Inside China's data center desert city — Meanwhile in Asia

[00:00:00] Simone McCarthy: - I mean, it's block after block of data centers. This is a place that was known as China's potato growing capital. And now it's just completely embedded in this very new, very high tech advanced industry. [00:00:13] Ivan Watson: - A total scam on the people of Ohio. - This became a potent political issue in the US. [00:00:20] Kunal Sagal: - But it's what happens when this enormous technological revolution starts showing up in people's backyards. - Okay, we'll do it from the top. - Hello, I'm Kunal Sagal. Welcome to Meanwhile in Asia, a weekly show where we take some of the stories that are shaping our region and try to spend a bit more time with them and give them more context. As you can tell, I'm not in this studio. I'm joining you from India. [00:00:48] Ivan Watson: - I'm Ivan Watson. I'm here in the studio. And Kunal and our viewers, what do you think about data centers? That's a question we're gonna be exploring in this episode. It's a big deal in the US in the midterm elections, but we're gonna be approaching this from the perspective of the world's second largest economy, China, which is in an AI race with the US right now, and taking a look at how it is dealing with this critical part of digital infrastructure. [00:01:21] Kunal Sagal: - Yeah, it's fascinating, Ivan, how differently this race is playing out in the two countries. In the US, like you said, data centers have become a huge source of controversy. Communities are pushing back over electricity bills, water consumption, and the impact on the environment. But in China, they're being built at astonishing speed, often in places you might least expect. We're talking enormous, hulking complexes rising out of farmland surrounded by fields of sunflowers and wind turbines. So places that until very recently had little to do with the global AI race are now suddenly becoming central to it. But we're joined today by Simone McCarthy, our senior China reporter who has been to one of those places. Simone, welcome to the show. [00:02:07] Simone McCarthy: - Thank you, Kunal, Ivan, happy to be here. [00:02:10] Ivan Watson: - You got rare access to data centers in this city, arguably in the middle of nowhere in Inner Mongolia. What did it look like? What did you see? [00:02:20] Simone McCarthy: - That's right. So this is an area that's like 200 miles north of Beijing and it's rolling plains, grassland, fields of sunflowers like Kunal said. There's about two million people in the region and they're slated to have 89 data centers built there. It's just become really this hotspot hub for this AI infrastructure. And this is a place that was known as China's potato growing capital. And now it's just completely embedded in this very new, very high tech advanced industry. [00:02:53] Ivan Watson: - That must be a bit of a culture clash there, right? [00:02:55] Simone McCarthy: - I mean, it's block after block of data centers construction workers are just lining the streets in and out of these areas, trucks bringing in supplies. I mean, you can see that this AI is really supercharging this area. And, you know, there were already data centers that were there, more like computing, Huawei, Apple. They've been there for a long time, but this is clearly this AI push in China, really this global drive for AI infrastructure is booming there. [00:03:22] Ivan Watson: - You've got a provincial city going from being the potato capital of China to arguably the data center capital of the country. - Absolutely. - That's a very dramatic change. [00:03:33] Simone McCarthy: - And it's very China too, right? [00:03:35] Ivan Watson: Because when China has- - Nobody builds like China. [00:03:37] Simone McCarthy: - Exactly. And when there's this kind of national ambition to push forward this data center build out, that's exactly when you're gonna see places that really latch onto this. And, you know, not every place is suitable for it, but this city, which is called Ulancab, it's basically 2.1 milliseconds to get to Beijing for this data speed. So, you know, it's pretty suitable, but it's just, it's a little bit surreal actually, seeing it on the ground. [00:04:03] Kunal Sagal: - Yes, Simel, like you said, it's astonishing how quickly China has built this. Whereas in the US, this AI infrastructure boom, I feel is running into a pretty remarkable obstacle. The people who are supposed to benefit from it are just the people that necessarily don't want it anywhere near them. Of course, it isn't necessarily opposition to AI itself, but it's what happens when this enormous technological revolution starts showing up in people's backyards, consuming electricity, water, potentially, you know, pushing up their bills as well. [00:04:37] Ivan Watson: - If somebody sticks an oil well in your backyard, you're not necessarily gonna get the proceeds of that oil. I mean, we've seen that in other kind of extractive industries in the past. - Yeah, exactly. - There are complaints that they make huge amounts of noise as well. And local communities were not necessarily consulted about that. And so, in a very short period of time, I mean, under two years, wouldn't you say, this became a potent political issue in the US. And then I think it's also combined with concern, stress, suspicion about artificial intelligence in general, right? [00:05:17] Simone McCarthy: - People in China have quite a different view of AI than people in the US, where there's much more of a thinking about how AI is gonna be used practically. There's also more of an understanding that the government will keep AI in check, at least in terms of the public viewpoint on it. So, even just that concept, yes, of course, there are concerns about job replacement and things like that, but there's not the same kind of broader conversation of like the existential threat of humanity in China that we see playing out in the US. [00:05:44] Ivan Watson: - Right, and then when it comes back to the data centers, the US, at the time we're taping it, we believe it has maybe 10 times more data centers, somewhere in the area of more than 5,000 data centers, whereas China believed to be approaching 500 of these facilities thus far. [00:06:02] Simone McCarthy: - And I think like the physical number is a little bit less important than like what's inside them. And because exactly as