WEBVTT - Nvidia Snaps Rout and Oracle Warns of TikTok Impact

0:00:01.400 --> 0:00:05.080
<v Speaker 1>From Mahard that we're innovation, money and power. Collie in

0:00:05.200 --> 0:00:10.120
<v Speaker 1>Silicon Valley, NBN. This is Bloomberg Technology with Caroline Hide

0:00:10.160 --> 0:00:11.559
<v Speaker 1>and Ed Lovelove.

0:00:25.480 --> 0:00:28.560
<v Speaker 2>A Meed Lovelow in San Francisco. Caroline hides off. This

0:00:28.600 --> 0:00:31.520
<v Speaker 2>is Bloomberg Technology coming up in video snaps. It's three

0:00:31.600 --> 0:00:34.440
<v Speaker 2>day route that wiped four hundred and thirty billion dollars

0:00:34.640 --> 0:00:38.519
<v Speaker 2>off its market cap. Full analysis ahead, plus Oracle warning

0:00:38.720 --> 0:00:42.479
<v Speaker 2>of a financial impact to its business should TikTok be

0:00:42.560 --> 0:00:45.279
<v Speaker 2>banned in the US? Details to come and we go

0:00:45.360 --> 0:00:49.479
<v Speaker 2>live to our Bloomberg invest Summit, uniting leaders in asset management, banking,

0:00:49.760 --> 0:00:51.599
<v Speaker 2>wealth and private markets.

0:00:51.960 --> 0:00:54.000
<v Speaker 3>Some key conversations to come.

0:00:54.120 --> 0:00:56.279
<v Speaker 2>Let's get to financial markets for the only thing that

0:00:56.400 --> 0:00:59.600
<v Speaker 2>everyone is talking about is in video. We have rebounded

0:00:59.640 --> 0:01:02.800
<v Speaker 2>up aroun three percent in the Tuesday session, but it

0:01:02.840 --> 0:01:05.880
<v Speaker 2>follows three straight days of declines which total more than

0:01:05.920 --> 0:01:10.119
<v Speaker 2>ten percent, so therefore a technical correction and also four

0:01:10.240 --> 0:01:13.240
<v Speaker 2>hundred and thirty billion dollars of market cap shed. There's

0:01:13.240 --> 0:01:17.240
<v Speaker 2>a lot of conversation narrative around valuation but also the

0:01:17.280 --> 0:01:20.760
<v Speaker 2>predictability of sales. How much money in video is going

0:01:20.800 --> 0:01:23.600
<v Speaker 2>to make? It's been really difficult, whether you're a sell

0:01:23.680 --> 0:01:28.000
<v Speaker 2>side analyst or even management to correctly say how much

0:01:28.000 --> 0:01:30.720
<v Speaker 2>revenue is going to come in, particularly with new generations

0:01:30.720 --> 0:01:34.160
<v Speaker 2>of AI accelerator coming online. So from a market cap perspective,

0:01:34.520 --> 0:01:36.600
<v Speaker 2>this is what we're supposed to look at in Video

0:01:36.680 --> 0:01:39.480
<v Speaker 2>down that's the same chart market cap. The point is

0:01:39.760 --> 0:01:43.000
<v Speaker 2>it's below Apple, it's below Microsoft. So in that time period,

0:01:43.480 --> 0:01:45.280
<v Speaker 2>in Video has gone from being the most valuable company

0:01:45.319 --> 0:01:47.480
<v Speaker 2>in the world down a few pegs. Jomping me into

0:01:47.480 --> 0:01:48.760
<v Speaker 2>my Bloomberg terminal and look.

0:01:48.600 --> 0:01:49.120
<v Speaker 3>At this chart.

0:01:49.160 --> 0:01:52.120
<v Speaker 2>Because of all the anxiety around the selloff, there is

0:01:52.200 --> 0:01:56.280
<v Speaker 2>also some I guess sanguine thinking. From a technical perspective,

0:01:56.520 --> 0:01:59.400
<v Speaker 2>it's a stock that continues to trade pretty handsomely above

0:01:59.520 --> 0:02:02.400
<v Speaker 2>it's fifty and one hundred day moving average. We're trading

0:02:02.400 --> 0:02:04.600
<v Speaker 2>at one hundred and twenty two dollars a share, and

0:02:04.640 --> 0:02:07.000
<v Speaker 2>people see one hundred and fifteen dollars a share for

0:02:07.040 --> 0:02:09.720
<v Speaker 2>those of you who follow the technicals as a key

0:02:09.760 --> 0:02:12.720
<v Speaker 2>level of resistance. Back from a short break, and I'm

0:02:12.720 --> 0:02:15.880
<v Speaker 2>delighted to be reunited on set with Bloomberg's Ian King.

0:02:16.240 --> 0:02:18.360
<v Speaker 2>I want to go back to that idea of how

0:02:18.360 --> 0:02:21.519
<v Speaker 2>difficult it is to forecast sales because over the weekend,

0:02:21.560 --> 0:02:25.040
<v Speaker 2>that's what we're all talking about. In part, it's because

0:02:25.520 --> 0:02:28.440
<v Speaker 2>since this all kicked off in twenty twenty three, every

0:02:28.480 --> 0:02:31.360
<v Speaker 2>time they book revenue, it always comes in pretty handsomely

0:02:31.400 --> 0:02:34.800
<v Speaker 2>above the midpoint of the guidance that Jensen gives, and

0:02:34.880 --> 0:02:37.440
<v Speaker 2>it's difficult for the cell size to say, well, what

0:02:37.520 --> 0:02:38.160
<v Speaker 2>happens next?

0:02:38.400 --> 0:02:39.839
<v Speaker 1>Yeah, no, you're absolutely right.

0:02:39.919 --> 0:02:44.240
<v Speaker 4>They are effectively over shooting expectations at every point, and

0:02:44.280 --> 0:02:46.400
<v Speaker 4>there are a number of factors that play into that,

0:02:46.440 --> 0:02:50.080
<v Speaker 4>including supply, including this kind of lumpiness of demand because

0:02:50.080 --> 0:02:53.359
<v Speaker 4>they only have really a small number of large customers.

0:02:53.520 --> 0:02:55.760
<v Speaker 1>So all of those things make it very difficult. But

0:02:55.840 --> 0:02:59.120
<v Speaker 1>at a certain point, you know, things will even out.

0:02:59.160 --> 0:03:02.800
<v Speaker 4>Supply will catch with demand, these big customers will decide

0:03:02.840 --> 0:03:06.079
<v Speaker 4>they've got enough infrastructure for the time being, and then

0:03:06.120 --> 0:03:08.839
<v Speaker 4>the story changes. Then other factors come in, and it's

0:03:08.880 --> 0:03:10.840
<v Speaker 4>when do we reach that transition point?

0:03:11.000 --> 0:03:13.640
<v Speaker 2>As you heard Ian over in New York City Bloomberg Investor,

0:03:13.760 --> 0:03:15.800
<v Speaker 2>it's underway. Some big names have been attendance and what

0:03:15.880 --> 0:03:17.560
<v Speaker 2>is it do you think they've been talking about?

0:03:17.639 --> 0:03:19.120
<v Speaker 3>It's in Vidia. Listen to this real quick.

0:03:20.600 --> 0:03:22.280
<v Speaker 5>I mean, if you look at the chart on in video,

0:03:22.320 --> 0:03:25.240
<v Speaker 5>you can barely see the correction, so I mean, you know,

0:03:25.320 --> 0:03:28.000
<v Speaker 5>on a percentage basis, it's down something like ten percent,

0:03:28.120 --> 0:03:30.520
<v Speaker 5>but you have to like peer really really carefully at

0:03:30.520 --> 0:03:34.839
<v Speaker 5>the chart to see that move. So I don't think

0:03:34.840 --> 0:03:36.760
<v Speaker 5>it means anything the people were on that stock, nor

0:03:36.840 --> 0:03:40.240
<v Speaker 5>what's expensive. But they're buying into a story, and as

0:03:40.280 --> 0:03:42.800
<v Speaker 5>long as the story is intact, like in video is

0:03:42.800 --> 0:03:46.800
<v Speaker 5>obviously intact, the story is going to continue.

0:03:47.400 --> 0:03:51.200
<v Speaker 2>Steve Wiseman of The Big Short Fame on Bloomberg's Avliance

0:03:51.200 --> 0:03:51.720
<v Speaker 2>earlier today.

0:03:51.720 --> 0:03:52.840
<v Speaker 3>He talked about the story.

0:03:53.240 --> 0:03:56.000
<v Speaker 2>The thing that I think investors are finding difficult is

0:03:56.480 --> 0:03:59.480
<v Speaker 2>the story of new products coming online. So H two

0:03:59.600 --> 0:04:03.680
<v Speaker 2>hundred is the kind of middle ground generation AI accelerator

0:04:03.720 --> 0:04:06.640
<v Speaker 2>that's ramping right now. But Jensen told us May twenty

0:04:06.640 --> 0:04:10.480
<v Speaker 2>second at earnings that Blackwell, which is the next generation accelerator,

0:04:10.520 --> 0:04:14.080
<v Speaker 2>will also ship this year, and so working out where

0:04:14.080 --> 0:04:15.840
<v Speaker 2>they're going and how much money they'll make from them

0:04:15.880 --> 0:04:16.440
<v Speaker 2>is really tough.

