:46,120
Speaker 4: Over one hundred and seventy five years, Siemens has been

16
00:00:46,159 --> 00:00:49,400
Speaker 4: at the forefront, if i'd say, several industrial revolutions. In

17
00:00:49,520 --> 00:00:54,160
Speaker 4: Vidia is at the forefront of the latest AI industrial revolution,

18
00:00:54,280 --> 00:00:57,600
Speaker 4: and it's a pleasure to have you both here with us. Gentlemen, Roland,

19
00:00:57,720 --> 00:00:59,959
Speaker 4: I want to try and understand how real this is.

20
00:01:00,320 --> 00:01:03,800
Speaker 4: You call it an industrial AI operating system. We're going

21
00:01:03,840 --> 00:01:06,080
Speaker 4: to talk about how you're joining forces on the software

22
00:01:06,080 --> 00:01:08,760
Speaker 4: side and the hardware side. But I think the most

23
00:01:08,800 --> 00:01:11,400
Speaker 4: useful place to start would be, could you outline a

24
00:01:11,520 --> 00:01:15,479
Speaker 4: timeline for when Semens goes beyond its own footprint around

25
00:01:15,520 --> 00:01:19,000
Speaker 4: the world using this operating system to customers actually doing

26
00:01:19,520 --> 00:01:21,679
Speaker 4: things at scale, because I think when you're on stage,

27
00:01:21,720 --> 00:01:23,119
Speaker 4: that's the kind of sense I was getting.

28
00:01:23,280 --> 00:01:24,880
Speaker 3: You want to accelerate to scale.

29
00:01:25,400 --> 00:01:28,720
Speaker 5: Well, I mean this technology is already in place and

30
00:01:28,800 --> 00:01:29,360
Speaker 5: it's working.

31
00:01:29,600 --> 00:01:31,280
Speaker 6: I mean we see there are many examples, and it

32
00:01:31,360 --> 00:01:35,480
Speaker 6: brought some on stage today where you have customers starting

33
00:01:35,480 --> 00:01:38,160
Speaker 6: with a digital twin of a product. They want to

34
00:01:38,200 --> 00:01:41,280
Speaker 6: manufacture a digital train of the manufacturing side and bring

35
00:01:41,319 --> 00:01:43,759
Speaker 6: that all together and not before that you.

36
00:01:43,720 --> 00:01:45,720
Speaker 5: Have it optimized them and build in the real world.

37
00:01:46,280 --> 00:01:49,240
Speaker 6: And AI is already working on the shop for they

38
00:01:49,240 --> 00:01:52,280
Speaker 6: call it machine learning. But the point is with the

39
00:01:52,360 --> 00:01:54,920
Speaker 6: new models, you can bring that to the next level.

40
00:01:55,120 --> 00:01:58,200
Speaker 6: That means you go away from giving an advice to

41
00:01:58,240 --> 00:02:01,560
Speaker 6: somebody with technology, but really act on your behalf. This

42
00:02:01,720 --> 00:02:05,640
Speaker 6: is when things go more autonomous or adaptive, and you

43
00:02:05,720 --> 00:02:06,960
Speaker 6: see that already starting.

44
00:02:07,400 --> 00:02:09,760
Speaker 5: The big thing is how can we scale it?

45
00:02:09,960 --> 00:02:12,800
Speaker 6: Because it requires a lot of skills from our customers,

46
00:02:12,840 --> 00:02:15,359
Speaker 6: a lot of technology, and it's still not that easy

47
00:02:15,400 --> 00:02:16,920
Speaker 6: to implement the working on it to.

48
00:02:16,919 --> 00:02:19,760
Speaker 7: Make it easy to deploy and easy to use, but

49
00:02:19,960 --> 00:02:23,280
Speaker 7: you see picking up momentum in each stuff we do,

50
00:02:23,560 --> 00:02:26,239
Speaker 7: and it's also great examples as a shipbuilding is a

51
00:02:26,360 --> 00:02:29,680
Speaker 7: short everlenge, but even start ups using our technology like

52
00:02:29,800 --> 00:02:30,840
Speaker 7: commentalst Fusion.

53
00:02:30,600 --> 00:02:35,120
Speaker 4: Systems Jensen, we've discussed the five layer cake often. You

54
00:02:35,160 --> 00:02:37,680
Speaker 4: know within video it started with the GPUs, but it's

55
00:02:37,720 --> 00:02:42,480
Speaker 4: now software, the element of simulation. On stage, you talked

56
00:02:42,480 --> 00:02:46,760
Speaker 4: about integrating the software side, in particular into DA Again,

57
00:02:46,800 --> 00:02:49,000
Speaker 4: a similar question for you, more of a timeline. What

58
00:02:49,160 --> 00:02:51,040
Speaker 4: is it you think you'll do first in how you

59
00:02:51,040 --> 00:02:52,080
Speaker 4: guys work closer together.

60
00:02:52,240 --> 00:02:55,000
Speaker 1: The first thing, let me just say very quickly, we're

61
00:02:55,000 --> 00:02:57,919
Speaker 1: announcing a big partnership between us. We've known each other

62
00:02:57,919 --> 00:03:00,120
Speaker 1: for a long time, but this is a partnership were

63
00:03:00,120 --> 00:03:01,480
Speaker 1: announcing is really a big deal.

64
00:03:02,240 --> 00:03:02,480
Speaker 5: One.

65
00:03:02,800 --> 00:03:08,560
Speaker 1: We're accelerating their EDA software, We're accelerating their simulation software.

