WEBVTT - A Deep Dive on DeepSeek

0:00:02.520 --> 0:00:07.040
<v Speaker 1>Bloomberg Audio Studios, Podcasts, radio News.

0:00:10.960 --> 0:00:14.280
<v Speaker 2>Welcome to the Bloomberg Daybreak Asia podcast. I'm Deck Chrisner.

0:00:14.640 --> 0:00:17.600
<v Speaker 2>On today's episode, we'll be taking a deep dive into

0:00:17.680 --> 0:00:21.920
<v Speaker 2>Deep Seek. This is a Chinese artificial intelligence startup and

0:00:22.040 --> 0:00:25.360
<v Speaker 2>its aimodel, known as R one, appears to rival the

0:00:25.440 --> 0:00:30.400
<v Speaker 2>performance of US products like Chat GPT. Perhaps its biggest

0:00:30.440 --> 0:00:33.279
<v Speaker 2>selling point is that the model was developed at a

0:00:33.320 --> 0:00:36.440
<v Speaker 2>fraction of the cost of competing products, and that could

0:00:36.440 --> 0:00:40.640
<v Speaker 2>make high valuations of some USAI related companies a little

0:00:40.640 --> 0:00:43.640
<v Speaker 2>tough to justify. Let me give you one example today

0:00:44.000 --> 0:00:47.559
<v Speaker 2>in US trading, shares in the AI chip leader in Nvidia,

0:00:47.600 --> 0:00:50.680
<v Speaker 2>we're down nearly seventeen percent, and that led to a

0:00:50.680 --> 0:00:53.960
<v Speaker 2>loss in the Nasdaq one hundred of three percent. In

0:00:54.000 --> 0:00:56.320
<v Speaker 2>a moment, we'll be speaking with Jamie Cox. He is

0:00:56.360 --> 0:00:59.720
<v Speaker 2>managing partner at Harris Financial Group. But let's begin in

0:00:59.800 --> 0:01:03.960
<v Speaker 2>home with Bloomberg Tech editor of Vlad Savov. It's always

0:01:04.000 --> 0:01:06.240
<v Speaker 2>a pleasure to chat with you've Lad. This feels like

0:01:06.280 --> 0:01:10.520
<v Speaker 2>a major inflection point in the development of AI models. First,

0:01:10.560 --> 0:01:13.199
<v Speaker 2>let me get your reaction and what you're hearing among

0:01:13.319 --> 0:01:15.760
<v Speaker 2>Hong Kongers. Now that follow the tech space.

0:01:16.319 --> 0:01:19.240
<v Speaker 3>Well, the interesting thing about deep Seek is it's a

0:01:19.280 --> 0:01:24.319
<v Speaker 3>company that has rocketed to global renown and recognition today

0:01:24.440 --> 0:01:26.640
<v Speaker 3>and over the past week, but it has been known

0:01:26.680 --> 0:01:29.880
<v Speaker 3>about ever since it's beginning in twenty twenty three. It's

0:01:29.920 --> 0:01:33.120
<v Speaker 3>one that our Bloomberg Intelligence colleagues have been taken a

0:01:33.160 --> 0:01:36.759
<v Speaker 3>close look at because it is developing such a hot

0:01:36.840 --> 0:01:40.440
<v Speaker 3>area of new software as AI. It has been developing

0:01:40.440 --> 0:01:44.320
<v Speaker 3>its own models, so people here have been aware of it.

0:01:44.319 --> 0:01:48.240
<v Speaker 3>It has been on the radar. But really the thing

0:01:48.280 --> 0:01:52.200
<v Speaker 3>that has happened is it hit a certain threshold of recognition.

0:01:53.200 --> 0:01:57.080
<v Speaker 3>Part of it really was Silicon Valley people, venture capitalists.

0:01:57.160 --> 0:02:00.520
<v Speaker 3>They took an interest, they tried it out. They found

0:02:00.560 --> 0:02:03.600
<v Speaker 3>out that it is, indeed, like you mentioned, as competitive,

0:02:03.640 --> 0:02:07.760
<v Speaker 3>as the company itself says. It has posted some benchmark

0:02:07.840 --> 0:02:10.800
<v Speaker 3>results on this website. It has claimed already that it

0:02:10.840 --> 0:02:14.440
<v Speaker 3>competes with open Ai, but frankly, every AI model developer

0:02:14.480 --> 0:02:17.360
<v Speaker 3>claims that they compete or they can rival open Ai.

0:02:17.840 --> 0:02:20.320
<v Speaker 3>The amazing thing, and the really really impressive thing is

0:02:20.320 --> 0:02:22.600
<v Speaker 3>that the people who know this stuff, which I can't

0:02:22.600 --> 0:02:25.320
<v Speaker 3>claim that I'm an expert in it, they all recognize

0:02:25.320 --> 0:02:27.720
<v Speaker 3>it as a breakthrough so we need to recognize that too.

