1
00:00:02,520 --> 00:00:13,360
Speaker 1: Bloomberg Audio Studios, podcasts, radio news. Bloomberg Tech is alive

2
00:00:13,440 --> 00:00:17,239
Speaker 1: from coast to coast with Caroline Hyde in New York

3
00:00:17,520 --> 00:00:19,520
Speaker 1: and Eva though in San Francisco.

4
00:00:21,600 --> 00:00:24,320
Speaker 2: This is Bloomberg Tech coming up. We go to AWS

5
00:00:24,400 --> 00:00:27,600
Speaker 2: we Invent in Las Vegas to discuss the cloud companies,

6
00:00:27,640 --> 00:00:31,480
Speaker 2: new chips, new models, and new AI agents. Becaus Michael

7
00:00:31,520 --> 00:00:35,280
Speaker 2: Dell donates an unprecedented six billion dollars aimed at jump

8
00:00:35,280 --> 00:00:38,760
Speaker 2: starting the investment accounts for twenty five million children in America,

9
00:00:39,440 --> 00:00:42,040
Speaker 2: and Warner Brothers Discovery receives a new round of bids,

10
00:00:42,280 --> 00:00:45,960
Speaker 2: with Netflix flashing mostly cash for its offer. We'll have

11
00:00:46,040 --> 00:00:48,400
Speaker 2: the details, but first let's check in on these markets

12
00:00:48,640 --> 00:00:50,519
Speaker 2: which are flashing green. We actually have a bit of

13
00:00:50,520 --> 00:00:52,960
Speaker 2: a reprieve after yesterday's sell off. We're back into risk

14
00:00:53,080 --> 00:00:56,040
Speaker 2: on mode tentatively. So across the benchmarks, we're looking at

15
00:00:56,080 --> 00:00:58,800
Speaker 2: the NaSTA one hundred nineteen percent of course in video

16
00:00:58,880 --> 00:01:01,120
Speaker 2: leading then in terms of points, but all of some

17
00:01:01,160 --> 00:01:02,640
Speaker 2: of the mag seven really dominating.

18
00:01:02,680 --> 00:01:03,160
Speaker 3: On the day.

19
00:01:03,400 --> 00:01:05,959
Speaker 2: You're looking at bitcoin, even getting a bit yesterday it

20
00:01:06,040 --> 00:01:09,400
Speaker 2: was woeful. Today we find some sort of stability where

21
00:01:09,440 --> 00:01:12,360
Speaker 2: up more than five percent in fact, largely Crypto is

22
00:01:12,400 --> 00:01:14,480
Speaker 2: in the green. We also turn our attention to some

23
00:01:14,520 --> 00:01:15,959
Speaker 2: of the key mags seven that we want to look

24
00:01:15,959 --> 00:01:18,280
Speaker 2: at in terms of their own events, their own announcements.

25
00:01:18,520 --> 00:01:20,479
Speaker 2: And I'm looking at what's happening with Amazon. We're currently

26
00:01:20,480 --> 00:01:23,240
Speaker 2: training up one point three percent. We're getting a little

27
00:01:23,280 --> 00:01:25,560
Speaker 2: nudge higher on some new news coming out about its

28
00:01:25,600 --> 00:01:28,840
Speaker 2: own large language models, about updates of course to its

29
00:01:28,840 --> 00:01:32,560
Speaker 2: own AI agentic focus. And key has got to be

30
00:01:32,959 --> 00:01:35,039
Speaker 2: Trainium three all about its chips.

31
00:01:35,120 --> 00:01:36,720
Speaker 3: Let's get straight over to Ed Ludlow.

32
00:01:36,760 --> 00:01:39,240
Speaker 2: You're in Las Vegas at the reinvent event.

33
00:01:41,959 --> 00:01:44,319
Speaker 4: Yeah, it's probably the Trainium three headlines that move the

34
00:01:44,360 --> 00:01:47,720
Speaker 4: needle right. This is a push forward acceleration of bringing

35
00:01:48,040 --> 00:01:52,480
Speaker 4: the latest generation accelerator to market called Trainium, but useful

36
00:01:52,480 --> 00:01:55,720
Speaker 4: in both the training and inference use case, and much

37
00:01:55,720 --> 00:01:59,360
Speaker 4: as Amazon has done in prygation generations of Trainium, it's

38
00:01:59,400 --> 00:02:02,840
Speaker 4: talking out the cost and performance metrics of it four

39
00:02:03,000 --> 00:02:06,280
Speaker 4: x the prior generation. On price performance, it was interesting

40
00:02:06,280 --> 00:02:07,600
Speaker 4: when the headline's hit, I mean, we're at one and

41
00:02:07,600 --> 00:02:11,120
Speaker 4: a half percent on Amazon, a small move lower or

42
00:02:11,160 --> 00:02:14,120
Speaker 4: pairing some of the gain on both Google and Nvidia.

43
00:02:14,200 --> 00:02:16,760
Speaker 4: And this is the story right. You know, Amazon wants

44
00:02:16,800 --> 00:02:22,320
Speaker 4: to expand the use of Trainium servers beyond one core customer,

45
00:02:22,320 --> 00:02:25,440
Speaker 4: which is anthropic, and they're giving evidence in this release.

46
00:02:25,520 --> 00:02:26,840
Speaker 4: We'll get to it later in the day when we

47
00:02:26,840 --> 00:02:31,120
Speaker 4: speak to Matt Garmon in particular of the savings that

48
00:02:31,160 --> 00:02:33,600
Speaker 4: those customers are made using it while also making sure

49
00:02:33,639 --> 00:02:36,640
Speaker 4: they have capacity that is based in video GPUs and

50
00:02:36,720 --> 00:02:40,080
Speaker 4: of course the open aideal, the AWS has thirty eight

51
00:02:40,120 --> 00:02:43,160
Speaker 4: billion dollars of it is predicated on availability of Nvidia.

52
00:02:43,200 --> 00:02:46,280
Speaker 2: I mean the eye the focus must be so trained

53
00:02:46,440 --> 00:02:50,960
Speaker 2: on what's just been seemingly a positive thrust for Google

54
00:02:51,200 --> 00:02:53,520
Speaker 2: when it all comes down to its own chips right now.

55
00:02:56,200 --> 00:02:58,680
Speaker 4: Yeah, in recent weeks and months when there have been

56
00:02:58,760 --> 00:03:03,360
Speaker 4: reports about Google TPU and the idea that Google Cloud

57
00:03:03,600 --> 00:03:06,640
Speaker 4: will have a big third party use it, it's moved

58
00:03:06,919 --> 00:03:11,920
Speaker 4: the needle on those shares. What universally right now, the

59
00:03:12,000 --> 00:03:16,200
Speaker 4: hyperscalers are experience the supply constraints right comply supply constraints

60
00:03:16,240 --> 00:03:18,840
Speaker 4: where they're basically saying, this is out there in the

61
00:03:18,800 --> 00:03:20,679
Speaker 4: real on Trainium three. It is out there in the

62
00:03:20,760 --> 00:03:23,519
Speaker 4: real world. By the way, general availability from today and

63
00:03:23,560 --> 00:03:26,320
Speaker 4: they've also talked about the pipeline through the Trainium four.

64
00:03:26,320 --> 00:03:29,600
Speaker 2: But you're right, it's all about TPU, it's all about chips.

65
00:03:29,600 --> 00:03:31,959
Speaker 2: It's also about their own innovations when it comes to

66
00:03:32,040 --> 00:03:34,400
Speaker 2: large language models lest we forget. Yes, they've got an

67
00:03:34,400 --> 00:03:37,520
Speaker 2: integration of open AI and anthropic Claude, but they've also

68
00:03:37,600 --> 00:03:39,440
Speaker 2: been movieving the needle on over two.

69
00:03:42,120 --> 00:03:46,080
Speaker 4: Yeah, an over two and updated year on version of

70
00:03:46,120 --> 00:03:49,880
Speaker 4: its first kind of in house foundation model, a variant

71
00:03:49,880 --> 00:03:53,280
Speaker 4: of which OMNI can take inputs MTI model inputs of text, video,

72
00:03:53,320 --> 00:03:56,440
Speaker 4: image responding kind. But that's going to be the question

73
00:03:56,480 --> 00:03:59,240
Speaker 4: the story today. You know, AWS is number one in

74
00:03:59,240 --> 00:04:03,560
Speaker 4: cloud and cloud computing through scale, through leveraging infrastructure, but

75
00:04:03,600 --> 00:04:05,880
Speaker 4: when does it become number one in AI? You know,

76
00:04:05,960 --> 00:04:08,440
Speaker 4: not just being a place that hosts others models, but

77
00:04:08,560 --> 00:04:11,640
Speaker 4: has evidence that its own foundation model is resulting in

78
00:04:11,720 --> 00:04:15,520
Speaker 4: something tangibly useful, particularly for enterprise customers. And I'm pretty

79
00:04:15,520 --> 00:04:17,480
Speaker 4: sure that that's what those that have joined the show

80
00:04:17,480 --> 00:04:18,360
Speaker 4: today want to talk about.

81
00:04:18,600 --> 00:04:21,559
Speaker 2: Yeah, we've got so many conversations come up in this show,

82
00:04:21,800 --> 00:04:24,080
Speaker 2: but you've got to be sticking around because we thank you, ed.

83
00:04:24,200 --> 00:04:27,360
Speaker 2: We're coming back to you because Throughout the show we'll

84
00:04:27,360 --> 00:04:29,320
Speaker 2: have interviews, but at the end we'll have a sit

85
00:04:29,360 --> 00:04:34,360
Speaker 2: down interview with AWSCO Matt Garmon at three pm Eastern time. Meanwhile, look,

86
00:04:34,360 --> 00:04:36,360
Speaker 2: this is all a story of infrastructure, about big tech

87
00:04:36,400 --> 00:04:39,560
Speaker 2: companies such as Amazon, but also Alphabet, Meta and Microsoft

88
00:04:39,880 --> 00:04:44,320
Speaker 2: spending heavily on AI with expected capital expenditure forever three

89
00:04:44,400 --> 00:04:47,520
Speaker 2: round eighty billion dollars combined in their current fiscal years. Now,

90
00:04:47,520 --> 00:04:50,799
Speaker 2: that's going against big tech's mantra for the past two decades.

91
00:04:50,839 --> 00:04:51,440
Speaker 3: Let's call it.

92
00:04:51,720 --> 00:04:54,599
Speaker 2: We are just more about delivering growth while keeping really

93
00:04:54,680 --> 00:04:55,800
Speaker 2: tight little spending.

94
00:04:56,120 --> 00:04:57,080
Speaker 3: It's a real flip reverse.

95
00:04:57,120 --> 00:04:59,839
Speaker 2: Let's talk about Oblimbug's tech equity reporter Carmen ryani Key,

96
00:05:00,080 --> 00:05:03,880
Speaker 2: and you've all been assessing how investor is tight. The

97
00:05:03,920 --> 00:05:06,240
Speaker 2: sudden wild capital expenditure.

98
00:05:06,600 --> 00:05:08,919
Speaker 5: Yeah, I mean, it has been the talking point and

99
00:05:08,920 --> 00:05:11,280
Speaker 5: the thing that we've been focusing on the most with

100
00:05:11,320 --> 00:05:13,719
Speaker 5: some of these big companies, and as you said, it

101
00:05:13,839 --> 00:05:17,039
Speaker 5: really represents a change from how they've operated over the

102
00:05:17,080 --> 00:05:20,720
Speaker 5: last two decades. They really had very capital light businesses

103
00:05:20,720 --> 00:05:24,520
Speaker 5: and were able to be very profitable, grow exponentially in

104
00:05:24,560 --> 00:05:27,360
Speaker 5: some ways because of that, and so now having this shift.