66
00:03:08,919 --> 00:03:14,280
Speaker 1: We're integrating AI technology, physical AI and agentic AI into

67
00:03:14,400 --> 00:03:19,040
Speaker 1: their team center and their factory automation operating system, and

68
00:03:19,080 --> 00:03:22,920
Speaker 1: so we're working together across this entire spectrum. When we

69
00:03:22,960 --> 00:03:25,480
Speaker 1: accelerate the software, then we'll get to use it to

70
00:03:25,520 --> 00:03:29,600
Speaker 1: design our chips and systems. When we accelerate their simulation software,

71
00:03:29,720 --> 00:03:32,560
Speaker 1: we'll use it in our AI factories to simulate the

72
00:03:32,560 --> 00:03:38,760
Speaker 1: thermal properties of our AI factories. When we integrate our

73
00:03:38,800 --> 00:03:44,400
Speaker 1: automation and agentic systems into their AI industrial operating system,

74
00:03:44,680 --> 00:03:47,320
Speaker 1: we can then use it in our factory floors with

75
00:03:47,440 --> 00:03:51,040
Speaker 1: our partners, for example, fox Con And so we're working

76
00:03:51,120 --> 00:03:53,280
Speaker 1: across this entire spectrum together and we're going to put

77
00:03:53,280 --> 00:03:56,200
Speaker 1: the technology to use basically as soon as we can.

78
00:03:56,320 --> 00:03:57,760
Speaker 3: What's the net effect for you, Jensen?

79
00:03:57,920 --> 00:04:02,400
Speaker 4: Is it improves efficient capital allocation. I know that might

80
00:04:02,440 --> 00:04:04,600
Speaker 4: sound a bit dry, but actually, right now, that's the

81
00:04:04,680 --> 00:04:08,080
Speaker 4: answer everyone's searching for. How is this investment in AI

82
00:04:08,560 --> 00:04:11,080
Speaker 4: and use of the technology actually changed things in the

83
00:04:11,120 --> 00:04:11,840
Speaker 4: real world for me?

84
00:04:12,120 --> 00:04:13,840
Speaker 5: Announced yesterday Vera Rubin.

85
00:04:14,080 --> 00:04:18,400
Speaker 1: It takes six different chips to integrate into this incredible

86
00:04:18,400 --> 00:04:22,240
Speaker 1: system called Vera Rubin. And when you're done, each one

87
00:04:22,279 --> 00:04:26,400
Speaker 1: of these Vera Rubin GPUs is two hundred and forty

88
00:04:26,760 --> 00:04:31,120
Speaker 1: thousand wants and it is ten times more energy efficient

89
00:04:31,120 --> 00:04:34,080
Speaker 1: than a last generation. It is ten times more cost

90
00:04:34,080 --> 00:04:37,200
Speaker 1: efficient than a last generation but it's still the technology

91
00:04:37,240 --> 00:04:42,640
Speaker 1: is insanely complicated. One hundred fifteen thousand engineering years came

92
00:04:42,680 --> 00:04:45,920
Speaker 1: together to build this system. And so when we accelerate

93
00:04:46,080 --> 00:04:49,200
Speaker 1: EDA tools, when we accelerate simulation tools, and when we

94
00:04:49,400 --> 00:04:53,920
Speaker 1: can eventually and I'm hoping very soon design entire vera

95
00:04:54,000 --> 00:04:58,960
Speaker 1: Ruben systems inside a Semen's digital twin, the chance the

96
00:04:59,080 --> 00:05:02,880
Speaker 1: ability for us to create much much more complex systems

97
00:05:02,920 --> 00:05:06,039
Speaker 1: will scale. We'll do it much more efficiently. And so

98
00:05:06,120 --> 00:05:08,520
Speaker 1: this is really about being able to do the impossible,

99
00:05:08,920 --> 00:05:12,120
Speaker 1: and being able to do it impossible the impossible right

100
00:05:12,320 --> 00:05:13,120
Speaker 1: the first time.

101
00:05:13,640 --> 00:05:18,839
Speaker 6: And and and and once they then realize that AI

102
00:05:19,080 --> 00:05:22,760
Speaker 6: creates real world impact, this is where it really deploys

103
00:05:22,800 --> 00:05:23,880
Speaker 6: the full power.

104
00:05:23,720 --> 00:05:25,239
Speaker 5: And also the economic power.

105
00:05:25,800 --> 00:05:27,560
Speaker 6: It's not only in the data centers are any eye

106
00:05:27,600 --> 00:05:29,880
Speaker 6: factories which we see, but also on the edge because

107
00:05:29,960 --> 00:05:32,640
Speaker 6: once you start influencing with low latency, you bring this

108
00:05:32,760 --> 00:05:35,599
Speaker 6: technology to the edge. This is a huge potential for

109
00:05:35,680 --> 00:05:39,760
Speaker 6: our customers to to deploy this technology. And this includes

110
00:05:39,800 --> 00:05:42,600
Speaker 6: of course in pluts hardware where we come from from

111
00:05:42,640 --> 00:05:43,080
Speaker 6: the JIPS.

112
00:05:43,839 --> 00:05:45,240
Speaker 5: It goes into the controllers.

113
00:05:45,400 --> 00:05:49,400
Speaker 6: Some of our controllers run on GPUs and then it goes.

114
00:05:49,320 --> 00:05:51,920
Speaker 3: All the way to the industrial PC exactly.

115
00:05:52,240 --> 00:05:57,480
Speaker 6: And these are aids. We super charge it and they

116
00:05:57,520 --> 00:06:00,599
Speaker 6: can now run algorithms train than the cloud.

117
00:06:00,640 --> 00:06:02,600
Speaker 5: They can run it on the shop floor and.

118
00:06:02,600 --> 00:06:05,680
Speaker 6: Do all that trick what we talked about it in

119
00:06:05,760 --> 00:06:09,400
Speaker 6: retime optimization and running in a planned and that makes

120
00:06:09,400 --> 00:06:10,040
Speaker 6: a huge difference.