0:02:28.240 --> 0:02:31.280
<v Speaker 3>And it's also worth mentioning here that the company has

0:02:31.280 --> 0:02:33.320
<v Speaker 3>put out a research paper of its own and they've

0:02:33.320 --> 0:02:36.680
<v Speaker 3>said that the costs of training this model are less

0:02:36.680 --> 0:02:39.320
<v Speaker 3>than six million dollars. Now, that excludes a whole bunch

0:02:39.360 --> 0:02:42.520
<v Speaker 3>of costs, so we can't be exactly sure about how

0:02:42.600 --> 0:02:46.160
<v Speaker 3>much more affordable it is for this small company to

0:02:46.200 --> 0:02:49.840
<v Speaker 3>develop it, but it is very clearly a difference maker

0:02:50.000 --> 0:02:52.160
<v Speaker 3>when it comes to our assessment of how much it

0:02:52.200 --> 0:02:54.560
<v Speaker 3>costs to develop a world leading AI model.

0:02:54.639 --> 0:02:57.200
<v Speaker 2>Yeah, I think to be fair, our analyst here in

0:02:57.240 --> 0:02:59.639
<v Speaker 2>New York, Man Deep Saying, pointed out that our one

0:02:59.760 --> 0:03:04.520
<v Speaker 2>was to developed using output from open Ai, Anthropic and

0:03:04.639 --> 0:03:07.320
<v Speaker 2>LAMA to train the models. So it's almost as though

0:03:07.480 --> 0:03:10.240
<v Speaker 2>AI is training AI a little bit. And I think

0:03:10.320 --> 0:03:14.120
<v Speaker 2>that what may have precipitated or at least engendered a

0:03:14.160 --> 0:03:16.880
<v Speaker 2>lot of the selling that we saw in the US

0:03:16.880 --> 0:03:19.600
<v Speaker 2>in the last session was the rating on the Apple

0:03:19.639 --> 0:03:22.959
<v Speaker 2>app Store. I think R one rose to the top level.

0:03:23.720 --> 0:03:28.160
<v Speaker 2>Can we agree though, that the current Western approach to AI.

0:03:28.800 --> 0:03:31.920
<v Speaker 2>I'm talking here about the reliance on those high end chips,

0:03:32.440 --> 0:03:37.120
<v Speaker 2>extensive computing power, and vast amounts of electricity. Is that

0:03:37.200 --> 0:03:39.200
<v Speaker 2>being challenged to some extent.

0:03:39.320 --> 0:03:42.080
<v Speaker 3>Yes, we can agree that it's challenged. We cannot agree

0:03:42.120 --> 0:03:45.200
<v Speaker 3>that it's a foregone conclusion that it's going to be

0:03:45.240 --> 0:03:47.920
<v Speaker 3>going away. One thing to bear in mind here, which

0:03:47.920 --> 0:03:51.240
<v Speaker 3>I think is very important, is that Deepsek's model has

0:03:51.280 --> 0:03:55.080
<v Speaker 3>delivered an impressive chatbot. It only works with text. So

0:03:55.440 --> 0:03:58.000
<v Speaker 3>what we have is a breakthrough and an impressive achievement,

0:03:58.120 --> 0:04:01.480
<v Speaker 3>but it's within one medium. One of the things that

0:04:01.520 --> 0:04:03.640
<v Speaker 3>you have to bear in mind is that the more

0:04:03.680 --> 0:04:06.000
<v Speaker 3>advanced chips, the ones that are making a video is

0:04:06.000 --> 0:04:09.040
<v Speaker 3>still a multi trillion dollar company, even with its massive

0:04:09.040 --> 0:04:12.080
<v Speaker 3>share price drop. The thing that drives it is that

0:04:12.240 --> 0:04:14.160
<v Speaker 3>you want those chips in order to get to the

0:04:14.200 --> 0:04:17.640
<v Speaker 3>next stage of AI, specifically generative AI. So at the

0:04:17.640 --> 0:04:21.680
<v Speaker 3>moment we're talking about generating text, but the next evolution

0:04:21.839 --> 0:04:24.599
<v Speaker 3>is going to be generating images. Generating video is stuff

0:04:24.600 --> 0:04:29.120
<v Speaker 3>that open ai and other services have already made major

0:04:29.120 --> 0:04:32.360
<v Speaker 3>advances in, and that's where deep seek is probably going

0:04:32.400 --> 0:04:34.400
<v Speaker 3>to need a lot more hardware than it's used just

0:04:34.440 --> 0:04:34.880
<v Speaker 3>for text.

0:04:34.920 --> 0:04:37.680
<v Speaker 2>So it may be too much to say that deep

0:04:37.760 --> 0:04:41.040
<v Speaker 2>seek has been able to engineer a way around the

0:04:41.080 --> 0:04:44.360
<v Speaker 2>export bands that were put in place during the Biden administration,

0:04:45.000 --> 0:04:48.279
<v Speaker 2>the bands that basically kept China from accessing some of

0:04:48.320 --> 0:04:53.559
<v Speaker 2>those advanced semiconductors manufactured by and Vidiot. We can't really

0:04:53.760 --> 0:04:55.000
<v Speaker 2>articulate that, can we?