105
00:05:27,440 --> 00:05:30,600
Speaker 5: We're seeing investors rewarded on summons and then really punish

106
00:05:30,600 --> 00:05:33,520
Speaker 5: it on others. So, you know, Microsoft, we saw you

107
00:05:33,880 --> 00:05:36,560
Speaker 5: jump after its latest starting support, but Meta sort of

108
00:05:36,600 --> 00:05:40,120
Speaker 5: got hit because Zuckerberg, you know, didn't explain enough about

109
00:05:40,120 --> 00:05:42,640
Speaker 5: the return on investment of this spending. The other thing

110
00:05:42,680 --> 00:05:45,640
Speaker 5: that's interesting is that this capital expenditures is a portion

111
00:05:45,720 --> 00:05:48,560
Speaker 5: of revenue have jumped, which to a level that's not

112
00:05:48,680 --> 00:05:52,520
Speaker 5: normal for tech companies. So for Microsoft, it's capital expenditures

113
00:05:52,520 --> 00:05:55,440
Speaker 5: are now twenty percent of its revenue. And in addition,

114
00:05:55,560 --> 00:05:58,840
Speaker 5: you know, Alphabet and Amazon, these spending to sales ratios

115
00:05:58,880 --> 00:06:00,800
Speaker 5: are some of the high in the market.

116
00:06:01,240 --> 00:06:05,680
Speaker 2: Okay, so we now have to perhaps digest a period

117
00:06:05,720 --> 00:06:08,840
Speaker 2: of higher capital expenditure. Were meanwhile worrying about some of

118
00:06:08,880 --> 00:06:10,800
Speaker 2: the financing that goes on within it, some of the

119
00:06:10,800 --> 00:06:14,040
Speaker 2: circular financing that's been the narrative. But today the market jumps.

120
00:06:14,640 --> 00:06:16,719
Speaker 3: Why what are you hearing from the investors you're talking to?

121
00:06:16,960 --> 00:06:19,799
Speaker 5: Yeah, I mean, I think that last week was very interesting.

122
00:06:19,800 --> 00:06:21,720
Speaker 5: I think we're seeing a little bit of buying, some

123
00:06:21,800 --> 00:06:25,640
Speaker 5: more optimism and enthusiasm coming back into the market. There's

124
00:06:25,640 --> 00:06:28,520
Speaker 5: a lot you know, in videos CFO was speaking this morning.

125
00:06:28,560 --> 00:06:31,040
Speaker 5: There's a there's just a lot of tech news sort

126
00:06:31,040 --> 00:06:32,880
Speaker 5: of coming back into the market that people are getting

127
00:06:32,880 --> 00:06:33,640
Speaker 5: excited about.

128
00:06:34,200 --> 00:06:35,360
Speaker 3: Well, i'll see if it holds.

129
00:06:35,560 --> 00:06:38,279
Speaker 2: Santa rally upon us potentially come and Ranicky, We so

130
00:06:38,360 --> 00:06:41,679
Speaker 2: appreciate your time. Let's talk about the crypto markets, because

131
00:06:41,720 --> 00:06:44,479
Speaker 2: they too are recovering today off that a heavy sell

132
00:06:44,560 --> 00:06:47,320
Speaker 2: of yesterday when bitcoin still as much as eight percent.

133
00:06:47,839 --> 00:06:49,800
Speaker 2: Let's take a look at strategy as well. The artist

134
00:06:49,880 --> 00:06:54,359
Speaker 2: formerly known as MicroStrategy, heavy investor in bitcoin gaining some

135
00:06:54,480 --> 00:06:57,359
Speaker 2: support after scrambling to calm its own investors, creating a

136
00:06:57,440 --> 00:07:00,159
Speaker 2: one point four billion dollar reserve to fund dividend and

137
00:07:00,240 --> 00:07:04,440
Speaker 2: interest payments rather than having to potentially sell bitcoin as

138
00:07:04,480 --> 00:07:07,159
Speaker 2: a last resort. And if I've seen a digital finance editor,

139
00:07:07,200 --> 00:07:10,400
Speaker 2: Anna Erera is joining us. What's more from London? Anna,

140
00:07:11,080 --> 00:07:12,400
Speaker 2: why today the did buying.

141
00:07:14,760 --> 00:07:17,400
Speaker 6: It's hard to know with crypto, as I'm sure you're

142
00:07:17,400 --> 00:07:19,520
Speaker 6: aware of, and it's crazy to think that yesterday we

143
00:07:19,520 --> 00:07:21,920
Speaker 6: were down eight percent and thinking of all the macro

144
00:07:22,680 --> 00:07:25,080
Speaker 6: condition that might have been leading to that drop. So

145
00:07:25,400 --> 00:07:28,400
Speaker 6: you know, we were speaking to traders today and investors,

146
00:07:28,440 --> 00:07:30,840
Speaker 6: and it seemed that although the price is up, the

147
00:07:30,920 --> 00:07:34,480
Speaker 6: sentiment is still a bit cautious, and volumes aren't as

148
00:07:34,560 --> 00:07:37,880
Speaker 6: up as they were before. I think people are sort

149
00:07:37,920 --> 00:07:41,160
Speaker 6: of assessing that crypto in reality has been on a

150
00:07:41,240 --> 00:07:45,320
Speaker 6: downward spiral of bit since October twelfth, the eleventh, when

151
00:07:45,320 --> 00:07:47,880
Speaker 6: the price crashed, And so while today we're a bit up,

152
00:07:47,960 --> 00:07:51,360
Speaker 6: in reality, you know, we're still pretty lower than we

153
00:07:51,360 --> 00:07:54,680
Speaker 6: were around a month ago when bitcoin reached its record highs.

154
00:07:54,760 --> 00:07:57,280
Speaker 2: I mean, negative bitcoin funding rate is still an anxiety,

155
00:07:57,320 --> 00:07:59,840
Speaker 2: extreme fear levels with coin market caps, fear and greed.

156
00:08:00,280 --> 00:08:04,240
Speaker 2: There's still a lot of comfort needed to an investor

157
00:08:04,240 --> 00:08:06,880
Speaker 2: base that's been beaten up, particularly the retail trade who

158
00:08:06,840 --> 00:08:09,400
Speaker 2: have tried to gain exposure not just by owning crypto,

159
00:08:09,680 --> 00:08:12,920
Speaker 2: but by owning crypto related assets such as what leveraged

160
00:08:13,720 --> 00:08:16,560
Speaker 2: bets on MicroStrategy now called Strategy and it talked to

161
00:08:16,640 --> 00:08:18,960
Speaker 2: us about how much retail of hurt here as well

162
00:08:18,960 --> 00:08:19,800
Speaker 2: as institutional.

163
00:08:20,840 --> 00:08:23,280
Speaker 6: Yes, obviously we know that people were trying to retail

164
00:08:23,280 --> 00:08:25,720
Speaker 6: investor are trying to get exposure into bitcoin and crypto

165
00:08:25,800 --> 00:08:29,800
Speaker 6: through buying ETFs. Thatt, we're supposed to strategy or strategy

166
00:08:29,840 --> 00:08:34,080
Speaker 6: stock itself, and that's fallen dramatically over the past year,

167
00:08:34,200 --> 00:08:36,079
Speaker 6: and so you know they've been hurting. But at the

168
00:08:36,120 --> 00:08:37,640
Speaker 6: same time we've been trying to wait to see if

169
00:08:37,640 --> 00:08:39,800
Speaker 6: more institutional buyers will come in and offset that. But

170
00:08:39,840 --> 00:08:41,600
Speaker 6: it seems to be from what we're hearing, a sort

171
00:08:41,600 --> 00:08:45,120
Speaker 6: of weight and see mode, partly waiting until the FED

172
00:08:45,120 --> 00:08:46,719
Speaker 6: decision next week. So it's kind of in a way,

173
00:08:46,800 --> 00:08:49,000
Speaker 6: same old with crypto. One day it goes ninety percent,

174
00:08:49,040 --> 00:08:51,280
Speaker 6: the next thaying it's a five percent. The asset class

175
00:08:51,280 --> 00:08:53,600
Speaker 6: seems to be maturing and there's more ways to invest

176
00:08:53,600 --> 00:08:56,160
Speaker 6: in it, but maybe not much has changed.

177
00:08:56,520 --> 00:08:57,400
Speaker 3: In Bergana or Era.

178
00:08:57,720 --> 00:08:59,400
Speaker 2: Always great to have you, Thank you very much, and

179
00:08:59,440 --> 00:09:02,120
Speaker 2: stay Withloomberg because there's a conversation coming up you won't

180
00:09:02,120 --> 00:09:05,040
Speaker 2: want to miss if you're in crypto Fondly Strategy President

181
00:09:05,160 --> 00:09:08,360
Speaker 2: CEO coming up in the next hour on Bloomberg Crypto.

182
00:09:08,679 --> 00:09:12,880
Speaker 2: Elsewhere in that ecosystem, not all is gaining support. We're

183
00:09:12,920 --> 00:09:16,480
Speaker 2: watching shares of American Bitcoin Corp. Look at that forty

184
00:09:16,480 --> 00:09:18,920
Speaker 2: four percent at one point down fifty percent. A bitcoin

185
00:09:19,000 --> 00:09:21,280
Speaker 2: minor which counts Eric Trump as a co founder and

186
00:09:21,400 --> 00:09:26,920
Speaker 2: chief strategy officer, seeing multiple trading halts amid the volatility

187
00:09:27,320 --> 00:09:30,200
Speaker 2: now coming up, we'll be joined by Colleen Aubrey, Senior

188
00:09:30,200 --> 00:09:33,319
Speaker 2: and Vice President of Applied AI Solutions over at AWS.

189
00:09:33,400 --> 00:09:36,079
Speaker 2: We are live from the Companies we Invent conference in

190
00:09:36,160 --> 00:09:36,760
Speaker 2: Las Vegas.

191
00:09:37,160 --> 00:09:38,120
Speaker 3: This is Bloomberg Tech.

192
00:09:53,240 --> 00:09:55,320
Speaker 4: Good morning from Las Vegas. So let's talk about some

193
00:09:55,320 --> 00:09:58,520
Speaker 4: of the announcements out from AWS. Colleen Aubrey is Senior

194
00:09:58,600 --> 00:10:03,000
Speaker 4: vice president of AWS, Applied AI Solutions and clean You

195
00:10:03,120 --> 00:10:06,360
Speaker 4: basically work with the full spectrum of the smallest startups

196
00:10:06,600 --> 00:10:09,080
Speaker 4: to the biggest enterprise customers that AWS has and the

197
00:10:09,120 --> 00:10:11,960
Speaker 4: public sector, and it's distilling it as far down as

198
00:10:11,960 --> 00:10:14,840
Speaker 4: I can. You basically help them to take technology and

199
00:10:14,960 --> 00:10:17,040
Speaker 4: innovate make whatever they do better.

200
00:10:17,280 --> 00:10:20,679
Speaker 7: Yeah, the mission I'm on is really to put AI

201
00:10:20,800 --> 00:10:22,800
Speaker 7: into the hands of the business on a day to

202
00:10:22,880 --> 00:10:25,319
Speaker 7: day basis and really make it work for their business

203
00:10:25,360 --> 00:10:28,840
Speaker 7: in production every day, helping them to deliver better experiences

204
00:10:28,880 --> 00:10:29,600
Speaker 7: to their customers.

205
00:10:29,640 --> 00:10:34,040
Speaker 4: We have a new generation of AWS Silicon and server design,

206
00:10:34,559 --> 00:10:37,400
Speaker 4: We have a new generation of Nova model, and we

207
00:10:37,440 --> 00:10:40,319
Speaker 4: have frontier agents. Right and I think you probably heard

208
00:10:40,640 --> 00:10:42,640
Speaker 4: me at the top of the show, but I think

209
00:10:42,800 --> 00:10:46,400
Speaker 4: the big question about AWS Reinvent is what is it

210
00:10:46,480 --> 00:10:50,719
Speaker 4: in this year's offering that takes aws beyond being the

211
00:10:50,800 --> 00:10:54,200
Speaker 4: sort of number one cloud computing platform for capacity to

212
00:10:54,640 --> 00:10:59,200
Speaker 4: giving something tangible, useful that can improve a company's own technology.

