1
00:00:02,480 --> 00:00:07,000
Speaker 1: Bloomberg Audio Studios, Podcasts, radio News.

2
00:00:07,800 --> 00:00:13,239
Speaker 2: Welcome back to Bloomberg Tech Lisa Sue Helios with m

3
00:00:13,320 --> 00:00:17,759
Speaker 2: I four to fifty five x AMD's first RAX scale

4
00:00:17,800 --> 00:00:21,600
Speaker 2: system solution. But inside it AMD's first in the world's

5
00:00:21,600 --> 00:00:23,759
Speaker 2: first two nanimeter.

6
00:00:23,800 --> 00:00:25,000
Speaker 3: Chip of that type.

7
00:00:25,560 --> 00:00:27,440
Speaker 2: A lot was made of it when you actually just

8
00:00:27,440 --> 00:00:29,319
Speaker 2: stood on stage and held it in your hand for

9
00:00:29,360 --> 00:00:30,000
Speaker 2: the first time.

10
00:00:30,320 --> 00:00:31,400
Speaker 3: Why is it significant?

11
00:00:31,520 --> 00:00:32,000
Speaker 4: Well, first of.

12
00:00:31,960 --> 00:00:34,360
Speaker 1: All, Ed, it's great to be here with you at CS.

13
00:00:34,440 --> 00:00:36,360
Speaker 1: I think CS is always a great way to kick

14
00:00:36,360 --> 00:00:38,800
Speaker 1: off the year because you get so much perspective. So

15
00:00:38,920 --> 00:00:42,360
Speaker 1: it was fun giving the keynote last night. Look, Helios is.

16
00:00:42,280 --> 00:00:44,760
Speaker 4: A massive system, you can see it in the.

17
00:00:44,680 --> 00:00:48,080
Speaker 1: Background here, and m I four fifty five is just

18
00:00:48,400 --> 00:00:52,479
Speaker 1: an incredibly powerful chip. And probably the context I would

19
00:00:52,680 --> 00:00:55,360
Speaker 1: give Ed is, you know, one of the things that

20
00:00:55,840 --> 00:00:58,920
Speaker 1: we're so clear about is that the demand for AI

21
00:00:59,000 --> 00:01:02,280
Speaker 1: compute is just continuing to increase. And you know, we

22
00:01:02,360 --> 00:01:04,280
Speaker 1: have seen that over the last five years. When you

23
00:01:04,319 --> 00:01:07,760
Speaker 1: think about just how much you know, new capabilities have

24
00:01:07,840 --> 00:01:11,240
Speaker 1: come on board, we've now seen a real inflection in

25
00:01:11,280 --> 00:01:13,800
Speaker 1: the number of people who are using AI. So if

26
00:01:13,840 --> 00:01:16,440
Speaker 1: you think today they're probably more than a billion active

27
00:01:16,520 --> 00:01:19,960
Speaker 1: users using AI and we expect that to scale to

28
00:01:20,160 --> 00:01:22,640
Speaker 1: over five billion users over the next five years. So

29
00:01:22,680 --> 00:01:25,160
Speaker 1: for all of that, you need compute and lots and

30
00:01:25,200 --> 00:01:28,080
Speaker 1: lots of compute. And from that standpoint, you know, m

31
00:01:28,120 --> 00:01:33,000
Speaker 1: I four fifty five is a significant leap forward in

32
00:01:33,080 --> 00:01:36,160
Speaker 1: terms of technology capability, made up of two and three

33
00:01:36,240 --> 00:01:40,440
Speaker 1: nanometer trips, three hundred and twenty billion transistors, just a

34
00:01:40,480 --> 00:01:41,840
Speaker 1: lot of performance.

35
00:01:41,360 --> 00:01:43,039
Speaker 2: And a lot of the timeline for it to be

36
00:01:43,120 --> 00:01:45,440
Speaker 2: deployed in the real world, and then who will be

37
00:01:45,480 --> 00:01:46,920
Speaker 2: the principal first user of it.

38
00:01:47,080 --> 00:01:48,800
Speaker 1: You'll see it in the second half of twenty six

39
00:01:49,040 --> 00:01:51,280
Speaker 1: and it will ramp from there. And you know, we

40
00:01:51,400 --> 00:01:55,400
Speaker 1: have very strong partnerships open AI. Greg Brockman was on

41
00:01:55,440 --> 00:01:58,880
Speaker 1: stage with us last night talking about all.

42
00:01:58,840 --> 00:02:00,520
Speaker 4: The use cases that they see.

43
00:02:00,920 --> 00:02:04,360
Speaker 1: We've announced a partnership with the Oracle, many others as well.

44
00:02:04,480 --> 00:02:07,920
Speaker 2: So given that it's two h it's in full production now,

45
00:02:07,960 --> 00:02:09,600
Speaker 2: it's getting ready to are.

46
00:02:09,680 --> 00:02:11,920
Speaker 4: We are absolutely getting ready to ship it.

47
00:02:12,360 --> 00:02:15,760
Speaker 2: That's at one end of the sort of scale and spectrum.

48
00:02:16,080 --> 00:02:18,760
Speaker 2: At the other you have m I four forty x,

49
00:02:18,840 --> 00:02:22,160
Speaker 2: which is for small data centers. I'm trying to simplify it,

50
00:02:22,160 --> 00:02:26,320
Speaker 2: but it's basically an enterprise product. What was it that

51
00:02:26,360 --> 00:02:28,400
Speaker 2: you were trying to solve for with that?

52
00:02:28,840 --> 00:02:31,240
Speaker 1: Yeah, I think what we're trying to solve for is,

53
00:02:31,520 --> 00:02:34,080
Speaker 1: you know, the world is a very heterogeneous world.

54
00:02:34,760 --> 00:02:36,640
Speaker 4: You have all kinds of use.

55
00:02:36,480 --> 00:02:39,200
Speaker 1: Cases for AI from you know, sort of the very

56
00:02:39,240 --> 00:02:42,240
Speaker 1: biggest cloud data centers that are doing you know, large

57
00:02:42,240 --> 00:02:47,440
Speaker 1: scale training and inference to enterprise applications as well as supercomputers,

58
00:02:47,720 --> 00:02:49,639
Speaker 1: and so we actually have a family of chips.

