1
00:00:02,520 --> 00:00:07,080
Speaker 1: Bloomberg Audio Studios, podcasts, radio news.

2
00:00:07,280 --> 00:00:10,640
Speaker 2: The third pillar around making sure that we are expediting

3
00:00:10,720 --> 00:00:14,640
Speaker 2: the export of our American AI stack. That is really

4
00:00:14,640 --> 00:00:15,880
Speaker 2: core to our mission at Armada.

5
00:00:16,120 --> 00:00:18,320
Speaker 3: Well, what I really like about the AI Action Plan

6
00:00:18,440 --> 00:00:20,320
Speaker 3: is I actually believe it's quite actionable.

7
00:00:22,800 --> 00:00:25,040
Speaker 2: That's just what a few of the AI leaders had

8
00:00:25,079 --> 00:00:28,680
Speaker 2: to say yesterday in Washington about President Trump's AI Action

9
00:00:28,880 --> 00:00:32,600
Speaker 2: Plan here with more in the administration's AI rollout is

10
00:00:32,640 --> 00:00:35,320
Speaker 2: one of the people that helped formulate write that policy

11
00:00:35,360 --> 00:00:39,960
Speaker 2: in that document, Shariram Krishnan, Whitehouse Senior Policy Advisor for AI,

12
00:00:40,080 --> 00:00:43,559
Speaker 2: to get here in person in Washington, d C. There

13
00:00:43,600 --> 00:00:45,960
Speaker 2: was a point towards the end of that that sequence

14
00:00:46,000 --> 00:00:52,240
Speaker 2: of comments about exporting the American technology stack, exporting chips

15
00:00:52,240 --> 00:00:56,720
Speaker 2: in particular. Your colleagues in the cabinet, mister Lutnick, for example,

16
00:00:56,760 --> 00:00:59,680
Speaker 2: talked about this at length on stage. Are we going

17
00:00:59,680 --> 00:01:02,040
Speaker 2: to get a sets of rules and how will we

18
00:01:02,160 --> 00:01:06,160
Speaker 2: decide the volume and specificity of the types of chips

19
00:01:06,160 --> 00:01:08,279
Speaker 2: that we can sell, particularly into China?

20
00:01:09,280 --> 00:01:12,560
Speaker 1: Thank you Ed for having me here. I think we're

21
00:01:12,560 --> 00:01:14,600
Speaker 1: going to work on that. I think if you look

22
00:01:14,640 --> 00:01:17,800
Speaker 1: at the executive order that President Trump signed, which is

23
00:01:17,840 --> 00:01:21,120
Speaker 1: about America setting the standard. What you're going to see

24
00:01:21,280 --> 00:01:24,720
Speaker 1: is we're going to work closely with our friends in commerce,

25
00:01:25,080 --> 00:01:29,240
Speaker 1: especially in BIS, and what we're going to figure out

26
00:01:29,520 --> 00:01:31,520
Speaker 1: is what those police are going to be. But the

27
00:01:31,560 --> 00:01:35,840
Speaker 1: metapoint here is President Trump said was very clear, we

28
00:01:35,880 --> 00:01:39,360
Speaker 1: need America to set the standard for how AI is

29
00:01:39,360 --> 00:01:43,119
Speaker 1: being used globally, from chips to models, to every part

30
00:01:43,160 --> 00:01:45,680
Speaker 1: of the stack. And what we've seen over the last

31
00:01:45,760 --> 00:01:48,960
Speaker 1: six months, be it with Huawei on the chip side,

32
00:01:49,120 --> 00:01:52,520
Speaker 1: or be it with models like deep Seak, we have competition,

33
00:01:53,000 --> 00:01:56,040
Speaker 1: and yesterday's plan sets a very clear note that we

34
00:01:56,240 --> 00:01:59,320
Speaker 1: need to be making sure that our standards exist worldwide.

