1
00:00:00,280 --> 00:00:02,480
Speaker 1: Let's just shift gears a little bit, talk chips more

2
00:00:02,480 --> 00:00:05,600
Speaker 1: broadly and how there's some new entrance into the AI

3
00:00:05,840 --> 00:00:08,240
Speaker 1: part of the equation. Jeff Bezos, for example, and Samsung

4
00:00:08,280 --> 00:00:11,880
Speaker 1: are betting big on AI chip startup ten Store, announcing

5
00:00:11,920 --> 00:00:14,200
Speaker 1: a seven hundred million dollar investment, which puts the company

6
00:00:14,200 --> 00:00:16,800
Speaker 1: now in a two point six billion dollar valuation. Joining

7
00:00:16,840 --> 00:00:19,520
Speaker 1: us to discuss is the CEO of the company, Jim Keller,

8
00:00:19,680 --> 00:00:22,720
Speaker 1: who is himself a legend and chip making world. Of course,

9
00:00:22,760 --> 00:00:26,279
Speaker 1: your resume includes know who's who of AMD, Apple, Tesla

10
00:00:26,320 --> 00:00:29,920
Speaker 1: and indeed Intel. Jim, the money that you're raising, what

11
00:00:30,040 --> 00:00:31,920
Speaker 1: is it that you're hoping to do. You're wanting to

12
00:00:31,920 --> 00:00:35,280
Speaker 1: make a more cheaper operation and option out there for

13
00:00:35,680 --> 00:00:36,880
Speaker 1: AI accelerator chips.

14
00:00:37,880 --> 00:00:40,760
Speaker 2: Yeah, that's right. We're building out the team, we're building

15
00:00:40,800 --> 00:00:43,879
Speaker 2: out our products, We're starting to ship our products, and

16
00:00:43,920 --> 00:00:46,280
Speaker 2: our real mission right now is to bring on developers

17
00:00:46,400 --> 00:00:47,840
Speaker 2: and early customers.

18
00:00:48,479 --> 00:00:50,800
Speaker 1: Let's talk about the customers and what you offer that

19
00:00:50,880 --> 00:00:52,640
Speaker 1: isn't in the market at the moment. Jim, what's your

20
00:00:52,680 --> 00:00:53,200
Speaker 1: selling point?

21
00:00:54,240 --> 00:00:56,800
Speaker 2: Well, we have a couple of things. We license our

22
00:00:56,800 --> 00:00:59,480
Speaker 2: IP and we're building a business on that for both

23
00:00:59,520 --> 00:01:03,680
Speaker 2: risk vice CPU and AI. We're selling low cost development

24
00:01:03,720 --> 00:01:06,600
Speaker 2: systems like I'm really interested in hitting a much lower

25
00:01:06,680 --> 00:01:09,959
Speaker 2: cost point. And then we open source to our whole

26
00:01:10,000 --> 00:01:13,720
Speaker 2: AI compiler stack and people like the transparency. They can

27
00:01:13,720 --> 00:01:17,080
Speaker 2: see exactly what the technology is doing. They actually it's

28
00:01:17,080 --> 00:01:20,440
Speaker 2: turned into a great recruiting vehicle for our software engineers.

29
00:01:21,680 --> 00:01:25,000
Speaker 1: Interesting. So let's just go first to the price points,

30
00:01:25,280 --> 00:01:27,680
Speaker 1: a cheap offering. A lot of that's done by having

31
00:01:27,720 --> 00:01:31,679
Speaker 1: less pricing components. For example, in videos, chips are incredibly expensive.

32
00:01:31,720 --> 00:01:34,600
Speaker 1: We know people pay it because of what they offer

33
00:01:34,760 --> 00:01:37,800
Speaker 1: the full stack in many ways, but they struggle from

34
00:01:37,840 --> 00:01:41,640
Speaker 1: a choke perspective on how foundwidth memory chips in particular,

35
00:01:41,760 --> 00:01:43,440
Speaker 1: you don't need them. Why don't you need them? Why

36
00:01:43,440 --> 00:01:44,400
Speaker 1: don't you want them?

37
00:01:44,840 --> 00:01:47,840
Speaker 2: Well, it's partly about choices. So there's two choices. One is,

38
00:01:48,200 --> 00:01:51,640
Speaker 2: if you pick a really expensive memory technology, you'll never

39
00:01:51,720 --> 00:01:54,920
Speaker 2: get the price down right, and so you need to

40
00:01:55,200 --> 00:01:58,880
Speaker 2: then make technology decisions to lower your costs. So we

41
00:01:59,120 --> 00:02:03,000
Speaker 2: have a tensor process or architecture, we keep the data

42
00:02:03,040 --> 00:02:06,640
Speaker 2: on chip. More, we have a software stack that's very

43
00:02:06,640 --> 00:02:10,960
Speaker 2: good at partitioning AI workloads across multiple chips, and then

44
00:02:11,000 --> 00:02:14,799
Speaker 2: they can send information and compute packets to each other

45
00:02:15,120 --> 00:02:18,240
Speaker 2: without having to go through memory, and that lowers the

46
00:02:18,240 --> 00:02:21,120
Speaker 2: memory VANDTH required and also lets it scale more naturally

47
00:02:21,160 --> 00:02:22,120
Speaker 2: across many chips.