59
00:02:49,919 --> 00:02:51,359
Speaker 4: At the highest end, is there.

60
00:02:51,400 --> 00:02:53,960
Speaker 1: M I four fifty five for the cloud environment, But

61
00:02:54,040 --> 00:02:57,960
Speaker 1: we announced last night a m I four forty which

62
00:02:58,000 --> 00:03:01,560
Speaker 1: is actually using the same basic building blocks, but is

63
00:03:01,639 --> 00:03:05,240
Speaker 1: now really focused on enterprise applications so that you can

64
00:03:05,280 --> 00:03:07,840
Speaker 1: go into you know, let's call it current data centers

65
00:03:07,919 --> 00:03:09,080
Speaker 1: with the new technology.

66
00:03:09,320 --> 00:03:10,880
Speaker 4: So we're excited about that as well.

67
00:03:11,440 --> 00:03:14,400
Speaker 1: You know, there is enterprises are starting to increase their

68
00:03:14,440 --> 00:03:17,480
Speaker 1: adoption of AI. In some cases they want their own

69
00:03:17,800 --> 00:03:20,840
Speaker 1: control of their data centers in terms of on prem environments.

70
00:03:21,000 --> 00:03:22,200
Speaker 3: What are they doing with it though?

71
00:03:22,320 --> 00:03:25,960
Speaker 2: I mean, you know, we've been so fixated on frontier

72
00:03:26,000 --> 00:03:29,200
Speaker 2: models with hundreds of billions of parameters and the scale

73
00:03:29,200 --> 00:03:32,200
Speaker 2: of infrastructure needed for that with MI I F forty,

74
00:03:32,480 --> 00:03:34,519
Speaker 2: we're talking about something slightly different. I think it's just

75
00:03:34,560 --> 00:03:36,960
Speaker 2: really interesting if you could explain what the demand is

76
00:03:37,480 --> 00:03:40,280
Speaker 2: from those enterprises, what they want with the technology.

77
00:03:40,400 --> 00:03:43,960
Speaker 1: Well, I think you see many enterprises now using AI

78
00:03:44,680 --> 00:03:49,960
Speaker 1: all throughout their business processes, whether you're talking about things

79
00:03:49,960 --> 00:03:54,560
Speaker 1: in their workflow. Even AMD we're using AI through every

80
00:03:54,640 --> 00:03:57,840
Speaker 1: part of our development process. We're seeing a lot of

81
00:03:57,840 --> 00:04:02,280
Speaker 1: applications in financial services and healthcare, and these are areas,

82
00:04:02,360 --> 00:04:06,000
Speaker 1: especially in financial services, that people actually don't want everything

83
00:04:06,080 --> 00:04:08,200
Speaker 1: necessarily in the cloud. They'd like to be able to

84
00:04:08,240 --> 00:04:13,280
Speaker 1: have their own on prem deployment or private cloud deployments.

85
00:04:13,560 --> 00:04:15,440
Speaker 1: And in this case, you don't want to have to

86
00:04:15,480 --> 00:04:18,320
Speaker 1: build a brand new data center for every new generation

87
00:04:18,400 --> 00:04:21,080
Speaker 1: of chip. M four point forty allows us to use

88
00:04:21,120 --> 00:04:24,680
Speaker 1: some of those existing data centers in upgrade with the

89
00:04:24,680 --> 00:04:25,680
Speaker 1: new capabilities.

90
00:04:26,160 --> 00:04:26,600
Speaker 3: Welcome.

91
00:04:26,760 --> 00:04:29,360
Speaker 2: If you're watching us on Bloomberg Television or you're listening

92
00:04:29,400 --> 00:04:32,240
Speaker 2: on Bloomberg Radio. We're live in Las Vegas and we're

93
00:04:32,240 --> 00:04:36,280
Speaker 2: with AMD CEO Lisa Sue, and we're talking about the

94
00:04:36,360 --> 00:04:42,000
Speaker 2: latest generation of accelerators. What makes this generation of AMD

95
00:04:42,120 --> 00:04:46,440
Speaker 2: accelerators the better option, particularly for on prem and at

96
00:04:46,480 --> 00:04:49,200
Speaker 2: the edge over what Nvidia is offering right now.

97
00:04:49,360 --> 00:04:52,200
Speaker 1: Well, the best way to think about it, ED is

98
00:04:52,839 --> 00:04:57,880
Speaker 1: we're in this place where AI is at a inflection point.

99
00:04:57,960 --> 00:05:01,360
Speaker 1: We're seeing AI now in every part of compute. We

100
00:05:01,400 --> 00:05:04,520
Speaker 1: see it in the largest models, you know, when you're

101
00:05:04,560 --> 00:05:07,279
Speaker 1: thinking about things like track, GPT and Gemini and Rock.

102
00:05:07,640 --> 00:05:10,240
Speaker 1: You know, we're also seeing you know, many use cases

103
00:05:10,440 --> 00:05:17,000
Speaker 1: in UH, you know, new capabilities like you know, video production, entertainment, healthcare,

104
00:05:17,160 --> 00:05:20,320
Speaker 1: where you're doing drug discovery, all of these various things.

105
00:05:20,480 --> 00:05:23,479
Speaker 1: You know, our claim to fame is really you know,

106
00:05:23,680 --> 00:05:28,080
Speaker 1: outstanding performance at you know, very advantage total cost of ownership.

107
00:05:28,360 --> 00:05:30,280
Speaker 1: And the other thing that you know we believe very

108
00:05:30,320 --> 00:05:34,560
Speaker 1: strongly in is an open ecosystem and deep partnerships, you know,

109
00:05:34,640 --> 00:05:37,400
Speaker 1: with our UH you know, with our overall.

110
00:05:37,240 --> 00:05:38,560
Speaker 4: Ecosystem coming together.

111
00:05:38,839 --> 00:05:41,560
Speaker 1: So when you put those things in perspective, I think

112
00:05:41,600 --> 00:05:46,640
Speaker 1: we have a great set of applications that will take

113
00:05:46,640 --> 00:05:48,600
Speaker 1: advantage of these newest generationships.