0:04:55.560 --> 0:04:58.240
<v Speaker 3>Yes, this is correct. Now, there are two things to

0:04:58.320 --> 0:05:03.279
<v Speaker 3>say here. Firstly, those bands and those trade cubs. Everyone

0:05:03.480 --> 0:05:06.000
<v Speaker 3>in the industry said, this is only going to encourage

0:05:06.120 --> 0:05:10.120
<v Speaker 3>China and Chinese companies to focus on more efficient methods.

0:05:10.320 --> 0:05:13.279
<v Speaker 3>So this is exactly what is being predicted by the industry.

0:05:13.520 --> 0:05:16.200
<v Speaker 3>They said, China will focus on more light, more small

0:05:16.240 --> 0:05:19.839
<v Speaker 3>scale models, and Deep Seek is the fruits of those efforts.

0:05:20.440 --> 0:05:23.120
<v Speaker 3>That being said, we cannot say that the sanctions are

0:05:23.120 --> 0:05:27.400
<v Speaker 3>not effective because Deep Seek CEO and founder has himself said,

0:05:27.800 --> 0:05:30.560
<v Speaker 3>the thing that Deep Seak requires right now is not

0:05:30.640 --> 0:05:34.680
<v Speaker 3>more money, is not more investment, is actually access to

0:05:34.800 --> 0:05:37.560
<v Speaker 3>those most advanced ships. The embargo is e cosed, the

0:05:37.640 --> 0:05:40.200
<v Speaker 3>ban on accessing those video chips is the thing that's

0:05:40.200 --> 0:05:41.200
<v Speaker 3>holding Deep Sea back.

0:05:41.480 --> 0:05:44.599
<v Speaker 2>Are we likely to see some type of fragmentation within

0:05:44.680 --> 0:05:48.039
<v Speaker 2>the artificial intelligence industry, which is to say that you

0:05:48.120 --> 0:05:51.719
<v Speaker 2>have certain products that have been produced at low cost

0:05:51.880 --> 0:05:56.320
<v Speaker 2>being rolled out at lower prices, and maybe those products

0:05:56.360 --> 0:05:59.440
<v Speaker 2>will see greater adoption and some of the more sophisticated

0:05:59.480 --> 0:06:02.919
<v Speaker 2>elements of artificial intelligence you'll have to pay a premium for,

0:06:03.600 --> 0:06:06.240
<v Speaker 2>and that segment of the market will be controlled by

0:06:07.000 --> 0:06:08.200
<v Speaker 2>very few companies.

0:06:08.640 --> 0:06:09.480
<v Speaker 4>Yes, I can very.

0:06:09.400 --> 0:06:11.440
<v Speaker 3>Much see that. One of the things that has been

0:06:11.560 --> 0:06:14.560
<v Speaker 3>the trend over the past few months is actually opening

0:06:14.640 --> 0:06:17.800
<v Speaker 3>Ey's revenue generation has accelerated, and it's one of the

0:06:17.839 --> 0:06:20.240
<v Speaker 3>trends that we're seeing. And if you had asked me

0:06:20.320 --> 0:06:22.239
<v Speaker 3>to predict what would be the big trend in AI

0:06:22.240 --> 0:06:25.560
<v Speaker 3>in twenty twenty five, it is exactly this taking all

0:06:25.600 --> 0:06:29.200
<v Speaker 3>that investment, taking all that developed technology, and turning into

0:06:29.240 --> 0:06:32.520
<v Speaker 3>products that you can sell subscriptions or in other ways

0:06:32.680 --> 0:06:35.440
<v Speaker 3>entices people to spend money on them. That is still

0:06:35.480 --> 0:06:38.040
<v Speaker 3>going to continue to be the case the premium products.

0:06:38.080 --> 0:06:42.400
<v Speaker 3>Open ai sells a two hundred dollars per month pro product.

0:06:42.960 --> 0:06:45.640
<v Speaker 3>There is an audience for that. But that being said,

0:06:46.160 --> 0:06:50.240
<v Speaker 3>the deep seek approach, it releases the software as an

0:06:50.240 --> 0:06:53.360
<v Speaker 3>open source software, so you as an individual can take

0:06:53.400 --> 0:06:55.520
<v Speaker 3>that and piggyback on top of it and decide that

0:06:55.560 --> 0:06:58.240
<v Speaker 3>you want to build your own specific chatbook. And a

0:06:58.240 --> 0:07:00.839
<v Speaker 3>good analogy that I have for this is when you

0:07:00.839 --> 0:07:03.919
<v Speaker 3>think about banking. You have the traditional banks, the ones

0:07:03.920 --> 0:07:06.800
<v Speaker 3>with branches and all those overheads and costs, and then

0:07:06.839 --> 0:07:09.359
<v Speaker 3>you have the digit banks, the ones that are mobile only,

0:07:09.440 --> 0:07:12.520
<v Speaker 3>app only, and because of that efficiency that those other

0:07:12.600 --> 0:07:16.280
<v Speaker 3>banks have, they can offer you better rates and better office.