48
00:02:22,760 --> 00:02:25,880
Speaker 1: So I'm a new investor. You brought me on board.

49
00:02:26,040 --> 00:02:29,400
Speaker 1: I'm interested Bezos Samsung. But you're saying, look, I'm going

50
00:02:29,440 --> 00:02:33,000
Speaker 1: to do this with more open architecture, with cheaper offerings.

51
00:02:33,040 --> 00:02:34,639
Speaker 1: How much of the market shed do you have to get?

52
00:02:34,680 --> 00:02:36,360
Speaker 1: What sort of price point are we looking at for

53
00:02:36,360 --> 00:02:37,160
Speaker 1: your chip offering?

54
00:02:38,760 --> 00:02:42,040
Speaker 2: Yeah, so this is really wild. The market is so big,

55
00:02:42,680 --> 00:02:44,840
Speaker 2: Like the percent we have to get is actually pretty

56
00:02:44,880 --> 00:02:47,839
Speaker 2: small to be a real value proposition. You know, next

57
00:02:47,880 --> 00:02:49,480
Speaker 2: year we'll die and go to heaven if we make

58
00:02:49,480 --> 00:02:52,280
Speaker 2: two hundred million dollars. We have already booked one hundred

59
00:02:52,280 --> 00:02:55,600
Speaker 2: and fifty million in revenue. I'm really interested in finding

60
00:02:55,760 --> 00:02:59,280
Speaker 2: small players, early players, people who want to own their technology.

61
00:03:00,040 --> 00:03:03,920
Speaker 2: People are excited about our software and hardware architecture, and

62
00:03:04,360 --> 00:03:06,960
Speaker 2: you know, I want to grow organically from there. I

63
00:03:07,000 --> 00:03:09,880
Speaker 2: think some people will focus on the big guys first.

64
00:03:10,400 --> 00:03:13,639
Speaker 2: That's not how new technology introductions go. You focus on

65
00:03:14,040 --> 00:03:17,680
Speaker 2: smaller players, people who need a differentiation, and that's what

66
00:03:17,720 --> 00:03:18,840
Speaker 2: we're really that's our mission.

67
00:03:19,240 --> 00:03:22,760
Speaker 1: So what therefore is the mission building in in terms

68
00:03:22,800 --> 00:03:26,400
Speaker 1: of long term value of generative AI of AI applications.

69
00:03:27,040 --> 00:03:28,480
Speaker 1: Do you think that we are in some sort of

70
00:03:28,560 --> 00:03:31,440
Speaker 1: hype cycle? Do you think that actually this market is

71
00:03:31,440 --> 00:03:33,040
Speaker 1: as big as manufacturing in.

72
00:03:34,000 --> 00:03:36,320
Speaker 2: Well, first of all, we are definitely in a hype cycle.

73
00:03:36,480 --> 00:03:39,280
Speaker 2: There's going to be multiple and people say, well, Jim,

74
00:03:39,360 --> 00:03:40,920
Speaker 2: you don't understand it's going to go down next year,

75
00:03:40,920 --> 00:03:42,000
Speaker 2: and then it's going to go up, and it's going

76
00:03:41,960 --> 00:03:45,440
Speaker 2: to down low pass filter all that stuff. Right, what's

77
00:03:45,480 --> 00:03:48,920
Speaker 2: really going to happen is AI is going to dominate

78
00:03:48,960 --> 00:03:52,280
Speaker 2: computing over the next ten years, and we're just starting

79
00:03:52,960 --> 00:03:55,120
Speaker 2: and the hardware is going to change a lot. The

80
00:03:55,160 --> 00:03:57,800
Speaker 2: software is going to change a lot. Like one day

81
00:03:57,840 --> 00:04:00,240
Speaker 2: you read in the papers lms can do anything, and

82
00:04:00,280 --> 00:04:03,360
Speaker 2: the next day you read they've hit a limit. Ignore

83
00:04:03,400 --> 00:04:06,160
Speaker 2: all that stuff. We're just starting this. There's going to

84
00:04:06,200 --> 00:04:09,600
Speaker 2: be massive transformations. So what I really wanted tense Torrent

85
00:04:09,880 --> 00:04:14,960
Speaker 2: to have is great computer designers, general purpose CPUs, AI processors,

86
00:04:15,320 --> 00:04:19,720
Speaker 2: chip design, system design, software design. And then the open

87
00:04:19,720 --> 00:04:22,440
Speaker 2: source stuff is really interesting because that engages a community.

