1
00:00:00,280 --> 00:00:07,200
Speaker 1: Bloomberg Audio, Studios, podcasts, radio news.

2
00:00:08,960 --> 00:00:12,559
Speaker 2: Investors woke up this morning to seismic losses for US

3
00:00:12,600 --> 00:00:13,680
Speaker 2: tech stocks.

4
00:00:13,680 --> 00:00:17,040
Speaker 3: Just creating a massive drag hundreds of billions of dollars

5
00:00:17,040 --> 00:00:18,200
Speaker 3: in market cap loss.

6
00:00:18,440 --> 00:00:21,040
Speaker 2: The Nasdaq fell as much as five point two percent

7
00:00:21,160 --> 00:00:24,560
Speaker 2: in overnight trading. Is Naszak futures and global tech stocks

8
00:00:24,840 --> 00:00:27,000
Speaker 2: extend their losses on fear of the AI race may

9
00:00:27,040 --> 00:00:29,639
Speaker 2: no longer be America's to lose. By the close of

10
00:00:29,680 --> 00:00:33,880
Speaker 2: a wild trading day, and Nvidia's stock fell by seventeen percent,

11
00:00:34,560 --> 00:00:37,839
Speaker 2: The chip maker saw five hundred and eighty nine billion

12
00:00:37,960 --> 00:00:42,000
Speaker 2: dollars erased from its market valuation, breaking its own record

13
00:00:42,120 --> 00:00:44,800
Speaker 2: for the biggest single day loss of market value on

14
00:00:44,880 --> 00:00:48,880
Speaker 2: the US Stock Exchange, and energy firms expected to profit

15
00:00:48,960 --> 00:00:52,239
Speaker 2: from AI demand sank two, led by a twenty one

16
00:00:52,320 --> 00:00:55,400
Speaker 2: percent drop in the share price of Constellation Energy Core.

17
00:00:56,160 --> 00:00:59,760
Speaker 3: This is a really stunning development, kind of coming out

18
00:00:59,760 --> 00:01:01,920
Speaker 3: of aware for the AI world.

19
00:01:02,600 --> 00:01:05,920
Speaker 2: This stunning development came in the form of a new

20
00:01:06,080 --> 00:01:12,520
Speaker 2: AI competitor, a Chinese company called deep Seek. Jackie Devalos,

21
00:01:12,560 --> 00:01:16,280
Speaker 2: who covers artificial intelligence for Bloomberg, says deep Seek has

22
00:01:16,360 --> 00:01:19,840
Speaker 2: disrupted the AI world by building a model that rivals

23
00:01:19,880 --> 00:01:23,000
Speaker 2: its competitors using a lot fewer resources.

24
00:01:23,520 --> 00:01:26,880
Speaker 3: Tech companies spend billions of dollars at this point on

25
00:01:26,959 --> 00:01:31,320
Speaker 3: training their AI models, And you had Meta last week

26
00:01:31,560 --> 00:01:35,680
Speaker 3: saying that it would spend upwards of sixty five billion

27
00:01:35,800 --> 00:01:39,240
Speaker 3: this year on AI development, So it's a lot of money.

28
00:01:40,080 --> 00:01:43,200
Speaker 2: Deep Sea claims it's spent closer to six million dollars

29
00:01:43,240 --> 00:01:46,080
Speaker 2: to train its most recent model, and it's made its

30
00:01:46,080 --> 00:01:49,920
Speaker 2: system open source, meaning anyone can access its code and

31
00:01:49,960 --> 00:01:54,040
Speaker 2: build on it in a matter of days. The company's chatbot,

32
00:01:54,160 --> 00:01:58,080
Speaker 2: which functions similarly to chat GPT, has rocketed to the

33
00:01:58,120 --> 00:02:01,720
Speaker 2: top of App store charts, and its sudden dominance has

34
00:02:01,760 --> 00:02:06,360
Speaker 2: some investors questioning the strategies and the valuations of the

35
00:02:06,480 --> 00:02:10,240
Speaker 2: US tech companies that are developing AI and the technology

36
00:02:10,240 --> 00:02:11,000
Speaker 2: that powers it.