0:04:16.839 --> 0:04:19.240
<v Speaker 4>It is, I mean, but that's really something that the

0:04:19.240 --> 0:04:19.960
<v Speaker 4>guys who run the.

0:04:20.080 --> 0:04:22.320
<v Speaker 1>Xpl you should be doing a good job with.

0:04:23.000 --> 0:04:25.320
<v Speaker 4>What I think is more difficult, and what causes more

0:04:25.360 --> 0:04:28.520
<v Speaker 4>concern is all of this infrastructure that's being put into place,

0:04:28.760 --> 0:04:31.479
<v Speaker 4>is it paying off, is it making money or is

0:04:31.520 --> 0:04:33.800
<v Speaker 4>it just being basically thrown at the wall to see

0:04:33.800 --> 0:04:36.240
<v Speaker 4>what will stick. And I think that's where if there

0:04:36.279 --> 0:04:40.440
<v Speaker 4>is a fundamental concern, we want to see AI translate

0:04:40.600 --> 0:04:43.960
<v Speaker 4>into real revenue and real profit at some of these

0:04:44.000 --> 0:04:45.720
<v Speaker 4>guys who are paying all of this money for all

0:04:45.720 --> 0:04:46.600
<v Speaker 4>of this in video.

0:04:46.400 --> 0:04:48.599
<v Speaker 2>Gear bloom Bezi and King who leads us and making

0:04:48.640 --> 0:04:51.280
<v Speaker 2>ducts a coverage at Bloomberg but for quite a long time.

0:04:51.400 --> 0:04:54.240
<v Speaker 2>Also like let our stocks team, we will continue talking

0:04:54.240 --> 0:04:56.880
<v Speaker 2>about in video. Bloomberg invest is now underway in the

0:04:56.880 --> 0:04:59.599
<v Speaker 2>heart of New York's financial district, where leaders in asset

0:04:59.680 --> 0:05:03.200
<v Speaker 2>managed and banking well from private markets are connecting with

0:05:03.320 --> 0:05:07.600
<v Speaker 2>world class journalists led by Bloomboa's Global Finance correspondent Shnali

0:05:07.680 --> 0:05:11.960
<v Speaker 2>Bassek to track the future's greatest risks and opportunities now.

0:05:11.960 --> 0:05:16.400
<v Speaker 2>Moments ago, Felipe Lafont of COTU was asked by David Rubinstein,

0:05:16.480 --> 0:05:19.760
<v Speaker 2>who would be the winner in AI, existing big tech

0:05:19.800 --> 0:05:22.160
<v Speaker 2>companies or smaller startups listen to.

0:05:22.080 --> 0:05:27.360
<v Speaker 6>This at the beginning of these new waves, the new

0:05:27.400 --> 0:05:30.280
<v Speaker 6>winners ten or twenty years out. Some of them are

0:05:30.400 --> 0:05:33.760
<v Speaker 6>existing companies and some of them are new. But the

0:05:33.839 --> 0:05:36.719
<v Speaker 6>new ones that get big, there's maybe one a year.

0:05:37.120 --> 0:05:37.840
<v Speaker 6>No more.

0:05:40.120 --> 0:05:43.239
<v Speaker 2>Delighted to say, as if by magic Bloomberg Shnale Bassek

0:05:43.320 --> 0:05:44.320
<v Speaker 2>joins us now with more.

0:05:44.360 --> 0:05:45.640
<v Speaker 3>This is a really big event.

0:05:45.760 --> 0:05:50.040
<v Speaker 2>And it's not surprising that whether you are in the

0:05:50.080 --> 0:05:53.920
<v Speaker 2>private equity or private growth equity area of the market,

0:05:54.080 --> 0:05:56.120
<v Speaker 2>or you're a hedge fund or whatever it is you are,

0:05:56.520 --> 0:05:59.160
<v Speaker 2>everyone seems to be talking about artificial intelligence.

0:06:01.600 --> 0:06:04.080
<v Speaker 7>It certainly does pervade a lot of the conversations we're

0:06:04.120 --> 0:06:06.440
<v Speaker 7>having here today. And of course Felipe Leafont is one

0:06:06.440 --> 0:06:11.120
<v Speaker 7>of the perfect people to talk about artificial intelligence too. Earlier,

0:06:11.160 --> 0:06:13.640
<v Speaker 7>as he was talking to David Rubinstein, David Rubinstein did

0:06:13.720 --> 0:06:15.799
<v Speaker 7>ask him if he thinks there's a lot.

0:06:15.600 --> 0:06:17.640
<v Speaker 3>Of hype around AI.

0:06:17.600 --> 0:06:21.080
<v Speaker 7>And he really leaned into the opportunities that are happening

0:06:21.160 --> 0:06:25.320
<v Speaker 7>in AI and really downplayed other types of opportunities, say

0:06:25.440 --> 0:06:28.240
<v Speaker 7>quantum computing, for example. So what is he seeing in

0:06:28.279 --> 0:06:32.000
<v Speaker 7>the AI opportunity? Some things got really wonky here, this

0:06:32.200 --> 0:06:37.480
<v Speaker 7>idea of artificial brains being implanted in a robot that

0:06:37.640 --> 0:06:42.480
<v Speaker 7>lives alongside humans, though that is probably about fifteen years away.

0:06:42.520 --> 0:06:46.200
<v Speaker 7>According to Kotoo's Lafonte, he's also seeing a lot of

0:06:46.320 --> 0:06:49.800
<v Speaker 7>opportunity here in power. That is something that a lot

0:06:49.839 --> 0:06:52.559
<v Speaker 7>of Wall Street is also watching, the banks the hedge

0:06:52.560 --> 0:06:55.720
<v Speaker 7>funds as well, and so certainly a more near term

0:06:55.800 --> 0:06:58.680
<v Speaker 7>opportunity a little earlier. Also on the AI front, I

0:06:58.720 --> 0:07:01.240
<v Speaker 7>would just say we also talked to Bollie of Eldridge

0:07:01.240 --> 0:07:04.880
<v Speaker 7>Industries and he talked about how it's changing media and entertainment.

0:07:05.120 --> 0:07:07.560
<v Speaker 7>The reason I bring this all up is because you

0:07:07.680 --> 0:07:10.800
<v Speaker 7>hear different types of investors ed really talking about the

0:07:10.800 --> 0:07:15.760
<v Speaker 7>AI opportunity, particularly in private markets as opposed to public markets,

0:07:15.960 --> 0:07:18.920
<v Speaker 7>and the growth opportunity you're seeing there, as well as

0:07:18.960 --> 0:07:23.040
<v Speaker 7>the governance structures that exist around companies like that that

0:07:23.080 --> 0:07:25.200
<v Speaker 7>would allow for that kind of experimentation.

0:07:25.920 --> 0:07:28.720
<v Speaker 2>As regular views of this show, No, I'm a longstanding

0:07:28.800 --> 0:07:32.160
<v Speaker 2>Chelsea season ticket holder, a member of Chelsea Football Club,

0:07:32.240 --> 0:07:35.120
<v Speaker 2>so I go watch the top early interview in full online.

0:07:35.200 --> 0:07:35.880
<v Speaker 8>What's to come?

0:07:35.920 --> 0:07:36.920
<v Speaker 3>You've put so much.

0:07:36.760 --> 0:07:39.440
<v Speaker 2>Work into Aditionale along with our colleagues from the Live

0:07:39.480 --> 0:07:41.840
<v Speaker 2>Events team. There are some big names to come in

0:07:41.880 --> 0:07:42.680
<v Speaker 2>the next few hours.

0:07:45.120 --> 0:07:47.600
<v Speaker 7>There certainly are shortly, we'll be talking to boaz Y

0:07:47.600 --> 0:07:49.520
<v Speaker 7>and Stein of SABA Capital, And I've got to say

0:07:49.520 --> 0:07:52.000
<v Speaker 7>one reason that's interesting, Ed is because we are just

0:07:52.200 --> 0:07:55.440
<v Speaker 7>in the center of his dispute with Blackrock. We know

0:07:55.560 --> 0:07:57.680
<v Speaker 7>that there are a couple of funds that he's targeting

0:07:57.720 --> 0:08:00.520
<v Speaker 7>as activists and closed end funds run by black Rock.

0:08:00.960 --> 0:08:03.560
<v Speaker 7>They have proxy voting that's going on in the last

0:08:03.600 --> 0:08:05.360
<v Speaker 7>couple of weeks. In the next couple of weeks, so

0:08:05.400 --> 0:08:07.640
<v Speaker 7>we don't really know yet if Saba is winning. We'll

0:08:07.680 --> 0:08:09.960
<v Speaker 7>find out a little more what he thinks on stage

0:08:10.000 --> 0:08:11.440
<v Speaker 7>in the afternoon. We're going to have a couple of

0:08:11.480 --> 0:08:15.080
<v Speaker 7>private credit conversations as well, John Zido, I'm looking forward

0:08:15.080 --> 0:08:17.240
<v Speaker 7>to Gary Gensler, the chairman.

0:08:16.960 --> 0:08:19.200
<v Speaker 3>Of the SEC. Of course, with Ann Marie Hordern.

0:08:19.440 --> 0:08:21.960
<v Speaker 7>We know the presidential debate is just days away, and

0:08:22.080 --> 0:08:24.800
<v Speaker 7>hearing what Gensler has to say about what a change

0:08:24.800 --> 0:08:27.920
<v Speaker 7>of the administration could mean if there is a change

0:08:27.920 --> 0:08:31.840
<v Speaker 7>in administration for the policies that he's set forth today.