114
00:05:48,960 --> 00:05:51,920
Speaker 2: You mentioned that Greg Brockman, who's the Opening Eye president,

115
00:05:51,960 --> 00:05:53,960
Speaker 2: was on stage with you last night, and one of

116
00:05:53,800 --> 00:05:56,400
Speaker 2: the basic points that he made was there are tools

117
00:05:56,440 --> 00:05:59,600
Speaker 2: and functions they would love to release and put out

118
00:05:59,640 --> 00:06:02,960
Speaker 2: into the world, but they are compute constrained. I often

119
00:06:03,000 --> 00:06:05,960
Speaker 2: ask you to quantify demand, but is there a way

120
00:06:06,000 --> 00:06:10,000
Speaker 2: to quantify the severity in the lack of compute, you know,

121
00:06:10,080 --> 00:06:12,040
Speaker 2: the deficit that's out there right now.

122
00:06:12,120 --> 00:06:14,240
Speaker 1: Well, let me just give you some numbers to kind

123
00:06:14,279 --> 00:06:18,320
Speaker 1: of ground what we think the demand environment is looking like.

124
00:06:18,400 --> 00:06:21,719
Speaker 1: So if you think, you know, today we have about

125
00:06:21,720 --> 00:06:24,719
Speaker 1: a billion active users and we're ramping that to you know,

126
00:06:24,800 --> 00:06:28,120
Speaker 1: five billion over the next five years, and we have

127
00:06:28,200 --> 00:06:30,560
Speaker 1: about let's call it, one hundred zet of flops of.

128
00:06:30,560 --> 00:06:32,360
Speaker 4: Compute you know, all around the world.

129
00:06:32,440 --> 00:06:36,640
Speaker 1: And that's just a generic number that that aggregates all

130
00:06:36,640 --> 00:06:38,679
Speaker 1: of that. You know, we think we have to increase

131
00:06:38,720 --> 00:06:43,680
Speaker 1: compute by another one hundred times as you go over

132
00:06:43,720 --> 00:06:46,520
Speaker 1: the next you know, four or five years. And I

133
00:06:46,560 --> 00:06:50,440
Speaker 1: introduced a term last night, the YadA flop. You know,

134
00:06:50,440 --> 00:06:52,800
Speaker 1: people are like, what is a YadA flop? A YadA

135
00:06:52,800 --> 00:06:56,800
Speaker 1: flop is actually ten to the twenty fourth in terms

136
00:06:56,800 --> 00:06:57,159
Speaker 1: of flop.

137
00:06:57,200 --> 00:06:59,320
Speaker 4: So that's a one followed by twenty four zeros.

138
00:06:59,800 --> 00:07:02,880
Speaker 1: And to give you, you know, just a view of

139
00:07:02,920 --> 00:07:05,720
Speaker 1: just how much things have really increased. I mean, that's

140
00:07:05,760 --> 00:07:09,440
Speaker 1: another one hundred times more compute than we have today.

141
00:07:09,760 --> 00:07:11,840
Speaker 1: So that gives you an idea. Now you think, what

142
00:07:11,880 --> 00:07:13,680
Speaker 1: are you going to use all that compute for? I

143
00:07:13,720 --> 00:07:16,560
Speaker 1: mean the truth is, the models that we have today

144
00:07:16,640 --> 00:07:19,640
Speaker 1: are great. I mean they do amazing things. You know,

145
00:07:19,680 --> 00:07:22,240
Speaker 1: we talked about a number of use cases. Uh you know,

146
00:07:22,280 --> 00:07:24,600
Speaker 1: perhaps you know one that's you know, very hits very

147
00:07:24,600 --> 00:07:27,480
Speaker 1: close to home is is writing software. Like you know,

148
00:07:27,520 --> 00:07:30,960
Speaker 1: people are using uh, the AI tools right now to

149
00:07:31,160 --> 00:07:34,520
Speaker 1: significantly enhance the productivity of software developers.

150
00:07:34,880 --> 00:07:37,880
Speaker 4: But it's good, but it can get so much better.

151
00:07:38,000 --> 00:07:39,760
Speaker 4: And I mean, I think that's the key point.

152
00:07:40,040 --> 00:07:40,240
Speaker 2: You know.

153
00:07:40,240 --> 00:07:43,880
Speaker 1: We we like to say that AI is really going

154
00:07:43,920 --> 00:07:46,640
Speaker 1: to be everywhere, and it's really for everyone, and it's

155
00:07:46,640 --> 00:07:50,520
Speaker 1: for each one of us to make our businesses more productive,

156
00:07:50,720 --> 00:07:53,600
Speaker 1: you know, each one of us more productive, you know,

157
00:07:53,680 --> 00:07:56,080
Speaker 1: going forward. And so we're still in the very early

158
00:07:56,120 --> 00:07:58,680
Speaker 1: innings of really unlocking the power of AI.

159
00:07:58,840 --> 00:08:01,440
Speaker 2: So where we stand is we Okay, there's a compute

160
00:08:01,440 --> 00:08:04,280
Speaker 2: deficit and software has kind of hit the limits of

161
00:08:04,280 --> 00:08:05,600
Speaker 2: what current generation.

162
00:08:05,320 --> 00:08:06,760
Speaker 3: Compute can offer.

163
00:08:07,240 --> 00:08:10,920
Speaker 2: Help us understand the bottlenecks and barriers that to deploying

164
00:08:11,000 --> 00:08:13,920
Speaker 2: that compute a lot at the moment about memory chips,

165
00:08:14,640 --> 00:08:19,880
Speaker 2: what else, energy, electricity, what's crossing your desk, Lisa that

166
00:08:20,840 --> 00:08:24,360
Speaker 2: gives you pause and say this is a problem right now.

167
00:08:24,560 --> 00:08:28,440
Speaker 1: Well, our job as a technology industry is to push

168
00:08:28,480 --> 00:08:29,240
Speaker 1: the bleeding edge.

169
00:08:29,320 --> 00:08:30,280
Speaker 4: I mean, that is our job.