37
00:02:11,440 --> 00:02:14,240
Speaker 3: And I think deep Seek is now putting a really

38
00:02:14,240 --> 00:02:16,720
Speaker 3: big focus on what can you come up within the

39
00:02:16,760 --> 00:02:19,560
Speaker 3: short term or are we just wasting our money here?

40
00:02:23,080 --> 00:02:25,800
Speaker 2: This is the big take from Bloomberg News. I'm Sarah

41
00:02:25,880 --> 00:02:29,600
Speaker 2: Holder today on the show The Rise of deep Seek.

42
00:02:29,880 --> 00:02:34,359
Speaker 2: How it could upend the USAI industry's competitive advantage or

43
00:02:34,480 --> 00:02:44,360
Speaker 2: push it to innovate. It's not what deep seek does

44
00:02:44,400 --> 00:02:48,520
Speaker 2: that's so revolutionary. Functionally, it's actually very similar to a

45
00:02:48,560 --> 00:02:50,360
Speaker 2: lot of other AI chatbots.

46
00:02:50,639 --> 00:02:55,960
Speaker 3: Yeah, it's a large language model that can handle reasoning,

47
00:02:56,200 --> 00:03:00,600
Speaker 3: and you can put in a question, it gives you answer.

48
00:03:01,440 --> 00:03:04,440
Speaker 3: Anyone can go to the app store and download it.

49
00:03:04,520 --> 00:03:08,040
Speaker 3: If you're a student looking for help on your essay,

50
00:03:08,600 --> 00:03:13,400
Speaker 3: or if you're a researcher looking to help on your

51
00:03:13,440 --> 00:03:17,040
Speaker 3: PhD dissertation, the same people that kind of go to

52
00:03:17,639 --> 00:03:22,200
Speaker 3: a chat GBT or an anthropics clod or a Google Gemini.

53
00:03:22,440 --> 00:03:26,720
Speaker 3: It's general purpose. And what you're also going to start

54
00:03:26,800 --> 00:03:30,720
Speaker 3: seeing is the coding community really delve into this. They're

55
00:03:31,000 --> 00:03:34,040
Speaker 3: kind of one of the bigger cohorts that are early

56
00:03:34,080 --> 00:03:38,120
Speaker 3: adopters of this because it is so crucial for coding.

57
00:03:38,160 --> 00:03:42,000
Speaker 3: It takes a lot of the mundane tasks from coding

58
00:03:42,080 --> 00:03:42,760
Speaker 3: out of the picture.

59
00:03:43,240 --> 00:03:46,360
Speaker 2: Why was deep Seek's rise seemingly such a surprise to

60
00:03:46,400 --> 00:03:47,120
Speaker 2: so many people.

61
00:03:47,680 --> 00:03:50,400
Speaker 3: Part of the surprise was that this startup sort of

62
00:03:50,440 --> 00:03:52,800
Speaker 3: came out of nowhere. It was founded in twenty twenty

63
00:03:52,840 --> 00:03:55,680
Speaker 3: three by a guy that used to run a hedge fund,

64
00:03:55,760 --> 00:03:57,880
Speaker 3: and it was kind of this spin out out of

65
00:03:57,880 --> 00:04:01,360
Speaker 3: this hedge fund and it really didn't come onto the

66
00:04:01,400 --> 00:04:06,000
Speaker 3: scene until this year when it released this model that

67
00:04:06,080 --> 00:04:09,680
Speaker 3: when it performed on benchmarks, it was pretty comparable to

68
00:04:10,120 --> 00:04:13,040
Speaker 3: what we've seen out of open AI and Thropic and

69
00:04:13,080 --> 00:04:16,720
Speaker 3: other leading models. And so part of the surprise was

70
00:04:16,920 --> 00:04:19,760
Speaker 3: how were they able to do this so quickly and

71
00:04:20,320 --> 00:04:24,080
Speaker 3: once we started seeing reports that not only was it faster,

72
00:04:24,520 --> 00:04:26,200
Speaker 3: but it was actually done at a.