121
00:06:10,040 --> 00:06:12,839
Speaker 4: So it de drives economy. The question that's being searched

122
00:06:12,839 --> 00:06:14,640
Speaker 4: for is how is this going to manifest in.

123
00:06:14,560 --> 00:06:15,039
Speaker 3: The real world.

124
00:06:15,080 --> 00:06:17,400
Speaker 4: You know, the emphasis that this is cs I think

125
00:06:17,520 --> 00:06:21,320
Speaker 4: is physical AI is not the manifestation of the final

126
00:06:21,360 --> 00:06:23,839
Speaker 4: stage of physical AI, just one giant robot that you

127
00:06:23,880 --> 00:06:27,120
Speaker 4: guys call a factory In the manufacturing context, is that

128
00:06:27,160 --> 00:06:29,920
Speaker 4: where you're seeing demand from actual customers, you know that

129
00:06:30,080 --> 00:06:32,880
Speaker 4: they need a factory that is automated one hundred percent

130
00:06:33,320 --> 00:06:35,159
Speaker 4: and genuinely autonomous in some degree.

131
00:06:35,480 --> 00:06:37,880
Speaker 6: Number one is if you want to build a factory,

132
00:06:38,080 --> 00:06:41,800
Speaker 6: and very often you're missing out on labor, and we

133
00:06:41,920 --> 00:06:44,880
Speaker 6: talk about on skilled labor, it's hard to find Number one.

134
00:06:44,920 --> 00:06:49,320
Speaker 6: Number two is once you build it autonomous and automated,

135
00:06:49,600 --> 00:06:51,880
Speaker 6: then you have a much much higher yield, but you

136
00:06:52,160 --> 00:06:54,560
Speaker 6: can generate you use lesser energy. By the way, at

137
00:06:54,600 --> 00:06:56,840
Speaker 6: the same time, before you optimize it in a way.

138
00:06:57,160 --> 00:06:59,760
Speaker 6: So therefore there's a lot of benefits. And if you

139
00:06:59,800 --> 00:07:03,080
Speaker 6: want to talking about the United States up manufacturing the

140
00:07:03,120 --> 00:07:06,440
Speaker 6: United States, you need to go as digital and as

141
00:07:06,560 --> 00:07:09,400
Speaker 6: automated and AI super changed as possible.

142
00:07:10,120 --> 00:07:16,320
Speaker 8: A factory is robotic and it is orchestrating robots that

143
00:07:16,360 --> 00:07:20,000
Speaker 8: are building systems that are also robotic, like for example,

144
00:07:20,040 --> 00:07:21,520
Speaker 8: self draging cars a robotic system.

145
00:07:22,240 --> 00:07:24,480
Speaker 5: And the reason why it's.

146
00:07:24,320 --> 00:07:27,160
Speaker 1: So hard to deploy robots today is because it's hard

147
00:07:27,160 --> 00:07:31,960
Speaker 1: to program these robotic systems. The software expertise that necessary

148
00:07:32,000 --> 00:07:34,440
Speaker 1: to customization necessary is really intense.

149
00:07:34,480 --> 00:07:36,119
Speaker 5: It's just too much.

150
00:07:36,760 --> 00:07:39,200
Speaker 1: And so the fact that we could now apply artificial

151
00:07:39,240 --> 00:07:44,520
Speaker 1: intelligence physical AI technology to these robotic systems make them

152
00:07:44,680 --> 00:07:49,080
Speaker 1: easier to teach. You show it a few demonstrations, and

153
00:07:49,080 --> 00:07:50,560
Speaker 1: the AI learned it by itself.

154
00:07:50,960 --> 00:07:52,880
Speaker 4: You think, would you guys are, say Jenson, that you've

155
00:07:52,880 --> 00:07:54,920
Speaker 4: solved for that software limitation.

156
00:07:55,440 --> 00:07:58,920
Speaker 1: That software limitation is the chat GPT moment of it

157
00:07:58,960 --> 00:08:01,400
Speaker 1: is now here. I think over the course of the

158
00:08:01,480 --> 00:08:03,480
Speaker 1: next couple two three years, we're going to make some

159
00:08:03,640 --> 00:08:04,640
Speaker 1: really big breakthroughs.

160
00:08:05,160 --> 00:08:08,480
Speaker 4: Let's please talk about energy, electricity, power supply, call it

161
00:08:08,520 --> 00:08:11,400
Speaker 4: what you will in turn. For each of you, how

162
00:08:11,400 --> 00:08:14,080
Speaker 4: worried are you about it as a bottleneck and what

163
00:08:14,160 --> 00:08:16,560
Speaker 4: is your experience day stay in running both companies.

164
00:08:17,160 --> 00:08:19,760
Speaker 1: In that respect, energy should always be a bottleneck for

165
00:08:19,800 --> 00:08:23,080
Speaker 1: any industry, and this is a new industry that's growing

166
00:08:23,200 --> 00:08:24,280
Speaker 1: incredibly fast.

167
00:08:24,640 --> 00:08:25,120
Speaker 5: As you know.

168
00:08:25,760 --> 00:08:28,960
Speaker 1: AI is both the technology that's going to revolutionize many applications,

169
00:08:29,000 --> 00:08:31,840
Speaker 1: and we're talking about some of them here, but the

170
00:08:31,880 --> 00:08:37,079
Speaker 1: AI industry itself, the manufacturing of the artificial intelligence takes energy.