35
00:01:59,600 --> 00:02:02,280
Speaker 2: I read all twenty three pages as many times as

36
00:02:02,280 --> 00:02:04,920
Speaker 2: I could think twenty three on the version I have,

37
00:02:05,040 --> 00:02:08,720
Speaker 2: but we can agree to disagree. There's a heavy emphasis

38
00:02:08,760 --> 00:02:12,200
Speaker 2: on deregulation and permitting that we get for the onshoring

39
00:02:12,240 --> 00:02:16,520
Speaker 2: component for exporting American AI. What is it tangibly What

40
00:02:16,639 --> 00:02:19,799
Speaker 2: action will be taken to open that up, be it

41
00:02:19,880 --> 00:02:21,959
Speaker 2: on the software end or the hardware end of the stack.

42
00:02:22,320 --> 00:02:26,200
Speaker 1: Yes, So if you prison. Trump signed three executive orders yesterday,

43
00:02:26,480 --> 00:02:29,280
Speaker 1: and one of them deals exactly with this and what

44
00:02:29,320 --> 00:02:31,320
Speaker 1: we're going to do is we're going to work with

45
00:02:31,480 --> 00:02:34,920
Speaker 1: industry to come up with a gold standard for what

46
00:02:34,960 --> 00:02:37,359
Speaker 1: the American stack should be. And there's going to be

47
00:02:37,440 --> 00:02:39,919
Speaker 1: a process and how we figured this out, and then

48
00:02:40,040 --> 00:02:42,640
Speaker 1: we're going to try and make sure that when we

49
00:02:42,680 --> 00:02:46,320
Speaker 1: work with our partners and our allied countries that they

50
00:02:46,360 --> 00:02:48,760
Speaker 1: are using our stack instead of the competitions.

51
00:02:49,120 --> 00:02:52,480
Speaker 3: For many would look at the four point two trillion

52
00:02:52,520 --> 00:02:56,600
Speaker 3: dollar market capitalization of Nvidia and wonder doesn't need any

53
00:02:56,600 --> 00:03:00,160
Speaker 3: more help in exporting its tech stack right now. How

54
00:03:00,280 --> 00:03:02,519
Speaker 3: much more support does an AMD or an Nvidia need.

55
00:03:02,440 --> 00:03:06,280
Speaker 1: You think, Well, the way we look at it is

56
00:03:06,400 --> 00:03:10,440
Speaker 1: we want America to win. We want American GPUs and

57
00:03:10,480 --> 00:03:15,079
Speaker 1: American models to be used. And Nvidia and AMD happened

58
00:03:15,080 --> 00:03:18,400
Speaker 1: to be great American companies along with many many others,

59
00:03:18,639 --> 00:03:22,080
Speaker 1: and so our job is to create the playing field

60
00:03:22,360 --> 00:03:25,880
Speaker 1: where company where Jensen or I saw doctor Lisa Sue

61
00:03:26,000 --> 00:03:29,000
Speaker 1: earlier can go sell their products worldwide.

62
00:03:29,360 --> 00:03:33,520
Speaker 3: Let's talk about that team, because there were some absences.

63
00:03:33,720 --> 00:03:36,240
Speaker 3: Let's say it we would think that Intel would be

64
00:03:36,600 --> 00:03:39,400
Speaker 3: a company about some report on its earnings that would

65
00:03:39,400 --> 00:03:42,560
Speaker 3: be integral to manufacturing here in the US. GSMC from

66
00:03:42,560 --> 00:03:44,480
Speaker 3: abroad coming in and putting money to work here in

67
00:03:44,480 --> 00:03:47,080
Speaker 3: the United States as well. What do we interpret by

68
00:03:47,080 --> 00:03:48,240
Speaker 3: those who weren't in the room.

69
00:03:50,280 --> 00:03:51,960
Speaker 1: I'm not sure I would read too much into that.