73
00:04:26,080 --> 00:04:27,279
Speaker 1: Fraction of the cost.

74
00:04:27,800 --> 00:04:31,279
Speaker 3: Remember, it's taken billions of dollars for these huge AI

75
00:04:31,400 --> 00:04:35,919
Speaker 3: companies to build and train and release these AI models

76
00:04:35,920 --> 00:04:39,080
Speaker 3: that power the chatbots that we all play around with.

77
00:04:39,480 --> 00:04:42,880
Speaker 2: Deepseek has said it spent only five point six million

78
00:04:42,920 --> 00:04:46,240
Speaker 2: dollars to train its latest AI model, though it also

79
00:04:46,400 --> 00:04:50,400
Speaker 2: cautioned that that figure didn't include all their costs, and

80
00:04:50,480 --> 00:04:54,320
Speaker 2: since it needs less computing power, it could have significantly

81
00:04:54,400 --> 00:04:56,000
Speaker 2: lower energy needs.

82
00:04:56,680 --> 00:04:57,640
Speaker 1: How did this.

83
00:04:57,600 --> 00:05:02,039
Speaker 2: Young startup develop this product so cheaply, Well, part.

84
00:05:01,880 --> 00:05:04,599
Speaker 3: Of it has to do with its roots as coming

85
00:05:04,600 --> 00:05:07,760
Speaker 3: out of this quantitative hedge fund, so it has this

86
00:05:07,960 --> 00:05:13,279
Speaker 3: deep knowledge in this computational framework. And not to get

87
00:05:13,279 --> 00:05:16,800
Speaker 3: too into the weeds, but deep Seak has focused on

88
00:05:16,839 --> 00:05:23,359
Speaker 3: what it calls maximizing software driven optimization. Basically, what is

89
00:05:23,560 --> 00:05:28,440
Speaker 3: able to do is create an architecture that ran more efficiently,

90
00:05:28,520 --> 00:05:35,039
Speaker 3: that didn't need these really intensive chips that are essential

91
00:05:35,160 --> 00:05:39,440
Speaker 3: for the models that run open Ai and Google and

92
00:05:39,640 --> 00:05:44,640
Speaker 3: Meta Islama. And it did this because there's been export

93
00:05:44,720 --> 00:05:49,200
Speaker 3: controls on chips going to China to prevent fresh AI

94
00:05:49,279 --> 00:05:53,440
Speaker 3: startups from coming about and challenging US dominance. And so

95
00:05:53,839 --> 00:05:56,520
Speaker 3: what we saw deep Seek do is kind of get

96
00:05:56,560 --> 00:06:01,640
Speaker 3: around these export controls by focus its models to run

97
00:06:01,680 --> 00:06:04,040
Speaker 3: in a different way, kind of in a different sort

98
00:06:04,080 --> 00:06:09,400
Speaker 3: of architecture. And Nvidia's H one hundreds obviously have propelled

99
00:06:09,480 --> 00:06:10,640
Speaker 3: the company.

100
00:06:10,240 --> 00:06:12,960
Speaker 1: To new heights, but those are really.

101
00:06:12,720 --> 00:06:15,680
Speaker 3: Expensive, so the fact that deep Seek doesn't have to

102
00:06:15,839 --> 00:06:18,520
Speaker 3: rely as much on them kind of takes a huge

103
00:06:18,600 --> 00:06:19,840
Speaker 3: chunk of that cost out.

104
00:06:20,480 --> 00:06:23,320
Speaker 2: When was deep Seek released and then when did it

105
00:06:23,400 --> 00:06:25,359
Speaker 2: start getting so popular.

106
00:06:25,760 --> 00:06:30,000
Speaker 3: Deep Seek was founded in twenty twenty three, but we

107
00:06:30,120 --> 00:06:33,520
Speaker 3: really started to see it make a splash in the

108
00:06:33,560 --> 00:06:38,960
Speaker 3: AI world when it released a report that outlined its

109
00:06:39,000 --> 00:06:42,120
Speaker 3: performance against other AI companies.