171
00:08:37,400 --> 00:08:40,480
Speaker 1: It takes energy, it takes AI factories. It's exactly the

172
00:08:40,520 --> 00:08:45,280
Speaker 1: reason why from Hopper to Blackwell we increased energy efficiency

173
00:08:46,040 --> 00:08:50,960
Speaker 1: by tenx. From Blackwell to Reubin, we increased energy efficiency

174
00:08:51,040 --> 00:08:55,720
Speaker 1: again by tenx. And that translates directly to our customer's

175
00:08:55,800 --> 00:09:00,760
Speaker 1: revenues because in the case of an AI factory, whatever

176
00:09:01,440 --> 00:09:04,600
Speaker 1: factory size you have, you're limited by the power, and

177
00:09:04,640 --> 00:09:07,480
Speaker 1: that power within that power constrained, you want to have

178
00:09:07,520 --> 00:09:11,160
Speaker 1: the most tokens or most AI per want that you

179
00:09:11,200 --> 00:09:15,319
Speaker 1: can possibly generate. And so every time we improve energy efficiency,

180
00:09:15,520 --> 00:09:20,040
Speaker 1: we're effectively improving both the AI capabilities for our customers

181
00:09:20,440 --> 00:09:23,839
Speaker 1: and their revenues because they're always constrained by power.

182
00:09:24,000 --> 00:09:26,280
Speaker 4: The response to your keinot and forgive me rowan one minute.

183
00:09:26,440 --> 00:09:29,440
Speaker 4: Was you know vera rubin ten x through put really

184
00:09:29,480 --> 00:09:34,720
Speaker 4: important tokens generated at one tenth. But they'll still say, hey, Jensen,

185
00:09:34,920 --> 00:09:37,920
Speaker 4: what if the electricity is just not there, it's not available.

186
00:09:38,120 --> 00:09:41,040
Speaker 5: Well, there's electricity, they're never enough electricity.

187
00:09:41,640 --> 00:09:43,920
Speaker 4: That's actually what is the anxiety level for you both

188
00:09:43,960 --> 00:09:45,080
Speaker 4: about that issue.

189
00:09:45,640 --> 00:09:49,280
Speaker 1: There's always energy, they're never enough energy. And this is

190
00:09:49,600 --> 00:09:53,040
Speaker 1: every and every industrial revolution will be energy constrained.

191
00:09:53,080 --> 00:09:55,920
Speaker 5: And this industrial revolution is also energy constrained.

192
00:09:56,559 --> 00:09:59,200
Speaker 1: Now one of the most important things here speaking about

193
00:09:59,200 --> 00:10:03,000
Speaker 1: the United States, if not for President Trump's pro energy

194
00:10:03,040 --> 00:10:06,040
Speaker 1: growth agenda, we would have a very hard time growing

195
00:10:06,040 --> 00:10:08,920
Speaker 1: at all. In order for our new industry to emerge,

196
00:10:08,960 --> 00:10:11,920
Speaker 1: you need energy, and so I think it's safe to

197
00:10:11,960 --> 00:10:14,600
Speaker 1: say that we wish we had more energy in United States.

198
00:10:14,840 --> 00:10:17,439
Speaker 1: You wi should have more energy. I think the world

199
00:10:17,480 --> 00:10:19,680
Speaker 1: all wish we had more energy, and so we have

200
00:10:19,720 --> 00:10:22,480
Speaker 1: to invest in all sorts of different forms of energy.

201
00:10:22,720 --> 00:10:26,120
Speaker 1: But whatever energy you have, you have to make it

202
00:10:26,160 --> 00:10:28,360
Speaker 1: as energy efficient as possible. And that's one of the

203
00:10:28,400 --> 00:10:32,000
Speaker 1: reasons why both of us drive our technology roadmap so hard,

204
00:10:32,240 --> 00:10:35,360
Speaker 1: because every single new generation of technology is more energy

205
00:10:35,360 --> 00:10:36,320
Speaker 1: efficient than the last.

206
00:10:36,840 --> 00:10:37,959
Speaker 5: That's kind of the nature of it.

207
00:10:38,040 --> 00:10:40,120
Speaker 4: You are in the AI factory. I won't say data

208
00:10:40,120 --> 00:10:42,480
Speaker 4: center is data centers where you store data. That you

209
00:10:42,559 --> 00:10:44,360
Speaker 4: are in the AI factory supply chain.

210
00:10:44,760 --> 00:10:45,439
Speaker 3: You must see it.

211
00:10:45,520 --> 00:10:48,080
Speaker 6: So what we see is number one is you see

212
00:10:48,120 --> 00:10:52,840
Speaker 6: that the energy demands is roughly scaling with the GDP growth.

213
00:10:52,960 --> 00:10:56,400
Speaker 6: It's the coupling because executive of this effect that we

214
00:10:56,440 --> 00:10:59,400
Speaker 6: are providing more and more efficient technology, So that means

215
00:10:59,720 --> 00:11:03,280
Speaker 6: the link. But still it grows along with economic growth.

216
00:11:03,880 --> 00:11:06,400
Speaker 6: And then on top comes a demand for data centers.

217
00:11:06,480 --> 00:11:09,240
Speaker 6: They demand high quality of energy at the same time,

218
00:11:10,320 --> 00:11:13,679
Speaker 6: which in some cases creates bottlenecks. Is it not from

219
00:11:13,720 --> 00:11:16,599
Speaker 6: power generation? And the talk is it either renewables or

220
00:11:16,679 --> 00:11:19,600
Speaker 6: gas turbines. There's a huge bottleneck for gas turbines as

221
00:11:19,600 --> 00:11:22,400
Speaker 6: we speak. It goes all the way to high voltage transformers,

222
00:11:22,880 --> 00:11:26,480
Speaker 6: to medium voltage into switch technologies. So therefore you see

223
00:11:26,520 --> 00:11:29,400
Speaker 6: along the whole supply chain, obviously there's a huge demand

224
00:11:29,720 --> 00:11:32,360
Speaker 6: we are serving that. There's a reason why this business

225
00:11:32,360 --> 00:11:35,440
Speaker 6: of ours is growing extremely fast. We can keep up

226
00:11:35,440 --> 00:11:37,640
Speaker 6: with the demand of our customers. But in some cases

227
00:11:37,679 --> 00:11:39,559
Speaker 6: you might end up in bottlenecks if you keep on

228
00:11:39,679 --> 00:11:41,160
Speaker 6: going as fast as possible.