170
00:08:30,600 --> 00:08:32,760
Speaker 1: And so you know, when we think about like the

171
00:08:33,040 --> 00:08:35,640
Speaker 1: four fifty five deploying you know two nanimeter and three

172
00:08:35,640 --> 00:08:39,439
Speaker 1: and nanimeter chips, having the latest generation memory high bendw

173
00:08:39,440 --> 00:08:43,120
Speaker 1: with memory that is out there, and really deploying these

174
00:08:43,120 --> 00:08:47,120
Speaker 1: big systems, the important thing is that the entire ecosystem

175
00:08:47,200 --> 00:08:50,760
Speaker 1: come together and we plan together for this next big

176
00:08:50,800 --> 00:08:52,280
Speaker 1: inflection in compute.

177
00:08:52,400 --> 00:08:54,840
Speaker 4: And that's exactly what we're doing right now.

178
00:08:55,000 --> 00:08:57,520
Speaker 1: I think we're working very closely with the entire supply

179
00:08:57,600 --> 00:09:01,960
Speaker 1: chain to ensure that we have the the resources to

180
00:09:02,080 --> 00:09:06,480
Speaker 1: expand this compute environment. And yes, you know some of

181
00:09:06,520 --> 00:09:10,480
Speaker 1: the things that you mentioned are let's call it constrained, but.

182
00:09:10,480 --> 00:09:13,480
Speaker 3: Which is most severely so, you know, I don't.

183
00:09:13,200 --> 00:09:14,640
Speaker 4: Think that's any one thing.

184
00:09:14,760 --> 00:09:16,600
Speaker 1: I think we're all looking at, you know, how do

185
00:09:16,640 --> 00:09:20,559
Speaker 1: we build faster? You know, certainly power is one of

186
00:09:20,640 --> 00:09:23,000
Speaker 1: those areas where you know, you see throughout the world,

187
00:09:23,640 --> 00:09:26,920
Speaker 1: you know, power is being brought online as fast as possible,

188
00:09:27,520 --> 00:09:30,360
Speaker 1: certainly from a silicon standpoint. You know, we're ramping our

189
00:09:30,400 --> 00:09:33,920
Speaker 1: production capabilities with our partners. From a memory standpoint, our

190
00:09:33,960 --> 00:09:37,559
Speaker 1: partners ramping as well. So it's not any one thing,

191
00:09:37,640 --> 00:09:39,480
Speaker 1: I think, it's all of these things have to go

192
00:09:40,280 --> 00:09:43,120
Speaker 1: sort of in tandem, and that's why partnership is just

193
00:09:43,240 --> 00:09:45,439
Speaker 1: so important in this business.

194
00:09:45,760 --> 00:09:49,560
Speaker 2: We started this conversation talking about helios first, RAC scale

195
00:09:50,520 --> 00:09:54,840
Speaker 2: architecture and infrastructure from AMD, could you talk about the

196
00:09:54,880 --> 00:09:57,200
Speaker 2: future and how much of the content you want to

197
00:09:57,240 --> 00:09:58,319
Speaker 2: own in a server?

198
00:09:58,880 --> 00:10:02,240
Speaker 3: You know, we start this story with the.

199
00:10:02,200 --> 00:10:05,400
Speaker 2: GPU, Frank k If you look at what Nvidia is doing,

200
00:10:05,480 --> 00:10:08,480
Speaker 2: they want to increasingly own all of what's inside the box.

201
00:10:08,960 --> 00:10:10,920
Speaker 2: Is that something that AMD's focused on too.

202
00:10:11,160 --> 00:10:14,480
Speaker 1: You know, what's most important for us is to ensure

203
00:10:14,600 --> 00:10:18,480
Speaker 1: that we have turnkey solutions that are very very easy

204
00:10:18,520 --> 00:10:21,040
Speaker 1: for our customers to deploy, because when you think about,

205
00:10:21,320 --> 00:10:23,839
Speaker 1: you know, how do you use all of this AI

206
00:10:23,920 --> 00:10:25,320
Speaker 1: compute most effectively?

207
00:10:25,720 --> 00:10:27,680
Speaker 4: You want it to go into the data.

208
00:10:27,480 --> 00:10:30,080
Speaker 1: Center and really be up and running on day one,

209
00:10:30,160 --> 00:10:32,520
Speaker 1: and for that you have to optimize a full system.

210
00:10:32,880 --> 00:10:35,800
Speaker 1: But from that standpoint, you know, we are very focused

211
00:10:35,840 --> 00:10:39,280
Speaker 1: on an open ecosystem. So yes, we designed the CPUs

212
00:10:39,320 --> 00:10:41,720
Speaker 1: and the GPUs and some of the networking elements, but

213
00:10:41,760 --> 00:10:44,800
Speaker 1: we also work you know, really with the broad ecosystem

214
00:10:44,840 --> 00:10:49,360
Speaker 1: of partners with industry standards. It's all about ensuring that

215
00:10:49,720 --> 00:10:51,880
Speaker 1: we get the best of all worlds when we put

216
00:10:52,240 --> 00:10:53,280
Speaker 1: our solutions together.

217
00:10:53,960 --> 00:10:57,560
Speaker 2: Looking ahead to m I five hundred twenty twenty seven,

218
00:10:58,600 --> 00:11:02,560
Speaker 2: that has one thousand times the performance of the I

219
00:11:02,559 --> 00:11:05,719
Speaker 2: three hundred generation. So your last generation of real well

220
00:11:05,760 --> 00:11:10,160
Speaker 2: deployed gear. Something's coming that's a thousand times better. How

221
00:11:10,200 --> 00:11:12,079
Speaker 2: did you make it a thousand times better?

222
00:11:12,440 --> 00:11:16,920
Speaker 1: It is just incredible engineering at every level. So m

223
00:11:16,960 --> 00:11:20,400
Speaker 1: I four fifty five is ten times better than the

224
00:11:20,480 --> 00:11:23,400
Speaker 1: trip that we just launched six months ago them I

225
00:11:23,440 --> 00:11:26,679
Speaker 1: three fifty five and m I five hundred is another

226
00:11:26,880 --> 00:11:29,439
Speaker 1: ten x you know. On top of that, we are

227
00:11:29,520 --> 00:11:32,480
Speaker 1: using the most advanced technology out there. We have a

228
00:11:33,120 --> 00:11:37,800
Speaker 1: very you know, very clear focus on you know, hardware, software,

229
00:11:38,520 --> 00:11:42,640
Speaker 1: system code design, and it is you know, clearly the

230
00:11:42,840 --> 00:11:44,800
Speaker 1: pushing the bleeding edge of capabilities.