213
00:10:59,280 --> 00:11:03,040
Speaker 7: Yeah, two areas that I'd like to talk about specifically

214
00:11:03,880 --> 00:11:06,319
Speaker 7: for me personally, one of my products is Amazon Connect

215
00:11:06,920 --> 00:11:10,520
Speaker 7: and this started as a contact center application, but today

216
00:11:10,520 --> 00:11:12,600
Speaker 7: you know, I see it really as an agentic product

217
00:11:12,640 --> 00:11:15,000
Speaker 7: that's actually looking at whole customer experience. And so for me,

218
00:11:15,080 --> 00:11:18,960
Speaker 7: this is where we're putting our infrastructure, our silicon, our models,

219
00:11:19,160 --> 00:11:21,560
Speaker 7: and now today like bringing that all together in a

220
00:11:21,600 --> 00:11:24,320
Speaker 7: product that can work for businesses. So we launched twenty

221
00:11:24,400 --> 00:11:27,120
Speaker 7: nine new features on Sunday we announced that, and I

222
00:11:27,120 --> 00:11:30,240
Speaker 7: would say there's four key gentic capabilities that come with that.

223
00:11:30,600 --> 00:11:34,200
Speaker 7: The first is actually AI Voice and allowing customers to

224
00:11:34,240 --> 00:11:37,720
Speaker 7: interact in a very natural way using Novasonic and having

225
00:11:37,800 --> 00:11:40,880
Speaker 7: agents actually resolve issues for them on the behalf in

226
00:11:40,920 --> 00:11:43,960
Speaker 7: the background. The second is actually putting AI to work

227
00:11:44,000 --> 00:11:47,480
Speaker 7: as a teammate next to customer service representatives, helping them

228
00:11:47,480 --> 00:11:50,800
Speaker 7: to actually get tasked on complete the paperwork the processes,

229
00:11:50,960 --> 00:11:53,720
Speaker 7: provide recommendations, and help them really have a better view

230
00:11:53,720 --> 00:11:56,200
Speaker 7: of the customer so that the conversation they have with

231
00:11:56,320 --> 00:12:00,000
Speaker 7: customers is much much richer. The third area is actually

232
00:12:00,200 --> 00:12:03,240
Speaker 7: combining what I call clickstream, which is the path that

233
00:12:03,280 --> 00:12:07,000
Speaker 7: customers will take through websites and profiles, to be able

234
00:12:07,040 --> 00:12:10,400
Speaker 7: to present a much more specific recommendation and what a

235
00:12:10,480 --> 00:12:13,079
Speaker 7: next step won't be for a customer. And finally, you know,

236
00:12:13,120 --> 00:12:15,720
Speaker 7: one of the big issues for companies is sort of

237
00:12:15,920 --> 00:12:18,760
Speaker 7: confidently putting AI to work in their business. And in

238
00:12:18,760 --> 00:12:22,240
Speaker 7: this case, we've added observability where you can expect, how

239
00:12:22,640 --> 00:12:25,599
Speaker 7: inspect how an AI is reasoning, how it's thinking, in

240
00:12:25,760 --> 00:12:29,199
Speaker 7: what tools it's using, so companies can really observe AI

241
00:12:29,240 --> 00:12:31,200
Speaker 7: in the same way they would think about the people

242
00:12:31,240 --> 00:12:33,040
Speaker 7: and their business working with customers.

243
00:12:33,040 --> 00:12:34,520
Speaker 4: I want to go to point number two. You use

244
00:12:34,559 --> 00:12:37,720
Speaker 4: the word teammates, yes, but you've in the past talked

245
00:12:37,760 --> 00:12:42,600
Speaker 4: about a hybrid workforce, people and agents. A lot of

246
00:12:42,640 --> 00:12:45,560
Speaker 4: what's come out since Sunday Night seems to reflect that

247
00:12:46,200 --> 00:12:50,439
Speaker 4: you want to increase the number of useful agents. That also, internally,

248
00:12:50,480 --> 00:12:52,680
Speaker 4: we can talk a little bit about what you're encouraging

249
00:12:52,720 --> 00:12:55,440
Speaker 4: Amazon teams to do, but the idea is that that

250
00:12:55,559 --> 00:12:59,760
Speaker 4: workforces will type should get comfortable working alongside agentic AI.

251
00:13:00,080 --> 00:13:00,320
Speaker 8: Yeah.

252
00:13:00,760 --> 00:13:03,720
Speaker 7: I see a future where actually everyone is managing a

253
00:13:03,760 --> 00:13:06,640
Speaker 7: team of AI agents. And I think our frontier agents

254
00:13:07,000 --> 00:13:11,160
Speaker 7: with software developments, sec ops and so DevOps and security

255
00:13:11,400 --> 00:13:13,480
Speaker 7: is sort of one of our first moves in that direction.

256
00:13:13,520 --> 00:13:17,959
Speaker 7: Where you actually have a teammate with a developer who's

257
00:13:18,000 --> 00:13:21,439
Speaker 7: able to address complex problems, able to work over hours and.

258
00:13:21,440 --> 00:13:23,360
Speaker 9: Days, be able to solve for whole.

259
00:13:23,160 --> 00:13:27,000
Speaker 7: Objectives, and to scale with that person, that the role changes.

260
00:13:27,080 --> 00:13:30,360
Speaker 7: I think we all are managing AI teammates, a team

261
00:13:30,840 --> 00:13:33,040
Speaker 7: of people that are all of AI, people that are

262
00:13:33,080 --> 00:13:36,520
Speaker 7: out able to we can delegate to, we can inspect

263
00:13:36,520 --> 00:13:39,079
Speaker 7: what they're doing, we can iterate with them, provide feedback,

264
00:13:39,360 --> 00:13:41,800
Speaker 7: and for me, that's where I think we end up going.

265
00:13:42,120 --> 00:13:44,320
Speaker 7: I think we're clearly moving that way in the customer

266
00:13:44,400 --> 00:13:47,520
Speaker 7: service direction, and I think the developer experience is also

267
00:13:47,559 --> 00:13:48,480
Speaker 7: clearly moving that direction.

268
00:13:48,559 --> 00:13:51,000
Speaker 4: You're in a leadership position at AWS, but you have

269
00:13:51,080 --> 00:13:54,400
Speaker 4: been with Amazon broadly entering your twenty first year. You're

270
00:13:54,440 --> 00:13:58,199
Speaker 4: part of the S team, the leadership team generally internally

271
00:13:58,240 --> 00:14:02,640
Speaker 4: within Amazon's many arms. Is this like, particularly in the

272
00:14:02,640 --> 00:14:05,920
Speaker 4: context of the technology you've released here Reinvent twenty twenty five,

273
00:14:06,000 --> 00:14:08,280
Speaker 4: Is this at a stage where Amazon teams are dog

274
00:14:08,320 --> 00:14:11,960
Speaker 4: fooding this or it's just inherent the use of technology.

275
00:14:12,040 --> 00:14:14,440
Speaker 4: You want your own teams to be. I think AI native.

276
00:14:14,600 --> 00:14:17,360
Speaker 7: Yeah, that's true, and I think what I observe within

277
00:14:17,440 --> 00:14:19,920
Speaker 7: Amazon certainly we're all very keen to get our hands

278
00:14:19,960 --> 00:14:22,320
Speaker 7: on our new frontier agents, and that is something we've

279
00:14:22,720 --> 00:14:24,800
Speaker 7: had in beta internally, and I think teams are very

280
00:14:24,800 --> 00:14:28,720
Speaker 7: excited to use that in production. We use connect within

281
00:14:28,760 --> 00:14:31,160
Speaker 7: our own customer service teams, our seller support teams, our

282
00:14:31,200 --> 00:14:34,160
Speaker 7: AWS support teams, for example. So we continue like to

283
00:14:34,160 --> 00:14:36,880
Speaker 7: iterate and learn from that. And I would say broadly

284
00:14:36,880 --> 00:14:42,240
Speaker 7: across Amazon, we've really tried to sort of embrace the chaos,

285
00:14:42,440 --> 00:14:45,320
Speaker 7: to put AI in the hands of every person in

286
00:14:45,360 --> 00:14:48,080
Speaker 7: the company and to see what they can do, how

287
00:14:48,160 --> 00:14:50,720
Speaker 7: they can transform how they work, what they learn. And

288
00:14:50,760 --> 00:14:53,880
Speaker 7: so you have a lot of teams sort of grassroots,

289
00:14:54,240 --> 00:14:56,240
Speaker 7: sort of trying to solve for what is the new

290
00:14:56,280 --> 00:14:59,359
Speaker 7: way of working using AI, and some of those experiments

291
00:14:59,480 --> 00:15:02,320
Speaker 7: don't lead in newhere, and some lead and a place

292
00:15:02,360 --> 00:15:04,040
Speaker 7: that's really meaningful. And then you have sort of this

293
00:15:04,120 --> 00:15:06,520
Speaker 7: social contagion that happens that people learn from each other.

294
00:15:06,560 --> 00:15:09,920
Speaker 4: Okay, aws number one in cloud computing and in the

295
00:15:09,960 --> 00:15:14,480
Speaker 4: sense of scale infrastructure deployments, is it number one in AI?

296
00:15:15,280 --> 00:15:18,080
Speaker 7: I think we have all the pieces in place and

297
00:15:18,120 --> 00:15:20,680
Speaker 7: we're well on our way. We're very, very focused, and

298
00:15:20,760 --> 00:15:22,960
Speaker 7: what I really like about our strategy is that we're

299
00:15:23,000 --> 00:15:25,480
Speaker 7: working all the way, all the way, the full stack,

300
00:15:25,720 --> 00:15:30,560
Speaker 7: you know, from clean energy, chips, data centers around the world,

301
00:15:31,560 --> 00:15:34,680
Speaker 7: bedrock with real great selection, including our own models. And

302
00:15:34,720 --> 00:15:37,320
Speaker 7: I think Roehet's making really fast progress on that. And

303
00:15:37,320 --> 00:15:39,360
Speaker 7: I think for Swami and I really trying to put

304
00:15:39,400 --> 00:15:43,280
Speaker 7: those all of those pieces to work where out of

305
00:15:43,320 --> 00:15:45,560
Speaker 7: the box a business can get value.

306
00:15:45,600 --> 00:15:47,480
Speaker 4: We've got a conversation with Swami coming up in a bit.

307
00:15:47,560 --> 00:15:51,400
Speaker 4: Colleen Aubrey, SVP of Applied AI Solutions to AWS, Thank

308
00:15:51,480 --> 00:15:52,720
Speaker 4: you so much, Carrot, what.

309
00:15:52,800 --> 00:15:56,160
Speaker 2: A great conversation. Let's stick with Amazon as well, because

310
00:15:56,240 --> 00:15:58,920
Speaker 2: the company, which is vast it plans to offer deliveries

311
00:15:58,920 --> 00:16:02,440
Speaker 2: of hundreds of house items, including some fresh groceries or

312
00:16:02,440 --> 00:16:05,880
Speaker 2: over the counter medicines within just thirty minutes and test program.

313
00:16:05,920 --> 00:16:08,840
Speaker 2: But it's begun in Philadelphia, we stand will begin there

314
00:16:08,880 --> 00:16:12,480
Speaker 2: and its home city of Seattle now coming up the

315
00:16:12,600 --> 00:16:16,640
Speaker 2: Dells make an unprecedented six point twenty five billion dollar

316
00:16:16,680 --> 00:16:18,600
Speaker 2: donation to the Kids of America or on that.

317
00:16:18,640 --> 00:16:20,000
Speaker 3: Next, this is Bloomberg Tech.