229
00:11:41,280 --> 00:11:42,520
Speaker 5: It's a regional dependence.

230
00:11:43,120 --> 00:11:46,280
Speaker 6: In some cases, where you have good programs, good policies,

231
00:11:46,640 --> 00:11:48,920
Speaker 6: you see that it's catching up along with the demand,

232
00:11:48,960 --> 00:11:51,560
Speaker 6: and others you might end up in a gap. But

233
00:11:51,640 --> 00:11:56,240
Speaker 6: you see the demand is higher than supplypply chain and we.

234
00:11:56,200 --> 00:11:59,800
Speaker 5: Are ramping up standing we're invested, which is investing ultimately

235
00:11:59,840 --> 00:12:00,880
Speaker 5: the condition we want.

236
00:12:01,440 --> 00:12:03,720
Speaker 1: We want a condition where the demand is very strong

237
00:12:03,760 --> 00:12:06,720
Speaker 1: because there's great demand for new technology.

238
00:12:07,080 --> 00:12:10,800
Speaker 4: Specific to you, Jensen, how severe is the memory bottleneck

239
00:12:10,920 --> 00:12:11,360
Speaker 4: right now?

240
00:12:12,480 --> 00:12:14,400
Speaker 5: Well, the memory bottleneck is severe.

241
00:12:14,920 --> 00:12:17,720
Speaker 1: But we're fortunate to have worked with all of you know,

242
00:12:17,760 --> 00:12:19,800
Speaker 1: first of all, and Video is the only company that

243
00:12:19,880 --> 00:12:23,400
Speaker 1: works with all three of the HBM suppliers, and all

244
00:12:23,520 --> 00:12:26,480
Speaker 1: three of them are major customers and major suppliers of us,

245
00:12:26,520 --> 00:12:30,560
Speaker 1: and so we had the good fortune of working with

246
00:12:30,559 --> 00:12:33,319
Speaker 1: them for a very long period time, and we've got

247
00:12:33,320 --> 00:12:34,160
Speaker 1: things all planned out.

248
00:12:34,440 --> 00:12:35,000
Speaker 5: It's gonna be ak.

249
00:12:35,240 --> 00:12:36,840
Speaker 4: You've been asked a lot about China in the last

250
00:12:36,840 --> 00:12:38,319
Speaker 4: twenty four hours. I'm not going to ask you a

251
00:12:38,360 --> 00:12:41,520
Speaker 4: slightly different question. What is the attitude of the Chinese

252
00:12:41,520 --> 00:12:44,800
Speaker 4: government in allowing h two hundred to go into the country.

253
00:12:44,880 --> 00:12:48,199
Speaker 4: I understand the position on licenses, the administration here in

254
00:12:48,240 --> 00:12:51,439
Speaker 4: the United States position, but what does China's government say

255
00:12:51,480 --> 00:12:53,120
Speaker 4: to you about how they want to approach it.

256
00:12:53,760 --> 00:12:56,360
Speaker 5: I haven't spoken directly to them, but.

257
00:12:57,880 --> 00:13:01,640
Speaker 1: Ultimately the way they'll commune ok through us to us,

258
00:13:01,720 --> 00:13:05,000
Speaker 1: we'll be through the companies. If the companies are allowed

259
00:13:05,040 --> 00:13:08,080
Speaker 1: to buy in video products in China, then they'll be

260
00:13:08,120 --> 00:13:10,599
Speaker 1: strong demand. And we're seeing strong demand, and so I

261
00:13:10,640 --> 00:13:14,440
Speaker 1: think indirectly they've communicated with their companies.

262
00:13:15,520 --> 00:13:19,679
Speaker 4: I've learned a lot about Semens's work in software, and

263
00:13:19,760 --> 00:13:23,400
Speaker 4: I've always thought the word Semens to be snall us

264
00:13:23,440 --> 00:13:27,800
Speaker 4: with industrial manufacturing, physical things. It's the same lesson we've

265
00:13:27,800 --> 00:13:30,480
Speaker 4: been through recently with in video, trying to understand.

266
00:13:30,160 --> 00:13:32,559
Speaker 3: Actually the software competencies is.

267
00:13:32,480 --> 00:13:35,000
Speaker 4: That an area where we might see you be active

268
00:13:35,040 --> 00:13:37,440
Speaker 4: in m and a think about what you else you

269
00:13:37,520 --> 00:13:39,520
Speaker 4: might need that you might not get from the Jensen

270
00:13:40,280 --> 00:13:41,319
Speaker 4: in video partnership.

271
00:13:41,840 --> 00:13:46,280
Speaker 6: So you might because we will increase our software competence

272
00:13:46,360 --> 00:13:50,360
Speaker 6: our software Tolio actually it's alibit short of thirty billion,

273
00:13:50,400 --> 00:13:53,800
Speaker 6: which Semens invested in MNA to below our software competence,

274
00:13:54,080 --> 00:13:57,960
Speaker 6: which brings us to the point that's missing today exactly

275
00:13:58,000 --> 00:14:02,040
Speaker 6: today we can build the most comprehensive physics based digital.