231
00:11:45,520 --> 00:11:48,840
Speaker 2: What is the status of a m d's ability to

232
00:11:48,960 --> 00:11:50,960
Speaker 2: sell products into China right now?

233
00:11:51,920 --> 00:11:54,240
Speaker 4: So you know, China is an important market for us.

234
00:11:54,679 --> 00:11:56,600
Speaker 1: You know, we actually sell a broad range of tips

235
00:11:56,600 --> 00:12:00,080
Speaker 1: into China, including our you know, our PCs as well

236
00:12:00,120 --> 00:12:03,160
Speaker 1: as you know other embedded chips.

237
00:12:02,559 --> 00:12:05,600
Speaker 3: In the data sets context of course, sorry, in the data.

238
00:12:05,400 --> 00:12:08,520
Speaker 1: Center context, we are you know, certainly we see China

239
00:12:08,559 --> 00:12:11,560
Speaker 1: as an important market. We were We did get some

240
00:12:11,640 --> 00:12:14,840
Speaker 1: licenses from the US government you know, late last year

241
00:12:15,080 --> 00:12:17,920
Speaker 1: as it relates to some of our previous generation are

242
00:12:18,080 --> 00:12:20,720
Speaker 1: m I three oh eight you know chips, and we

243
00:12:20,760 --> 00:12:24,000
Speaker 1: are in the process of applying for new licenses with

244
00:12:24,200 --> 00:12:30,440
Speaker 1: ourm I three twenty five chips that were recently allowed

245
00:12:30,559 --> 00:12:33,600
Speaker 1: to apply for licenses. We haven't gotten those licenses yet,

246
00:12:33,760 --> 00:12:36,520
Speaker 1: but we continue to view China as an important market.

247
00:12:36,559 --> 00:12:38,560
Speaker 2: For the reason I ask about it is in part

248
00:12:38,640 --> 00:12:41,200
Speaker 2: because a lot of the work that's being done in

249
00:12:41,240 --> 00:12:44,240
Speaker 2: open source models and bridging the gap between open and

250
00:12:44,280 --> 00:12:46,800
Speaker 2: closed it is being done in China to some extent.

251
00:12:47,880 --> 00:12:49,800
Speaker 2: There's been a lot of discussion about the demands being

252
00:12:49,840 --> 00:12:52,120
Speaker 2: there in China, But could you reflect a little bit

253
00:12:52,120 --> 00:12:55,520
Speaker 2: on that demand, but also what the Chinese government's attitude

254
00:12:55,600 --> 00:13:00,680
Speaker 2: is to you taking a later gen neration of tech

255
00:13:00,720 --> 00:13:01,319
Speaker 2: to the country.

256
00:13:01,679 --> 00:13:04,520
Speaker 1: Well, I do think the demand for you know, AI

257
00:13:04,600 --> 00:13:07,600
Speaker 1: in general and in China is high for all the

258
00:13:07,640 --> 00:13:09,480
Speaker 1: reasons that we talked about. I think we are in

259
00:13:09,520 --> 00:13:14,880
Speaker 1: a demand environment where more compute is beneficial across the world.

260
00:13:15,320 --> 00:13:17,600
Speaker 1: We think, you know, China is an important market for US,

261
00:13:17,720 --> 00:13:20,839
Speaker 1: and it's very active in having our solutions deployed, so

262
00:13:21,080 --> 00:13:23,520
Speaker 1: you know, we continue to view it as something that's important.

263
00:13:23,840 --> 00:13:25,520
Speaker 4: We're working with the US government as.

264
00:13:25,440 --> 00:13:28,240
Speaker 1: Well as our Chinese customers, you know, to find good

265
00:13:28,240 --> 00:13:29,280
Speaker 1: solutions there.

266
00:13:29,200 --> 00:13:31,840
Speaker 2: And there are signs from both governments that the licensed

267
00:13:31,840 --> 00:13:34,800
Speaker 2: process is moving. Commerce is kind of notorious for things

268
00:13:34,840 --> 00:13:36,480
Speaker 2: sitting on a desk for quite a long time.

269
00:13:36,559 --> 00:13:39,360
Speaker 1: I think we are optimistic that, you know, we'll have

270
00:13:39,400 --> 00:13:41,920
Speaker 1: an opportunity to get some of those licenses.

271
00:13:41,920 --> 00:13:45,840
Speaker 2: Granted you're watching Bloomberg Television, you're listening to Bloomberg Radio,

272
00:13:45,960 --> 00:13:48,320
Speaker 2: this is Bloomberg Tech, and we live in Las Vegas

273
00:13:48,679 --> 00:13:53,160
Speaker 2: with the A M D CEO, Lisa Sue. Last question,

274
00:13:53,200 --> 00:13:56,080
Speaker 2: really in the data center context, is the markets and

275
00:13:56,120 --> 00:14:00,360
Speaker 2: investors want data and signs that you're taking markets share.

276
00:14:00,880 --> 00:14:03,800
Speaker 2: What would the metrics be that you'd point to, either

277
00:14:03,840 --> 00:14:06,720
Speaker 2: that already exist or over the coming twelve months that

278
00:14:06,760 --> 00:14:08,120
Speaker 2: would evidence that well.

279
00:14:08,200 --> 00:14:12,280
Speaker 1: I think M I four fifty five is a clear

280
00:14:12,360 --> 00:14:16,440
Speaker 1: inflection point in both our technology capability as well as

281
00:14:16,440 --> 00:14:19,680
Speaker 1: the deep partnerships that we have across the industry. So

282
00:14:19,720 --> 00:14:21,960
Speaker 1: we're excited about, you know, what we see in front

283
00:14:22,000 --> 00:14:25,120
Speaker 1: of us. And you know, we've talked about you know,

284
00:14:25,200 --> 00:14:29,360
Speaker 1: tens of billions of dollars in an AI revenue as

285
00:14:29,400 --> 00:14:31,960
Speaker 1: we get into twenty twenty seven, and I think these

286
00:14:31,960 --> 00:14:34,520
Speaker 1: are important metrics, you know, for us as a company

287
00:14:35,120 --> 00:14:37,240
Speaker 1: when we think about the AI potential.