70
00:03:52,080 --> 00:03:54,520
Speaker 1: I think if you look at yesterday's event, it was

71
00:03:54,680 --> 00:03:59,400
Speaker 1: amazingly well attended. We had industry leaders from across the stack,

72
00:03:59,480 --> 00:04:03,200
Speaker 1: from the semi conductor space to energy, to AI models

73
00:04:03,240 --> 00:04:07,160
Speaker 1: to AI applications, and as you said, like you know,

74
00:04:07,280 --> 00:04:10,880
Speaker 1: we really want American some my conductor industry to win,

75
00:04:11,280 --> 00:04:14,160
Speaker 1: and you saw today's event being extremely well represented.

76
00:04:14,200 --> 00:04:17,440
Speaker 2: When it came to that, you have advised the president

77
00:04:17,600 --> 00:04:20,640
Speaker 2: and collectively as a group, looked at policy. But there

78
00:04:20,680 --> 00:04:22,520
Speaker 2: was a moment that we have to address yesterday where

79
00:04:22,560 --> 00:04:25,880
Speaker 2: the President, after making various people stand up, said that

80
00:04:25,920 --> 00:04:28,680
Speaker 2: he had looked at breaking up in video, but then

81
00:04:28,760 --> 00:04:32,000
Speaker 2: came to understand how crucial they were to the American

82
00:04:32,040 --> 00:04:35,240
Speaker 2: technology stack and that his aid so basically said to him,

83
00:04:35,320 --> 00:04:38,919
Speaker 2: Jensen's great. Can I ask you, did you ever discuss

84
00:04:39,000 --> 00:04:41,239
Speaker 2: with the president the idea of breaking up in video

85
00:04:41,520 --> 00:04:43,800
Speaker 2: and why that was included in his speech?

86
00:04:44,080 --> 00:04:46,520
Speaker 1: No, I wasn't in the meeting with the president, but

87
00:04:46,560 --> 00:04:49,120
Speaker 1: I think what the President was referring to, is that

88
00:04:49,360 --> 00:04:52,919
Speaker 1: Nvidia has just done a phenomenal job in building products

89
00:04:52,960 --> 00:04:56,960
Speaker 1: and applications and GPUs that the world is using, and

90
00:04:57,000 --> 00:04:59,039
Speaker 1: that they're so ahead of the curve. So if you

91
00:04:59,120 --> 00:05:01,320
Speaker 1: listen to the second lot of what the President said,

92
00:05:01,560 --> 00:05:04,080
Speaker 1: he talked about how it's going to take several years

93
00:05:04,120 --> 00:05:06,719
Speaker 1: for people to catch up with them. And I think

94
00:05:06,760 --> 00:05:08,880
Speaker 1: that's kind of one of the pieces that's put America

95
00:05:08,920 --> 00:05:11,560
Speaker 1: in this pole position when it comes to the GPU,

96
00:05:11,640 --> 00:05:14,120
Speaker 1: is that everybody in the world wants to get access.

97
00:05:13,839 --> 00:05:18,719
Speaker 2: To We appreciate the clarity issue round open source, very prominent,

98
00:05:18,920 --> 00:05:22,000
Speaker 2: very high up in the AI Action Plan. Why that's

99
00:05:22,040 --> 00:05:23,799
Speaker 2: something you're particularly focused on as well.

100
00:05:24,000 --> 00:05:26,600
Speaker 1: Yes, And this is an issue near and dear to

101
00:05:26,680 --> 00:05:28,760
Speaker 1: my heart. And this is something which I think the

102
00:05:28,800 --> 00:05:32,320
Speaker 1: Biden administration got totally wrong, where they tried to basically

103
00:05:32,360 --> 00:05:35,799
Speaker 1: scare people about open source. We believe that open source

104
00:05:35,839 --> 00:05:39,880
Speaker 1: is the integral part of innovation, of how academic research

105
00:05:40,000 --> 00:05:44,200
Speaker 1: is done, how you get little tech to innovate quickly