110
00:06:42,560 --> 00:06:45,640
Speaker 2: What stood out to investors in those reports.

111
00:06:45,440 --> 00:06:47,080
Speaker 1: You know, it performed really well.

112
00:06:47,160 --> 00:06:51,800
Speaker 3: So its latest model reasoning capabilities were similar to that

113
00:06:51,960 --> 00:06:55,599
Speaker 3: of chat GPT. Deep Seek also said that it's better

114
00:06:55,800 --> 00:07:00,760
Speaker 3: at various tax including coding and math and and chat

115
00:07:00,839 --> 00:07:04,360
Speaker 3: GPT is sort of this leader in conversational AI. It's

116
00:07:04,360 --> 00:07:10,280
Speaker 3: pretty versatile, but deepseeak really targets more specific application that

117
00:07:10,360 --> 00:07:13,800
Speaker 3: developers might be looking for that are more customize and

118
00:07:13,920 --> 00:07:14,800
Speaker 3: cost effective.

119
00:07:15,440 --> 00:07:19,040
Speaker 1: You still lack some of that, you know, polish.

120
00:07:19,320 --> 00:07:22,520
Speaker 3: It's not as polished as you might see a chat GPT,

121
00:07:23,400 --> 00:07:27,680
Speaker 3: but in terms of its speed and just the way

122
00:07:27,760 --> 00:07:31,400
Speaker 3: that it's able to handle a complex question or query

123
00:07:32,160 --> 00:07:34,760
Speaker 3: is really surprising given you know, this company was just

124
00:07:34,800 --> 00:07:35,840
Speaker 3: founded about a year ago.

125
00:07:36,240 --> 00:07:38,160
Speaker 2: So when deep seak rose to the top of the

126
00:07:38,200 --> 00:07:41,760
Speaker 2: app charts, when it released this information showing how it worked,

127
00:07:42,040 --> 00:07:45,640
Speaker 2: how it's performing, it had the seismic impact on US markets.

128
00:07:46,080 --> 00:07:48,640
Speaker 2: What do experts think about the market reaction? Was it

129
00:07:48,760 --> 00:07:50,560
Speaker 2: overblown appropriate?

130
00:07:50,960 --> 00:07:54,600
Speaker 3: I think you're seeing some market experts issue and I

131
00:07:54,720 --> 00:07:58,680
Speaker 3: told you so that there was reason to doubt that

132
00:07:58,960 --> 00:08:01,640
Speaker 3: so many billions being poured into.

133
00:08:01,440 --> 00:08:02,840
Speaker 1: The space was justified.

134
00:08:03,240 --> 00:08:08,240
Speaker 3: You also have other experts kind of saying, hold up,

135
00:08:08,640 --> 00:08:12,280
Speaker 3: let's wait and see what this startup is all about.

136
00:08:12,760 --> 00:08:14,160
Speaker 1: We need to look under the hood.

137
00:08:14,240 --> 00:08:18,760
Speaker 3: These are just the first initial days of getting to

138
00:08:18,840 --> 00:08:23,720
Speaker 3: compare what deep Seeks model does how much more runway

139
00:08:23,960 --> 00:08:25,360
Speaker 3: has to go as well.

140
00:08:25,400 --> 00:08:27,160
Speaker 1: If you remember the early.

141
00:08:26,920 --> 00:08:31,280
Speaker 3: Days of CHATGBT felt like we've never seen anything like

142
00:08:31,320 --> 00:08:34,680
Speaker 3: it before, and so I do think people are still

143
00:08:34,679 --> 00:08:39,120
Speaker 3: trying to wrap their heads around how good is this model.

144
00:08:39,400 --> 00:08:42,479
Speaker 2: They're also trying to figure out whether deep seeks reports

145
00:08:42,480 --> 00:08:46,160
Speaker 2: of lower costs and greater efficiency are as impressive as

146
00:08:46,200 --> 00:08:47,240
Speaker 2: they initially seen.

