WEBVTT - China’s Got Big Plans for AI — In the Desert

0:00:02.720 --> 0:00:07.200
<v Speaker 1>Bloomberg Audio Studios, podcasts, radio news.

0:00:08.360 --> 0:00:11.320
<v Speaker 2>Earlier this year, our colleague James Mager took a trip

0:00:11.360 --> 0:00:13.560
<v Speaker 2>to the middle of the desert in the remote region

0:00:13.640 --> 0:00:16.560
<v Speaker 2>of Shinjong, China.

0:00:16.760 --> 0:00:22.919
<v Speaker 3>Shinjung is the western third of China. Basically it borders Mongolia, Kazakhstan,

0:00:23.239 --> 0:00:27.520
<v Speaker 3>Uzebekistan and where we went on this trip is situated

0:00:27.520 --> 0:00:30.880
<v Speaker 3>in a river valley. It's got these lovely snow capped

0:00:30.920 --> 0:00:35.280
<v Speaker 3>mountains overlooking the town. Really was a stunningly beautiful place,

0:00:35.760 --> 0:00:37.080
<v Speaker 3>but very hard to get to.

0:00:37.960 --> 0:00:40.680
<v Speaker 2>James was there on a lead generated by Bloomberg's data

0:00:40.720 --> 0:00:44.479
<v Speaker 2>journalism team that a handful of Chinese companies were building

0:00:44.520 --> 0:00:47.879
<v Speaker 2>some forty data centers across the region and planned to

0:00:47.960 --> 0:00:51.960
<v Speaker 2>power them with tens of thousands of Nvidia chips. Chips

0:00:52.040 --> 0:00:55.000
<v Speaker 2>that the US banned from being exported to China in

0:00:55.040 --> 0:00:58.160
<v Speaker 2>twenty twenty two is as well. China has criticized the

0:00:58.280 --> 0:01:02.400
<v Speaker 2>US moves to expand restrictions one is access to sympiconductor technology,

0:01:02.520 --> 0:01:05.640
<v Speaker 2>saying that they will harm supply chains and the world economy.

0:01:05.920 --> 0:01:10.520
<v Speaker 1>They want to build a AI industry that can go

0:01:10.640 --> 0:01:12.120
<v Speaker 1>head to head with the US.

0:01:13.040 --> 0:01:16.440
<v Speaker 2>Bloomberg data reporter Andy Linn led the team's investigation.

0:01:17.040 --> 0:01:21.680
<v Speaker 1>In modern warfares, AI is playing a larger role, so

0:01:22.160 --> 0:01:26.440
<v Speaker 1>the US is worried about China developing the AI capacity

0:01:26.880 --> 0:01:30.600
<v Speaker 1>with developing their high end military must be.

0:01:30.640 --> 0:01:36.920
<v Speaker 2>In Since the Biding Administration's initial restrictions on advanced AI chips,

0:01:37.200 --> 0:01:40.520
<v Speaker 2>the US has tightened export controls, putting them at the

0:01:40.600 --> 0:01:45.080
<v Speaker 2>center of tensions with Beijing. Just last week, Bloomberg learned

0:01:45.080 --> 0:01:48.200
<v Speaker 2>of plans by the Trump administration to further restrict the

0:01:48.240 --> 0:01:52.160
<v Speaker 2>shipments of AI chips to Thailand and Malaysia, part of

0:01:52.200 --> 0:01:55.840
<v Speaker 2>an effort to prevent them from being potentially smuggled into China.

0:01:56.440 --> 0:01:59.880
<v Speaker 1>So by building these data centers and build up the

0:02:00.040 --> 0:02:05.360
<v Speaker 1>computing power, they aim to build a local, domestic EI

0:02:05.400 --> 0:02:08.560
<v Speaker 1>industry that can go ahead to have with opening AI

0:02:08.680 --> 0:02:11.280
<v Speaker 1>Alphabet Amazon Meta in the US.

0:02:15.200 --> 0:02:18.840
<v Speaker 2>This is the Big Take Asia from Bloomberg News. I'm Wanha.