318
00:16:34,920 --> 00:16:38,120
Speaker 2: Michael and Susan Dell are donating six point twenty five

319
00:16:38,160 --> 00:16:41,360
Speaker 2: billion dollars, gifting twenty five million children in America two

320
00:16:41,400 --> 00:16:43,760
Speaker 2: hundred and fifty dollars each in an effort to jumpstart

321
00:16:43,800 --> 00:16:47,000
Speaker 2: their investment accounts for the future. The proceeds build on

322
00:16:47,040 --> 00:16:50,600
Speaker 2: the Invest America Initiative or Trump accounts as they're known,

323
00:16:50,720 --> 00:16:52,880
Speaker 2: which we'll see one thousand dollars for every child born

324
00:16:52,920 --> 00:16:55,200
Speaker 2: from twenty twenty five to twenty twenty eight. Let's talk

325
00:16:55,200 --> 00:16:57,920
Speaker 2: about all of it with bost Tom Maloney and how

326
00:16:58,000 --> 00:17:01,240
Speaker 2: unprecedented is this? This SI is a philanthropic gift.

327
00:17:02,280 --> 00:17:06,119
Speaker 10: It's pretty unprecedtentded I mean, there've been big philanthropic gifts before,

328
00:17:06,200 --> 00:17:09,320
Speaker 10: but something going to cover twenty five million children in

329
00:17:09,359 --> 00:17:12,840
Speaker 10: the cover really is pretty unheard of. And a gift

330
00:17:12,920 --> 00:17:15,040
Speaker 10: to go through the federal government. I mean, he's donating

331
00:17:15,080 --> 00:17:18,040
Speaker 10: the money to the US Department of Treasury. That's pretty

332
00:17:18,119 --> 00:17:21,440
Speaker 10: unusual too, So it's quite an unprecedented gift.

333
00:17:21,880 --> 00:17:23,680
Speaker 2: Let's go back to what all of this is sort

334
00:17:23,680 --> 00:17:27,280
Speaker 2: of sitting alongside, which is invest in America. We sort

335
00:17:27,280 --> 00:17:29,879
Speaker 2: of saw that roundtable alongside other CEOs, a lot of

336
00:17:29,880 --> 00:17:32,640
Speaker 2: them tech who have been thinking about how to get

337
00:17:32,680 --> 00:17:35,399
Speaker 2: money into the hands of children to invest in the

338
00:17:35,480 --> 00:17:35,960
Speaker 2: longer term.

339
00:17:36,000 --> 00:17:38,120
Speaker 4: Tom, that's right.

340
00:17:38,200 --> 00:17:40,120
Speaker 10: Yeah, those were created as part of the One Big

341
00:17:40,160 --> 00:17:43,800
Speaker 10: Beautiful Bill, as you mentioned, one thousand dollars for kids

342
00:17:43,800 --> 00:17:46,040
Speaker 10: born between twenty twenty five and twenty twenty eight. And

343
00:17:46,119 --> 00:17:48,920
Speaker 10: what Della is doing is kind of covering the gap

344
00:17:49,000 --> 00:17:52,280
Speaker 10: for children ten under who aren't covered under that program

345
00:17:52,440 --> 00:17:55,520
Speaker 10: and giving them two hundred and fifty dollars to be

346
00:17:55,640 --> 00:17:59,000
Speaker 10: used for things like college or startup, starting up their

347
00:17:59,040 --> 00:18:02,760
Speaker 10: own business, or home deposit when they're eighteen or older.

348
00:18:03,320 --> 00:18:06,520
Speaker 2: Now, Michael Dell and his family and he's the eleventh

349
00:18:06,840 --> 00:18:11,200
Speaker 2: wealthiest person in the world, but it's other tech moneyed

350
00:18:11,560 --> 00:18:13,679
Speaker 2: people who have been putting this initiative to work. And

351
00:18:13,720 --> 00:18:16,520
Speaker 2: I think about really Brad Gerstner, who has been the

352
00:18:16,600 --> 00:18:19,920
Speaker 2: driving force initially behind what is this sort of invest

353
00:18:20,000 --> 00:18:22,760
Speaker 2: America outreach. He set it up back in twenty twenty three.

354
00:18:22,800 --> 00:18:24,360
Speaker 2: Of course we know him from Ultimeter.

355
00:18:25,080 --> 00:18:27,040
Speaker 10: Yeah, that's right. He's been working on this for a

356
00:18:27,080 --> 00:18:30,879
Speaker 10: couple of years. As you mentioned it predates Trump obviously,

357
00:18:32,160 --> 00:18:36,680
Speaker 10: you know, the name Trump Accounts certainly got his attention

358
00:18:37,000 --> 00:18:39,480
Speaker 10: and I think helped put it into the One Big

359
00:18:39,520 --> 00:18:43,159
Speaker 10: Beautiful Bill. But yeah, it's kind of an issue that

360
00:18:44,880 --> 00:18:47,560
Speaker 10: Democrats and Republicans have been interested in something like this,

361
00:18:47,680 --> 00:18:51,280
Speaker 10: a product like this that gets everyday Americans invested in

362
00:18:51,359 --> 00:18:54,159
Speaker 10: the stock market from a very young age. It's kind

363
00:18:54,200 --> 00:18:55,879
Speaker 10: of a bipartisan issue.

364
00:18:56,320 --> 00:18:58,800
Speaker 2: Yeah, And Brad over at Altameter, he's been saying that

365
00:18:58,880 --> 00:19:01,520
Speaker 2: this is a platform for every company in America to

366
00:19:01,560 --> 00:19:04,000
Speaker 2: think about how they reward employee's own children.

367
00:19:04,040 --> 00:19:05,439
Speaker 3: But it doesn't need to just be a company.

368
00:19:05,480 --> 00:19:08,520
Speaker 2: It could be mums and dad's, churches, synagogue's. Many to

369
00:19:08,600 --> 00:19:10,920
Speaker 2: be putting it aside in a sort of tax efficient way.

370
00:19:10,960 --> 00:19:11,240
Speaker 9: Tom.

371
00:19:11,480 --> 00:19:12,480
Speaker 3: What's interesting is, of.

372
00:19:12,400 --> 00:19:15,919
Speaker 2: Course Dell Stock has moved on all of this, and

373
00:19:15,960 --> 00:19:18,439
Speaker 2: this comes after a truth social post.

374
00:19:18,680 --> 00:19:19,360
Speaker 3: Yeah, that's right.

375
00:19:19,640 --> 00:19:22,159
Speaker 10: Dell stock was up about four percent last time I looked.

376
00:19:22,880 --> 00:19:26,280
Speaker 10: Trump obviously singling out Michael and Susan Dell for their

377
00:19:26,280 --> 00:19:29,600
Speaker 10: donation this morning, he was seemingly very happy about it.

378
00:19:29,640 --> 00:19:33,800
Speaker 10: And you know, when Trump tweets about something, it tends

379
00:19:33,840 --> 00:19:35,760
Speaker 10: to move stuff in the market. So we're kind of

380
00:19:35,760 --> 00:19:37,520
Speaker 10: saying that I don't know if it's going to last,

381
00:19:37,880 --> 00:19:38,840
Speaker 10: but there you go.

382
00:19:39,440 --> 00:19:41,760
Speaker 2: And of course that contributes to the networth of the

383
00:19:41,800 --> 00:19:44,080
Speaker 2: Dell family. On the back of that, Bloomberg's Tom Maloney,

384
00:19:44,280 --> 00:19:46,679
Speaker 2: it's been fascinating talking to you, Thank you very much. Indeed,

385
00:19:46,960 --> 00:19:49,600
Speaker 2: let's turn now to Warner Brothers Discovery. The company has

386
00:19:49,600 --> 00:19:52,680
Speaker 2: received a new round of bids, with one from Netflix

387
00:19:52,720 --> 00:19:56,080
Speaker 2: consisting of a mostly cash offers, according to sources. For more,

388
00:19:56,119 --> 00:19:59,520
Speaker 2: Bloomberg's Michelle Davis, who covers M and A Immediate joins us.

389
00:19:59,560 --> 00:20:00,880
Speaker 3: Now, so can we.

390
00:20:00,840 --> 00:20:03,120
Speaker 2: Break down how these is starting to note because they're

391
00:20:03,119 --> 00:20:03,840
Speaker 2: going to be sweetened.

392
00:20:04,560 --> 00:20:07,600
Speaker 11: They've been sweetened, yes, so the bidding is heating up.

393
00:20:07,720 --> 00:20:10,840
Speaker 11: Yesterday Warner Brothers had asked for sweetened offers from all

394
00:20:10,840 --> 00:20:12,960
Speaker 11: of the companies, so we know they received offers from

395
00:20:13,119 --> 00:20:16,080
Speaker 11: Netflix mostly cash. That was a big surprise because people

396
00:20:16,080 --> 00:20:18,720
Speaker 11: were expecting it to be stock. You know, Netflix's stock

397
00:20:18,720 --> 00:20:21,159
Speaker 11: has been doing pretty well. They have great currency, but

398
00:20:21,160 --> 00:20:24,000
Speaker 11: they're also an investment grade company, so they can afford

399
00:20:24,040 --> 00:20:26,240
Speaker 11: to raise a lot of money. We hear they're in

400
00:20:26,280 --> 00:20:28,520
Speaker 11: the market talking to banks about tens of billions of

401
00:20:28,560 --> 00:20:31,640
Speaker 11: dollars for a bridgeland. So there's Netflix, there's Comcast which

402
00:20:31,680 --> 00:20:34,520
Speaker 11: has also bid it only wants to streaming, and the

403
00:20:34,560 --> 00:20:37,879
Speaker 11: studios of Warner Brothers similar to Netflix. We're still trying

404
00:20:37,920 --> 00:20:40,439
Speaker 11: to glean the details of that offer, but we have

405
00:20:40,600 --> 00:20:43,080
Speaker 11: heard that it likely includes a little more stock than

406
00:20:43,119 --> 00:20:45,639
Speaker 11: the cash offered by Netflix. And finally, Paramount, the one

407
00:20:45,640 --> 00:20:47,800
Speaker 11: who kicked all of this off. They're the only bidders

408
00:20:47,800 --> 00:20:50,160
Speaker 11: who have actually made a play for the whole company.

409
00:20:50,640 --> 00:20:53,560
Speaker 11: We're hearing that their offers also is also completely cash.

410
00:20:53,680 --> 00:20:56,919
Speaker 11: They have backing from Larry Ellison, David Ellison, you know,

411
00:20:57,040 --> 00:20:59,760
Speaker 11: some of the rich, wealthiest people on Earth and Middle

412
00:20:59,800 --> 00:21:01,680
Speaker 11: East earn money as well as Apollo er putting in

413
00:21:01,800 --> 00:21:03,160
Speaker 11: some of the financing for their offer.

414
00:21:03,240 --> 00:21:05,600
Speaker 2: And it feels as though that's been the narrative that

415
00:21:06,000 --> 00:21:08,960
Speaker 2: the Edison bid might be one that is more approved

416
00:21:09,000 --> 00:21:11,560
Speaker 2: of by the administration. But are there any blocks? Are

417
00:21:11,600 --> 00:21:14,680
Speaker 2: there any issues from a regulatory perspective for Paramounts guide

418
00:21:14,680 --> 00:21:16,640
Speaker 2: downs just getting any big O or any of these

419
00:21:16,640 --> 00:21:18,119
Speaker 2: players getting any BIGA.

420
00:21:18,280 --> 00:21:21,200
Speaker 11: Yes, So the Warner Brothers board is really I mean,

421
00:21:21,280 --> 00:21:23,440
Speaker 11: they are in a position where they have lots to assess,

422
00:21:23,440 --> 00:21:27,400
Speaker 11: but it's not apples to apples, because they're evaluating, you know,

423
00:21:27,840 --> 00:21:29,680
Speaker 11: a bid for the whole company in cash, a bid

424
00:21:29,720 --> 00:21:32,280
Speaker 11: that might include stock and cash but only for part

425
00:21:32,320 --> 00:21:35,679
Speaker 11: and then you know other structures that are unclear for

426
00:21:35,760 --> 00:21:38,600
Speaker 11: Paramount and comcasts. There's clearly lots of synergies that they

427
00:21:38,640 --> 00:21:41,040
Speaker 11: could achieve by combining the companies, but there's a view

428
00:21:41,040 --> 00:21:43,359
Speaker 11: that doing that, you know, to achieve those synergies, they

429
00:21:43,400 --> 00:21:45,840
Speaker 11: would cut a lot of jobs. I mean, maybe thousands

430
00:21:45,840 --> 00:21:49,240
Speaker 11: of jobs at the studios by combining these studios Netflix

431
00:21:49,760 --> 00:21:51,920
Speaker 11: Their argument might be, we're not going to do something

432
00:21:51,960 --> 00:21:53,639
Speaker 11: like that, and so it'll play better, you know, with

433
00:21:53,760 --> 00:21:58,240
Speaker 11: labor unions and regulators. But Netflix also already has a

434
00:21:58,240 --> 00:22:00,880
Speaker 11: lot of power. So it's a bit wildcard to see

435
00:22:00,880 --> 00:22:02,200
Speaker 11: how the regulators are going to view that.