0:08:32.200 --> 0:08:34.480
<v Speaker 7>Of course, a lot of people in tech and in

0:08:34.520 --> 0:08:37.959
<v Speaker 7>general markets looking at the potential for an etheream ETF.

0:08:38.120 --> 0:08:41.320
<v Speaker 7>There's some real potential for news later this afternoon.

0:08:41.720 --> 0:08:45.319
<v Speaker 2>Bluemokes Shinali Bassic. Just an incredible lineup. Stay tuned. The

0:08:45.360 --> 0:08:48.040
<v Speaker 2>Bloomberg invest Summit is taking place all of today and

0:08:48.120 --> 0:08:51.320
<v Speaker 2>tomorrow right there in New York City. Don't forget to

0:08:51.400 --> 0:08:52.760
<v Speaker 2>tune in, check it out.

0:08:53.440 --> 0:08:55.720
<v Speaker 3>We'll be right back. This has been bug technology.

0:09:11.559 --> 0:09:14.920
<v Speaker 2>China's President's using ping is urging the nation to step

0:09:15.000 --> 0:09:19.520
<v Speaker 2>up innovation because other countries dominate certain key technologies.

0:09:19.840 --> 0:09:20.559
<v Speaker 3>While he said.

0:09:20.440 --> 0:09:24.080
<v Speaker 2>China had quote made great progress in science, he added

0:09:24.120 --> 0:09:28.120
<v Speaker 2>that original innovation ability is still relatively weak. Some key

0:09:28.200 --> 0:09:31.160
<v Speaker 2>core technologies are controlled by others, and there is a

0:09:31.200 --> 0:09:36.600
<v Speaker 2>shortage of top scientific and technological talents sticking with China.

0:09:36.679 --> 0:09:40.599
<v Speaker 2>After initial reporting in May from Bloomberg that Huawei was

0:09:40.679 --> 0:09:45.840
<v Speaker 2>secretly sponsoring a research competition run by Optica, a review

0:09:46.200 --> 0:09:50.200
<v Speaker 2>of internal Optica corporate records shows the alliance ran far

0:09:50.360 --> 0:09:55.400
<v Speaker 2>deeper than publicly known, blossoming over decades even as US

0:09:55.520 --> 0:09:59.920
<v Speaker 2>China tensions over technology intensified. Let's bring in Bloomberg's Mike Chef,

0:10:00.520 --> 0:10:04.199
<v Speaker 2>who leads our coverage at the intersection of politics and technology.

0:10:04.600 --> 0:10:08.760
<v Speaker 2>You know, the original Bloomberg report in May was astonishing, frankly,

0:10:09.400 --> 0:10:12.720
<v Speaker 2>and the further details in this sort of a polisical environment,

0:10:13.000 --> 0:10:16.520
<v Speaker 2>and the headlines of the last twelve months are almost unbelievable.

0:10:16.559 --> 0:10:17.240
<v Speaker 3>What have we learned?

0:10:18.320 --> 0:10:18.520
<v Speaker 9>Well?

0:10:18.559 --> 0:10:21.520
<v Speaker 10>At first, let me commend you on the segue into

0:10:21.559 --> 0:10:25.520
<v Speaker 10>this segment, because it really highlights the importance of this story.

0:10:25.920 --> 0:10:28.839
<v Speaker 10>It shows that China is really on the quest. It's

0:10:28.880 --> 0:10:32.599
<v Speaker 10>on the hunt for the technological advances. It needs to

0:10:32.880 --> 0:10:37.400
<v Speaker 10>bolster its economy and gain that technological edge and key areas,

0:10:37.720 --> 0:10:41.040
<v Speaker 10>and this shows one way it's been doing it. Huawei,

0:10:41.080 --> 0:10:45.040
<v Speaker 10>which is such an industrial champion in China, was not

0:10:45.080 --> 0:10:49.480
<v Speaker 10>only sponsoring secretly this competition, but it was forging a

0:10:49.600 --> 0:10:52.760
<v Speaker 10>much deeper relationship with Optica, and it was a two

0:10:52.760 --> 0:10:57.319
<v Speaker 10>way street. Optica is a century old organization that whose

0:10:57.360 --> 0:11:01.480
<v Speaker 10>members twenty four thousand members worldwide do research and sensitive

0:11:01.520 --> 0:11:04.800
<v Speaker 10>areas in light that apply to things like semiconductors that

0:11:04.840 --> 0:11:08.240
<v Speaker 10>are essential to everything in the modern economy. So the

0:11:08.280 --> 0:11:14.199
<v Speaker 10>consequences of this are significant. And Keith O'Keefe, whose initial

0:11:14.240 --> 0:11:16.840
<v Speaker 10>report really broke the ground on this and her latest

0:11:16.880 --> 0:11:20.679
<v Speaker 10>reporting advanced the story further. She found that there was

0:11:20.720 --> 0:11:24.440
<v Speaker 10>a whistleblower who flagged the risks that some of the

0:11:24.600 --> 0:11:28.560
<v Speaker 10>work being done by US scientists could be exported to

0:11:28.640 --> 0:11:29.760
<v Speaker 10>China as a result of this.

0:11:30.880 --> 0:11:32.560
<v Speaker 2>I don't want to simplify it too much, but the

0:11:32.559 --> 0:11:35.960
<v Speaker 2>basic idea is that that money came from Huawei. And

0:11:36.440 --> 0:11:39.560
<v Speaker 2>in response to this reporting by Kato o'keef, who's had

0:11:39.559 --> 0:11:44.160
<v Speaker 2>a great impacts ince joining Bloomberg, Obtica's spokesperson basically says

0:11:44.280 --> 0:11:48.960
<v Speaker 2>that that group Obtica's confident that they acted properly and

0:11:49.040 --> 0:11:52.280
<v Speaker 2>in good faith. Were showing the response of Liz Rogan,

0:11:52.280 --> 0:11:55.559
<v Speaker 2>who's the OCTICA CEO on the screen, Mic, I.

0:11:55.480 --> 0:11:57.080
<v Speaker 3>Guess what happens next.

0:11:57.160 --> 0:12:00.400
<v Speaker 2>You know, when stories like this develop, sometimes law makers

0:12:00.440 --> 0:12:03.040
<v Speaker 2>take notice, sometimes regulators take notice.

0:12:03.559 --> 0:12:07.760
<v Speaker 10>Well, lawmakers have taken notice, and the top Republican and

0:12:07.880 --> 0:12:11.120
<v Speaker 10>Democrat on how Science Committee are looking into this. They

0:12:11.120 --> 0:12:15.719
<v Speaker 10>expressed grave concern that US scientific research could be exposed

0:12:15.800 --> 0:12:19.400
<v Speaker 10>or even exported unwittingly to China as a result of

0:12:19.440 --> 0:12:22.880
<v Speaker 10>a relationship like this. Now the organization is saying that

0:12:22.920 --> 0:12:26.000
<v Speaker 10>it acted in good faith and to be fair, it

0:12:26.040 --> 0:12:30.160
<v Speaker 10>did not violate any rules per se in allowing this

0:12:30.280 --> 0:12:34.000
<v Speaker 10>fund this competition to be funded anonymously, and yet the

0:12:34.040 --> 0:12:38.000
<v Speaker 10>appearance was complicated for one, for the organization, but it

0:12:38.040 --> 0:12:41.000
<v Speaker 10>was also a risk for the scientists who accepted the

0:12:41.080 --> 0:12:44.200
<v Speaker 10>money without knowing that it came from Huawei. You see,

0:12:44.240 --> 0:12:48.200
<v Speaker 10>many researchers in the US have to disclose to other

0:12:48.360 --> 0:12:52.600
<v Speaker 10>government and university funding agencies where their other sources of

0:12:52.720 --> 0:12:56.679
<v Speaker 10>money are coming from. And if they don't tell those

0:12:56.880 --> 0:13:00.480
<v Speaker 10>other funders, if they don't disclose it, hey, some money

0:13:00.480 --> 0:13:03.760
<v Speaker 10>from Wawei by the way, there could create a problem

0:13:03.880 --> 0:13:05.640
<v Speaker 10>for them down the road as well.

0:13:06.559 --> 0:13:10.720
<v Speaker 2>Blnberg's Mike Shephard, really appreciate your insight and reporting here

0:13:10.720 --> 0:13:11.720
<v Speaker 2>on Boombog Technology.

0:13:11.760 --> 0:13:12.120
<v Speaker 3>Thank you.

0:13:12.520 --> 0:13:15.400
<v Speaker 2>We continue to talk about China. China based Byte Dance

0:13:15.520 --> 0:13:17.960
<v Speaker 2>is facing a ban in the US in the form

0:13:18.000 --> 0:13:22.400
<v Speaker 2>of TikTok, and Oracle is now warning that that would

0:13:22.440 --> 0:13:26.000
<v Speaker 2>adversely affect its revenues and profits. This is a story

0:13:26.000 --> 0:13:28.559
<v Speaker 2>that caught my eye pretty late yesterday. I want to

0:13:28.559 --> 0:13:31.440
<v Speaker 2>bring in Bloomberg Intelligence and this Satura Irana and you've

0:13:31.640 --> 0:13:35.320
<v Speaker 2>basically constantly crunched the numbers on what the impact would be.