147
00:08:48,000 --> 00:08:51,080
Speaker 3: You look at the other AI companies open Ai and

148
00:08:51,200 --> 00:08:55,160
Speaker 3: Meta and Google, their models take a longer time to

149
00:08:55,240 --> 00:08:55,680
Speaker 3: come out.

150
00:08:55,760 --> 00:08:57,200
Speaker 1: They're big tech giants.

151
00:08:57,200 --> 00:09:01,040
Speaker 3: They're naturally going to move a little bit slower when

152
00:09:01,080 --> 00:09:04,120
Speaker 3: it comes to releasing their technology to the billions.

153
00:09:03,679 --> 00:09:05,440
Speaker 1: Of people it has captive already.

154
00:09:06,000 --> 00:09:09,240
Speaker 3: And that's the benefit of being a small startup that

155
00:09:09,280 --> 00:09:12,040
Speaker 3: you get to release things quicker and make a big splash.

156
00:09:12,040 --> 00:09:15,839
Speaker 1: But whether that splash really has legs, whether it lasts, is.

157
00:09:15,760 --> 00:09:17,440
Speaker 3: Still yet to be seen, and I think that's what

158
00:09:18,360 --> 00:09:21,760
Speaker 3: many people in the AI field are wondering about.

159
00:09:22,200 --> 00:09:25,360
Speaker 2: On January twenty fourth, deep Seat got a high profile

160
00:09:25,480 --> 00:09:28,800
Speaker 2: not of approval when Mark Andresen posted on x that

161
00:09:28,880 --> 00:09:31,640
Speaker 2: the company's latest version is quote one of the most

162
00:09:31,720 --> 00:09:35,240
Speaker 2: amazing and impressive breakthroughs he's ever seen once.

163
00:09:35,280 --> 00:09:39,520
Speaker 3: He tweeted that coming from a tech titan like himself,

164
00:09:39,760 --> 00:09:43,520
Speaker 3: he is very familiar with the AI space his firm,

165
00:09:43,559 --> 00:09:46,960
Speaker 3: and Resent Horowitz is a big investor in AI companies,

166
00:09:47,360 --> 00:09:51,080
Speaker 3: and so kind of getting his shock and awe at

167
00:09:51,080 --> 00:09:54,400
Speaker 3: how good this Deep Seek model was, I think sent

168
00:09:54,520 --> 00:09:56,240
Speaker 3: further reverberations throughout the.

169
00:09:56,200 --> 00:10:00,920
Speaker 2: Market coming up. What is the emergence of this Chinese

170
00:10:00,920 --> 00:10:05,199
Speaker 2: firm mean for US dominance in artificial intelligence? And could

171
00:10:05,240 --> 00:10:09,120
Speaker 2: Deep Seek's ultra low cost model put pressure on soaring

172
00:10:09,240 --> 00:10:21,360
Speaker 2: American AI budgets. The US has been a clear leader

173
00:10:21,440 --> 00:10:24,560
Speaker 2: in the artificial intelligence space, a position it's tried to

174
00:10:24,559 --> 00:10:28,320
Speaker 2: protect by restricting the sale of AI chips and technology abroad.

175
00:10:28,960 --> 00:10:33,400
Speaker 2: So I asked Bloomberg's Jackie Devalos how American policymakers might

176
00:10:33,520 --> 00:10:36,160
Speaker 2: react to the China based Deep Seeks rise.

177
00:10:37,000 --> 00:10:41,080
Speaker 3: I think it's an opportunity for people to take a

178
00:10:41,120 --> 00:10:44,200
Speaker 3: step back and evaluate.

179
00:10:43,640 --> 00:10:45,439
Speaker 1: Kind of where we are now, what.