0:02:19.040 --> 0:02:21.520
<v Speaker 2>Every week we take you inside some of the world's

0:02:21.520 --> 0:02:25.280
<v Speaker 2>biggest and most powerful economies and the markets, tycoons, and

0:02:25.360 --> 0:02:29.600
<v Speaker 2>businesses that drive this ever shifting region. Today on the show,

0:02:30.000 --> 0:02:40.919
<v Speaker 2>China's AI ambition rises from the desert. China is fast

0:02:40.960 --> 0:02:44.520
<v Speaker 2>becoming a world leader in artificial intelligence it's one of

0:02:44.520 --> 0:02:49.000
<v Speaker 2>the top consumers of semiconductors, and as Deepseek's latest model shows,

0:02:49.280 --> 0:02:53.360
<v Speaker 2>Chinese companies are hungry to create top tier AI technology

0:02:53.639 --> 0:02:58.799
<v Speaker 2>to rival international competitors. But Andy says the US government's

0:02:58.880 --> 0:03:02.519
<v Speaker 2>export controls have created a real problem for China by

0:03:02.520 --> 0:03:06.040
<v Speaker 2>limiting its access to Nvidia's priced semiconductors.

0:03:07.280 --> 0:03:12.919
<v Speaker 1>Nvidia is the industrial standard for any AI training operations,

0:03:12.960 --> 0:03:16.120
<v Speaker 1>so for example, Open Ai, the most famous AI companies

0:03:16.120 --> 0:03:19.560
<v Speaker 1>in the world, use Nvidia chips for training the chat,

0:03:19.600 --> 0:03:25.359
<v Speaker 1>GPT models and whatever. You want to build a sophisticated

0:03:25.680 --> 0:03:29.000
<v Speaker 1>AI model, you need to use a lot of Nvidia chips.

0:03:29.480 --> 0:03:32.600
<v Speaker 1>So MVA chip is sought after across all the world

0:03:32.760 --> 0:03:36.720
<v Speaker 1>for all the AI development because every country wants their

0:03:36.840 --> 0:03:38.760
<v Speaker 1>own AI.

0:03:38.080 --> 0:03:42.200
<v Speaker 2>Industry, and he says China's ultimate goal is to have

0:03:42.240 --> 0:03:45.800
<v Speaker 2>its own source of advanced AI chips, ones that are

0:03:45.800 --> 0:03:50.680
<v Speaker 2>on par with Nvidia's offerings. For now, though, in semiconductors

0:03:50.760 --> 0:03:54.040
<v Speaker 2>are the best product on the market and Chinese companies

0:03:54.080 --> 0:03:55.840
<v Speaker 2>are eager to access them.

0:03:56.440 --> 0:04:01.480
<v Speaker 1>Since the US introduced these export controls, I heard sources

0:04:01.520 --> 0:04:04.440
<v Speaker 1>in China saying they are having a hard time getting

0:04:05.240 --> 0:04:10.200
<v Speaker 1>large volumes of Nvidia chips for their operations. I think

0:04:10.280 --> 0:04:14.960
<v Speaker 1>this slow China's access to these chips, but not stop

0:04:15.080 --> 0:04:17.120
<v Speaker 1>them from getting them outright.

0:04:18.520 --> 0:04:22.039
<v Speaker 2>The US government has no official consensus on how many

0:04:22.080 --> 0:04:26.279
<v Speaker 2>restricted Nvidia chips are currently in China. Two senior Biding

0:04:26.320 --> 0:04:30.200
<v Speaker 2>officials estimated that China had around twenty five thousand chips,

0:04:30.800 --> 0:04:33.080
<v Speaker 2>but most of the people Bloomberg spoke to say there

0:04:33.120 --> 0:04:36.479
<v Speaker 2>isn't an agreed upon estimate. That's why when Andy found

0:04:36.480 --> 0:04:40.080
<v Speaker 2>documents laying out the volume of Nvidia chips that companies

0:04:40.200 --> 0:04:42.800
<v Speaker 2>hope to obtain for these data centers, he was.