276
00:14:01,720 --> 00:14:03,480
Speaker 5: Twin of whatever you want to build.

277
00:14:04,559 --> 00:14:07,839
Speaker 6: Still, there are pieces and operational software software which runs

278
00:14:07,880 --> 00:14:11,280
Speaker 6: plans where you can look into that I gives new

279
00:14:11,320 --> 00:14:14,359
Speaker 6: spaces obviously be invested in dogmatics.

280
00:14:14,440 --> 00:14:18,240
Speaker 5: It's life science, so it's about molecules. We can imagine

281
00:14:18,280 --> 00:14:21,680
Speaker 5: doing more there also in simulation. But as we're building.

282
00:14:22,920 --> 00:14:26,000
Speaker 6: A data backbound for this life science industry super elevent.

283
00:14:26,040 --> 00:14:29,240
Speaker 6: We did it for other industries with Team Center. We

284
00:14:29,360 --> 00:14:31,640
Speaker 6: repeat it now with Luma for life science, which is

285
00:14:32,040 --> 00:14:35,080
Speaker 6: again shortening cightful times and reducing costs for life science

286
00:14:35,080 --> 00:14:38,880
Speaker 6: and drugs and medicines. So and I mean the beauty

287
00:14:38,880 --> 00:14:41,160
Speaker 6: of it is that we live in both worlds. We

288
00:14:41,320 --> 00:14:43,480
Speaker 6: have to domain know how we know how to run things,

289
00:14:43,520 --> 00:14:48,200
Speaker 6: how to operate, how to automate plans, buildings, grades, trains,

290
00:14:49,200 --> 00:14:52,280
Speaker 6: and we have their software competence too, and that makes

291
00:14:52,320 --> 00:14:54,840
Speaker 6: us in a unique position. Then and then we need

292
00:14:54,840 --> 00:14:59,440
Speaker 6: strong partners which have complimentary technology, which is so amazing.

293
00:15:00,040 --> 00:15:02,400
Speaker 6: If that comes together, you'll see matching things happen well.

294
00:15:02,440 --> 00:15:04,760
Speaker 4: In video will have access to new technology on the

295
00:15:04,800 --> 00:15:08,400
Speaker 4: infant side through GROC. I'm going to do my classic DNSEN.

296
00:15:08,440 --> 00:15:11,680
Speaker 4: Could you please clarify for me question, which is is

297
00:15:11,720 --> 00:15:14,720
Speaker 4: this an acquisition or is it a licensing deal spread

298
00:15:14,760 --> 00:15:17,880
Speaker 4: over time. Because Jonathan and Sunny joined.

299
00:15:18,120 --> 00:15:21,240
Speaker 1: We hired four hundred or something like that a little

300
00:15:21,240 --> 00:15:28,320
Speaker 1: bit less incredible engineers, and we also license their technology.

301
00:15:27,680 --> 00:15:33,200
Speaker 1: They designed an architecture that's very, very different than what

302
00:15:33,240 --> 00:15:38,000
Speaker 1: we've done, and it's focused on low latency token generations

303
00:15:39,200 --> 00:15:42,440
Speaker 1: and video is incredibly good at inference, and we're great

304
00:15:42,480 --> 00:15:45,880
Speaker 1: at training post training as well as the inference phase

305
00:15:46,400 --> 00:15:50,680
Speaker 1: and test time scaling of AI. And so we've got

306
00:15:50,400 --> 00:15:54,720
Speaker 1: that space covered. I'm excited about some of the work

307
00:15:54,720 --> 00:15:56,400
Speaker 1: that we might be able to do together to invent

308
00:15:56,480 --> 00:16:00,000
Speaker 1: a new segment that might be able to address future use.

309
00:16:01,240 --> 00:16:03,600
Speaker 1: I haven't described it to people, but the time will come.

310
00:16:03,760 --> 00:16:07,240
Speaker 4: So you know, Jonathan is kind of behind the TPU

311
00:16:07,360 --> 00:16:10,560
Speaker 4: and the LPU, and you know, before of this happened,

312
00:16:10,560 --> 00:16:14,360
Speaker 4: there was Rock V two and V three and you

313
00:16:14,440 --> 00:16:16,280
Speaker 4: know all of that, and it's leading to me to ask,

314
00:16:16,480 --> 00:16:18,480
Speaker 4: what would that new platform or segment be.

315
00:16:18,840 --> 00:16:20,920
Speaker 3: What is it that you're trying to build on.

316
00:16:20,920 --> 00:16:23,680
Speaker 5: With those I haven't told anybody yet, but I'm here, Jens.

317
00:16:23,840 --> 00:16:24,480
Speaker 5: You can't tell me.

318
00:16:24,760 --> 00:16:26,080
Speaker 1: You know, I'm going to look in the camera and

319
00:16:26,120 --> 00:16:29,840
Speaker 1: tell you that that come to a GtC conference in

320
00:16:29,840 --> 00:16:32,600
Speaker 1: the future, and I will tell you all the secrets.

321
00:16:33,480 --> 00:16:35,760
Speaker 4: I'm going to make a hard pivot because I have

322
00:16:35,840 --> 00:16:42,920
Speaker 4: to data centers in space I reported just before the holiday.

323
00:16:43,000 --> 00:16:46,840
Speaker 4: Is that the reason SpaceX wants to go public, Bear

324
00:16:46,880 --> 00:16:49,160
Speaker 4: with me, Bear with me. The reason that SpaceX wants

325
00:16:49,160 --> 00:16:51,880
Speaker 4: to raise thirty forty billion dollars whatever that is needs

326
00:16:51,880 --> 00:16:55,920
Speaker 4: to buy the GPUs. Have you discussed data centers in

327
00:16:55,960 --> 00:16:58,560
Speaker 4: space with Elon and SpaceX, Jensen.