288
00:14:38,360 --> 00:14:41,160
Speaker 2: For all the focus on data centers, some forget that

289
00:14:41,360 --> 00:14:44,920
Speaker 2: a m D is leader in PC in many respects.

290
00:14:46,600 --> 00:14:50,680
Speaker 2: The forecasters have very different opinions of what will happen

291
00:14:50,720 --> 00:14:54,640
Speaker 2: this year. Some see, you know, shrinking market, some see

292
00:14:54,920 --> 00:14:59,560
Speaker 2: modest growth driven literally by just AIPC. You've been able

293
00:14:59,600 --> 00:15:03,120
Speaker 2: to tape market share and grow irrespective of what the

294
00:15:03,120 --> 00:15:06,400
Speaker 2: broader conditions are, but it haven't been great. How have

295
00:15:06,520 --> 00:15:09,640
Speaker 2: you done that and do you expect that to continue

296
00:15:09,640 --> 00:15:10,360
Speaker 2: to be the case.

297
00:15:10,840 --> 00:15:14,560
Speaker 1: Well, the PC market is a very good market for us.

298
00:15:14,640 --> 00:15:16,640
Speaker 1: You know, we grew a ton in the PC market

299
00:15:16,800 --> 00:15:20,400
Speaker 1: in twenty twenty five, and that really came from the

300
00:15:20,400 --> 00:15:22,000
Speaker 1: strength of our product portfolio.

301
00:15:22,400 --> 00:15:24,560
Speaker 4: We bet early in aipcs, so.

302
00:15:24,560 --> 00:15:27,560
Speaker 1: It was a clear area where we believe that the

303
00:15:27,560 --> 00:15:31,200
Speaker 1: technology would generate demand. We also went through a refresh

304
00:15:31,280 --> 00:15:34,280
Speaker 1: cycle with Windows eleven, and as we go into twenty

305
00:15:34,320 --> 00:15:37,080
Speaker 1: twenty six, I think we'll want to see how a

306
00:15:37,120 --> 00:15:41,360
Speaker 1: few quarters play out. I think the general demand for

307
00:15:41,720 --> 00:15:45,560
Speaker 1: computing is certainly there. There are some supplies chain constraints

308
00:15:45,600 --> 00:15:47,360
Speaker 1: that you know, we're working through and we want to

309
00:15:47,520 --> 00:15:50,280
Speaker 1: you know, watch going forward. But you know, our case

310
00:15:50,400 --> 00:15:54,440
Speaker 1: is one where we are still underrepresented in parts of

311
00:15:54,440 --> 00:15:55,000
Speaker 1: the market.

312
00:15:55,160 --> 00:15:57,760
Speaker 4: You know, we are very very strong in gaming, we're

313
00:15:57,840 --> 00:15:58,880
Speaker 4: very strong in consumer.

314
00:15:59,120 --> 00:16:03,080
Speaker 1: I think we're under represented in enterprise laptops, and we view.

315
00:16:02,920 --> 00:16:04,400
Speaker 4: This as a growth area for US.

316
00:16:04,560 --> 00:16:06,000
Speaker 3: Is AI PC change that.

317
00:16:06,160 --> 00:16:11,600
Speaker 1: Aipcs absolutely help in terms of, you know, just the

318
00:16:11,720 --> 00:16:15,760
Speaker 1: upgrade cycle coming in. We're excited about some of our

319
00:16:15,800 --> 00:16:19,160
Speaker 1: work with AI development systems as well. We announced a

320
00:16:19,200 --> 00:16:22,280
Speaker 1: new AI development system last night that you know, we

321
00:16:22,280 --> 00:16:24,080
Speaker 1: think will be also very attractive.

322
00:16:24,400 --> 00:16:27,240
Speaker 2: Those constraints you were talking about in the PC context

323
00:16:27,560 --> 00:16:29,960
Speaker 2: are specifically dram or it's broader than that.

324
00:16:30,360 --> 00:16:31,720
Speaker 4: It's more around the memory side.

325
00:16:31,760 --> 00:16:34,720
Speaker 1: So when you think about memory overall, I think we

326
00:16:34,840 --> 00:16:38,280
Speaker 1: have so much demand coming from let's call it AI

327
00:16:38,400 --> 00:16:41,480
Speaker 1: Data center compute that we want to see how it

328
00:16:41,560 --> 00:16:43,480
Speaker 1: impacts you know, sort of the rest of the memory

329
00:16:43,520 --> 00:16:44,720
Speaker 1: market out there.

330
00:16:45,200 --> 00:16:48,800
Speaker 3: One of the other areas that you discussed with Greg.

331
00:16:48,600 --> 00:16:52,040
Speaker 2: Brockman on stage from Open AI was sort of the

332
00:16:52,120 --> 00:16:56,000
Speaker 2: net or broad economic impact of AI, not just for

333
00:16:56,840 --> 00:17:01,760
Speaker 2: the companies. I think you were talking more about economy again,

334
00:17:02,040 --> 00:17:06,119
Speaker 2: very difficulty, so that how does one measure progress in

335
00:17:06,200 --> 00:17:10,000
Speaker 2: whether AI has or has not had a direct positive

336
00:17:10,000 --> 00:17:12,400
Speaker 2: economic impact around the world in any given year.

337
00:17:13,000 --> 00:17:15,720
Speaker 1: You know, It's true, it's hard to deconvolve all of

338
00:17:15,800 --> 00:17:18,320
Speaker 1: the things that are happening. But I think from a

339
00:17:18,440 --> 00:17:21,520
Speaker 1: sense of you know, what we see in the business,

340
00:17:22,240 --> 00:17:25,080
Speaker 1: and you know, many people want to see direct return

341
00:17:25,119 --> 00:17:29,199
Speaker 1: on investment for a particular a set of investments, what

342
00:17:29,280 --> 00:17:33,159
Speaker 1: I would say is that we know that AI.