180
00:10:45,480 --> 00:10:51,280
Speaker 3: It's taken, the level of investment, and ultimately what people

181
00:10:51,360 --> 00:10:55,320
Speaker 3: want to see going forward. A lot of the attention

182
00:10:55,520 --> 00:10:59,800
Speaker 3: right now, especially going into this new administration with President

183
00:11:00,280 --> 00:11:06,160
Speaker 3: has been around national competitiveness against adversaries like China. And

184
00:11:06,480 --> 00:11:09,920
Speaker 3: it really is shocking when you consider the fact that

185
00:11:09,960 --> 00:11:12,000
Speaker 3: the US has done a lot of work to make

186
00:11:12,040 --> 00:11:15,720
Speaker 3: sure it can restrict the supply of these high powered

187
00:11:15,840 --> 00:11:20,800
Speaker 3: chips to China, citing these national security concerns, and then

188
00:11:20,840 --> 00:11:23,959
Speaker 3: you have deep Seak coming onto the scene, which means

189
00:11:23,960 --> 00:11:26,680
Speaker 3: that they were able to achieve this low cost model

190
00:11:27,000 --> 00:11:30,520
Speaker 3: without relying on these high power chips.

191
00:11:30,559 --> 00:11:32,600
Speaker 1: So clearly there was a work around.

192
00:11:32,320 --> 00:11:36,880
Speaker 3: Here that the AI community missed, but also the US

193
00:11:36,920 --> 00:11:37,920
Speaker 3: missed well.

194
00:11:38,000 --> 00:11:40,320
Speaker 2: Right the fact that deep Seek was developed so cheaply

195
00:11:40,440 --> 00:11:44,200
Speaker 2: for millions, not billions, What kind of pressure does that

196
00:11:44,240 --> 00:11:47,280
Speaker 2: put on US companies and their spending. Are they going

197
00:11:47,360 --> 00:11:50,400
Speaker 2: to try and adopt any of the strategies that deep

198
00:11:50,440 --> 00:11:52,360
Speaker 2: Seek used to bring those costs down.

199
00:11:53,240 --> 00:11:55,839
Speaker 3: That's an open question because we've already seen a lot

200
00:11:55,880 --> 00:12:00,520
Speaker 3: of concern about the supply of chips. Are we getting enough,

201
00:12:00,640 --> 00:12:03,480
Speaker 3: are we getting them fast enough? Who are they going

202
00:12:03,520 --> 00:12:06,400
Speaker 3: to in Vidias coming out with new models. Then you

203
00:12:06,440 --> 00:12:09,520
Speaker 3: have in Vidia competitors coming out with new models, and

204
00:12:09,600 --> 00:12:13,160
Speaker 3: so this deep Seek development really raises the question of well,

205
00:12:13,160 --> 00:12:15,240
Speaker 3: wait a second, do we need this many chips to

206
00:12:15,280 --> 00:12:19,760
Speaker 3: begin with? You might see AI companies get more creative

207
00:12:19,800 --> 00:12:22,920
Speaker 3: in how they're building these models, perhaps changing up their

208
00:12:23,000 --> 00:12:27,680
Speaker 3: architecture and framework. You might see investors starting to perhaps

209
00:12:27,840 --> 00:12:32,240
Speaker 3: dole out less money and demand that you know, they

210
00:12:32,320 --> 00:12:35,680
Speaker 3: find other cheaper ways to make this. So I think

211
00:12:35,800 --> 00:12:38,320
Speaker 3: it's all going to start kind of coming together with

212
00:12:38,400 --> 00:12:41,440
Speaker 3: a bit more urgency now that we know that clearly

213
00:12:41,760 --> 00:12:45,840
Speaker 3: this Chinese competitor is giving US companies a run for

214
00:12:45,880 --> 00:12:46,360
Speaker 3: their money.

215
00:12:46,600 --> 00:12:49,920
Speaker 2: Could this be good news for the AI industry overall

216
00:12:49,960 --> 00:12:53,719
Speaker 2: if it pushes this kind of innovation and investment, or

217
00:12:53,800 --> 00:12:56,360
Speaker 2: does it feel like more of a threat.