328
00:16:58,520 --> 00:17:01,520
Speaker 1: I can't discuss what I have told everybody. I've discussed

329
00:17:01,600 --> 00:17:02,280
Speaker 1: with anybody.

330
00:17:03,280 --> 00:17:06,320
Speaker 4: You think it's a viable technology platform, sure?

331
00:17:06,680 --> 00:17:09,360
Speaker 1: Sure, I mean there's lots of energy in space, right

332
00:17:09,440 --> 00:17:14,639
Speaker 1: and and the cooling is abundant in space, and so

333
00:17:14,760 --> 00:17:21,760
Speaker 1: that the challenges of AI factories are different out in space.

334
00:17:21,760 --> 00:17:24,639
Speaker 4: It would be AI factories in space. Yeah, forgive the

335
00:17:24,720 --> 00:17:27,280
Speaker 4: ignorance almost, but are we literally talking about the same

336
00:17:28,040 --> 00:17:30,880
Speaker 4: architecture for the GPU that goes into an AI factory

337
00:17:31,359 --> 00:17:32,919
Speaker 4: still just being able to go into kind of a

338
00:17:32,960 --> 00:17:34,160
Speaker 4: satellite form factor?

339
00:17:34,440 --> 00:17:34,640
Speaker 5: Yeah?

340
00:17:34,720 --> 00:17:37,200
Speaker 1: Sure, but the way that you would cool it empower

341
00:17:37,240 --> 00:17:40,320
Speaker 1: it would be very different, and so the system design

342
00:17:40,359 --> 00:17:41,439
Speaker 1: will be radically different.

343
00:17:41,840 --> 00:17:42,920
Speaker 5: The chips will be the same.

344
00:17:43,200 --> 00:17:45,800
Speaker 4: You're pretty concerned with data centers or WAI factories here

345
00:17:45,840 --> 00:17:49,000
Speaker 4: on Earth to seem and see the need for or

346
00:17:49,119 --> 00:17:50,879
Speaker 4: viability of such a technology.

347
00:17:51,480 --> 00:17:54,240
Speaker 6: I don't know whether we explored or potentials on Earth yet,

348
00:17:54,600 --> 00:17:58,240
Speaker 6: but there's one beauty in this idea. Think about any

349
00:17:58,320 --> 00:18:01,760
Speaker 6: kind of manufacturing, which want you want to bring to space.

350
00:18:02,560 --> 00:18:05,440
Speaker 6: Everything but you produce you want to bring down to Earth.

351
00:18:05,760 --> 00:18:08,960
Speaker 6: It's hard if you do energy, how to bring it

352
00:18:09,000 --> 00:18:11,000
Speaker 6: down to Earth. It's hard if you produce any kind

353
00:18:11,000 --> 00:18:14,600
Speaker 6: of hardware how to bring it down. But tokens intelligence,

354
00:18:14,760 --> 00:18:17,280
Speaker 6: I mean, you can transfer easily to Earth. So therefore,

355
00:18:17,520 --> 00:18:20,880
Speaker 6: if I would starting produce something in space, I would

356
00:18:20,880 --> 00:18:21,360
Speaker 6: start there.

357
00:18:22,640 --> 00:18:24,359
Speaker 3: We should talk about what's on Homas driving.

358
00:18:24,840 --> 00:18:28,720
Speaker 4: Have you seen Elon Musk's response to your keynote yesterday?

359
00:18:29,520 --> 00:18:32,240
Speaker 3: What do you say part one in short ways, Well,

360
00:18:32,520 --> 00:18:33,080
Speaker 3: we were.

361
00:18:32,920 --> 00:18:35,199
Speaker 4: Already doing that smiley emoji on.

362
00:18:35,440 --> 00:18:38,159
Speaker 3: X the the other part of it, by.

363
00:18:38,000 --> 00:18:39,200
Speaker 5: The way, I would be surprised.

364
00:18:40,760 --> 00:18:43,080
Speaker 1: First of all, I think I think the Tesla stack

365
00:18:43,240 --> 00:18:46,200
Speaker 1: is the most advanced AV stack in the world, and

366
00:18:48,520 --> 00:18:51,120
Speaker 1: I think the Tesla AV operations is the most advanced

367
00:18:51,160 --> 00:18:57,720
Speaker 1: in the world. And I'm fairly certain that that they

368
00:18:57,880 --> 00:19:01,880
Speaker 1: were already using n ai yes, and whether whether their

369
00:19:02,160 --> 00:19:05,800
Speaker 1: their AI also did reasoning or not, it is somewhat

370
00:19:05,840 --> 00:19:07,840
Speaker 1: secondary to that first part.

371
00:19:08,440 --> 00:19:11,000
Speaker 4: His point was that the first ninety nine is hard enough,

372
00:19:11,040 --> 00:19:13,200
Speaker 4: but the long tail thereafter is important.

373
00:19:13,200 --> 00:19:14,600
Speaker 3: You know, it's very difficult.

374
00:19:14,760 --> 00:19:16,679
Speaker 4: The question I got from the audience that was at

375
00:19:16,680 --> 00:19:19,840
Speaker 4: your keynote was on a dollar per mile basis, what

376
00:19:19,960 --> 00:19:22,520
Speaker 4: is the fundamental difference on your stack and your software

377
00:19:22,560 --> 00:19:25,120
Speaker 4: approach to Tesla's vision based approach?

378
00:19:28,359 --> 00:19:29,760
Speaker 5: Ours is also vision based.

379
00:19:30,200 --> 00:19:32,280
Speaker 1: You know, of course we have we have in addition

380
00:19:32,400 --> 00:19:34,600
Speaker 1: to to vision, we also have radar and light r

381
00:19:35,880 --> 00:19:39,920
Speaker 1: but but otherwise otherwise the approach is rather similar.