88
00:04:22,480 --> 00:04:25,320
Speaker 2: People are really passionate about that to contribute back.

89
00:04:26,600 --> 00:04:29,960
Speaker 1: This is where you're thinking about your own talent team.

90
00:04:30,000 --> 00:04:32,559
Speaker 1: You've managed talent, as we said, sort of across the board.

91
00:04:33,040 --> 00:04:35,559
Speaker 1: You began like you were integral to the chip design

92
00:04:35,560 --> 00:04:37,440
Speaker 1: process in the US, and you think back to DC

93
00:04:37,680 --> 00:04:39,840
Speaker 1: and then the work that you then did an AMD,

94
00:04:40,279 --> 00:04:42,680
Speaker 1: moving to Apple, you were at Intel, and I just

95
00:04:42,720 --> 00:04:44,600
Speaker 1: have to go there on this day that Pat Gelsinger

96
00:04:44,640 --> 00:04:48,560
Speaker 1: hands in his well, hands his hat on off, he's

97
00:04:48,600 --> 00:04:51,000
Speaker 1: going to re resigning. Many would say it's because they

98
00:04:51,080 --> 00:04:53,960
Speaker 1: didn't get into AI quickly enough. That was prior to him.

99
00:04:54,200 --> 00:04:55,880
Speaker 1: What do you make of the transition.

100
00:04:56,839 --> 00:04:59,240
Speaker 2: Well, it's hard to say. So. I like Pat personally.

101
00:04:59,240 --> 00:05:01,760
Speaker 2: I've met him a number of times. He's a smart guy.

102
00:05:01,839 --> 00:05:06,719
Speaker 2: He really cares. I think the market focuses too much

103
00:05:06,720 --> 00:05:10,880
Speaker 2: on stuff like execution and AI transition. The way you

104
00:05:11,200 --> 00:05:14,600
Speaker 2: make money, get customers to build great products, like Steve

105
00:05:14,680 --> 00:05:17,600
Speaker 2: Jobs is the best at this, focus on the product,

106
00:05:17,760 --> 00:05:20,760
Speaker 2: make the best product you possibly can. And I think

107
00:05:21,000 --> 00:05:24,080
Speaker 2: they need somebody way more hands on, really focus on product,

108
00:05:24,600 --> 00:05:27,560
Speaker 2: and they need to stop talking about AI execution all

109
00:05:27,640 --> 00:05:30,440
Speaker 2: the nonsense. Right when they make the best product, the

110
00:05:30,440 --> 00:05:33,680
Speaker 2: best fab, the best chip, the best GP accelerator. They'll

111
00:05:33,720 --> 00:05:34,120
Speaker 2: do fine.

112
00:05:35,320 --> 00:05:38,200
Speaker 1: Jim. When you're thinking about all of the errors, you

113
00:05:38,279 --> 00:05:41,159
Speaker 1: want to own. How much does US and China factor

114
00:05:41,200 --> 00:05:42,039
Speaker 1: into your growth story?

115
00:05:44,320 --> 00:05:49,400
Speaker 2: So, first of all, we are building open source software,

116
00:05:49,760 --> 00:05:54,880
Speaker 2: risk five based technology. We think that's an open platform.

117
00:05:54,960 --> 00:05:57,320
Speaker 2: There's a lot of people interested in that. At the

118
00:05:57,360 --> 00:06:00,800
Speaker 2: same time, you know, there are you know, restrictions by

119
00:06:00,839 --> 00:06:03,320
Speaker 2: the US government and trade rules. We have really good

120
00:06:03,400 --> 00:06:06,440
Speaker 2: lawyers and trade compliance people. We work very closely with

121
00:06:06,480 --> 00:06:09,960
Speaker 2: those those guys. We want to do the right thing

122
00:06:10,080 --> 00:06:12,600
Speaker 2: both for the business and for you know, being compliant

123
00:06:12,600 --> 00:06:15,920
Speaker 2: to what's going on. And it's a changing world and

124
00:06:16,279 --> 00:06:20,800
Speaker 2: we're very involved in that. Short run, our business possibilities

125
00:06:20,839 --> 00:06:24,120
Speaker 2: are really big, and you know, if China plays in that,

126
00:06:24,120 --> 00:06:26,880
Speaker 2: that'll be interesting. If it doesn't for you know, various reasons,

127
00:06:26,880 --> 00:06:27,560
Speaker 2: that's fine.

128
00:06:28,480 --> 00:06:31,560
Speaker 1: Got a big evaluation come back when you're executing on

129
00:06:31,600 --> 00:06:34,080
Speaker 1: the latest round. Thank you, Jim Keller of ted Stone.

130
00:06:34,120 --> 00:06:35,320
Speaker 1: All right, great to have you.