218
00:12:56,600 --> 00:12:58,000
Speaker 1: I think it depends on who you are.

219
00:12:58,400 --> 00:13:01,920
Speaker 3: If you're the US government, and this definitely feels more

220
00:13:02,040 --> 00:13:06,840
Speaker 3: like a threat. If you are an innovator who is

221
00:13:07,679 --> 00:13:11,160
Speaker 3: enthralled by reading through some of these technical reports that

222
00:13:11,240 --> 00:13:14,760
Speaker 3: Deep Seek put out, it's a really exciting time, and

223
00:13:15,520 --> 00:13:18,640
Speaker 3: I think if you're an AI company, then this is

224
00:13:19,600 --> 00:13:23,439
Speaker 3: a good time to perhaps like light a fire under

225
00:13:23,480 --> 00:13:26,400
Speaker 3: everyone and say, hey, how can we be more creative.

226
00:13:26,480 --> 00:13:29,680
Speaker 3: I think investors are always far more sensitive to these

227
00:13:29,720 --> 00:13:30,520
Speaker 3: things than.

228
00:13:30,840 --> 00:13:33,040
Speaker 1: Perhaps the industry itself.

229
00:13:33,520 --> 00:13:36,040
Speaker 3: But on the whole, I think we're going to see

230
00:13:36,120 --> 00:13:41,959
Speaker 3: far more urgency in releasing products, trying out new features,

231
00:13:42,040 --> 00:13:45,480
Speaker 3: kind of keeping people engaged, because the more people use this,

232
00:13:45,559 --> 00:13:48,040
Speaker 3: the more data goes in, the better these models get.

233
00:13:48,559 --> 00:13:52,120
Speaker 2: By Monday afternoon, and Video released a statement calling deep

234
00:13:52,120 --> 00:13:56,280
Speaker 2: Seek an excellent AI advancement, one that would require a

235
00:13:56,280 --> 00:13:59,880
Speaker 2: lot of in video products in the future. Microsoft ce

236
00:14:00,240 --> 00:14:04,240
Speaker 2: Satya Nadella also reacted with a post on LinkedIn, saying

237
00:14:04,559 --> 00:14:07,760
Speaker 2: as AI gets more efficient and accessible, we will see

238
00:14:07,760 --> 00:14:11,040
Speaker 2: it's use skyrocket, turning it into a commodity we just

239
00:14:11,120 --> 00:14:14,840
Speaker 2: can't get enough of. So Deepseek has this edge, the

240
00:14:14,960 --> 00:14:17,760
Speaker 2: edge being it can do all this a lot cheaper

241
00:14:17,800 --> 00:14:20,920
Speaker 2: than competitors. Did you expect to see such a strong

242
00:14:21,160 --> 00:14:24,560
Speaker 2: reaction in the markets, this intense of a reaction.

243
00:14:24,600 --> 00:14:29,560
Speaker 3: This intensive a reaction, is a huge surprise, because the

244
00:14:29,680 --> 00:14:34,320
Speaker 3: evidence that this AI startup might dethrone the leading AI

245
00:14:34,480 --> 00:14:39,360
Speaker 3: tech companies isn't completely solid yet. This has just kind

246
00:14:39,400 --> 00:14:42,480
Speaker 3: of been some early signs that they've gotten traction on

247
00:14:42,560 --> 00:14:46,440
Speaker 3: the app Store, that they're performing while on benchmarks, and

248
00:14:46,680 --> 00:14:49,240
Speaker 3: sure that you know that this can be built faster

249
00:14:49,320 --> 00:14:53,320
Speaker 3: and more cheaply, But I think the overall concern that

250
00:14:53,400 --> 00:14:58,080
Speaker 3: we're clearly seeing, especially going into tech earnings over the

251
00:14:58,120 --> 00:15:03,320
Speaker 3: next couple weeks, is question whether are the billions of

252
00:15:03,360 --> 00:15:07,960
Speaker 3: dollars that are going into these models that companies have

253
00:15:08,040 --> 00:15:12,400
Speaker 3: been telling investors it's worth it. We are going to

254
00:15:12,480 --> 00:15:16,920
Speaker 3: monetize this and all of this investment will eventually.