382
00:19:40,640 --> 00:19:41,000
Speaker 5: I think.

383
00:19:41,160 --> 00:19:44,280
Speaker 1: I think Elon's approach is about as state of the

384
00:19:44,440 --> 00:19:50,119
Speaker 1: art as anybody knows of a tunnel of striving robotics,

385
00:19:50,200 --> 00:19:52,880
Speaker 1: and so it's it's a it's a stack that's hard

386
00:19:52,880 --> 00:19:53,440
Speaker 1: to criticize.

387
00:19:53,600 --> 00:19:54,520
Speaker 5: I wouldn't criticize it.

388
00:19:54,520 --> 00:19:57,720
Speaker 1: I would just encourage them to continue to do what

389
00:19:57,760 --> 00:19:58,160
Speaker 1: they're doing.

390
00:19:58,200 --> 00:19:58,960
Speaker 5: They're doing a great.

391
00:19:58,880 --> 00:20:02,199
Speaker 4: Job physically for you is and forgive my pronunciation, but

392
00:20:02,359 --> 00:20:07,159
Speaker 4: er langan manufacturing site. When's that real? It's the same question.

393
00:20:07,760 --> 00:20:11,040
Speaker 6: I mean, in contrast to the side you was talking

394
00:20:11,040 --> 00:20:13,520
Speaker 6: about before this one exists. I mean, we are we

395
00:20:13,560 --> 00:20:15,760
Speaker 6: are ramping up, we are automating it as we speak,

396
00:20:15,920 --> 00:20:19,399
Speaker 6: using technology. I showcase some of that, and we are

397
00:20:19,440 --> 00:20:21,520
Speaker 6: we are really deploying step by step now on that

398
00:20:21,680 --> 00:20:25,520
Speaker 6: very same side, this technology we was talking about, the iebrain,

399
00:20:25,640 --> 00:20:30,879
Speaker 6: the digital trim composer, real worlds and really time connectivity

400
00:20:31,000 --> 00:20:33,640
Speaker 6: to what happens on the shop floor, so you can

401
00:20:33,720 --> 00:20:37,720
Speaker 6: act and drive basically dynamically what's happened on the shop floor.

402
00:20:38,040 --> 00:20:41,640
Speaker 6: So and we will talk about more along two thousand.

403
00:20:41,320 --> 00:20:45,639
Speaker 1: And there Semens using technology in the Semens factory, Epion

404
00:20:45,760 --> 00:20:48,359
Speaker 1: is using Semens technology in our funds find and Rea factory,

405
00:20:48,880 --> 00:20:51,480
Speaker 1: and so you know, it's a great partnership.

406
00:20:51,800 --> 00:20:55,120
Speaker 4: Jensen, you are one of the most important and biggest

407
00:20:55,119 --> 00:20:58,520
Speaker 4: employers in Silicon Valley. Have been down to Santa Clarency.

408
00:20:58,720 --> 00:21:01,960
Speaker 4: You are the leader of the technology industry. Right now,

409
00:21:02,480 --> 00:21:04,840
Speaker 4: the industry and those of us that live in California

410
00:21:04,920 --> 00:21:08,439
Speaker 4: are reviewing the billionaires tax. The question that a lot

411
00:21:08,480 --> 00:21:11,640
Speaker 4: of people submitted to me to ask you is, how

412
00:21:11,680 --> 00:21:14,879
Speaker 4: does that impact that talent pool and the industry and

413
00:21:14,960 --> 00:21:15,720
Speaker 4: Silicon Valley.

414
00:21:15,800 --> 00:21:18,639
Speaker 3: Is it something that's concerned to you or it is not.

415
00:21:20,440 --> 00:21:21,840
Speaker 5: I haven't thought about it even once.

416
00:21:23,359 --> 00:21:26,960
Speaker 1: We work in Silicon Valley because because that's where the

417
00:21:27,000 --> 00:21:31,359
Speaker 1: talent pool is, and and we have offices all over

418
00:21:31,400 --> 00:21:34,439
Speaker 1: the world. Wherever there's talent, we have offices, we have office.

419
00:21:34,480 --> 00:21:37,080
Speaker 1: Some in Germany we have office, and you know, all

420
00:21:37,119 --> 00:21:40,720
Speaker 1: over the world. And so so we chose to live

421
00:21:40,720 --> 00:21:47,040
Speaker 1: in Silicon Valley and and what whatever taxes I guess

422
00:21:47,480 --> 00:21:48,119
Speaker 1: they would like to.

423
00:21:49,840 --> 00:21:51,399
Speaker 5: Apply, sy be it.

424
00:21:51,560 --> 00:21:54,360
Speaker 1: Maybe I'm perfectly fine with it.

425
00:21:54,359 --> 00:21:55,800
Speaker 5: It didn't. It never crossed my mind once.

426
00:21:55,960 --> 00:21:56,880
Speaker 3: I appreciate the answer.

427
00:21:56,920 --> 00:21:59,880
Speaker 4: And again, you know it's right now in the technology industry,

428
00:22:00,040 --> 00:22:01,359
Speaker 4: that's what people are reflecting on.

429
00:22:01,960 --> 00:22:04,359
Speaker 3: We've had it really not this person, not this person.

430
00:22:04,560 --> 00:22:06,320
Speaker 5: This person's trying to build the future of AI.

431
00:22:06,960 --> 00:22:08,560
Speaker 3: And that's what the conversation was about today.

432
00:22:08,880 --> 00:22:12,320
Speaker 4: Roland Bush, Siemens CEO, Jensen Wong and Video CEO