343
00:17:33,080 --> 00:17:35,160
Speaker 4: Is making a difference in the productivity of companies.

344
00:17:35,200 --> 00:17:37,280
Speaker 1: We know that I can see that, you know, within

345
00:17:37,320 --> 00:17:40,639
Speaker 1: a MD in terms of as we deploy AI, you know,

346
00:17:40,680 --> 00:17:43,440
Speaker 1: we're able to get products to market faster, We're able

347
00:17:43,440 --> 00:17:46,800
Speaker 1: to you know, significantly improve some of our business processes.

348
00:17:46,880 --> 00:17:49,280
Speaker 1: So you know, as we go forward over the next

349
00:17:49,640 --> 00:17:51,360
Speaker 1: several years, I think you're going to see that much

350
00:17:51,400 --> 00:17:55,400
Speaker 1: broader enterprises. Every CEO that I talked to is talking

351
00:17:55,440 --> 00:17:58,320
Speaker 1: about AI. It is front and center in terms of

352
00:17:58,359 --> 00:18:00,199
Speaker 1: how to build a better company, how to build the

353
00:18:00,240 --> 00:18:03,720
Speaker 1: better portfolio, and so you know, I think what you

354
00:18:03,760 --> 00:18:06,680
Speaker 1: know Greg was talking about is when you aggregate all

355
00:18:06,720 --> 00:18:11,919
Speaker 1: of that, AI has to impact you know the world

356
00:18:12,040 --> 00:18:15,600
Speaker 1: at a GDP level, and we'll see that over the

357
00:18:15,600 --> 00:18:16,280
Speaker 1: next few years.

358
00:18:16,880 --> 00:18:20,200
Speaker 2: You're watching Bloomberg Television, you're listening to Bloomberg Radio. This

359
00:18:20,240 --> 00:18:22,760
Speaker 2: is Bloomberg Tech and we're live in Las Vegas. We're

360
00:18:22,760 --> 00:18:26,640
Speaker 2: speaking to A m d's CEO, Lisa Sue. You are

361
00:18:26,720 --> 00:18:32,000
Speaker 2: an investor in Generative baronyx's also technology partner, and they

362
00:18:32,000 --> 00:18:35,320
Speaker 2: have unveiled a humanoid robot here at Las Vegas cees.

363
00:18:35,440 --> 00:18:37,760
Speaker 2: In fact, if the magic of television can happen, and

364
00:18:37,800 --> 00:18:38,440
Speaker 2: we cut to the why.

365
00:18:38,400 --> 00:18:40,320
Speaker 3: Do we have to see it in the background? Right?

366
00:18:40,880 --> 00:18:46,440
Speaker 2: You know, this is the first tangible sign I feel

367
00:18:46,440 --> 00:18:48,560
Speaker 2: we've seen from AM and D on how you intend

368
00:18:48,560 --> 00:18:52,840
Speaker 2: to play in physical AI. Yes, explain your strategy. It

369
00:18:52,960 --> 00:18:55,360
Speaker 2: is the next big market, right, Yes.

370
00:18:55,160 --> 00:18:58,040
Speaker 1: And I wouldn't say it's the first time, but it's

371
00:18:58,080 --> 00:19:01,439
Speaker 1: probably one of the areas where we don't highlight as

372
00:19:01,520 --> 00:19:04,000
Speaker 1: much because there's so much focused on data center and

373
00:19:04,040 --> 00:19:07,760
Speaker 1: cloud and the opportunities there are you know, very much.

374
00:19:07,600 --> 00:19:08,240
Speaker 4: In front of us.

375
00:19:08,359 --> 00:19:10,520
Speaker 1: But when we look at physical AI, you know, starting

376
00:19:10,600 --> 00:19:14,359
Speaker 1: from all of the work we've done in FPGAs and

377
00:19:14,440 --> 00:19:17,560
Speaker 1: embedded real time you know capability, we have been in

378
00:19:17,600 --> 00:19:19,680
Speaker 1: this space for a long time. You know, we already

379
00:19:19,720 --> 00:19:23,280
Speaker 1: power a lot of robotic applications you know out there.

380
00:19:23,760 --> 00:19:27,880
Speaker 1: But I think as we go into the humanoid capability,

381
00:19:27,920 --> 00:19:30,719
Speaker 1: and you know, we're excited about our partnership with you know,

382
00:19:31,040 --> 00:19:34,840
Speaker 1: g Bionics and the work with on June one, I

383
00:19:34,840 --> 00:19:37,240
Speaker 1: think that takes us to another level in terms of

384
00:19:37,359 --> 00:19:40,560
Speaker 1: capability and intelligence and what we're trying to do so.

385
00:19:40,840 --> 00:19:43,280
Speaker 2: Is the business model to be all things the brain

386
00:19:43,400 --> 00:19:46,600
Speaker 2: inside of the humanoid robot and inference side that, the

387
00:19:46,680 --> 00:19:49,200
Speaker 2: underlying software being traded on a trained on a m

388
00:19:49,280 --> 00:19:51,840
Speaker 2: D accelerators. Just I don't what's to go to the market,

389
00:19:51,840 --> 00:19:53,800
Speaker 2: I guess is what I'm asking you should.

390
00:19:53,560 --> 00:19:57,840
Speaker 1: Expect that our partnerships extend all through all of those levels,

391
00:19:57,840 --> 00:20:02,320
Speaker 1: so we have the components that can power the humanoid robots,

392
00:20:02,359 --> 00:20:06,040
Speaker 1: you know, sort of real time local capability, which is

393
00:20:06,080 --> 00:20:08,119
Speaker 1: a very very important and then we also have the

394
00:20:08,520 --> 00:20:11,040
Speaker 1: technology behind that in terms of, you know, how to

395
00:20:11,080 --> 00:20:14,880
Speaker 1: train and inference on these humanoids.

396
00:20:15,320 --> 00:20:18,199
Speaker 3: When last we met in person, it was in Washington,

397
00:20:18,280 --> 00:20:18,480
Speaker 3: d C.