0:07:16.040 --> 0:07:19.400
<v Speaker 2>So to go back to the hardware software issue, it

0:07:19.480 --> 0:07:23.920
<v Speaker 2>seems like Chinese AI engineers have found a way to

0:07:24.160 --> 0:07:28.080
<v Speaker 2>work with the software given the limitations of the hardware

0:07:28.120 --> 0:07:30.560
<v Speaker 2>that they have been forced to use because of those

0:07:31.040 --> 0:07:34.920
<v Speaker 2>US export controls. But I'm wondering whether on the semiconductor

0:07:34.920 --> 0:07:38.320
<v Speaker 2>industry side of things in China that we may see

0:07:38.360 --> 0:07:41.200
<v Speaker 2>some sort of breakthrough when it comes to advanced AI

0:07:41.320 --> 0:07:43.960
<v Speaker 2>chips in the near future. Is that a possibility.

0:07:44.280 --> 0:07:46.880
<v Speaker 3>Yes, There's two things to say here. Firstly, from what

0:07:46.920 --> 0:07:49.200
<v Speaker 3>little we know about Deep Seek, and bear in mind

0:07:49.320 --> 0:07:52.080
<v Speaker 3>it's a very very young startup. It came out of

0:07:52.080 --> 0:07:55.800
<v Speaker 3>an AI hedge fund AI driven hedge fund, so we

0:07:55.840 --> 0:07:58.760
<v Speaker 3>still don't know that much about it. But from interviews

0:07:58.760 --> 0:08:00.840
<v Speaker 3>with the founder, he has said that most of his

0:08:00.960 --> 0:08:05.520
<v Speaker 3>engineers are graduates from China's universities. They are entirely domestic

0:08:05.720 --> 0:08:10.000
<v Speaker 3>Chinese engineers, so credit to the company is educational system

0:08:10.360 --> 0:08:14.400
<v Speaker 3>now on the semiconductor front, there are technical limitations and challenges.

0:08:14.520 --> 0:08:18.360
<v Speaker 3>The classic one is EUV machines that only ASML makes

0:08:18.520 --> 0:08:22.560
<v Speaker 3>and is prevented from selling to China. Solving that issue.

0:08:22.600 --> 0:08:25.040
<v Speaker 3>We haven't seen much evidence of China doing it, but

0:08:25.080 --> 0:08:27.400
<v Speaker 3>then again, it may be something along the lines of

0:08:27.440 --> 0:08:29.720
<v Speaker 3>deep Seek. We never saw deep Seek. We were only

0:08:29.760 --> 0:08:31.920
<v Speaker 3>aware that is building a model. We never saw it

0:08:32.320 --> 0:08:35.319
<v Speaker 3>doing a breakthrough until today, until it was actually ready

0:08:35.360 --> 0:08:36.280
<v Speaker 3>to come out there.

0:08:36.480 --> 0:08:38.440
<v Speaker 2>Flight, We'll leave it there, Thank you so much. Let's

0:08:38.480 --> 0:08:41.520
<v Speaker 2>have of Bloomberg Tech editor in Asia joining us from

0:08:41.559 --> 0:08:51.439
<v Speaker 2>Hong Kong here on the Daybreak Asia podcast. Welcome back

0:08:51.440 --> 0:08:55.040
<v Speaker 2>to the Daybreak Asia Podcast. I'm Doug Chrisner. We continue

0:08:55.040 --> 0:08:58.480
<v Speaker 2>with our deep dive on deep Seek. News of the

0:08:58.559 --> 0:09:01.960
<v Speaker 2>Chinese startups new Chat bought r one, sparked to sell

0:09:01.960 --> 0:09:05.600
<v Speaker 2>off in many AI related names globally, and it wiped

0:09:05.640 --> 0:09:08.600
<v Speaker 2>out nearly a trillion dollars in market cap from US

0:09:08.600 --> 0:09:12.120
<v Speaker 2>and European tech names. So now a key question is

0:09:12.200 --> 0:09:15.920
<v Speaker 2>on capex spending plans for some of America's biggest companies.

0:09:16.160 --> 0:09:18.480
<v Speaker 2>Joining me now for a closer look. As Jamie Cox

0:09:19.000 --> 0:09:22.520
<v Speaker 2>is managing partner at Harris Financial Group. Jamie, thank you

0:09:22.559 --> 0:09:25.000
<v Speaker 2>so much for being with us. Certainly was an interesting

0:09:25.080 --> 0:09:28.360
<v Speaker 2>day where AI was concerned. How do you make sense

0:09:28.600 --> 0:09:31.520
<v Speaker 2>of the market reaction to this Deep Seek story.