0:04:42.760 --> 0:04:48.240
<v Speaker 1>Surprised because most of the people I talked to would

0:04:48.320 --> 0:04:52.440
<v Speaker 1>say that they assume some developments of these data standard

0:04:52.560 --> 0:04:58.320
<v Speaker 1>using Nvidia chips is ongoing, but not like declaring to

0:04:58.960 --> 0:05:01.880
<v Speaker 1>the world that they're going to use band Nvidia chips

0:05:01.880 --> 0:05:03.120
<v Speaker 1>in this large volume.

0:05:03.520 --> 0:05:06.400
<v Speaker 2>In the fourth quarter of last year, the Chinese government

0:05:06.440 --> 0:05:09.799
<v Speaker 2>approved a total of thirty nine data center investment projects

0:05:09.839 --> 0:05:13.480
<v Speaker 2>in Shinjan and the neighboring Chinghai province. They've built out

0:05:13.560 --> 0:05:16.000
<v Speaker 2>plans to use more than one hundred and fifteen thousand

0:05:16.200 --> 0:05:17.320
<v Speaker 2>Nvidia processors.

0:05:17.760 --> 0:05:20.599
<v Speaker 1>If wanted to compare this scale to what the US

0:05:20.640 --> 0:05:24.320
<v Speaker 1>have usually the tech companies there will claim if you

0:05:24.440 --> 0:05:28.360
<v Speaker 1>have more than one undred thousand Nvidia chips in one place,

0:05:28.839 --> 0:05:32.719
<v Speaker 1>then you'ing in a place to develop a state of

0:05:32.800 --> 0:05:38.400
<v Speaker 1>the art models which can possibly compete with Gemini Chat JBT.

0:05:38.640 --> 0:05:42.080
<v Speaker 2>These kind of models, there's no explanation in the official

0:05:42.200 --> 0:05:45.400
<v Speaker 2>documents of how companies plan to acquire the band in

0:05:45.480 --> 0:05:49.680
<v Speaker 2>Nvidia chips, and it could all be aspirational. In a

0:05:49.720 --> 0:05:52.880
<v Speaker 2>response from Nvidia, the company said that posting a web

0:05:52.920 --> 0:05:57.560
<v Speaker 2>page about restricted products is not the same as successfully licensing,

0:05:57.760 --> 0:06:01.160
<v Speaker 2>building and operating a data center, and that Nvidia does

0:06:01.200 --> 0:06:05.440
<v Speaker 2>not provide any support or repairs for restricted products. In

0:06:05.480 --> 0:06:08.880
<v Speaker 2>the meantime, the companies listed in the filings, state officials

0:06:08.920 --> 0:06:13.040
<v Speaker 2>and central government representatives in Beijing all declined to comment

0:06:13.080 --> 0:06:18.080
<v Speaker 2>when asked to explain. Still the building continues in the desert.

0:06:19.480 --> 0:06:22.120
<v Speaker 2>Most of these data centers are located in a single

0:06:22.200 --> 0:06:25.560
<v Speaker 2>compound in Ewu County, set up by the local government

0:06:25.680 --> 0:06:30.200
<v Speaker 2>in Shinjug. These data centers house big computers, servers, and

0:06:30.279 --> 0:06:34.440
<v Speaker 2>other components and are used to process, store, and distribute data.

0:06:35.000 --> 0:06:37.640
<v Speaker 2>And while they might be located far from big cities

0:06:37.680 --> 0:06:41.360
<v Speaker 2>like Shanghai or Shenzen, startups in these cities can send

0:06:41.400 --> 0:06:44.560
<v Speaker 2>their models to the data centers for training and optimization.

0:06:45.440 --> 0:06:50.400
<v Speaker 1>Theoretically, any company in China can access that computing power. Essentially,

0:06:50.400 --> 0:06:53.039
<v Speaker 1>you can send a request to the data centers in

0:06:53.080 --> 0:06:56.360
<v Speaker 1>Shinjioan saying that, hey, we want to use how many

0:06:56.520 --> 0:07:00.839
<v Speaker 1>chips to train our models in the eastern regions, and

0:07:00.880 --> 0:07:03.640
<v Speaker 1>then the data centers and Shinjong will produce a result

0:07:03.680 --> 0:07:07.359
<v Speaker 1>to you, maybe in a few minutes or even a

0:07:07.400 --> 0:07:11.560
<v Speaker 1>few seconds. Yeah, so these are all connected because you

0:07:11.680 --> 0:07:14.520
<v Speaker 1>don't need to be present at the data center use

0:07:14.720 --> 0:07:15.640
<v Speaker 1>the computing power.