398
00:20:19,160 --> 00:20:24,040
Speaker 2: And the President had just outlined a broad strategy for

399
00:20:24,200 --> 00:20:29,760
Speaker 2: America in AI and it really centered around infrastructure deregulation

400
00:20:30,119 --> 00:20:34,280
Speaker 2: allowing those building the infrastructure to move faster. That was

401
00:20:34,359 --> 00:20:36,600
Speaker 2: kind of in the second half of last year. In

402
00:20:36,640 --> 00:20:39,200
Speaker 2: the months that have followed, have you seen any signs

403
00:20:39,240 --> 00:20:42,520
Speaker 2: that it worked and anything that you could point to

404
00:20:43,000 --> 00:20:46,280
Speaker 2: that says, yeah, people are able to build faster maybe

405
00:20:46,280 --> 00:20:49,120
Speaker 2: to address some of the compute deficits we discussed.

406
00:20:49,200 --> 00:20:52,160
Speaker 1: Well, I can say for sure, you know, the President's

407
00:20:52,200 --> 00:20:54,600
Speaker 1: AI action plan. You know, when we met, and I

408
00:20:54,680 --> 00:20:56,720
Speaker 1: think this was back in July when it came out,

409
00:20:57,400 --> 00:21:01,920
Speaker 1: I was very optimistic about having a really forward leaning

410
00:21:01,960 --> 00:21:06,560
Speaker 1: strategy from you know, sort of the whole view of

411
00:21:06,640 --> 00:21:08,760
Speaker 1: what does it take for the US to lead an AI.

412
00:21:09,240 --> 00:21:12,159
Speaker 1: And I think we've made a ton of progress along

413
00:21:12,200 --> 00:21:15,600
Speaker 1: the way. And you know, I had Michael Kratzios joined

414
00:21:15,680 --> 00:21:18,119
Speaker 1: us last night on stage as well to talk about

415
00:21:18,160 --> 00:21:20,720
Speaker 1: the Genesis Mission, which is you know, another you know,

416
00:21:20,800 --> 00:21:25,040
Speaker 1: sort of public private partnership approach to really advanced science

417
00:21:25,640 --> 00:21:27,679
Speaker 1: in the United States, and when you look at you know,

418
00:21:27,720 --> 00:21:31,199
Speaker 1: all of these things, you know, building faster, ensuring that

419
00:21:31,280 --> 00:21:34,960
Speaker 1: we have you know, the right export controls so that

420
00:21:35,000 --> 00:21:38,840
Speaker 1: we were able to have the US stack adopted.

421
00:21:38,440 --> 00:21:40,200
Speaker 3: Across the controls.

422
00:21:40,280 --> 00:21:42,240
Speaker 4: Currently, we are certainly working.

423
00:21:42,119 --> 00:21:45,520
Speaker 1: Very closely with uh, you know, the the various parties

424
00:21:45,520 --> 00:21:47,480
Speaker 1: in the US government to ensure that we have the

425
00:21:47,560 --> 00:21:50,840
Speaker 1: right balance there. And we also have you know, this

426
00:21:50,960 --> 00:21:54,160
Speaker 1: notion of how do we invest more here and ensure

427
00:21:54,280 --> 00:21:56,520
Speaker 1: that in the United States that we are you know,

428
00:21:57,320 --> 00:22:01,040
Speaker 1: running as fast as possible to bring you AI capacity

429
00:22:01,200 --> 00:22:04,440
Speaker 1: you know online, to help us in you know, science

430
00:22:04,600 --> 00:22:07,800
Speaker 1: and you know sort of the broader you know, economic benefits.

431
00:22:08,600 --> 00:22:13,000
Speaker 2: Lisa, what what happens in twenty twenty six, what happens

432
00:22:13,040 --> 00:22:14,840
Speaker 2: in the world of AI, and what do you think

433
00:22:14,920 --> 00:22:17,719
Speaker 2: will define this year in terms of the progress that

434
00:22:17,760 --> 00:22:19,200
Speaker 2: your industry hopes to make.

435
00:22:19,440 --> 00:22:22,760
Speaker 1: Well, I started our keynote last night with the sentence

436
00:22:22,840 --> 00:22:24,800
Speaker 1: that you know, you ain't seen nothing yet.

437
00:22:25,240 --> 00:22:26,440
Speaker 4: That's really how I feel.

438
00:22:26,480 --> 00:22:30,199
Speaker 1: I Mean, we're sitting here in January and it's just

439
00:22:30,400 --> 00:22:34,159
Speaker 1: amazing how much progress is made, you know, every week

440
00:22:34,200 --> 00:22:36,960
Speaker 1: and every month when we see how these models are developing,

441
00:22:37,080 --> 00:22:39,639
Speaker 1: when we see how the use cases are developing, and

442
00:22:39,680 --> 00:22:43,639
Speaker 1: then when we see the tangible results on businesses and outcomes,

443
00:22:44,240 --> 00:22:46,600
Speaker 1: I believe that, you know, we saw a good amount

444
00:22:46,800 --> 00:22:50,600
Speaker 1: of that, you know, come to Fruition in twenty twenty five.

445
00:22:50,640 --> 00:22:52,600
Speaker 1: We're going to see much more of that in twenty

446
00:22:52,640 --> 00:22:55,879
Speaker 1: twenty six, So that everyone should understand that, you know,

447
00:22:56,160 --> 00:22:59,040
Speaker 1: AI is not just you know, hype out there. It's

448
00:22:59,080 --> 00:23:01,159
Speaker 1: not just you know, sort of things that people are

449
00:23:01,160 --> 00:23:03,960
Speaker 1: talking about in the investment community. It's things that people

450
00:23:04,040 --> 00:23:07,920
Speaker 1: are using every day, real time and feeling like, Hey,

451
00:23:08,080 --> 00:23:11,080
Speaker 1: my life is better because I have this technology.

452
00:23:11,119 --> 00:23:12,920
Speaker 4: And I think we're going to see that in twenty

453
00:23:12,960 --> 00:23:13,440
Speaker 4: twenty six.

454
00:23:13,720 --> 00:23:17,600
Speaker 2: Lisa suit AMD CEO AMD with it's in the world's

455
00:23:17,600 --> 00:23:20,760
Speaker 2: first two animeter chip going into Helos, its first rack

456
00:23:20,840 --> 00:23:22,399
Speaker 2: scale system solution