0:13:36.040 --> 0:13:40.199
<v Speaker 2>But it's not straightforward, you know, for our audience's benefit.

0:13:40.840 --> 0:13:43.640
<v Speaker 3>Oracle was at the heart of TikTok's.

0:13:43.120 --> 0:13:49.760
<v Speaker 2>Compromise by on shoring US user data in Oracle servers. Basically, Now,

0:13:49.760 --> 0:13:52.439
<v Speaker 2>if that's banned, what's the point of the Oracle relationship.

0:13:52.520 --> 0:13:56.920
<v Speaker 11>Yeah, it's hosting the IQ application. So their infrastructure has

0:13:56.960 --> 0:13:59.280
<v Speaker 11>been used to do host that, so which is why

0:13:59.400 --> 0:14:01.280
<v Speaker 11>if it goes they're not going to be getting that

0:14:01.360 --> 0:14:04.040
<v Speaker 11>much revenue out Now they've never disclosed what the exact

0:14:04.080 --> 0:14:07.160
<v Speaker 11>revenue amount is. But you know, I think one thing

0:14:07.800 --> 0:14:10.200
<v Speaker 11>that's happening good on the article side is there's a

0:14:10.200 --> 0:14:13.680
<v Speaker 11>massive demand for their infrastructure as a service business. They are,

0:14:13.760 --> 0:14:17.439
<v Speaker 11>you know, helping out Microsoft there, They're helping out or

0:14:17.520 --> 0:14:19.200
<v Speaker 11>to Google and the I mean, they just really deal

0:14:19.240 --> 0:14:21.840
<v Speaker 11>with them. So that business is growing over forty percent.

0:14:22.040 --> 0:14:24.040
<v Speaker 11>It's running at a rundrate of about two billion a

0:14:24.120 --> 0:14:27.120
<v Speaker 11>quarter right now, so annualized rundrate of about eight billion.

0:14:27.200 --> 0:14:29.480
<v Speaker 11>So I mean, you know, it will be a setback

0:14:29.520 --> 0:14:31.360
<v Speaker 11>I would say for a quarter or two, but they

0:14:31.360 --> 0:14:34.520
<v Speaker 11>can make up that sales with other demand that's out there.

0:14:34.560 --> 0:14:37.160
<v Speaker 2>Stucks down a percentage point in the session in the moment,

0:14:37.360 --> 0:14:39.400
<v Speaker 2>up more than thirty percent year today. But what you

0:14:39.440 --> 0:14:41.440
<v Speaker 2>said is interesting, Right, we don't have a number, we

0:14:41.480 --> 0:14:44.360
<v Speaker 2>don't know what the value of the relationship is, but

0:14:44.440 --> 0:14:47.640
<v Speaker 2>You've covered this industry long enough, right you know that

0:14:47.720 --> 0:14:51.920
<v Speaker 2>if a company issues a statement saying this is substantive

0:14:52.280 --> 0:14:55.640
<v Speaker 2>material is the word, then it's worth looking into.

0:14:56.520 --> 0:14:57.320
<v Speaker 1>No, no fair amount.

0:14:57.400 --> 0:14:59.440
<v Speaker 11>I completely agree with you, But what I'm trying to

0:14:59.480 --> 0:15:02.000
<v Speaker 11>say is the demand for these kinds of work is

0:15:02.080 --> 0:15:04.560
<v Speaker 11>so strong right now. So maybe it's a setback for

0:15:04.640 --> 0:15:07.200
<v Speaker 11>Oracle for six months or nine months, but they will

0:15:07.200 --> 0:15:09.600
<v Speaker 11>make up that revenue from other sources. It's not as

0:15:09.600 --> 0:15:13.080
<v Speaker 11>if it's lost forever that particular revenue stream, you know.

0:15:13.120 --> 0:15:15.240
<v Speaker 11>In the in the defense of Oracle, right now, they

0:15:15.240 --> 0:15:20.160
<v Speaker 11>are really becoming the fourth largest cloud provider out there,

0:15:20.400 --> 0:15:25.640
<v Speaker 11>you know, right behind Amazon, Microsoft and Google Cloud Platform,

0:15:25.960 --> 0:15:28.360
<v Speaker 11>because this kind of work is really the biggest thing

0:15:28.360 --> 0:15:32.000
<v Speaker 11>in demand right now. Application software is struggling, but infrastructure

0:15:32.040 --> 0:15:35.440
<v Speaker 11>software is doing well because of cloud resources.

0:15:35.440 --> 0:15:37.440
<v Speaker 2>And that was the focus of the research out this

0:15:37.520 --> 0:15:40.680
<v Speaker 2>morning from you and Blue mg Intelligence, narrowing the gap

0:15:40.720 --> 0:15:43.480
<v Speaker 2>when it comes to cloud Ana rag Rana, Bloomberg Intelligence,

0:15:43.520 --> 0:15:44.200
<v Speaker 2>Thank you so much.

0:15:44.360 --> 0:15:44.640
<v Speaker 3>Okay.

0:15:44.680 --> 0:15:49.200
<v Speaker 2>Staying with China, It's spacecraft carrying samples collected on the

0:15:49.240 --> 0:15:52.720
<v Speaker 2>far side of the Moon return to Earth today, marking

0:15:52.880 --> 0:15:57.560
<v Speaker 2>an important milestone in China's path for further lunar exploration.

0:15:58.040 --> 0:16:01.200
<v Speaker 3>The change is six space landed.

0:16:00.880 --> 0:16:03.760
<v Speaker 2>In a grasslands region of Inner Mongolia, and was the

0:16:03.800 --> 0:16:06.840
<v Speaker 2>second mission by China to the more distant part of

0:16:06.840 --> 0:16:10.720
<v Speaker 2>the Moon, following the twenty nineteen landing by Chang E four.

0:16:11.120 --> 0:16:14.120
<v Speaker 2>No other country has sent a spacecraft to the far side,

0:16:14.120 --> 0:16:17.840
<v Speaker 2>although NASA has plans to support missions there by a

0:16:17.920 --> 0:16:28.760
<v Speaker 2>US spacecraft. Okay, it's signed for Talking Tech and first up.

0:16:28.760 --> 0:16:32.200
<v Speaker 2>In yesterday's show, we discussed Wall Street Journal report that

0:16:32.280 --> 0:16:36.160
<v Speaker 2>Meta and Apple had held talks on an AI partnership.

0:16:36.200 --> 0:16:40.680
<v Speaker 2>After the show, Bloomberg's Mark German reported or clarified the

0:16:40.720 --> 0:16:45.800
<v Speaker 2>real situation. Apple dismissed Meta's AI chatbot months ago. According

0:16:45.880 --> 0:16:49.760
<v Speaker 2>to sources, the iPhone maker held brief discussions about using

0:16:49.800 --> 0:16:53.480
<v Speaker 2>Meta's Lama chatbot, but ultimately decided not to move forward,

0:16:53.760 --> 0:16:58.520
<v Speaker 2>citing privacy practices that weren't stringent enough. Plus Open ai

0:16:58.600 --> 0:17:01.800
<v Speaker 2>issues are warning to China. According to local media reports,

0:17:02.040 --> 0:17:05.480
<v Speaker 2>open ai says it will begin blocking access to tools

0:17:05.480 --> 0:17:07.680
<v Speaker 2>and software in the region beginning in July.

0:17:08.240 --> 0:17:10.240
<v Speaker 3>It's unclear what prompted.

0:17:09.800 --> 0:17:12.919
<v Speaker 2>The move, and while open ai is unofficially available in

0:17:12.960 --> 0:17:17.000
<v Speaker 2>the country, many can access its features through VPN and

0:17:17.160 --> 0:17:20.359
<v Speaker 2>US defense startup Anderil has secured one point five billion

0:17:20.400 --> 0:17:24.600
<v Speaker 2>dollars of new capital, boosting its valuation to about fourteen

0:17:24.680 --> 0:17:27.600
<v Speaker 2>billion dollars. That's according to a report by The Information.

0:17:27.720 --> 0:17:31.639
<v Speaker 2>The startup, backed by Peter Teel, makes drones and autonomous

0:17:31.640 --> 0:17:35.000
<v Speaker 2>weapons systems, and recently one a major contract to develop

0:17:35.160 --> 0:17:39.520
<v Speaker 2>unmanned fighter jets for the US Air Force. Another story retracking,

0:17:39.560 --> 0:17:43.000
<v Speaker 2>the health platform Function reached fifty three million dollars in

0:17:43.040 --> 0:17:46.600
<v Speaker 2>its Series A funding round. The platform focuses on preventative

0:17:46.600 --> 0:17:51.119
<v Speaker 2>approaches to help individuals manage their well being. Delighted to

0:17:51.160 --> 0:17:55.200
<v Speaker 2>say that Function's co founder Mark Heyman joins us. Now

0:17:55.400 --> 0:17:58.879
<v Speaker 2>you know it's a significant summer money. Congratulations Mark on

0:17:59.160 --> 0:18:02.000
<v Speaker 2>the Series A. But actually, what caught my eye in

0:18:02.040 --> 0:18:06.440
<v Speaker 2>the first instance is how many users you've onboarded. Yeah,

0:18:06.480 --> 0:18:09.600
<v Speaker 2>and how big the weight list is relative to those

0:18:09.680 --> 0:18:10.040
<v Speaker 2>on board.