0:09:31.880 --> 0:09:34.960
<v Speaker 4>I think the market's reaction was quite positive. I mean,

0:09:35.000 --> 0:09:35.959
<v Speaker 4>there were only.

0:09:35.800 --> 0:09:39.040
<v Speaker 1>Eight dollstocks that were down today. Sixty percent of the

0:09:39.320 --> 0:09:41.960
<v Speaker 1>S and P five hundred companies were up. So this

0:09:42.120 --> 0:09:46.800
<v Speaker 1>was very much an isolated day for you know, chip stocks.

0:09:46.840 --> 0:09:50.040
<v Speaker 1>This was sort of the moment in time where you

0:09:50.120 --> 0:09:53.959
<v Speaker 1>see disruption take hold, and when you start to see

0:09:53.960 --> 0:10:01.520
<v Speaker 1>things like AI disruption and the possibility that maybe companies

0:10:01.559 --> 0:10:05.600
<v Speaker 1>are overinvesting in AI infrastructure, you have to sit back

0:10:05.600 --> 0:10:08.160
<v Speaker 1>and think a minute and ask yourself, you know, if

0:10:08.280 --> 0:10:11.200
<v Speaker 1>Nvidia can legitimately sell thirty thousand dollars chips or not.

0:10:11.600 --> 0:10:14.280
<v Speaker 1>So I think it made a lot of investors think.

0:10:14.559 --> 0:10:17.679
<v Speaker 1>But a lot of investors have been worried about overvaluation in.

0:10:17.640 --> 0:10:18.559
<v Speaker 4>These stocks anyway.

0:10:18.600 --> 0:10:23.000
<v Speaker 1>So I think that the Deep Seek situation sort of precipitated.

0:10:23.480 --> 0:10:26.560
<v Speaker 1>Was the precipitating event for things that may have happened anyway.

0:10:26.720 --> 0:10:28.800
<v Speaker 1>So I think that's one thing to consider.

0:10:28.880 --> 0:10:31.280
<v Speaker 4>The second is I actually think this is a good thing.

0:10:32.040 --> 0:10:36.480
<v Speaker 1>It's very important that we not get COMPLACENTECH in our

0:10:36.520 --> 0:10:39.720
<v Speaker 1>technology prowess in the United States. We need to realize

0:10:39.720 --> 0:10:42.920
<v Speaker 1>that these technologies can be created at scale all over

0:10:42.920 --> 0:10:43.360
<v Speaker 1>the world.

0:10:44.000 --> 0:10:45.360
<v Speaker 4>And what Deep.

0:10:45.160 --> 0:10:47.880
<v Speaker 1>Seek basically did was kick off what I believe will

0:10:47.920 --> 0:10:51.880
<v Speaker 1>be an arms race, maybe a spot Nick moment for

0:10:51.960 --> 0:10:55.000
<v Speaker 1>the United States where we get really serious in the

0:10:55.720 --> 0:10:58.760
<v Speaker 1>prospect of competition from around the world will actually propel

0:10:59.240 --> 0:11:02.960
<v Speaker 1>AI developed forward. So I'm actually very positive both on

0:11:03.600 --> 0:11:08.840
<v Speaker 1>the development of AI and actually, and ironically, even though

0:11:08.840 --> 0:11:10.959
<v Speaker 1>the stocks we took a beating today, I actually think

0:11:11.000 --> 0:11:14.920
<v Speaker 1>that it further illustrates the need to build out AI

0:11:15.040 --> 0:11:19.000
<v Speaker 1>infrastructure because what we're seeing now is disruption in basically

0:11:19.080 --> 0:11:21.839
<v Speaker 1>AI one point zero, which will bring about the possibility

0:11:21.880 --> 0:11:25.000
<v Speaker 1>of AI two point zero, which will make it universally

0:11:25.080 --> 0:11:28.280
<v Speaker 1>usable and much more widely adopted than what we see

0:11:28.360 --> 0:11:31.120
<v Speaker 1>right now. So I'm pretty excited about what we see.

0:11:31.280 --> 0:11:33.160
<v Speaker 1>This is just a moment in time in the market,

0:11:33.640 --> 0:11:35.840
<v Speaker 1>and we'll be looking back on this as one of

0:11:35.840 --> 0:11:38.680
<v Speaker 1>the best buying opportunities in these stocks at quite some time.

0:11:38.960 --> 0:11:41.480
<v Speaker 2>So one of the things that I've found very compelling

0:11:41.520 --> 0:11:44.760
<v Speaker 2>here is that deep Seek's chatbot, known as R one

0:11:45.840 --> 0:11:50.400
<v Speaker 2>was apparently trained using less than cutting edge semiconductors, and

0:11:50.480 --> 0:11:55.480
<v Speaker 2>that in turn required much less energy consumption. So I

0:11:55.520 --> 0:11:57.280
<v Speaker 2>hear what you're saying when it comes to the sell

0:11:57.320 --> 0:12:00.960
<v Speaker 2>off that we had in certain semiconductor names today. But

0:12:01.080 --> 0:12:04.640
<v Speaker 2>I was also struck by the fact that Constellation Energy

0:12:04.720 --> 0:12:08.520
<v Speaker 2>was down twenty percent. How does the Deep Seek story

0:12:08.679 --> 0:12:10.640
<v Speaker 2>change the conversation around power?