436
00:22:02,359 --> 00:22:05,000
Speaker 2: Yea, many are going to say, is YouTube count being

437
00:22:05,040 --> 00:22:06,760
Speaker 2: counted in the numbers? How we think about this from

438
00:22:06,800 --> 00:22:11,280
Speaker 2: a regulatory perspective, from a state and federal perspective, timeline, any.

439
00:22:11,080 --> 00:22:12,080
Speaker 3: Sort of idea.

440
00:22:12,240 --> 00:22:14,639
Speaker 11: We're hearing that they want to get this sorted before

441
00:22:14,640 --> 00:22:16,800
Speaker 11: the end of the year, and that makes some sense

442
00:22:16,840 --> 00:22:19,800
Speaker 11: because for the banks who are underwriting the loans, they

443
00:22:19,800 --> 00:22:22,080
Speaker 11: don't want the loans on their books at your end,

444
00:22:22,160 --> 00:22:23,720
Speaker 11: because it's going to cost them a lot more money,

445
00:22:23,720 --> 00:22:25,280
Speaker 11: which means the bidders won't be able to pay as

446
00:22:25,320 --> 00:22:27,399
Speaker 11: much for these assets. So we've heard that there's a

447
00:22:27,440 --> 00:22:30,640
Speaker 11: deadline of before the holidays. A decision could be made

448
00:22:30,680 --> 00:22:32,800
Speaker 11: as early as this week. You know, the binding offers

449
00:22:32,800 --> 00:22:34,720
Speaker 11: are in, so that means Warner Brothers could move quickly,

450
00:22:34,960 --> 00:22:37,560
Speaker 11: but they're also giving themselves room to ask for more bids.

451
00:22:38,200 --> 00:22:40,200
Speaker 11: You know, later on this month, CHERL.

452
00:22:40,320 --> 00:22:42,560
Speaker 2: Davis will be across the story. We appreciate it.

453
00:22:42,840 --> 00:22:43,040
Speaker 9: Now.

454
00:22:43,040 --> 00:22:44,919
Speaker 3: Coming up, we're going to be going back to.

455
00:22:44,920 --> 00:22:48,120
Speaker 2: AWS Reinvent the conferences on in Las Vegas. We're gonna

456
00:22:48,119 --> 00:22:50,919
Speaker 2: sit down with Swami sieve us of Romanian, vice president

457
00:22:50,960 --> 00:22:52,399
Speaker 2: of a Gentki for that company.

458
00:22:52,520 --> 00:23:02,880
Speaker 3: This is pretty big tech. Welcome back to Bloomberg Tech.

459
00:23:02,880 --> 00:23:04,439
Speaker 2: We check in on these markets that are losing a

460
00:23:04,480 --> 00:23:06,480
Speaker 2: bit of their earlier esteem. We're still holding on to

461
00:23:06,560 --> 00:23:08,959
Speaker 2: three ten percent higher, but remember we're up more than

462
00:23:08,960 --> 00:23:11,480
Speaker 2: a percentage point in earlier trading. Is in video perhaps

463
00:23:11,520 --> 00:23:14,000
Speaker 2: loses some of its edge in terms of points perspective.

464
00:23:14,200 --> 00:23:16,720
Speaker 2: Apple leading the charge right now, Intel as well. I'm

465
00:23:16,720 --> 00:23:19,320
Speaker 2: looking at Bitcoin though, still holding onto its games. Best

466
00:23:19,440 --> 00:23:23,359
Speaker 2: day since May. We're now at the highest level and

467
00:23:23,400 --> 00:23:26,320
Speaker 2: ninety nine since the end of November, so a few

468
00:23:26,400 --> 00:23:28,639
Speaker 2: days we're up four point percent as we see a

469
00:23:28,720 --> 00:23:30,720
Speaker 2: little bit of shifting in sentiment. Let's look at a

470
00:23:30,760 --> 00:23:32,720
Speaker 2: key name that has shifted in sentiment. It was beaten

471
00:23:32,800 --> 00:23:35,520
Speaker 2: up in November. Now we're still up two percent. Let's

472
00:23:35,520 --> 00:23:38,480
Speaker 2: call it. We're rolling over somewhat. But Pan Andeer, in

473
00:23:38,600 --> 00:23:41,600
Speaker 2: another AI names, have seen a bit of love today.

474
00:23:41,640 --> 00:23:43,920
Speaker 2: It seems as though Pan Andeer could be a trillion

475
00:23:43,960 --> 00:23:46,480
Speaker 2: dollar company. That's what Dan I've been saying on the day.

476
00:23:46,520 --> 00:23:48,639
Speaker 2: But really he's looking at big Tech and MAG seven

477
00:23:48,880 --> 00:23:51,040
Speaker 2: to add twenty to twenty five percent in terms of

478
00:23:51,040 --> 00:23:53,000
Speaker 2: share growth for twenty twenty six.

479
00:23:53,359 --> 00:23:54,800
Speaker 3: But for now, we return to one of.

480
00:23:54,760 --> 00:23:56,919
Speaker 2: The key MAG seven in that pile, and it is

481
00:23:56,960 --> 00:23:59,560
Speaker 2: Amazon and it's Aws. Even is upon us in Las

482
00:23:59,640 --> 00:24:00,920
Speaker 2: Vegas and take it away.

483
00:24:03,359 --> 00:24:04,200
Speaker 9: Yeah, thank you very much.

484
00:24:04,280 --> 00:24:08,919
Speaker 4: Karas talk more about Amazon's agentic plans with Swami Siva Supermanian.

485
00:24:08,960 --> 00:24:12,359
Speaker 4: He's vice president for a Genta ki at AWS and

486
00:24:12,400 --> 00:24:15,760
Speaker 4: you've basically brought out and released three from what you

487
00:24:15,800 --> 00:24:16,920
Speaker 4: call frontier agents.

488
00:24:17,000 --> 00:24:19,160
Speaker 9: That's right, across.

489
00:24:19,200 --> 00:24:24,960
Speaker 4: An autonomous agent security DevOps and considering where aw sits

490
00:24:24,960 --> 00:24:27,800
Speaker 4: in the market and everyone that attends reinvent, probably the

491
00:24:28,520 --> 00:24:30,840
Speaker 4: question most people would have is if I'm a large

492
00:24:30,960 --> 00:24:35,040
Speaker 4: enterprise and I have multiple teams, multiple deaths, doing multiple things,

493
00:24:35,320 --> 00:24:38,320
Speaker 4: how do I deploy three of these frontier agents?

494
00:24:38,400 --> 00:24:39,240
Speaker 9: Yeah?

495
00:24:39,560 --> 00:24:43,320
Speaker 8: Father MEAs out of it, saying here at daw As

496
00:24:43,320 --> 00:24:46,480
Speaker 8: we are of building the foundation for billions of agents,

497
00:24:46,920 --> 00:24:49,879
Speaker 8: and that's an understatement. So what do we launch with

498
00:24:49,920 --> 00:24:54,360
Speaker 8: these frontier agents? Is essentially a new category of agents,

499
00:24:54,480 --> 00:24:58,199
Speaker 8: because while so much our activity is happening in agents,

500
00:24:58,560 --> 00:25:02,880
Speaker 8: there are historically being more assistance to humans. And while

501
00:25:02,920 --> 00:25:05,760
Speaker 8: we deploy it, even with an Amazon Christen sew thousand

502
00:25:05,800 --> 00:25:09,720
Speaker 8: set developers to help with software development, we were actually

503
00:25:09,760 --> 00:25:13,280
Speaker 8: wanting to push the envelopement being able to actually do

504
00:25:13,400 --> 00:25:17,439
Speaker 8: a lot more where they are actually autonomous, where you

505
00:25:17,520 --> 00:25:20,560
Speaker 8: don't need humans to constantly steer them, and they are

506
00:25:20,600 --> 00:25:22,720
Speaker 8: massively scalable and they can run along.

507
00:25:22,800 --> 00:25:26,280
Speaker 4: To be autonomous, they need to have an end goal,

508
00:25:26,520 --> 00:25:28,560
Speaker 4: that's right, How does that work? So it's a case

509
00:25:28,560 --> 00:25:32,240
Speaker 4: of saying, by the way you use the term AI assistant, Yeah,

510
00:25:32,280 --> 00:25:34,720
Speaker 4: what we're now really talking about is AI co worker.

511
00:25:34,880 --> 00:25:37,320
Speaker 9: That's right, and that is the key thing here.

512
00:25:37,560 --> 00:25:40,560
Speaker 8: It's just like what these frontier agents are doing is

513
00:25:40,600 --> 00:25:44,680
Speaker 8: completely transforming how software is getting done in these teams.

514
00:25:45,440 --> 00:25:49,119
Speaker 8: So much activity happens primarily in like software development, but

515
00:25:49,400 --> 00:25:53,840
Speaker 8: now these agentic teammates they show up as another teammate

516
00:25:55,000 --> 00:25:59,439
Speaker 8: for software development. They pull up actually jobs from like

517
00:25:59,520 --> 00:26:03,879
Speaker 8: guitar task and start coding and clearing backlog. And if

518
00:26:03,920 --> 00:26:06,040
Speaker 8: you're in the middle of the night actually getting page

519
00:26:06,080 --> 00:26:10,040
Speaker 8: to deal with an issue happening with your website, this

520
00:26:10,119 --> 00:26:12,880
Speaker 8: agent first takes the first page and says, oh, let

521
00:26:12,880 --> 00:26:15,920
Speaker 8: me look into what's happening and unpack and then find

522
00:26:15,920 --> 00:26:18,199
Speaker 8: out the root costs so that you can actually quickly

523
00:26:18,280 --> 00:26:20,480
Speaker 8: resolve it and go back to bed, like I've been

524
00:26:20,520 --> 00:26:22,560
Speaker 8: on call and I know I could use that help

525
00:26:22,960 --> 00:26:26,479
Speaker 8: and be proactive so that those issues don't happen. And

526
00:26:26,680 --> 00:26:30,280
Speaker 8: third is actually a security teammate, so it shows up again.

527
00:26:30,600 --> 00:26:34,040
Speaker 8: And most of security has always been about afterthought. One

528
00:26:34,080 --> 00:26:36,160
Speaker 8: of the things we are changing the game here is

529
00:26:36,280 --> 00:26:40,560
Speaker 8: actually meeting developers even before they write a single line

530
00:26:40,560 --> 00:26:43,800
Speaker 8: of code when they're designing software. We catch them right

531
00:26:43,840 --> 00:26:45,960
Speaker 8: there and say here are the things you should be

532
00:26:46,000 --> 00:26:48,560
Speaker 8: aware of. And when they write code, we say, oh,

533
00:26:48,640 --> 00:26:52,400
Speaker 8: don't do that, do this and actually do penetration testing.

534
00:26:52,080 --> 00:26:53,160
Speaker 9: All before shipping.

535
00:26:53,320 --> 00:26:56,280
Speaker 8: So now you can view it as like you have

536
00:26:56,400 --> 00:27:00,960
Speaker 8: the best in class developer, OPS and security engineers showing

537
00:27:01,040 --> 00:27:03,040
Speaker 8: up as virtual teammates for every.