0:18:10.119 --> 0:18:11.199
<v Speaker 3>It just explain that to me.

0:18:11.440 --> 0:18:12.560
<v Speaker 1>Yeah, Well, thank you for having me.

0:18:12.800 --> 0:18:16.640
<v Speaker 12>This is an incredible moment in medical history because we're

0:18:16.640 --> 0:18:20.640
<v Speaker 12>seeing consumer driven healthcare exploding and we hit a nerve,

0:18:20.720 --> 0:18:23.080
<v Speaker 12>a pain point where people want access to their own data,

0:18:23.359 --> 0:18:25.080
<v Speaker 12>their own biology. They want to look under the hood

0:18:25.080 --> 0:18:27.520
<v Speaker 12>and not just rely on the healthcare system to provide

0:18:27.520 --> 0:18:29.600
<v Speaker 12>access to their own health information. And so we've gotten

0:18:29.600 --> 0:18:33.040
<v Speaker 12>fifty thousand people as members in the first year in

0:18:33.160 --> 0:18:35.920
<v Speaker 12>our data and two hundred thousand people on the waitlist,

0:18:35.960 --> 0:18:38.040
<v Speaker 12>and there's just a hunger for people to know about

0:18:38.040 --> 0:18:40.679
<v Speaker 12>their health, to be empowered with the tools and have

0:18:40.760 --> 0:18:44.360
<v Speaker 12>their own agency over their health, using their data from

0:18:44.359 --> 0:18:46.960
<v Speaker 12>their own body and having ownership of it, which right

0:18:47.000 --> 0:18:49.520
<v Speaker 12>now is sort of held within the healthcare system in

0:18:49.560 --> 0:18:51.720
<v Speaker 12>a very tight way that's hard to access for people,

0:18:51.720 --> 0:18:54.800
<v Speaker 12>hard to understand, and we make it easy accessible, and

0:18:54.840 --> 0:18:57.560
<v Speaker 12>it's the first personalized health platform that allows you to

0:18:57.600 --> 0:18:59.800
<v Speaker 12>access to your own health data through initial one hundred

0:18:59.800 --> 0:19:01.920
<v Speaker 12>and ten and bound markers for four hundred and ninety

0:19:01.960 --> 0:19:04.840
<v Speaker 12>nine dollars. It's about fifteen thousand dollars worth. Lab tests

0:19:05.040 --> 0:19:08.240
<v Speaker 12>and provides deep insights from world scientific literature from top

0:19:08.280 --> 0:19:11.679
<v Speaker 12>knowledge experts, and provides guidance on what likely things that

0:19:11.680 --> 0:19:13.120
<v Speaker 12>are going on with you and what to do about

0:19:13.119 --> 0:19:15.360
<v Speaker 12>them in order to empower you to change your behaviors

0:19:15.400 --> 0:19:17.040
<v Speaker 12>and your lifestyle, which is really where more of these

0:19:17.080 --> 0:19:17.439
<v Speaker 12>right is.

0:19:18.960 --> 0:19:21.720
<v Speaker 2>Mark just explain the basics of the technology you know,

0:19:22.520 --> 0:19:25.639
<v Speaker 2>every time we hear about a data driven platform like this,

0:19:26.320 --> 0:19:28.720
<v Speaker 2>we immediately start talking about our official intelligence. But I

0:19:28.720 --> 0:19:32.640
<v Speaker 2>think in this case function has problem that the expression

0:19:32.640 --> 0:19:34.960
<v Speaker 2>a more simple function just to explain the basics.

0:19:35.000 --> 0:19:38.679
<v Speaker 12>Absolutely, no, you know, it's hard to believe, but medicine

0:19:38.760 --> 0:19:41.240
<v Speaker 12>is still in the dark ages. It's very analog. We

0:19:41.600 --> 0:19:43.800
<v Speaker 12>don't have pen and paper. We have electronic medical records,

0:19:43.800 --> 0:19:46.720
<v Speaker 12>but essentially pen and paper put in a PDF for

0:19:46.840 --> 0:19:50.560
<v Speaker 12>electronic medical record with no relational content, with no understanding

0:19:50.560 --> 0:19:54.480
<v Speaker 12>of the interactions between the data points, with no correlation.

0:19:54.640 --> 0:19:57.960
<v Speaker 12>And so it's really a unique moment in history where

0:19:58.000 --> 0:20:00.880
<v Speaker 12>we can gather your own data from the blood from

0:20:00.920 --> 0:20:04.840
<v Speaker 12>eventually biom bios sensors and wearables, from your electronic records

0:20:05.240 --> 0:20:07.960
<v Speaker 12>from oh mix and into your own health platform allows

0:20:07.960 --> 0:20:09.959
<v Speaker 12>you to understand what's going on with your body. So

0:20:10.080 --> 0:20:12.639
<v Speaker 12>it's a very simple process to sign up. It's five minutes.

0:20:12.840 --> 0:20:15.200
<v Speaker 12>You go to functionhealth dot com. You can sign up

0:20:15.760 --> 0:20:18.680
<v Speaker 12>and you get a scheduled appointment for your lab test

0:20:18.680 --> 0:20:21.360
<v Speaker 12>at one of the two thousand Quest laboratories. Your blood's drawn.

0:20:21.400 --> 0:20:22.480
<v Speaker 1>It takes about fifteen minutes.

0:20:22.520 --> 0:20:24.520
<v Speaker 12>It's frictionless, you don't have to go through insurance, your

0:20:24.520 --> 0:20:27.280
<v Speaker 12>doctor gets away from the patronalistic healthcare system. It allows

0:20:27.280 --> 0:20:29.760
<v Speaker 12>people to be agents of their own health and the

0:20:29.800 --> 0:20:31.199
<v Speaker 12>seeo your own health, which I think is what we

0:20:31.200 --> 0:20:33.840
<v Speaker 12>all need to be given the epidemic of chronic disease

0:20:34.200 --> 0:20:34.800
<v Speaker 12>that's now based.

0:20:34.800 --> 0:20:37.160
<v Speaker 2>Okay, just want to jump in on that insurance point.

0:20:37.200 --> 0:20:39.240
<v Speaker 2>As you can tell you, I come from the United Kingdom,

0:20:39.240 --> 0:20:42.560
<v Speaker 2>where your healthcaret is very different. Six years in, I'm

0:20:42.640 --> 0:20:45.760
<v Speaker 2>used to the insurance basis here. How does your business

0:20:45.800 --> 0:20:48.520
<v Speaker 2>model work within the confines of that system.

0:20:48.560 --> 0:20:50.720
<v Speaker 12>Well, it's very interesting because people are going outside the

0:20:50.760 --> 0:20:54.680
<v Speaker 12>system to actually learn more about their health, to get

0:20:54.760 --> 0:20:56.840
<v Speaker 12>data about their health, whether it's home testing or whether

0:20:56.880 --> 0:20:59.920
<v Speaker 12>it's wearables. Now this is successible to your blood test

0:21:00.080 --> 0:21:01.760
<v Speaker 12>and you don't have to go through insurance. It's four

0:21:01.800 --> 0:21:03.600
<v Speaker 12>hundred nine nine dollars a year. It's a dollar thirty

0:21:03.640 --> 0:21:05.800
<v Speaker 12>seven cents a day, which is less than your dunkin

0:21:05.880 --> 0:21:08.359
<v Speaker 12>Donuts a cup of coffee, and it allows you to

0:21:08.400 --> 0:21:11.560
<v Speaker 12>have deep insights about your health across the spectrum of

0:21:11.560 --> 0:21:13.960
<v Speaker 12>your health, from nutritions to hormone to metabolic health, to

0:21:14.040 --> 0:21:19.440
<v Speaker 12>toxins to cardiovascer risk to all your biomarkers that relate

0:21:19.480 --> 0:21:22.040
<v Speaker 12>to your overall well being and something that was never

0:21:22.040 --> 0:21:23.200
<v Speaker 12>accessible to people before.

0:21:23.320 --> 0:21:24.680
<v Speaker 1>And so when you go through your.

0:21:24.640 --> 0:21:27.920
<v Speaker 12>Doctor insurance, there's a big resistance, a firewall where they say, no,

0:21:27.960 --> 0:21:29.760
<v Speaker 12>you don't need these tests, you don't need that test,

0:21:29.840 --> 0:21:31.879
<v Speaker 12>or they just want these small number of tests. But

0:21:31.880 --> 0:21:33.440
<v Speaker 12>it's better to look at things early on in the

0:21:33.480 --> 0:21:36.159
<v Speaker 12>trajectory of disease. And we've seen just by the simple approach,

0:21:36.200 --> 0:21:39.600
<v Speaker 12>we've saved lives. One man was in his forties, forty

0:21:39.600 --> 0:21:41.720
<v Speaker 12>four years old did the panel, found out he had

0:21:41.760 --> 0:21:44.160
<v Speaker 12>prostate cancer, was able to treat it early, get cured,

0:21:44.200 --> 0:21:46.399
<v Speaker 12>and was able to then live on a healthy life

0:21:46.400 --> 0:21:48.640
<v Speaker 12>with his young kids as opposed to being at.

0:21:48.640 --> 0:21:49.160
<v Speaker 3>Risk for death.

0:21:49.160 --> 0:21:51.800
<v Speaker 12>So I think we are seeing incredible uptake. We're seeing

0:21:51.840 --> 0:21:56.040
<v Speaker 12>incredible results from the users and incredible stories of lives changed.