0:12:11.559 --> 0:12:14.200
<v Speaker 4>So there's two parts of that question. I'll address the

0:12:14.200 --> 0:12:16.319
<v Speaker 4>power one first. There have been.

0:12:17.720 --> 0:12:23.520
<v Speaker 1>Reports of inability for companies to get adequate power sources

0:12:23.559 --> 0:12:29.160
<v Speaker 1>to data centers. So companies like Microsoft Amazon are putting

0:12:29.160 --> 0:12:34.080
<v Speaker 1>in personal nuclear reactors to actually power data centers, and

0:12:34.160 --> 0:12:37.800
<v Speaker 1>companies like Constellation Energy provide those services. So if the

0:12:37.920 --> 0:12:39.880
<v Speaker 1>data centers are not going to be as plentible and

0:12:39.920 --> 0:12:43.520
<v Speaker 1>therefore the power sources will not be needed, that's why

0:12:43.520 --> 0:12:45.199
<v Speaker 1>you see these knock on effects. It's one of the

0:12:45.280 --> 0:12:49.319
<v Speaker 1>reasons why the power story had been one of the

0:12:49.679 --> 0:12:53.320
<v Speaker 1>sort of tertiary ways to play the AI story without

0:12:53.360 --> 0:12:56.120
<v Speaker 1>actually buying the tech stocks. So that's why they're sort

0:12:56.160 --> 0:12:59.600
<v Speaker 1>of connected together in a trade.

0:12:59.880 --> 0:13:03.120
<v Speaker 4>I don't believe that.

0:13:02.320 --> 0:13:05.160
<v Speaker 1>Either story will turn out to be, you know, the

0:13:05.200 --> 0:13:06.760
<v Speaker 1>sell off today will be the right one.

0:13:06.800 --> 0:13:08.600
<v Speaker 4>I think it's actually the wrong choice.

0:13:08.960 --> 0:13:12.480
<v Speaker 1>But to go back to talk about the clever technology

0:13:12.559 --> 0:13:17.360
<v Speaker 1>or the clever engineering that made this possible, where the

0:13:17.600 --> 0:13:21.000
<v Speaker 1>R one, you know, learning module is able to use.

0:13:22.440 --> 0:13:25.200
<v Speaker 4>Just a just a It's a very small change, but

0:13:25.240 --> 0:13:26.520
<v Speaker 4>it's very very impactful.

0:13:26.559 --> 0:13:29.840
<v Speaker 1>So if you think about like disruptive technologies, you have

0:13:30.000 --> 0:13:34.840
<v Speaker 1>incumbent players who basically iterate they have built this fantastic

0:13:35.440 --> 0:13:38.440
<v Speaker 1>black box, and they just iterate small and small pieces,

0:13:38.480 --> 0:13:41.560
<v Speaker 1>Whereas these disruptive players come come around and they actually

0:13:41.600 --> 0:13:44.480
<v Speaker 1>change the entire game. And that's what Deep Seek has done.

0:13:44.679 --> 0:13:47.720
<v Speaker 1>They basically took the AI engine which is ever present

0:13:47.800 --> 0:13:52.720
<v Speaker 1>in open ai, where the technology is ready for any

0:13:52.800 --> 0:13:56.920
<v Speaker 1>question on any topic, whereas in the in the deep

0:13:56.920 --> 0:14:01.240
<v Speaker 1>Seek infrastructure, it would basically take the input and only

0:14:01.320 --> 0:14:04.040
<v Speaker 1>turn on the pieces of the infrastructure that were needed

0:14:04.080 --> 0:14:04.800
<v Speaker 1>to answer.

0:14:04.480 --> 0:14:05.600
<v Speaker 4>The question or to iterate.

0:14:05.840 --> 0:14:09.600
<v Speaker 1>So it basically didn't have the entire system running, you know,

0:14:09.679 --> 0:14:13.440
<v Speaker 1>all the time. And that will be readily adopted by

0:14:13.520 --> 0:14:18.960
<v Speaker 1>all AI providers. For sure, Beta and for sure Open

0:14:19.000 --> 0:14:22.080
<v Speaker 1>AI will adopt that clever engineering and they too will

0:14:22.120 --> 0:14:24.160
<v Speaker 1>have it. And so the reason why I think that

0:14:24.240 --> 0:14:27.479
<v Speaker 1>this is just the beginning is that now that infrastructure

0:14:27.840 --> 0:14:32.280
<v Speaker 1>that was being built to handle an ever present AI

0:14:32.440 --> 0:14:36.280
<v Speaker 1>infrastructure now could be used for other things. And I

0:14:36.280 --> 0:14:39.720
<v Speaker 1>think that's why we're going to see this AI infrastructure

0:14:39.760 --> 0:14:43.160
<v Speaker 1>actually blossom as a result of this disruption, not be

0:14:43.280 --> 0:14:43.880
<v Speaker 1>torn apart.