0:07:16.280 --> 0:07:20.040
<v Speaker 2>Shinjong is an ideal location for large scale data centers.

0:07:20.600 --> 0:07:24.000
<v Speaker 2>Land is cheap and rental costs are low. The region

0:07:24.160 --> 0:07:27.680
<v Speaker 2>also has cool weather, which helps offset the heat generated

0:07:27.720 --> 0:07:31.760
<v Speaker 2>by the servers. Most importantly, it's a hub for green energy.

0:07:32.600 --> 0:07:34.720
<v Speaker 2>On the way to look for the data centers, James

0:07:34.720 --> 0:07:40.480
<v Speaker 2>Mager stumbled across miles and miles of renewable energy installations.

0:07:40.320 --> 0:07:45.040
<v Speaker 3>The desert was just a dusty, rocky plane, and scattered

0:07:45.040 --> 0:07:48.679
<v Speaker 3>across the desert was just windmill after windmill, after solar

0:07:48.720 --> 0:07:51.960
<v Speaker 3>panel installation after solar panel installation, and then there was

0:07:52.000 --> 0:07:56.800
<v Speaker 3>also this power generation, which is where the company uses

0:07:56.920 --> 0:08:00.360
<v Speaker 3>mirrors to concentrate sunlight on tooward tower, which is filled

0:08:00.400 --> 0:08:03.880
<v Speaker 3>with molten sodium, and then that heat is generated by

0:08:04.240 --> 0:08:07.760
<v Speaker 3>concentrating all that sunlight is then used to generate electricity.

0:08:07.960 --> 0:08:10.200
<v Speaker 3>And you can see this like twenty thirty kilometers away

0:08:10.240 --> 0:08:13.960
<v Speaker 3>because the sunlight is so blinding as it's concentrated onto

0:08:13.960 --> 0:08:14.520
<v Speaker 3>this tower.

0:08:15.760 --> 0:08:19.680
<v Speaker 2>This combination of green energy and advanced computing fits in

0:08:19.720 --> 0:08:24.840
<v Speaker 2>with Beijing's economic push for sustainable development. After the break,

0:08:25.000 --> 0:08:28.560
<v Speaker 2>what is China's master plan to build these data centers?

0:08:29.040 --> 0:08:44.080
<v Speaker 2>We zoom in on one company to find out. Bloomberg's

0:08:44.120 --> 0:08:47.560
<v Speaker 2>James Mager spent three days in Shinjung tracking down the

0:08:47.640 --> 0:08:52.800
<v Speaker 2>data centers. Andy Linn and his team had found on paper, Iwu.

0:08:52.720 --> 0:08:56.080
<v Speaker 3>Is pretty remote, even in Shinjong, which is pretty remote

0:08:56.080 --> 0:08:58.959
<v Speaker 3>in China. To get there, it was a four hour

0:08:59.320 --> 0:09:03.040
<v Speaker 3>plane ride from Beijing to the city of Hami, and

0:09:03.080 --> 0:09:05.280
<v Speaker 3>then from Hami to Iwu is another three and a

0:09:05.360 --> 0:09:08.600
<v Speaker 3>half for our car ride up into the mountains.

0:09:09.320 --> 0:09:12.480
<v Speaker 2>Using the data team's coordinates, James was able to locate

0:09:12.640 --> 0:09:16.360
<v Speaker 2>massive building complexes in the desert right where Andy said

0:09:16.400 --> 0:09:18.400
<v Speaker 2>the data centers were being built.