106
00:05:44,240 --> 00:05:46,839
Speaker 1: and be on a level playing field. And we think

107
00:05:46,880 --> 00:05:49,880
Speaker 1: America should win. And you know, the first day when

108
00:05:49,880 --> 00:05:52,640
Speaker 1: I took this job, we had Deep Sea come out

109
00:05:52,960 --> 00:05:57,240
Speaker 1: and really, you know, send a signal that America wasn't

110
00:05:57,279 --> 00:05:59,240
Speaker 1: actually leading in open source at the time, and we

111
00:05:59,279 --> 00:06:01,880
Speaker 1: believe that needs we rectified. And that's why in the

112
00:06:01,880 --> 00:06:04,560
Speaker 1: action plan we send a very strong signal that open

113
00:06:04,600 --> 00:06:06,600
Speaker 1: source or open weights America is going to have a

114
00:06:06,680 --> 00:06:07,440
Speaker 1: leadership position.

115
00:06:07,880 --> 00:06:10,280
Speaker 3: Another thing near in dat to your heart is pro wrestling,

116
00:06:11,080 --> 00:06:13,520
Speaker 3: and it feels like we're in a wrestling match between

117
00:06:13,640 --> 00:06:16,080
Speaker 3: China and the US, and I'm really interested as to

118
00:06:16,120 --> 00:06:19,080
Speaker 3: how you navigate the rules of that wrestling match. For example,

119
00:06:19,480 --> 00:06:22,479
Speaker 3: China gives US rare earth metal access in return, we

120
00:06:22,520 --> 00:06:25,440
Speaker 3: give them some H twenties. But how does Vialion navigate

121
00:06:25,480 --> 00:06:28,120
Speaker 3: if China stops and chokes that will they have to

122
00:06:28,160 --> 00:06:30,640
Speaker 3: be limited in H twenty How do they get certainty

123
00:06:30,640 --> 00:06:31,360
Speaker 3: in this environment?

124
00:06:32,720 --> 00:06:34,880
Speaker 1: Well, first of all, if you're talking pro wrestling, let's

125
00:06:34,880 --> 00:06:37,360
Speaker 1: speak clear. We have the face, we have the baby

126
00:06:37,400 --> 00:06:39,880
Speaker 1: face who's going to ultimately go over for all of

127
00:06:39,880 --> 00:06:42,680
Speaker 1: you pro wrestling fans. But look, you know, the way

128
00:06:42,720 --> 00:06:46,039
Speaker 1: we think about it is we want the American tech

129
00:06:46,040 --> 00:06:49,039
Speaker 1: stack to win. One of the metrics I talk about

130
00:06:49,320 --> 00:06:53,159
Speaker 1: is every token that is inferenced worldwide, we want it

131
00:06:53,200 --> 00:06:56,520
Speaker 1: to be on an American GPU or using an American model,

132
00:06:56,520 --> 00:06:59,360
Speaker 1: how many our Quaterllion tokens there are, And when it

133
00:06:59,400 --> 00:07:02,760
Speaker 1: comes to the head twenty, I think the way we

134
00:07:02,800 --> 00:07:06,159
Speaker 1: think about it is we would much rather have Nvideo

135
00:07:06,440 --> 00:07:11,160
Speaker 1: or AMD be the GPU of choice rather using Huawei

136
00:07:11,320 --> 00:07:12,840
Speaker 1: or somebody else's. And that's how we.

137
00:07:12,800 --> 00:07:15,640
Speaker 3: See it in video is a babyface as a fan favorite.

138
00:07:15,680 --> 00:07:16,120
Speaker 2: I love it.

139
00:07:16,240 --> 00:07:18,440
Speaker 3: Sharam krish Nan has been great having time with you.

140
00:07:18,520 --> 00:07:21,360
Speaker 3: Thank you, White House Senior Policy Advisor for AI