255
00:15:17,840 --> 00:15:18,680
Speaker 1: Be fruitful.

256
00:15:19,120 --> 00:15:22,880
Speaker 3: That's coming into question, and what we're going to see

257
00:15:22,880 --> 00:15:26,400
Speaker 3: over the next couple of days is wondering what is

258
00:15:26,440 --> 00:15:29,040
Speaker 3: going on kind of behind the scenes, how is this

259
00:15:29,120 --> 00:15:30,120
Speaker 3: being monetized?

260
00:15:30,240 --> 00:15:32,280
Speaker 1: Does it have to be this expensive?

261
00:15:33,000 --> 00:15:35,680
Speaker 2: Jackie? There are still a lot of unknowns here right

262
00:15:36,120 --> 00:15:38,320
Speaker 2: What are you going to be watching out for over

263
00:15:38,360 --> 00:15:39,560
Speaker 2: the next couple of weeks.

264
00:15:40,400 --> 00:15:45,000
Speaker 3: There's so much excitement that goes into a newcomer really

265
00:15:45,480 --> 00:15:49,840
Speaker 3: plowing into the space. When you have this landscape already

266
00:15:49,880 --> 00:15:54,160
Speaker 3: really cemented, It's almost this like David and Goliath energy

267
00:15:54,280 --> 00:15:57,480
Speaker 3: coming in and I'm going to be looking out for

268
00:15:57,920 --> 00:16:03,240
Speaker 3: improvements and deep seeks apps how it uses this momentum.

269
00:16:03,840 --> 00:16:08,000
Speaker 3: I'm going to be looking for any reactions from President Trump.

270
00:16:08,520 --> 00:16:11,560
Speaker 3: You know, Chat Gibt didn't exist when President Trump was

271
00:16:12,200 --> 00:16:15,840
Speaker 3: in office the first time, and I'm curious if this

272
00:16:15,960 --> 00:16:21,040
Speaker 3: is going to inform how he goes about policy with

273
00:16:21,200 --> 00:16:25,360
Speaker 3: China in other ways that could have reverberations inside the

274
00:16:25,400 --> 00:16:30,040
Speaker 3: tech industry and the AI community. Those two worlds, the

275
00:16:30,120 --> 00:16:34,840
Speaker 3: policy in Washington and the AI community and technology in

276
00:16:34,880 --> 00:16:38,840
Speaker 3: Silicon Valley, are far more intertwined today than they ever

277
00:16:38,920 --> 00:16:39,280
Speaker 3: have been.

278
00:16:45,640 --> 00:16:48,840
Speaker 2: This is the Big Take from Bloomberg News. I'm Sarah Holder.

279
00:16:49,200 --> 00:16:52,160
Speaker 2: This episode was produced by Alex Tie. It was edited

280
00:16:52,160 --> 00:16:55,600
Speaker 2: by Tracy Samuelson and Lynn Dwong. It was mixed and

281
00:16:55,640 --> 00:16:58,440
Speaker 2: sound designed by Alex Sugia. It was fact checked by

282
00:16:58,440 --> 00:17:02,120
Speaker 2: Adrian A. Tapia. Our senior producer is Naomi Shaven. Our

283
00:17:02,160 --> 00:17:06,280
Speaker 2: senior editor is Elizabeth Ponso. Our executive producer is Nicole Beemster.

284
00:17:06,359 --> 00:17:10,040
Speaker 2: Border Sage Bauman is Bloomberg's head of podcasts. If you

285
00:17:10,119 --> 00:17:12,760
Speaker 2: liked this episode, make sure to subscribe and review The

286
00:17:12,800 --> 00:17:15,840
Speaker 2: Big Take wherever you listen to podcasts. It helps people

287
00:17:15,880 --> 00:17:19,200
Speaker 2: find the show. Thanks for listening. We'll be back tomorrow