0:09:19.920 --> 0:09:22.319
<v Speaker 3>There was a lot of construction in Eyu, and there

0:09:22.360 --> 0:09:24.760
<v Speaker 3>was some construction going on up in the area with

0:09:24.800 --> 0:09:28.440
<v Speaker 3>all the solar panels. I've never seen a data center before.

0:09:28.920 --> 0:09:30.200
<v Speaker 3>I mean, this is the first time I've seen a

0:09:30.240 --> 0:09:33.280
<v Speaker 3>data center. But they look like what I imagine a data center.

0:09:33.040 --> 0:09:33.600
<v Speaker 1>Would look like.

0:09:34.160 --> 0:09:36.640
<v Speaker 3>You have a large building which is three, four or

0:09:36.640 --> 0:09:40.199
<v Speaker 3>five stories tall, has almost no windows. They obviously don't

0:09:40.240 --> 0:09:42.120
<v Speaker 3>need a lot of light. Having a lot of windows

0:09:42.160 --> 0:09:44.400
<v Speaker 3>is going to make it harder to keep the temperature

0:09:44.400 --> 0:09:47.880
<v Speaker 3>inside the data center controlled at the proper level.

0:09:48.120 --> 0:09:50.640
<v Speaker 2>It sounds like they had certainly all the hallmarks of

0:09:50.720 --> 0:09:51.600
<v Speaker 2>being a data center.

0:09:51.920 --> 0:09:52.199
<v Speaker 1>Yes.

0:09:53.320 --> 0:09:57.520
<v Speaker 2>Among all the projects examined in the investigation, one caught

0:09:57.520 --> 0:10:01.359
<v Speaker 2>the team's attention. It involves a company controlled by Nyocore,

0:10:01.679 --> 0:10:04.720
<v Speaker 2>an energy firm that's partially owned by the state government

0:10:05.040 --> 0:10:08.960
<v Speaker 2>based in Tianjin in northern China. Niacore's main business is

0:10:08.960 --> 0:10:12.640
<v Speaker 2>supplying solar and wind power. And I asked Andy, why

0:10:12.679 --> 0:10:16.000
<v Speaker 2>a green energy firm, of all companies, would be building

0:10:16.120 --> 0:10:16.840
<v Speaker 2>data centers.

0:10:17.600 --> 0:10:20.920
<v Speaker 1>Yeah. So you don't usually think that green energy companies

0:10:21.040 --> 0:10:25.360
<v Speaker 1>delve into the data standard business, but in China, Beijing

0:10:25.400 --> 0:10:29.480
<v Speaker 1>has been encouraging green energy companies to invest into AI

0:10:29.640 --> 0:10:34.040
<v Speaker 1>data centers because they want to lower the carbon foot prints.

0:10:34.120 --> 0:10:36.400
<v Speaker 1>That will create a win win.

0:10:37.160 --> 0:10:41.200
<v Speaker 2>According to the documents, Bloomberg found Nyacore plans to build

0:10:41.240 --> 0:10:43.960
<v Speaker 2>a data center with six hundred and twenty five servers.

0:10:44.360 --> 0:10:46.840
<v Speaker 1>It is one of the largest projects we found in

0:10:46.880 --> 0:10:51.040
<v Speaker 1>the batch of the documents. Based on their investment approvals,

0:10:51.400 --> 0:10:56.360
<v Speaker 1>they need two thousand, one hundred GPU chips. This a

0:10:56.440 --> 0:11:00.960
<v Speaker 1>lot because if you imagine one at one hundred chips

0:11:01.000 --> 0:11:05.840
<v Speaker 1>cost around twenty thousand dollars, then well is a very

0:11:05.880 --> 0:11:11.440
<v Speaker 1>big amount of money just for mid sized green energy

0:11:11.440 --> 0:11:12.240
<v Speaker 1>company in China.