0:21:56.240 --> 0:21:58.000
<v Speaker 8>And really this is all about ending sight.

0:21:58.840 --> 0:22:01.439
<v Speaker 2>Mak Himan Kai, founder Health, thank you for joining us

0:22:01.480 --> 0:22:03.480
<v Speaker 2>on the Programer're coming up. We're going to be joined

0:22:03.520 --> 0:22:06.240
<v Speaker 2>by the CEO of x Scaviniberti on how the startup

0:22:06.280 --> 0:22:08.520
<v Speaker 2>plans to take on in Vidia.

0:22:08.560 --> 0:22:09.879
<v Speaker 3>That's coming up next, a.

0:22:09.880 --> 0:22:12.960
<v Speaker 2>Quick look at shares of Microsoft the European Commission has

0:22:13.040 --> 0:22:17.720
<v Speaker 2>issued it a statement of objections alleging that Microsoft's teams

0:22:17.920 --> 0:22:21.600
<v Speaker 2>an office three sixty five fel flou of the EU's

0:22:21.680 --> 0:22:24.800
<v Speaker 2>competition rules. The stock up four ten percent. A story

0:22:24.800 --> 0:22:38.159
<v Speaker 2>we'll dig into this is Bloomberg Technology. Welcome back to

0:22:38.160 --> 0:22:41.399
<v Speaker 2>Bloomberg Technology at Ludlow in San Francisco. Let's take a

0:22:41.480 --> 0:22:45.080
<v Speaker 2>quick look at the markets. Broadly speaking, US equities are

0:22:45.119 --> 0:22:47.520
<v Speaker 2>higher then as that one hundred, up a four percentage point.

0:22:47.560 --> 0:22:49.600
<v Speaker 2>A big part of that story, of course, is in

0:22:49.720 --> 0:22:53.159
<v Speaker 2>Vidia now up five percent in this session. Is the

0:22:53.280 --> 0:22:57.440
<v Speaker 2>rebound from that technical correction three straight days of declines

0:22:57.920 --> 0:23:02.120
<v Speaker 2>exceeding ten percent until now, and there's no real news

0:23:02.119 --> 0:23:06.280
<v Speaker 2>flow around in video. This is about some serious gains.

0:23:06.320 --> 0:23:08.320
<v Speaker 2>It was about a forty three percent jump from May

0:23:08.359 --> 0:23:11.520
<v Speaker 2>twenty second when in Video reported earnings. And the conversation

0:23:11.600 --> 0:23:15.080
<v Speaker 2>around how hard it is to predict sales growth because

0:23:15.200 --> 0:23:17.399
<v Speaker 2>you're jumping from H one hundred to H two hundred.

0:23:17.640 --> 0:23:19.640
<v Speaker 2>Blackwell will start shipping at the end of the year,

0:23:19.640 --> 0:23:21.800
<v Speaker 2>and the street is basically saying we can't keep up

0:23:21.800 --> 0:23:24.520
<v Speaker 2>with a management's forecast. What's going on. I also take

0:23:24.560 --> 0:23:27.159
<v Speaker 2>a quick look at bitcoin trades twenty four to seven.

0:23:27.760 --> 0:23:30.080
<v Speaker 2>We focused on it big time in yesterday's show, but

0:23:30.119 --> 0:23:34.080
<v Speaker 2>we're back nearer to sixty two thousand dollars per token,

0:23:34.560 --> 0:23:38.119
<v Speaker 2>noting that there have been pretty substantive outflows from a

0:23:38.320 --> 0:23:42.040
<v Speaker 2>dedicated bitcoin spot. Ets I want to linger on the

0:23:42.080 --> 0:23:45.640
<v Speaker 2>AI accelerator market for a moment. Etched is a young

0:23:45.720 --> 0:23:48.679
<v Speaker 2>AI chip company making a big bet on AI. They

0:23:48.720 --> 0:23:51.960
<v Speaker 2>spent the past two years building soho the world's first

0:23:52.000 --> 0:23:55.639
<v Speaker 2>specialized AI chip for transformers, or so they claim, and

0:23:55.680 --> 0:23:58.480
<v Speaker 2>that bet is paying off. The company just nabbed a

0:23:58.520 --> 0:24:01.720
<v Speaker 2>one hundred and twenty million dollar Series A round led Briy,

0:24:01.800 --> 0:24:05.480
<v Speaker 2>Primary Venture Partners and Positive some Ventures. Let's bring in

0:24:05.520 --> 0:24:07.919
<v Speaker 2>Gavin Aberti, one of the company's two co founders and

0:24:07.960 --> 0:24:08.760
<v Speaker 2>its CEO.

0:24:09.400 --> 0:24:11.399
<v Speaker 3>Okay, let's go back to basics on this. There were

0:24:11.400 --> 0:24:11.720
<v Speaker 3>a lot of.

0:24:11.720 --> 0:24:16.440
<v Speaker 2>Big claims in that introduction. Your main point is that

0:24:16.960 --> 0:24:21.200
<v Speaker 2>the specifically designed chip for the development to transform transformers

0:24:21.240 --> 0:24:24.240
<v Speaker 2>and models is better than the GPU. But what is

0:24:24.280 --> 0:24:26.679
<v Speaker 2>it that you're building? What is so well so?

0:24:26.720 --> 0:24:29.520
<v Speaker 13>Who's a kind of aioship that is only able to

0:24:29.560 --> 0:24:34.560
<v Speaker 13>do one very specific thing. An accelerator from Nvidio or

0:24:34.600 --> 0:24:38.280
<v Speaker 13>Google is flexible. It can be programmed to run many

0:24:38.280 --> 0:24:43.840
<v Speaker 13>different kinds of AI models like convolutional networks, LSTMs or transformers.

0:24:43.920 --> 0:24:45.560
<v Speaker 8>Right, our chips are different.

0:24:46.320 --> 0:24:49.320
<v Speaker 13>They are only able to run this one very neurow

0:24:49.320 --> 0:24:53.160
<v Speaker 13>class of model, which we call transformers. But when running

0:24:53.200 --> 0:24:57.080
<v Speaker 13>these kinds of models, they are vastly more performant, more

0:24:57.119 --> 0:24:59.960
<v Speaker 13>than twenty times higher throughput than eight one hundred.

0:25:00.080 --> 0:25:02.960
<v Speaker 2>So what is it from an intellectual property standpoint or

0:25:02.960 --> 0:25:07.040
<v Speaker 2>a technology standpoint unique to H that allows for that?

0:25:07.080 --> 0:25:09.760
<v Speaker 2>What is the secret source behind? So that makes it

0:25:09.800 --> 0:25:13.359
<v Speaker 2>any different from a performance perspective or a design perspective

0:25:13.400 --> 0:25:15.320
<v Speaker 2>relative to H one hundred eight two hundred.

0:25:15.800 --> 0:25:18.960
<v Speaker 13>Well, we're able to burn the transformer algorithm into the

0:25:19.040 --> 0:25:22.480
<v Speaker 13>chip we're building, our own silicon and our own servers,

0:25:22.920 --> 0:25:25.359
<v Speaker 13>and this enables us to get more than twenty times

0:25:25.400 --> 0:25:28.880
<v Speaker 13>higher throughput in terms of output words per second from

0:25:28.960 --> 0:25:33.440
<v Speaker 13>models like check GPT or LAMA. But you're also able

0:25:33.480 --> 0:25:36.600
<v Speaker 13>to get much better latency numbers. If you measure the

0:25:36.760 --> 0:25:39.479
<v Speaker 13>time to the first token, we can cut that by

0:25:39.520 --> 0:25:40.920
<v Speaker 13>a factor of twenty as well.

0:25:41.960 --> 0:25:46.119
<v Speaker 2>You told my colleague Piles Vary that you think H

0:25:46.200 --> 0:25:50.359
<v Speaker 2>can be the world's biggest company. That's quite a claim

0:25:50.440 --> 0:25:54.840
<v Speaker 2>to make outline the pathway to that For.

0:25:54.920 --> 0:25:57.119
<v Speaker 8>Me, well, we are taking a bet.

0:25:57.840 --> 0:26:00.920
<v Speaker 13>Unlike every other company in the market, our chips can

0:26:01.000 --> 0:26:06.440
<v Speaker 13>only run transformers. So if transformers change dramatically or go away,

0:26:07.000 --> 0:26:10.600
<v Speaker 13>then we'll be in a bad place. But if we're

0:26:10.680 --> 0:26:14.160
<v Speaker 13>right and transformers keep being the dominant way that AI

0:26:14.240 --> 0:26:17.280
<v Speaker 13>models work, will be the most performance chip on the

0:26:17.320 --> 0:26:19.480
<v Speaker 13>market by an order of magnitude.

0:26:19.600 --> 0:26:24.080
<v Speaker 2>So let's talk about who your target customer is because transformers,

0:26:24.160 --> 0:26:27.440
<v Speaker 2>or let's let's use another term, large language model in

0:26:27.480 --> 0:26:28.840
<v Speaker 2>the natural language context.

0:26:28.520 --> 0:26:29.639
<v Speaker 8>For us a misconception.

0:26:29.760 --> 0:26:31.919
<v Speaker 3>Actually, we'll fix the misconception.