0:14:44.040 --> 0:14:46.480
<v Speaker 2>It is a critical week for tech earnings. This week.

0:14:46.520 --> 0:14:50.800
<v Speaker 2>Four of the mag seven are reporting. That's Apple, Microsoft, Tesla,

0:14:51.040 --> 0:14:54.160
<v Speaker 2>and Meta Platforms. What are your expectations here and what

0:14:54.200 --> 0:14:57.560
<v Speaker 2>do these companies need to say about the future in

0:14:57.680 --> 0:15:01.360
<v Speaker 2>order for investors to really kind of not lose heart.

0:15:02.120 --> 0:15:04.560
<v Speaker 4>Well, two of the four of them had a very

0:15:04.560 --> 0:15:05.240
<v Speaker 4>good day to day.

0:15:05.280 --> 0:15:07.960
<v Speaker 1>I mean, Apple and Meta really bucked the trend, and

0:15:08.040 --> 0:15:11.760
<v Speaker 1>it was largely because they're consumers of these more expensive,

0:15:12.360 --> 0:15:16.000
<v Speaker 1>you know, chips, and I think if anything that lowers

0:15:16.000 --> 0:15:20.280
<v Speaker 1>the infrastructure costs a cruise directly to players like Meta

0:15:20.560 --> 0:15:24.520
<v Speaker 1>or Apple. Apple's unique story because there have been some

0:15:25.120 --> 0:15:29.720
<v Speaker 1>concerns about the inability to produce or sell the iPhone.

0:15:30.200 --> 0:15:32.960
<v Speaker 1>Maybe sales had declined, maybe people weren't upgrading at the

0:15:33.000 --> 0:15:36.720
<v Speaker 1>same rate. But the story on Apple is less hardware

0:15:36.800 --> 0:15:39.840
<v Speaker 1>these days and definitely more about.

0:15:39.200 --> 0:15:40.680
<v Speaker 4>You know, the ancillary services.

0:15:40.680 --> 0:15:42.400
<v Speaker 1>So I think Apple is going to do quite well

0:15:42.440 --> 0:15:45.440
<v Speaker 1>and I think the stock definitely had a fantastic day

0:15:45.480 --> 0:15:47.640
<v Speaker 1>to day on the prospects of them, you know, having

0:15:47.720 --> 0:15:50.800
<v Speaker 1>higher margins because they're spending less on their data infrastructure

0:15:51.080 --> 0:15:54.360
<v Speaker 1>meta the same thing. I think Microsoft kind of got

0:15:54.400 --> 0:15:57.080
<v Speaker 1>caught up in the AI given its investment in the

0:15:57.160 --> 0:16:01.280
<v Speaker 1>in the open AI platform. But I think that Microsoft

0:16:01.320 --> 0:16:04.800
<v Speaker 1>is actually, of the three, the best position to continue

0:16:04.800 --> 0:16:07.640
<v Speaker 1>to grow earnings year over year, and so I think

0:16:07.680 --> 0:16:09.920
<v Speaker 1>three of the four are doing well. I'm not I

0:16:09.920 --> 0:16:13.680
<v Speaker 1>don't follow Tesla that much, but I feel like Elon

0:16:13.800 --> 0:16:16.360
<v Speaker 1>Musk may be in a better position to understand whether

0:16:16.440 --> 0:16:20.680
<v Speaker 1>or not this deep seek technology is actually you know,

0:16:20.680 --> 0:16:23.320
<v Speaker 1>whether or not the technology being deployed or the things

0:16:23.360 --> 0:16:26.000
<v Speaker 1>that the company has said that they've been able to do,

0:16:26.120 --> 0:16:29.520
<v Speaker 1>or actual reality because Tesla has one of the largest

0:16:29.520 --> 0:16:33.680
<v Speaker 1>AI infrastructures powered by all the automobiles that Tesla makes,

0:16:33.880 --> 0:16:37.440
<v Speaker 1>so their neural net has been around for some time

0:16:37.520 --> 0:16:41.160
<v Speaker 1>and has been generating AI content for generation of software

0:16:41.240 --> 0:16:43.800
<v Speaker 1>with the cars. So I think maybe we might learn

0:16:43.840 --> 0:16:47.240
<v Speaker 1>a thing or two in the conference call that may

0:16:47.360 --> 0:16:50.200
<v Speaker 1>be related to deep seek in some of this AI disruption.

0:16:50.280 --> 0:16:52.720
<v Speaker 1>I think that will be the story for the for

0:16:52.880 --> 0:16:54.320
<v Speaker 1>the Tesla conference call this.