0:11:13.320 --> 0:11:17.319
<v Speaker 2>Niacord is selling its computer power to infinitgens Ai, one

0:11:17.360 --> 0:11:21.520
<v Speaker 2>of the largest AI infrastructure companies in China. The company

0:11:21.559 --> 0:11:24.280
<v Speaker 2>has raised one billion yun or almost one hundred and

0:11:24.360 --> 0:11:27.280
<v Speaker 2>forty million dollars in funding since it was founded in

0:11:27.280 --> 0:11:31.839
<v Speaker 2>twenty twenty three. Niacord declined to comment and infinogens Ai

0:11:32.080 --> 0:11:35.360
<v Speaker 2>couldn't be reached for a response, but last year, the

0:11:35.360 --> 0:11:38.960
<v Speaker 2>CEO of infinogens Ai said in an interview with local

0:11:39.000 --> 0:11:42.200
<v Speaker 2>media that the goal is to make computing power more

0:11:42.280 --> 0:11:45.400
<v Speaker 2>accessible to AI companies all across China.

0:11:46.080 --> 0:11:50.960
<v Speaker 1>Their CEO said they aim to create a computing power

0:11:51.080 --> 0:11:54.960
<v Speaker 1>system that allows AI developers across the main line to

0:11:55.400 --> 0:11:59.080
<v Speaker 1>just log into their system and get all the computing

0:11:59.160 --> 0:12:02.520
<v Speaker 1>power they want, just like opening a like water tap

0:12:02.600 --> 0:12:05.959
<v Speaker 1>and then you can get the water you want.

0:12:06.120 --> 0:12:09.960
<v Speaker 2>Computer power on tap. But in order to make computer

0:12:10.080 --> 0:12:14.400
<v Speaker 2>power accessible anywhere, anytime, there's an issue that needs to

0:12:14.440 --> 0:12:19.440
<v Speaker 2>be addressed first. Shinjong is China's most western region. Most

0:12:19.480 --> 0:12:22.280
<v Speaker 2>of the AI companies that need the computing power are

0:12:22.280 --> 0:12:25.640
<v Speaker 2>based in big cities like Shanghai, far in the east.

0:12:26.280 --> 0:12:29.240
<v Speaker 2>That's more than two thousand miles apart, about the distance

0:12:29.240 --> 0:12:33.040
<v Speaker 2>from Chicago to Los Angeles. Andy says. The Chinese government

0:12:33.160 --> 0:12:36.360
<v Speaker 2>plans to bridge this supplied demand gap by building what

0:12:36.400 --> 0:12:38.880
<v Speaker 2>they call computing power corridors.

0:12:39.440 --> 0:12:42.559
<v Speaker 1>You can think about this as the water resource management

0:12:42.720 --> 0:12:46.400
<v Speaker 1>system in China. They have been moving the water from

0:12:46.440 --> 0:12:49.240
<v Speaker 1>the southern region to the north region, which is suffering

0:12:49.280 --> 0:12:52.800
<v Speaker 1>from draft. So they have been moving crucial resources around

0:12:52.840 --> 0:12:56.000
<v Speaker 1>in the mainland, and now they want to move the

0:12:56.120 --> 0:12:56.960
<v Speaker 1>computing power.

0:12:57.720 --> 0:13:01.120
<v Speaker 2>So in some ways they're thinking of a computing power

0:13:01.280 --> 0:13:04.320
<v Speaker 2>like a utility, right, and the computing power corridor would

0:13:04.360 --> 0:13:05.560
<v Speaker 2>be like the power grid.

0:13:06.080 --> 0:13:09.040
<v Speaker 1>Yeah, yeah, that's what they are aiming for. They're building

0:13:09.040 --> 0:13:13.520
<v Speaker 1>infrastructure like more cables along the codors, so that they

0:13:13.600 --> 0:13:18.160
<v Speaker 1>can encourage more AI developers to build infrastructure in the

0:13:18.200 --> 0:13:19.080
<v Speaker 1>western regions.

0:13:19.800 --> 0:13:23.320
<v Speaker 2>This is all part of China's plan to dominate in AI.