538
00:27:02,800 --> 00:27:06,720
Speaker 4: Software you say best in class? What evidence does AWS

539
00:27:06,760 --> 00:27:10,360
Speaker 4: have that in real world deployments? Running these three agents

540
00:27:10,400 --> 00:27:15,639
Speaker 4: in parallel autonomously has a clear benefit. Yeah, their efficiency, productivity.

541
00:27:16,000 --> 00:27:16,960
Speaker 4: What data can you share?

542
00:27:17,040 --> 00:27:19,159
Speaker 9: Yeah, I'll give actually a couple of examples.

543
00:27:19,200 --> 00:27:23,240
Speaker 8: One is in the DevOps agent that we launched, we

544
00:27:23,400 --> 00:27:28,760
Speaker 8: ran it across thousands of escalations that happened this year alone,

545
00:27:28,840 --> 00:27:33,399
Speaker 8: on incidents and so forth. These agents actually correctly identified

546
00:27:33,400 --> 00:27:35,440
Speaker 8: the root cost eighty six percent of time.

547
00:27:35,960 --> 00:27:39,600
Speaker 9: That is like really impresses eighty six eighty six eighty

548
00:27:39,640 --> 00:27:40,320
Speaker 9: six percent.

549
00:27:40,520 --> 00:27:43,800
Speaker 8: And the second one is Commonwealth Bank of Australia.

550
00:27:43,920 --> 00:27:46,480
Speaker 9: They actually have a huge cloud.

551
00:27:46,200 --> 00:27:50,600
Speaker 8: Infrastructure running across thousand, seven hundred accounts and they put

552
00:27:50,600 --> 00:27:54,600
Speaker 8: this DevOps agent test to test and for a very

553
00:27:54,640 --> 00:27:58,119
Speaker 8: complex networking stack to debug it. And what they said,

554
00:27:58,160 --> 00:28:01,159
Speaker 8: what did to take them? Like our Diba this was

555
00:28:01,240 --> 00:28:04,000
Speaker 8: able to do it in minutes and these are just

556
00:28:04,359 --> 00:28:08,240
Speaker 8: few simple examples, same thing. What's going on at security agent.

557
00:28:08,359 --> 00:28:13,320
Speaker 8: Smart Bug is a great photo company and there I'm

558
00:28:13,359 --> 00:28:17,400
Speaker 8: a big customer, and they are completely automating their security

559
00:28:18,280 --> 00:28:22,480
Speaker 8: operations altogether using like AWS Security Agent. And these are

560
00:28:22,520 --> 00:28:25,320
Speaker 8: all the beginning of how this is going to change

561
00:28:25,320 --> 00:28:26,280
Speaker 8: the game in.

562
00:28:26,240 --> 00:28:26,800
Speaker 9: A big way.

563
00:28:26,920 --> 00:28:29,800
Speaker 8: Because what do you want these agentic teammates to be

564
00:28:30,040 --> 00:28:34,040
Speaker 8: is always be there and you delegates on the boring

565
00:28:34,160 --> 00:28:39,080
Speaker 8: drudgery associated with like ops and security and backlog on

566
00:28:39,200 --> 00:28:43,560
Speaker 8: developments so that developers are empowered to be extremely creative

567
00:28:43,720 --> 00:28:47,240
Speaker 8: and you will start seeing fighter ten x improvement in

568
00:28:47,320 --> 00:28:48,560
Speaker 8: productivity in a big way.

569
00:28:48,920 --> 00:28:51,880
Speaker 4: To reach out its three agents, Kero is software development.

570
00:28:52,000 --> 00:28:55,440
Speaker 4: You have the AWS Security Agent, the AWS DevOps agent.

571
00:28:55,560 --> 00:28:59,680
Speaker 4: They are designed to run for hours autonomously. To sum

572
00:28:59,760 --> 00:29:02,880
Speaker 4: that disconcerting, right, And I go back to my other question,

573
00:29:03,800 --> 00:29:07,680
Speaker 4: how much concern is there in the last segment Clean

574
00:29:08,080 --> 00:29:12,400
Speaker 4: talk to me about observability, understanding in real time what's doing.

575
00:29:12,400 --> 00:29:14,880
Speaker 4: But that sounds to me like it's human supervision.

576
00:29:15,480 --> 00:29:16,320
Speaker 9: It's a great question.

577
00:29:16,440 --> 00:29:19,840
Speaker 8: I mean, this is the nuance and the art of balance.

578
00:29:19,840 --> 00:29:22,200
Speaker 8: We had to do one. You want these agents to

579
00:29:22,280 --> 00:29:26,240
Speaker 8: be autonomous and massively scalable, but you also want them

580
00:29:26,280 --> 00:29:29,239
Speaker 8: to not go off the rails. So first we are

581
00:29:29,240 --> 00:29:32,080
Speaker 8: built it with the right guard rails. And second even

582
00:29:32,120 --> 00:29:35,440
Speaker 8: a spart a development workflow, these agents can go actually

583
00:29:35,840 --> 00:29:39,200
Speaker 8: take a goal from a developer to say, hey, go

584
00:29:39,240 --> 00:29:40,600
Speaker 8: work on upgrading.

585
00:29:40,160 --> 00:29:43,280
Speaker 9: This piece of code to the latest systyk and then

586
00:29:43,320 --> 00:29:44,160
Speaker 9: start upgrading.

587
00:29:44,440 --> 00:29:47,320
Speaker 8: But then as a final step, it might send it

588
00:29:47,360 --> 00:29:49,560
Speaker 8: for a code review to the human and say take

589
00:29:49,560 --> 00:29:51,960
Speaker 8: a look and make sure I'm okay. And once they

590
00:29:51,960 --> 00:29:54,240
Speaker 8: say I'm okay, I think it's fine, or they can

591
00:29:54,360 --> 00:29:56,239
Speaker 8: even configure it to say if it pass us all

592
00:29:56,280 --> 00:29:58,880
Speaker 8: these tests, you're good to ship it. So that you

593
00:29:59,120 --> 00:30:02,760
Speaker 8: have built in s either through human or automated tests.

594
00:30:02,800 --> 00:30:05,760
Speaker 8: And these are the kind of things now we are

595
00:30:05,920 --> 00:30:09,479
Speaker 8: able to actually innovate with the customers as well. And

596
00:30:09,520 --> 00:30:13,360
Speaker 8: even there we have done some pure innovation like introduced

597
00:30:13,400 --> 00:30:19,800
Speaker 8: automated mathematically provable property based testing, so that means now

598
00:30:20,120 --> 00:30:24,080
Speaker 8: developers can specify their goal and we can mathematically prove

599
00:30:24,160 --> 00:30:28,040
Speaker 8: with things like kiro that whatever it is generating can

600
00:30:28,120 --> 00:30:30,800
Speaker 8: be tested exactly that way and it's accurate. So that

601
00:30:30,960 --> 00:30:32,280
Speaker 8: is like another game changer.

602
00:30:33,200 --> 00:30:35,920
Speaker 4: SWAM we see the supermanian vice president for a gens

603
00:30:36,040 --> 00:30:39,840
Speaker 4: Ki at aws three new Frontier agents carry We're talking

604
00:30:39,840 --> 00:30:41,200
Speaker 4: about here in Las Vegas, back.

605
00:30:41,040 --> 00:30:43,600
Speaker 2: To you, loving it back here in New York. Ed's

606
00:30:43,680 --> 00:30:46,880
Speaker 2: time for talking tech. First up, Apple's head of AI

607
00:30:47,120 --> 00:30:50,080
Speaker 2: John Jian Andrea, Well, he's coming down with mans to

608
00:30:50,160 --> 00:30:52,680
Speaker 2: lead the company entirely in the spring. The move caps

609
00:30:52,720 --> 00:30:55,840
Speaker 2: are pretty tumultuous tenure that included a fumbled entry into

610
00:30:55,880 --> 00:31:00,480
Speaker 2: General's AI. Apple won't be directly replacing Jian Andrea, instead

611
00:31:00,680 --> 00:31:04,120
Speaker 2: opting to break up the AIT plus Samsung It's unveiled

612
00:31:04,160 --> 00:31:08,920
Speaker 2: its first trifle smartphone. It's called Galaxy Z Trifle and

613
00:31:08,960 --> 00:31:12,320
Speaker 2: the device contains two hinges allow it to transform into

614
00:31:12,360 --> 00:31:14,760
Speaker 2: a larger tablet like device. The phone is set to

615
00:31:14,800 --> 00:31:17,120
Speaker 2: be released first in South Korea on December the twelfth,

616
00:31:17,160 --> 00:31:20,600
Speaker 2: with a price of about four hundred and fifty dollars,

617
00:31:20,760 --> 00:31:23,640
Speaker 2: and the US Commerce Department has agreed to invest as

618
00:31:23,720 --> 00:31:26,720
Speaker 2: much as one hundred and fifty million dollars into x Light.

619
00:31:27,000 --> 00:31:30,360
Speaker 2: It's a chip startup where former Intel CEO Pat Gelsinger

620
00:31:30,560 --> 00:31:33,080
Speaker 2: serves as executive chairman. So the latest moved by the

621
00:31:33,080 --> 00:31:36,520
Speaker 2: Trump administration to bring chip making capabilities back to the

622
00:31:36,600 --> 00:31:39,840
Speaker 2: United States. Now coming up, guess while we go back

623
00:31:39,880 --> 00:31:43,160
Speaker 2: to AWS in Vegas. This time it's a conversation.

624
00:31:42,720 --> 00:31:44,560
Speaker 3: With Sanjay Back to Cone.

625
00:31:44,680 --> 00:31:47,840
Speaker 2: Nas chief Product and Technology Officer is a blueber tech.

626
00:32:04,480 --> 00:32:07,720
Speaker 2: Luma AI, known for its flagship creative product, dream Machine

627
00:32:07,800 --> 00:32:11,160
Speaker 2: video generator, taking a step forward in global expansion with

628
00:32:11,200 --> 00:32:14,719
Speaker 2: opening its first international office and it's in London. This

629
00:32:14,800 --> 00:32:17,720
Speaker 2: follows its recent nine hundred million dollar Series C investment,

630
00:32:17,720 --> 00:32:20,240
Speaker 2: which was led by Humane and Met Jane Is with

631
00:32:20,320 --> 00:32:23,000
Speaker 2: US co founder CEO of Luma AI and so.

632
00:32:23,000 --> 00:32:23,680
Speaker 3: Pleased to join you.

633
00:32:23,680 --> 00:32:25,840
Speaker 2: I can't wait to get into reasoning video models, into

634
00:32:25,840 --> 00:32:26,960
Speaker 2: the future of world models.

635
00:32:27,000 --> 00:32:28,719
Speaker 3: But first am it? Why London?

636
00:32:31,120 --> 00:32:33,120
Speaker 12: We have a huge pipeline. Actually, by the way, thanks

637
00:32:33,160 --> 00:32:36,600
Speaker 12: for having me. I'm very excited to be here. We

638
00:32:36,640 --> 00:32:42,240
Speaker 12: have a huge pipeline of researchers, engineers from Europe and

639
00:32:42,520 --> 00:32:46,840
Speaker 12: also deep Mind that want to join luma And also

640
00:32:47,080 --> 00:32:50,400
Speaker 12: London is the gateway for the business in Europe as

641
00:32:50,440 --> 00:32:53,200
Speaker 12: well as Middle East, so it seems to be the

642
00:32:53,320 --> 00:32:56,280
Speaker 12: right place to have our second office outside of Palo

643
00:32:56,320 --> 00:32:58,760
Speaker 12: Alto and Bay Area. So yeah, today is our launch

644
00:32:58,800 --> 00:32:59,760
Speaker 12: of the London office.

645
00:33:00,000 --> 00:33:02,560
Speaker 2: I have a feeling demos hasibus over at deep Mind

646
00:33:02,600 --> 00:33:05,560
Speaker 2: and the AI labs at Google we'll be having is

647
00:33:05,560 --> 00:33:08,800
Speaker 2: is pricked by that, amit? So why would people currently

648
00:33:08,840 --> 00:33:11,080
Speaker 2: working or having been trained at deep Mind and the

649
00:33:11,280 --> 00:33:13,720
Speaker 2: like want a jump ship to help build Ray three

650
00:33:13,760 --> 00:33:14,880
Speaker 2: your reasoning video model.