0:26:32.600 --> 0:26:37.000
<v Speaker 13>Transformers used to be for just large language models. Other

0:26:37.080 --> 0:26:40.920
<v Speaker 13>kinds of models like image generators and video generators used

0:26:40.960 --> 0:26:44.919
<v Speaker 13>to be what's called units, but that's changed over the

0:26:45.000 --> 0:26:48.640
<v Speaker 13>past year. It's turned out that transformers work better than

0:26:48.680 --> 0:26:52.080
<v Speaker 13>other kinds of AI models for generating images and videos too.

0:26:52.760 --> 0:26:54.200
<v Speaker 8>I don't know if you saw Sora.

0:26:54.000 --> 0:26:57.399
<v Speaker 13>From open AIS, but that's a transformer as well, powered

0:26:57.440 --> 0:27:00.879
<v Speaker 13>by the same underlying kind of models check GPT.

0:27:01.080 --> 0:27:03.560
<v Speaker 2>But the point I'd like to get to it with

0:27:03.720 --> 0:27:07.040
<v Speaker 2>regards to who your customers or potential customers would be

0:27:07.160 --> 0:27:13.200
<v Speaker 2>is that some transformers have parameters in the hundreds of billions,

0:27:13.880 --> 0:27:17.160
<v Speaker 2>SOA may be a good example of that. What's happening

0:27:17.240 --> 0:27:22.520
<v Speaker 2>now in AI development is more modest ambitions, lots of

0:27:22.560 --> 0:27:27.520
<v Speaker 2>companies working on models with scale a much lower scale

0:27:27.760 --> 0:27:28.879
<v Speaker 2>in terms of parameters.

0:27:29.200 --> 0:27:31.880
<v Speaker 3>Where do you want to go that we.

0:27:31.880 --> 0:27:34.879
<v Speaker 8>Are selling to the big transformer models.

0:27:35.640 --> 0:27:37.480
<v Speaker 13>We have tens of millions and pre orders from the

0:27:37.480 --> 0:27:41.080
<v Speaker 13>customers you'd expect, but I expect that we are going

0:27:41.119 --> 0:27:45.080
<v Speaker 13>to get much much bigger AI models. The way we've

0:27:45.119 --> 0:27:48.760
<v Speaker 13>gone from GPT two and twenty nineteen to GPT four

0:27:48.840 --> 0:27:53.439
<v Speaker 13>today is not by inventing new algorithms. The transformer algorithm

0:27:53.480 --> 0:27:57.440
<v Speaker 13>is almost the same. Instead, we have made the models

0:27:57.680 --> 0:28:02.280
<v Speaker 13>far bigger or one point five billion parameters like GPT

0:28:02.320 --> 0:28:03.760
<v Speaker 13>two at one point in.

0:28:03.720 --> 0:28:06.240
<v Speaker 3>The tens or one hundreds of billions or trillions.

0:28:06.080 --> 0:28:07.640
<v Speaker 8>One point trullion for GPT four.

0:28:07.800 --> 0:28:09.640
<v Speaker 2>Let me ask you this though one hundred and twenty

0:28:09.680 --> 0:28:13.680
<v Speaker 2>million dollars series A, it seems like an eye watering

0:28:13.720 --> 0:28:17.320
<v Speaker 2>sum of money, but we've learned that this is a

0:28:17.359 --> 0:28:19.800
<v Speaker 2>capital intensive enterprise from an R and D perspective of

0:28:19.880 --> 0:28:23.040
<v Speaker 2>talent perspective, One thing that was of interest to me

0:28:23.320 --> 0:28:25.320
<v Speaker 2>and wanting used to come on the show is that

0:28:25.800 --> 0:28:30.920
<v Speaker 2>your contract manufacturing partner is TSMC, and not only TSMC,

0:28:31.040 --> 0:28:34.680
<v Speaker 2>but their emerging business group where they basically say, okay,

0:28:34.880 --> 0:28:36.919
<v Speaker 2>you're just starting out, but we can help you to

0:28:36.920 --> 0:28:37.920
<v Speaker 2>scale this product.

0:28:38.160 --> 0:28:39.560
<v Speaker 3>What's that experience been like?

0:28:40.320 --> 0:28:43.160
<v Speaker 13>TSMC is ay a pleasure to work with working on

0:28:43.280 --> 0:28:46.240
<v Speaker 13>their foreign nimeter node. One of their state of the

0:28:46.360 --> 0:28:48.680
<v Speaker 13>art processes, same one that in the video is making

0:28:48.680 --> 0:28:52.200
<v Speaker 13>their black Well GPS on. They were building a eoretical

0:28:52.240 --> 0:28:55.040
<v Speaker 13>sized diet, so a very big chip that is a

0:28:55.200 --> 0:28:58.480
<v Speaker 13>complicated kind of chip to build well.

0:28:58.360 --> 0:29:00.840
<v Speaker 2>On Blackwell's case, they had to literally split it in two,

0:29:01.160 --> 0:29:01.640
<v Speaker 2>which that's a.

0:29:01.560 --> 0:29:03.240
<v Speaker 8>Great point that for Blackwell.

0:29:03.640 --> 0:29:06.800
<v Speaker 13>Normally, what happens in chips is as you move from

0:29:06.960 --> 0:29:09.880
<v Speaker 13>generation to generation, you move to a more advanced kind

0:29:09.920 --> 0:29:12.800
<v Speaker 13>of chip technology. So from the A one hundred to

0:29:12.800 --> 0:29:15.760
<v Speaker 13>the H one hundred, they moved from seven nimeter to

0:29:16.200 --> 0:29:16.960
<v Speaker 13>five an animeter.

0:29:17.880 --> 0:29:18.959
<v Speaker 8>But that's changed.

0:29:19.600 --> 0:29:22.440
<v Speaker 13>They couldn't move to the new technology for the Blackwell

0:29:22.560 --> 0:29:24.959
<v Speaker 13>chips and as a result they had to put two

0:29:25.000 --> 0:29:25.600
<v Speaker 13>of them together.

0:29:26.240 --> 0:29:27.240
<v Speaker 8>And if you look at the.

0:29:27.160 --> 0:29:30.280
<v Speaker 13>Progress of GPUs over the past four or five years,

0:29:30.800 --> 0:29:34.800
<v Speaker 13>they haven't gotten better, They've just gotten bigger performance per

0:29:34.840 --> 0:29:36.000
<v Speaker 13>area stayed constant.

0:29:36.160 --> 0:29:40.760
<v Speaker 2>Yeah, I mean you are an incredibly confident found a

0:29:40.840 --> 0:29:42.800
<v Speaker 2>co founder CEO. It's a pleasure to have you on

0:29:42.880 --> 0:29:48.360
<v Speaker 2>the program, and you're clearly ambitious. The difference between nvideor

0:29:48.680 --> 0:29:52.600
<v Speaker 2>Inexsynvidior is producing at volume. You have one hundred and

0:29:52.640 --> 0:29:54.640
<v Speaker 2>twenty million dollars in the bank. Now, what do you

0:29:54.720 --> 0:29:56.360
<v Speaker 2>need to use that cash?

0:29:56.440 --> 0:29:56.800
<v Speaker 3>Four?

0:29:57.360 --> 0:30:02.160
<v Speaker 2>And how capital intensive is the process of scaling this product?

0:30:02.600 --> 0:30:06.280
<v Speaker 2>Even if you have a partner like TSMC, this is

0:30:06.320 --> 0:30:07.760
<v Speaker 2>a capital intensive business.

0:30:08.520 --> 0:30:10.720
<v Speaker 13>A majority of our Series A is going to go

0:30:10.840 --> 0:30:13.840
<v Speaker 13>into building these products and scaling.

0:30:13.480 --> 0:30:16.960
<v Speaker 8>Up the business. That is where the money goes.

0:30:17.720 --> 0:30:20.080
<v Speaker 13>And I think the big difference too is again we

0:30:20.120 --> 0:30:23.680
<v Speaker 13>are very specialized that we are inherently taking a bet

0:30:23.720 --> 0:30:26.720
<v Speaker 13>on the transformer architecture. A lot of folks in the

0:30:26.720 --> 0:30:30.360
<v Speaker 13>industry will say that, especially transformers were invented, there could

0:30:30.360 --> 0:30:32.800
<v Speaker 13>be something else on the horizon that will put us

0:30:32.840 --> 0:30:35.240
<v Speaker 13>out of business, and we are gambling.

0:30:34.840 --> 0:30:37.120
<v Speaker 2>That's not just very quickly, we just have ten seconds.

0:30:37.200 --> 0:30:39.360
<v Speaker 2>Have you shown Jensen your product?

0:30:40.720 --> 0:30:42.600
<v Speaker 8>I imagaine Jensen's Road of Us by now.