0:13:23.840 --> 0:13:26.199
<v Speaker 2>But in order for China to get there, it needs

0:13:26.280 --> 0:13:29.160
<v Speaker 2>all these things to come together. Its own high tech

0:13:29.240 --> 0:13:32.840
<v Speaker 2>chips that can rival in NVIDIAs, computer power on tap,

0:13:33.240 --> 0:13:37.120
<v Speaker 2>and an industry with cutting edge innovations. And Andy says

0:13:37.120 --> 0:13:41.280
<v Speaker 2>building such a national network could encourage more Chinese companies

0:13:41.360 --> 0:13:44.040
<v Speaker 2>to jump into the AI space without needing to build

0:13:44.040 --> 0:13:47.440
<v Speaker 2>their own data centers. That would help startups come up

0:13:47.440 --> 0:13:51.240
<v Speaker 2>with more innovations like Deepseek did with its chatbot, and

0:13:51.400 --> 0:13:56.239
<v Speaker 2>ultimately narrow the gap between the US and China.

0:13:56.280 --> 0:13:59.200
<v Speaker 1>I think one of the most powerful aspect of this

0:13:59.480 --> 0:14:03.800
<v Speaker 1>utility for AI computing powers that it will not be

0:14:03.920 --> 0:14:08.320
<v Speaker 1>controlled by a couple of tech giants only. So if

0:14:08.320 --> 0:14:11.680
<v Speaker 1>you make a comparison between US and China. In the US,

0:14:11.840 --> 0:14:15.959
<v Speaker 1>computing power is mostly owned by a handful of tech giants.

0:14:16.200 --> 0:14:20.960
<v Speaker 1>Open ai madea alphabet, but if you make it as

0:14:20.960 --> 0:14:26.200
<v Speaker 1>a utility, every developer across the mainline can just log

0:14:26.280 --> 0:14:29.920
<v Speaker 1>into a system and then got as much computing power

0:14:30.000 --> 0:14:33.520
<v Speaker 1>as they want. Then it will create a very distributed

0:14:33.600 --> 0:14:38.640
<v Speaker 1>system for AI innovations, so that anyone who have the

0:14:38.720 --> 0:14:42.960
<v Speaker 1>idea can just create its own model, and then we'll

0:14:43.000 --> 0:14:47.320
<v Speaker 1>have a more collaborative environment for these developments, which is

0:14:47.400 --> 0:14:49.600
<v Speaker 1>totally different from what we have in the US.

0:14:50.360 --> 0:14:53.640
<v Speaker 2>There's a lot of uncertainty about this model. Will China

0:14:53.720 --> 0:14:56.440
<v Speaker 2>be able to create a cheap and reliable power source

0:14:56.720 --> 0:14:59.560
<v Speaker 2>that everyone can tap into, will it be able to

0:14:59.560 --> 0:15:03.760
<v Speaker 2>acquire computer chips that AI development demands, and will China

0:15:03.840 --> 0:15:06.680
<v Speaker 2>be able to make those chips itself rather than relying

0:15:06.720 --> 0:15:08.480
<v Speaker 2>on a US company Like in video.

0:15:09.200 --> 0:15:13.360
<v Speaker 1>What we found in Shinjiai and Shanghai for these scale

0:15:13.440 --> 0:15:17.280
<v Speaker 1>of data CENTTA development shows that Beijing is trying very

0:15:17.320 --> 0:15:21.880
<v Speaker 1>hard to nurture is on domestic AI industry so that

0:15:21.920 --> 0:15:33.320
<v Speaker 1>they can match the tech giants from the wests.

0:15:39.120 --> 0:15:42.840
<v Speaker 2>This is The Big Take Asia from Bloomberg News. I'm wanha.

0:15:42.920 --> 0:15:45.680
<v Speaker 2>To get more from The Big Take and unlimited access

0:15:45.720 --> 0:15:48.680
<v Speaker 2>to all of Bloomberg dot Com, subscribe today at Bloomberg

0:15:48.720 --> 0:15:52.200
<v Speaker 2>dot com slash podcast offer. If you like the episode,

0:15:52.280 --> 0:15:54.720
<v Speaker 2>make sure to subscribe and review The Big Take Asia

0:15:54.880 --> 0:15:58.120
<v Speaker 2>wherever you listen to podcasts. It really helps people find

0:15:58.160 --> 0:16:00.480
<v Speaker 2>the show. Thanks for listening, See next time.