0:16:54.320 --> 0:16:57.160
<v Speaker 2>Year, Jamie. Before I let you go, We've also have

0:16:57.200 --> 0:17:00.920
<v Speaker 2>a FED decision on Wednesday, Powell news conference right after

0:17:00.960 --> 0:17:04.240
<v Speaker 2>we get the decision. Is it likely that we're going

0:17:04.280 --> 0:17:06.600
<v Speaker 2>to get kind of a hawkish message right now?

0:17:07.160 --> 0:17:10.680
<v Speaker 4>No, I think the FED is on hold.

0:17:11.400 --> 0:17:15.160
<v Speaker 1>They may discuss, as they did in the press conference

0:17:15.200 --> 0:17:19.399
<v Speaker 1>and December, the need to be cautious about lowering rates

0:17:19.480 --> 0:17:24.119
<v Speaker 1>too fast. But it's going to be interesting to watch

0:17:24.119 --> 0:17:26.320
<v Speaker 1>over the next couple of months because the President is

0:17:26.359 --> 0:17:28.720
<v Speaker 1>going to put enormous pressure on the FED to do

0:17:28.760 --> 0:17:32.960
<v Speaker 1>something about mortgage rates. That has been a particular concern

0:17:33.280 --> 0:17:36.200
<v Speaker 1>to President Trump, and he is going to start beating

0:17:36.200 --> 0:17:38.919
<v Speaker 1>the drum on that very loudly. So I have a

0:17:38.920 --> 0:17:41.120
<v Speaker 1>feeling we'll start to see the FED start to talk

0:17:41.200 --> 0:17:43.040
<v Speaker 1>about how they're going to deal with the balance sheet

0:17:43.560 --> 0:17:44.800
<v Speaker 1>and how they could.

0:17:44.680 --> 0:17:47.120
<v Speaker 4>Help consumers on the mortgage side.

0:17:47.840 --> 0:17:49.960
<v Speaker 1>So I don't think this FED meeting will be much

0:17:49.960 --> 0:17:52.480
<v Speaker 1>on the FED fund rate in terms of news. I mean,

0:17:52.480 --> 0:17:55.080
<v Speaker 1>the FED fund futures basically are ninety nine percent that

0:17:55.160 --> 0:17:59.840
<v Speaker 1>they hold. What is not built into the model is

0:18:00.040 --> 0:18:02.440
<v Speaker 1>what they talk about on the ballot sheet, and there

0:18:02.480 --> 0:18:06.439
<v Speaker 1>has been runoff at a fairly decent equipped over the

0:18:06.480 --> 0:18:08.480
<v Speaker 1>last year and a half, and I think that may

0:18:08.520 --> 0:18:11.600
<v Speaker 1>be ready to slow down. And if the FED would

0:18:12.080 --> 0:18:15.040
<v Speaker 1>you slow its mortgage backed securities runoff, I think that

0:18:15.119 --> 0:18:17.200
<v Speaker 1>might help mortgage rates some. So I think that's where

0:18:17.200 --> 0:18:19.119
<v Speaker 1>you're going to see some of the action. And it

0:18:19.200 --> 0:18:22.280
<v Speaker 1>would not surprise me at all if the President is out,

0:18:22.440 --> 0:18:25.440
<v Speaker 1>you know, with comments in and around the FED meeting,

0:18:25.440 --> 0:18:28.000
<v Speaker 1>that would you know, really encourage them to be more

0:18:28.160 --> 0:18:30.960
<v Speaker 1>aggressive in helping mortgage rates to.

0:18:30.920 --> 0:18:33.719
<v Speaker 2>Con Jamie will leave it there. Thank you so much

0:18:33.760 --> 0:18:36.720
<v Speaker 2>for being with us. Jimmy Cox there, managing partner at

0:18:36.760 --> 0:18:40.280
<v Speaker 2>Harris Financial Group, joining from Richmond, Virginia. Here on the

0:18:40.359 --> 0:18:46.000
<v Speaker 2>Daybreak Asia Podcast. Thanks for listening to today's episode of

0:18:46.040 --> 0:18:50.160
<v Speaker 2>the Bloomberg Daybreak Asia Edition podcast. Each weekday, we look

0:18:50.160 --> 0:18:53.960
<v Speaker 2>at the story shaping markets, finance, and geopolitics in the

0:18:54.000 --> 0:18:57.240
<v Speaker 2>Asia Pacific. You can find us on Apple, Spotify, the

0:18:57.280 --> 0:19:01.280
<v Speaker 2>Bloomberg Podcast YouTube channel, or anywhere else you listen. Join

0:19:01.400 --> 0:19:04.399
<v Speaker 2>us again tomorrow for insight on the market moves from

0:19:04.400 --> 0:19:08.879
<v Speaker 2>Hong Kong to Singapore and Australia. I'm Doug Prisoner and

0:19:09.000 --> 0:19:10.160
<v Speaker 2>this is Bloomberg