651
00:33:15,880 --> 00:33:18,360
Speaker 12: Yeah, I think that's a great question. So Luma already

652
00:33:18,400 --> 00:33:22,080
Speaker 12: has actually a deep bench of researchers from from n

653
00:33:22,160 --> 00:33:26,440
Speaker 12: video from from deep Mind, from you know, schools like MIT,

654
00:33:26,840 --> 00:33:32,040
Speaker 12: Stanford and Berkeley. And the reason that these incredibly exceptional

655
00:33:32,080 --> 00:33:36,120
Speaker 12: people actually join Luma is because, one, while it is

656
00:33:36,200 --> 00:33:39,960
Speaker 12: a very very well resourced AGI lab, we are only

657
00:33:39,960 --> 00:33:42,440
Speaker 12: about one hundred and fifty people, so we get to

658
00:33:42,600 --> 00:33:45,520
Speaker 12: do and get to have resources per person that, like,

659
00:33:45,560 --> 00:33:47,200
Speaker 12: you know, it's unheard of in the rest of the

660
00:33:47,200 --> 00:33:51,640
Speaker 12: industry actually. And second, Luma is so ultra aligned on

661
00:33:51,680 --> 00:33:54,640
Speaker 12: this one goal, which is to build multi model AGI.

662
00:33:54,720 --> 00:33:58,360
Speaker 12: There's actually no second There is no second project happening

663
00:33:58,360 --> 00:33:59,080
Speaker 12: at Luma.

664
00:33:59,160 --> 00:33:59,959
Speaker 9: This is what we do.

665
00:34:00,320 --> 00:34:06,680
Speaker 12: So people researchers, engineers and product people who believe strongly

666
00:34:07,200 --> 00:34:09,880
Speaker 12: in this mission, Luma is the best place in the

667
00:34:09,920 --> 00:34:10,680
Speaker 12: world for them to be.

668
00:34:11,400 --> 00:34:14,680
Speaker 2: Looking at some of the amazing video that you create

669
00:34:15,239 --> 00:34:17,200
Speaker 2: I can see how it goes into the creative sphere,

670
00:34:17,239 --> 00:34:20,279
Speaker 2: into advertising and the like, but it feels as though

671
00:34:20,360 --> 00:34:23,960
Speaker 2: this model, these world models, has moved to perhaps more

672
00:34:23,960 --> 00:34:27,120
Speaker 2: physical AI applications must open up different industries.

673
00:34:28,120 --> 00:34:32,480
Speaker 12: Yeah so, I mean video and video is actually the

674
00:34:32,520 --> 00:34:34,640
Speaker 12: path to AGI. And let me tell you just very

675
00:34:34,640 --> 00:34:37,480
Speaker 12: briefly why that is the case. Language gives us reasoning,

676
00:34:37,640 --> 00:34:40,360
Speaker 12: and language gives us the human abstraction that is necessary.

677
00:34:40,680 --> 00:34:43,120
Speaker 12: Video shows us the entire universe. Right, you know how

678
00:34:43,400 --> 00:34:46,480
Speaker 12: things behave, how water behaves, how the laws of physics work.

679
00:34:46,719 --> 00:34:51,320
Speaker 12: Then you combine them together. Then video, audio and language

680
00:34:51,360 --> 00:34:55,080
Speaker 12: together is basically a chance to build a universal simulator.

681
00:34:55,520 --> 00:34:57,800
Speaker 12: So one application of that, as you point out, is

682
00:34:58,000 --> 00:35:01,080
Speaker 12: obviously for entertainment and to generate video to automate or

683
00:35:01,160 --> 00:35:04,240
Speaker 12: or to to make digital the act of creating video.

684
00:35:04,880 --> 00:35:07,759
Speaker 12: But in addition to that, it is the gateway to

685
00:35:07,800 --> 00:35:11,960
Speaker 12: be able to build actually general purpose robotics. That is

686
00:35:12,000 --> 00:35:14,160
Speaker 12: the second area that we are very deeply focused on

687
00:35:14,200 --> 00:35:16,360
Speaker 12: and now starting to build out a team and the

688
00:35:16,400 --> 00:35:20,600
Speaker 12: research area of that, and we have if we want

689
00:35:20,640 --> 00:35:24,319
Speaker 12: to have any shot at building general purpose robots, then

690
00:35:24,320 --> 00:35:26,960
Speaker 12: the only way that happens is by giving them this

691
00:35:27,120 --> 00:35:30,040
Speaker 12: level of general understanding of the universe so that they can,

692
00:35:30,440 --> 00:35:32,640
Speaker 12: you know, reason in their head. They can actually simulate

693
00:35:32,680 --> 00:35:35,000
Speaker 12: every scenario right in their head, like, Okay, what would

694
00:35:35,000 --> 00:35:36,480
Speaker 12: happen if I do this? What would happen if I

695
00:35:36,480 --> 00:35:40,480
Speaker 12: do something else? This level of learning and reasoning is essential,

696
00:35:40,480 --> 00:35:43,320
Speaker 12: so video as it advances and as we scale these models,

697
00:35:43,440 --> 00:35:45,440
Speaker 12: they will not only get better and better, they will

698
00:35:45,440 --> 00:35:48,440
Speaker 12: get more accurate in simulating physics. And this is the

699
00:35:48,480 --> 00:35:51,359
Speaker 12: path to building physical intelligence. And that's why Luma's model

700
00:35:52,120 --> 00:35:56,520
Speaker 12: mission is to build multimodel agi that can generate, understand,

701
00:35:56,760 --> 00:35:59,160
Speaker 12: and operate in the physical world. And that is the

702
00:35:59,239 --> 00:35:59,680
Speaker 12: end goal.

703
00:36:00,600 --> 00:36:02,359
Speaker 2: What is the biggest blocker to that or the thing

704
00:36:02,400 --> 00:36:03,759
Speaker 2: that keeps you up Because at the moment, you've got

705
00:36:03,800 --> 00:36:05,440
Speaker 2: the money, Boy, do you have the money nine hundred

706
00:36:05,480 --> 00:36:08,080
Speaker 2: million raised, But we've also got promise of compute coming

707
00:36:08,080 --> 00:36:08,880
Speaker 2: from Saudi Arabia.

708
00:36:09,000 --> 00:36:09,880
Speaker 3: Is it about compute?

709
00:36:09,920 --> 00:36:12,560
Speaker 2: Is it more about the talent that is what you

710
00:36:12,600 --> 00:36:14,360
Speaker 2: need to push through to get to your mission.

711
00:36:15,280 --> 00:36:18,359
Speaker 12: Yeah, So there are two really important things for LUMA.

712
00:36:19,400 --> 00:36:22,560
Speaker 12: Over the last couple of years, we worked tirelessly to

713
00:36:23,000 --> 00:36:25,080
Speaker 12: solve the research problem for like you know, really really

714
00:36:25,120 --> 00:36:27,920
Speaker 12: great video. Now the next frontier for us are these

715
00:36:27,960 --> 00:36:30,080
Speaker 12: omni models that are able to like you know, reason

716
00:36:30,120 --> 00:36:33,759
Speaker 12: an audio, video, language text together. So the research is

717
00:36:34,440 --> 00:36:36,439
Speaker 12: like you know, the first one that we pay mental

718
00:36:36,400 --> 00:36:38,879
Speaker 12: amount of attention to now to solve the research of course,

719
00:36:38,920 --> 00:36:42,160
Speaker 12: you know, people and really really brilliant people is one

720
00:36:42,160 --> 00:36:44,120
Speaker 12: of those bottlencks. So we hire from some of the

721
00:36:44,160 --> 00:36:46,880
Speaker 12: best places. See, we don't need thousands of people. We

722
00:36:46,960 --> 00:36:48,960
Speaker 12: need like, you know, maybe two hundred or three hundred

723
00:36:49,040 --> 00:36:52,719
Speaker 12: really really brilliant people to work on these problems. So yeah,

724
00:36:52,760 --> 00:36:57,240
Speaker 12: that's that's the first one. The talent and great aligned people,

725
00:36:57,320 --> 00:36:59,719
Speaker 12: that is the first one. Second is obviously compute. So

726
00:36:59,760 --> 00:37:02,719
Speaker 12: alongside with the SERIOC announcement, we announced that we will

727
00:37:02,760 --> 00:37:07,280
Speaker 12: build with Humane and collaboration with Humane, a two gigawat

728
00:37:07,440 --> 00:37:11,799
Speaker 12: compute cluster that is the largest compute build out in

729
00:37:11,840 --> 00:37:14,279
Speaker 12: our space of world models and video models and like

730
00:37:14,280 --> 00:37:16,960
Speaker 12: you know, this kind of physical AI. And it is

731
00:37:17,000 --> 00:37:20,400
Speaker 12: also one of the largest compute built out in AI period, right,

732
00:37:20,560 --> 00:37:23,160
Speaker 12: So compute is one of the other things that is

733
00:37:23,239 --> 00:37:27,920
Speaker 12: a big limitation. Ultimately, multi model AI will be a

734
00:37:27,960 --> 00:37:30,839
Speaker 12: superset of llms, will be a superset of the AI

735
00:37:30,880 --> 00:37:33,640
Speaker 12: we have today, it will require more compute than llms

736
00:37:33,719 --> 00:37:36,960
Speaker 12: do right now. So compute is the second very important

737
00:37:36,960 --> 00:37:39,200
Speaker 12: input to our business that, like you know, we are

738
00:37:39,280 --> 00:37:41,759
Speaker 12: working to not only shore up but solve in a

739
00:37:41,760 --> 00:37:44,359
Speaker 12: way we can serve these models economically to a lot

740
00:37:44,400 --> 00:37:44,800
Speaker 12: of people.

741
00:37:45,160 --> 00:37:47,440
Speaker 2: I mean, Jane, come back and tell us how you'll

742
00:37:47,440 --> 00:37:48,400
Speaker 2: continuing on that mission.

743
00:37:48,520 --> 00:37:50,560
Speaker 3: Go found a CEO of Luma AI.

744
00:37:51,000 --> 00:37:51,360
Speaker 9: Thank you.

745
00:37:52,040 --> 00:37:55,440
Speaker 2: Coming up, we're talking more about AI and infrastructure. We're

746
00:37:55,440 --> 00:37:58,640
Speaker 2: going to aws's event in Vegas. Were a conversation with

747
00:37:58,760 --> 00:38:02,400
Speaker 2: Sanjay Blackta Conde, NAS chief product and technology officer is

748
00:38:02,400 --> 00:38:03,040
Speaker 2: a blue big.

749
00:38:02,920 --> 00:38:21,759
Speaker 4: Tech Sanday back to Conde, NAS chief Product and Technology

750
00:38:21,800 --> 00:38:25,239
Speaker 4: officer joins us from AWS reinvent in Las Vegas. You're

751
00:38:25,320 --> 00:38:27,640
Speaker 4: kind of bucking a trend that we've been talking about

752
00:38:28,120 --> 00:38:31,160
Speaker 4: throughout the year in the context of AI. Actually on

753
00:38:31,320 --> 00:38:34,319
Speaker 4: prem has kind of been back when AI is being

754
00:38:34,400 --> 00:38:36,959
Speaker 4: run at the edge. You're going the other way from

755
00:38:36,960 --> 00:38:39,560
Speaker 4: on prem, relying heavily on AWS.

756
00:38:39,640 --> 00:38:41,240
Speaker 9: The rationale, well.

757
00:38:41,080 --> 00:38:43,440
Speaker 13: We are actually users of AI. We are not like

758
00:38:43,480 --> 00:38:47,120
Speaker 13: some of these LLLM companies that are creating and training

759
00:38:47,160 --> 00:38:50,040
Speaker 13: their own models. Although we have trained models in the past,

760
00:38:50,080 --> 00:38:53,000
Speaker 13: but at a much smaller scale. So for us, investing

761
00:38:53,000 --> 00:38:55,719
Speaker 13: in infrastructure and building out data centers for the size

762
00:38:55,760 --> 00:38:57,280
Speaker 13: of our operation doesn't make sense.