0:30:43.200 --> 0:30:46.239
<v Speaker 2>Gavin ABERTI co founder and CEO of etched. A lot

0:30:46.320 --> 0:30:49.120
<v Speaker 2>of fighting talk, a lot of confidence. Thank you for

0:30:49.200 --> 0:30:59.040
<v Speaker 2>joining us in the program. Okay, back to the story

0:30:59.080 --> 0:31:02.360
<v Speaker 2>we were discussing, open ai has worn developers in China

0:31:02.840 --> 0:31:05.240
<v Speaker 2>it will begin blocking their access to its sools and

0:31:05.280 --> 0:31:08.800
<v Speaker 2>software from July. That's according to local media reports suggesting

0:31:09.040 --> 0:31:12.520
<v Speaker 2>the chat GPT creators taking a more active stance to

0:31:12.640 --> 0:31:16.160
<v Speaker 2>bar users from nations where it doesn't offer services. Bloombased,

0:31:16.240 --> 0:31:19.800
<v Speaker 2>Jackiedavlos leading the AIB joins me at a DC. This

0:31:19.840 --> 0:31:22.720
<v Speaker 2>is really interesting because it's kind of an old school

0:31:22.800 --> 0:31:26.040
<v Speaker 2>tech thing that a lot of the access to open

0:31:26.080 --> 0:31:29.600
<v Speaker 2>ai in China was through VPN, you know, which is

0:31:29.800 --> 0:31:31.800
<v Speaker 2>what people do to stream things in different places.

0:31:31.840 --> 0:31:33.840
<v Speaker 3>But give me more details in this story.

0:31:34.880 --> 0:31:38.200
<v Speaker 9>Well, what we know ed is that Chinese developers will

0:31:38.200 --> 0:31:41.360
<v Speaker 9>have this access cut off starting in July. As you mentioned,

0:31:41.800 --> 0:31:46.760
<v Speaker 9>this really originated through local media reporting, which cited screenshots

0:31:46.800 --> 0:31:49.719
<v Speaker 9>that had this open Ai memo to developers saying they

0:31:49.760 --> 0:31:52.800
<v Speaker 9>were going to have this access removed. Open Ai has

0:31:52.840 --> 0:31:56.280
<v Speaker 9>now confirmed the decision to Bloomberg News earlier this morning

0:31:56.280 --> 0:32:00.160
<v Speaker 9>that it is in fact taking quote additional steps to

0:32:00.200 --> 0:32:04.160
<v Speaker 9>block API traffic from regions where we do not support

0:32:04.200 --> 0:32:07.160
<v Speaker 9>access to open AI's services. Now, this is not the

0:32:07.200 --> 0:32:10.920
<v Speaker 9>first time that China has come against the crosshairs of

0:32:12.040 --> 0:32:15.880
<v Speaker 9>American tech companies and even regulators here and US officials

0:32:15.880 --> 0:32:19.600
<v Speaker 9>who believe that China is using this technology for more

0:32:19.640 --> 0:32:24.200
<v Speaker 9>nefarious purposes. Now open ai really started ratching it up

0:32:24.240 --> 0:32:28.800
<v Speaker 9>its own actions against who it believes is bad actors

0:32:28.840 --> 0:32:34.160
<v Speaker 9>and using its technology, perhaps to manipulate public opinion. For example,

0:32:34.200 --> 0:32:38.360
<v Speaker 9>remember that report in May ed that said Russia, China,

0:32:38.440 --> 0:32:41.960
<v Speaker 9>Iran and Israel we're using open ai to manipulate public opinion.

0:32:42.080 --> 0:32:44.680
<v Speaker 9>So you really see these actions really starting to take

0:32:44.720 --> 0:32:46.520
<v Speaker 9>shape here.

0:32:46.400 --> 0:32:48.840
<v Speaker 2>There's two sides of the story, right, there's the use

0:32:48.920 --> 0:32:53.800
<v Speaker 2>of the technology by malicious actors or on authorized actors

0:32:53.840 --> 0:32:56.320
<v Speaker 2>and also i think a data access point. But there's

0:32:56.360 --> 0:32:59.680
<v Speaker 2>also a competition side of the story. Right, you have

0:32:59.720 --> 0:33:03.160
<v Speaker 2>to sick champions in China like bay Do that are

0:33:03.200 --> 0:33:08.040
<v Speaker 2>building large language models, big large language models to support

0:33:08.080 --> 0:33:11.680
<v Speaker 2>generative AI technology, give me that side of the story.

0:33:12.720 --> 0:33:15.719
<v Speaker 9>Well, they were quick to pounce on the fact that

0:33:15.880 --> 0:33:19.840
<v Speaker 9>now there's this wide swath of developers who need AI

0:33:19.960 --> 0:33:22.680
<v Speaker 9>technology to turn to, and so what some of these

0:33:22.720 --> 0:33:26.080
<v Speaker 9>local reports were saying is that they started appealing to

0:33:26.480 --> 0:33:29.720
<v Speaker 9>developers and saying, hey, come switch to our tech. Really

0:33:29.760 --> 0:33:32.360
<v Speaker 9>goes to show that there is this not just this

0:33:32.440 --> 0:33:36.800
<v Speaker 9>technology battle, but there's a need to really start ramping

0:33:36.880 --> 0:33:41.000
<v Speaker 9>up its own internal AI capabilities. This is exactly what

0:33:41.280 --> 0:33:45.280
<v Speaker 9>lawmakers here in Washington need are concerned about that AI

0:33:45.400 --> 0:33:48.080
<v Speaker 9>is really ramping up their efforts not just to have

0:33:48.160 --> 0:33:52.520
<v Speaker 9>AI play a role in it's economic capabilities, but military

0:33:53.040 --> 0:33:55.000
<v Speaker 9>and so of course, this isn't just you know, about

0:33:55.000 --> 0:33:57.040
<v Speaker 9>what consumers have access to, but how it can be

0:33:57.200 --> 0:34:00.560
<v Speaker 9>used against kind of national security. And this is exactly

0:34:00.560 --> 0:34:04.120
<v Speaker 9>where we're seeing here in Washington actions on behalf of

0:34:04.200 --> 0:34:09.759
<v Speaker 9>for example, the Treasury Department imposing rules to curb investment

0:34:09.840 --> 0:34:12.880
<v Speaker 9>that would support AI investment in some of these areas,

0:34:13.160 --> 0:34:16.200
<v Speaker 9>and also, of course, you know curbs on chip exports.

0:34:16.239 --> 0:34:18.839
<v Speaker 9>We're really starting to see both of these things really

0:34:18.840 --> 0:34:19.480
<v Speaker 9>come together.

0:34:20.760 --> 0:34:23.640
<v Speaker 2>I think that what they're doing from a technical standpoint,

0:34:23.680 --> 0:34:26.399
<v Speaker 2>I'm just reiterating what an open ai spokesperson told us

0:34:26.440 --> 0:34:29.960
<v Speaker 2>is that they are taking additional sets to block API traffic.

0:34:30.200 --> 0:34:32.920
<v Speaker 2>I still find it astonishing that the origin of this

0:34:32.960 --> 0:34:35.640
<v Speaker 2>story is people being able to access open AI from

0:34:35.719 --> 0:34:38.160
<v Speaker 2>China through VPN. Just give me a little bit more

0:34:38.160 --> 0:34:41.160
<v Speaker 2>on what you just said about the DC side of this.

0:34:41.320 --> 0:34:42.320
<v Speaker 3>You know Joe Biden.

0:34:42.560 --> 0:34:48.160
<v Speaker 2>President Biden has some policy towards restricting Chinese products entering

0:34:48.280 --> 0:34:51.080
<v Speaker 2>the United States or tariffs, and the Treasury.

0:34:50.680 --> 0:34:52.120
<v Speaker 3>Department you mentioned as well.

0:34:52.440 --> 0:34:55.400
<v Speaker 2>Are we expecting a sort of expansion of that activity.

0:34:56.520 --> 0:34:58.000
<v Speaker 3>That's a key question. I think.

0:34:58.080 --> 0:35:01.799
<v Speaker 9>If there's one area where lawmakers here in Washington have

0:35:02.000 --> 0:35:06.160
<v Speaker 9>really come together, it's around the national security argument that

0:35:06.320 --> 0:35:10.240
<v Speaker 9>China has capabilities in artificial intelligence and it will continue

0:35:10.239 --> 0:35:13.520
<v Speaker 9>to bolster them. Now you saw this with TikTok. Really

0:35:13.640 --> 0:35:18.399
<v Speaker 9>the concern that Chinese influence was coming through somehow in

0:35:18.800 --> 0:35:22.480
<v Speaker 9>TikTok and influencing public opinion here through that tech. But

0:35:22.560 --> 0:35:25.279
<v Speaker 9>then you look at artificial intelligence and what can be

0:35:25.360 --> 0:35:28.879
<v Speaker 9>done with drones for example, really things on the battlefield.

0:35:28.920 --> 0:35:31.280
<v Speaker 9>This is where a lot of that concern is coming from.

0:35:31.840 --> 0:35:34.399
<v Speaker 9>And so of course you're seeing chip manufacturing is kind

0:35:34.400 --> 0:35:37.960
<v Speaker 9>of the hallmark example for the Biden administration really starting

0:35:37.960 --> 0:35:40.319
<v Speaker 9>to take the lead here. It's all part of this

0:35:40.400 --> 0:35:45.880
<v Speaker 9>broader strategy to really undercut Beijing's efforts to have AI

0:35:46.280 --> 0:35:48.359
<v Speaker 9>kind of bolster its military capabilities.

0:35:49.160 --> 0:35:53.520
<v Speaker 2>Bloomba's Jackiedavolos, this is a pretty fast developing story. Grateful

0:35:53.520 --> 0:35:55.640
<v Speaker 2>for you keeping us honest on it out of Washington,

0:35:55.719 --> 0:36:00.399
<v Speaker 2>d C. Thank you, Bloombo Technology.

0:36:00.920 --> 0:36:03.800
<v Speaker 1>The Bay open every day, open