763
00:38:57,640 --> 00:39:01,799
Speaker 4: The work with AWS is focused on deli ring personalized content, right,

764
00:39:02,320 --> 00:39:04,799
Speaker 4: And is that at the foundation model level you know

765
00:39:04,840 --> 00:39:07,800
Speaker 4: AWS out with Nova to today, or it's just simply

766
00:39:07,880 --> 00:39:11,480
Speaker 4: that you're leveraging their scale the multitude of data platforms

767
00:39:11,480 --> 00:39:11,719
Speaker 4: they have.

768
00:39:11,880 --> 00:39:14,600
Speaker 13: Yeah, for that particular use case, we're actually just leveraging

769
00:39:14,600 --> 00:39:17,680
Speaker 13: their infrastructure and scale. We've built all the personalization models

770
00:39:17,680 --> 00:39:20,759
Speaker 13: ourselves and we've trained them in house, and we've been

771
00:39:20,760 --> 00:39:24,759
Speaker 13: doing that for the last probably three four years, long

772
00:39:24,800 --> 00:39:27,759
Speaker 13: before LLMS became a thing. We've had our own data

773
00:39:27,760 --> 00:39:30,520
Speaker 13: science team and we've been training models. So we use

774
00:39:30,560 --> 00:39:33,480
Speaker 13: our own homegrown models for most of the personalization and

775
00:39:33,560 --> 00:39:34,600
Speaker 13: recommendation workload.

776
00:39:34,680 --> 00:39:38,440
Speaker 4: So here in lies the question of AWS reinvent AWS

777
00:39:38,640 --> 00:39:42,200
Speaker 4: number one in cloud computing, but they want to do more,

778
00:39:42,400 --> 00:39:44,560
Speaker 4: They want to be number one in AI. You know,

779
00:39:44,640 --> 00:39:48,440
Speaker 4: their own foundation models, the agentic tools that they release today.

780
00:39:48,960 --> 00:39:51,680
Speaker 4: What would it take from Amazon in terms of the

781
00:39:51,760 --> 00:39:54,400
Speaker 4: utility of that technology for you to rely on them

782
00:39:54,400 --> 00:39:54,880
Speaker 4: more heavily.

783
00:39:55,200 --> 00:39:59,600
Speaker 13: Well, actually, for some of the generative use cases you

784
00:39:59,600 --> 00:40:02,320
Speaker 13: know that are LLLM based. We are using Amazon Bedrock.

785
00:40:03,080 --> 00:40:05,800
Speaker 13: You know, we have a contracts management, rights clearance system

786
00:40:05,840 --> 00:40:09,799
Speaker 13: that we just launched. We are using Amazon's Bedrock capability

787
00:40:09,880 --> 00:40:13,560
Speaker 13: for some of our moderation AI based moderation of user

788
00:40:13,600 --> 00:40:17,000
Speaker 13: generated content. So we're looking more and more now to

789
00:40:17,000 --> 00:40:21,280
Speaker 13: towards using out of the box capabilities that Amazon provides

790
00:40:21,440 --> 00:40:23,560
Speaker 13: rather than build and train our own models, which we.

791
00:40:23,640 --> 00:40:24,560
Speaker 9: Used to do in the past.

792
00:40:24,600 --> 00:40:27,600
Speaker 13: I think the need for that is becoming less and less.

793
00:40:28,160 --> 00:40:31,279
Speaker 4: Conde has a relationship with open AI. Condy was one

794
00:40:31,320 --> 00:40:33,160
Speaker 4: of the first to make a deal with open Ai

795
00:40:33,719 --> 00:40:36,080
Speaker 4: early days, it seems. But how is that progressing?

796
00:40:36,440 --> 00:40:37,120
Speaker 9: That's going well.

797
00:40:37,160 --> 00:40:40,680
Speaker 13: Actually, we're starting to use chat GPT quite widely within

798
00:40:40,760 --> 00:40:46,080
Speaker 13: the enterprise internally, yes, and for external use case. We've

799
00:40:46,080 --> 00:40:50,840
Speaker 13: actually launched an AI based recipe search on bond Appetite

800
00:40:52,000 --> 00:40:53,960
Speaker 13: on our on our website, and also it's going to

801
00:40:54,000 --> 00:40:56,279
Speaker 13: come out in our app which allows customers to go

802
00:40:56,360 --> 00:40:58,680
Speaker 13: in and do natural language search and also be able

803
00:40:58,760 --> 00:41:01,840
Speaker 13: to modify the it has peace according to their taste.

804
00:41:02,160 --> 00:41:03,960
Speaker 13: So that's the first use case we are looking at

805
00:41:03,960 --> 00:41:05,399
Speaker 13: others with opening I as well.

806
00:41:06,120 --> 00:41:09,680
Speaker 4: The broad theme of the program today here at AWS

807
00:41:09,760 --> 00:41:12,759
Speaker 4: has been about companies of all sizes moving from using

808
00:41:12,840 --> 00:41:17,040
Speaker 4: AI assistance internally to what awso call the AI co worker,

809
00:41:17,520 --> 00:41:20,600
Speaker 4: you know, a hybrid workforce of people and energentic AI.

810
00:41:21,239 --> 00:41:25,640
Speaker 4: Conde has done layoffs in the past two years, financial years.

811
00:41:25,960 --> 00:41:30,120
Speaker 4: How much has that been about AI tools changing productivity,

812
00:41:30,560 --> 00:41:33,560
Speaker 4: eliminating certain roles, changing certain roles.

813
00:41:33,800 --> 00:41:35,680
Speaker 13: Well, I don't think it's most of it has been

814
00:41:35,719 --> 00:41:38,279
Speaker 13: about AI. Actually, some of it, you know, especially in tech.

815
00:41:38,360 --> 00:41:41,520
Speaker 13: You know, we've had we use extensively, We use AI

816
00:41:41,600 --> 00:41:43,480
Speaker 13: for our work work on a day to day basis,

817
00:41:43,520 --> 00:41:46,640
Speaker 13: which eliminates the need for some roles. We can do

818
00:41:46,719 --> 00:41:50,520
Speaker 13: more things with fewer people. But other than that, across

819
00:41:50,520 --> 00:41:53,400
Speaker 13: the organization, we've not really had any AI based impacts.

820
00:41:54,160 --> 00:41:58,320
Speaker 4: Got to ask in a different part of the Amazon universe.

821
00:41:58,960 --> 00:42:02,440
Speaker 4: I've been experimenting at home with Alexa and Alexa plus.

822
00:42:02,840 --> 00:42:05,719
Speaker 4: Do we get some kind of Conde nas Alexa integration.

823
00:42:06,120 --> 00:42:07,520
Speaker 9: Well, that's work going on there.

824
00:42:07,880 --> 00:42:10,640
Speaker 13: We have actually already integrated with Amazon Alexa.

825
00:42:10,640 --> 00:42:12,440
Speaker 9: So how does that work available?

826
00:42:12,760 --> 00:42:15,280
Speaker 13: So you know, you can query and you can get content,

827
00:42:15,360 --> 00:42:17,040
Speaker 13: you know, read out to you from some of our

828
00:42:17,040 --> 00:42:20,120
Speaker 13: publications not all so, but I think it's pretty cool.

829
00:42:20,120 --> 00:42:20,799
Speaker 13: We should try it out.

830
00:42:21,200 --> 00:42:22,600
Speaker 4: At the heart of that question is something a bit

831
00:42:22,600 --> 00:42:28,000
Speaker 4: more existential. Conday sees itself more as maybe entertainment, and

832
00:42:28,040 --> 00:42:30,399
Speaker 4: if you look at the revenue streams very different from

833
00:42:30,400 --> 00:42:33,640
Speaker 4: saying you're a news organization. Just explain how you see

834
00:42:33,680 --> 00:42:35,239
Speaker 4: this company transitioning right now?

835
00:42:35,320 --> 00:42:37,759
Speaker 13: Yeah, I think they're definitely in the entertainment space because

836
00:42:37,760 --> 00:42:39,760
Speaker 13: if you look at our brands, you know we mostly

837
00:42:39,840 --> 00:42:43,960
Speaker 13: cover leisure, fashion, lifestyle. We're not a daily news outlet,

838
00:42:44,000 --> 00:42:46,160
Speaker 13: so people don't come to us on an everyday basis.

839
00:42:46,239 --> 00:42:49,960
Speaker 13: So we are competing with other entertainment outlets, whether it

840
00:42:50,040 --> 00:42:53,200
Speaker 13: is you know, streaming media like Netflix or Hulu or

841
00:42:53,719 --> 00:42:57,960
Speaker 13: Amazon Prime, or social media you know TikTok and Instagram

842
00:42:58,000 --> 00:43:00,640
Speaker 13: and others. So people have limited spare time, so we

843
00:43:00,680 --> 00:43:03,200
Speaker 13: are competing for a slice of that. So we definitely

844
00:43:03,239 --> 00:43:06,319
Speaker 13: have to do way better in terms of our personalization

845
00:43:06,400 --> 00:43:09,640
Speaker 13: and user experience and also our great journalism that we have,

846
00:43:09,800 --> 00:43:11,000
Speaker 13: I think is what is going to take.

847
00:43:10,960 --> 00:43:13,080
Speaker 4: Us forward, Sunjay, in the year ahead, what's the one

848
00:43:13,120 --> 00:43:15,279
Speaker 4: big change you want to make on the technology side,

849
00:43:15,280 --> 00:43:17,360
Speaker 4: the one thing you're still yet to do be AI

850
00:43:17,560 --> 00:43:18,160
Speaker 4: or something else?

851
00:43:18,360 --> 00:43:22,480
Speaker 13: Yes, I think it's mostly definitely around AI. We have

852
00:43:22,560 --> 00:43:25,520
Speaker 13: a pretty solid data infrastructure that we've built on Amazon

853
00:43:25,560 --> 00:43:28,239
Speaker 13: with data bricks, and our next task is really to

854
00:43:28,239 --> 00:43:30,600
Speaker 13: figure out how do we vectorize all of our content

855
00:43:31,120 --> 00:43:34,440
Speaker 13: and make it readily available in real time to lllams.

856
00:43:34,920 --> 00:43:36,840
Speaker 13: I think that is probably our number one challenge.

857
00:43:37,000 --> 00:43:39,560
Speaker 4: Sanjay Bakta Conde Nas, thank you so much for joining

858
00:43:39,640 --> 00:43:40,600
Speaker 4: us here in Las Vegas.

859
00:43:40,640 --> 00:43:41,399
Speaker 9: Carry back to you.

860
00:43:41,760 --> 00:43:45,040
Speaker 2: Fascinating set of conversations across the gamut of all things

861
00:43:45,080 --> 00:43:46,560
Speaker 2: AI and entertainment edge.

862
00:43:46,600 --> 00:43:47,160
Speaker 3: We thank you.

863
00:43:47,160 --> 00:43:49,080
Speaker 2: You got so much more coming up that does it

864
00:43:49,239 --> 00:43:52,160
Speaker 2: this edition of Bloomberg Tech, But do stick around. It

865
00:43:52,239 --> 00:43:54,759
Speaker 2: is going to sit down interview with AWS CEO Matt

866
00:43:54,800 --> 00:43:59,160
Speaker 2: Garman's three pm Eastern twelve pm Pacific first right here

867
00:43:59,160 --> 00:44:00,000
Speaker 2: on Boomog Television.

868
00:44:00,080 --> 00:44:02,359
Speaker 3: And who all, don't forget to check out a podcast you're.

869
00:44:02,239 --> 00:44:04,600
Speaker 2: Finding on the terminal as well as online on Apples, Spotify,

870
00:44:04,640 --> 00:44:06,600
Speaker 2: and iHeart from New York from Las Vegas.

871
00:44:07,040 --> 00:44:07,919
Speaker 3: This is a blue Bag Tech