you said, Ivan, China does have this tremendous capability to ramp up the building of infrastructure. And a lot of that goes back to these kind of systemic issues that are different in the US in terms of permitting. I mean, I cannot understate how much of a national priority AI development is for Beijing. It basically sees this high tech transformation of its economy, specifically embedding AI into manufacturing, into healthcare, into all kinds of products as well as sectors, industries. I mean, that really is how it sees itself moving forward its economy instead of being the world's factory for these low end goods. It's more about becoming really moving way up the supply chain and also saving its own economy. And, you know, as we've talked about before, the declining population as well. I mean, this is something that AI automation are all meant to offset in China. So this is like top tier priority for Beijing. [00:07:03] Ivan Watson: - Beijing is putting its money where its mouth is, announcing a $560 billion plan in September. That's more than half a trillion dollars, the equivalent to scale up data infrastructure. [00:07:15] Kunal Sagal: - Exactly. - Yeah, I think that's where the contrast with China gets really interesting. I mean, Simone, why is this so important to Beijing? What is the economic imperative? [00:07:26] Simone McCarthy: - China right now is facing a slowing economy. That's natural for a country that had double digit growth, you know, 15 years ago. China also wants to be able to be a global leader in this technology too. And so that's where I think the AI infrastructure is so critical because there's 700 million AI users in China right now. So this infrastructure is supporting that. It's an incredibly big picture for Beijing. [00:07:48] Ivan Watson: - That said, there's also a big difference when it comes to the scale of this in the two countries, right? China can build things about as fast as any country in the world. I mean, it's really remarkable to see up close, but the US is far ahead when it comes to data centers, right? [00:08:05] Simone McCarthy: - Yeah, I mean, so that's something that we need to be really put front and center on the table here when we're talking about how the US and China are approaching this because China lacks something that the US has, I think it's fair to say in abundance, which is the highest end chips. And you can build as many physical data centers and racks as you want, but you need that high tech hardware to fill it. And so that's the thing that we're really watching very closely because as you guys well know, the US has put restrictions on Chinese tech firms' access to the highest end chips. And then China also has its own restrictions on imports of some of those not quite as high end, but slightly lower level chips. And so those are both hampering the amount of this really critical hardware that the Chinese tech firms can get. [00:08:56] Ivan Watson: - And that's deliberate. This is a US federal government policy. - These are aimed at slowing down the AI revolution in China. [00:09:05] Simone McCarthy: - And you're talking about the midterms. I mean, this is something that I think there is concern within China that when the Democrats are now having control of both houses of Congress, that there could even be more of a push to put more restrictions on some of those, you know, return some of those restrictions that the Trump administration has rolled out, put more restrictions on some of these chips. - I mean, there's this crazy stat that I was looking at when I was preparing to do this piece and travel to Ulan Cobb, which is basically that four of the US top tech firms have more computing power or more access to chips that they can use for AI than all of China's tech companies combined. [00:09:45] Ivan Watson: - And look, we have reports that some $31 million were spent on political ads in the US in these midterms talking about data centers. - Wow, $31 million. - Yeah, tens of millions of dollars and anecdotally about very angry kind of local town council meetings where people are just really, really-- [00:10:06] Simone McCarthy: - I mean, it's a huge issue. [00:10:07] Ivan Watson: - In the US, China can control dissent on a scale that the US can't even begin to. The level of censorship and state security presence is massive. But I mean, what did you hear from people in this former potato capital of China when they had all these data centers sprouting up next door? [00:10:28] Simone McCarthy: - I was very curious to hear what people had to say. - And so we spoke to several business owners who were operating grocery stores, guest houses, restaurants, just right around this area where there's this massive AI data center build out. And honestly, their response was, look, the economy has been bad in recent years. And we are so happy for the extra business that's coming in from this deluge of construction workers, many of whom are migrant workers, I should say, that are coming into the area that are doing this construction. And they were really happy to have, you know, double the business that they had a couple of years ago. [00:11:10] Ivan Watson: - Totally juicing the local economy. [00:11:12] Simone McCarthy: - Totally. But I mean, at the same time, we should say that that construction may not continue forever. So are they going to see those same levels when these data centers are complete? It's hard to say. [00:11:23] Ivan Watson: - But on a whole, I mean, you live in Beijing. You don't hear the suspicion about AI, the way-- [00:11:29] Simone McCarthy: - Or the backlash against data centers either. I mean, both of those things. And I think they're absolutely related. And of course, I mean, the space for public discussion in China, and especially dissent in China, is vastly different from the United States. But there's also a sense, I think it's fair to say, when there is this kind of national priority, people also just realize, this is what Beijing wants to get done. This is going to go forward because this is the way that, you know, this is the plan that the government has for the future, for the economy. [00:11:59] Kunal Sagal: - Is China's biggest advantage here not necessarily the technology? Of course, China's AI is extremely advanced, but is their primary advantage not necessarily the technology, but its ability to get things done because of the system. [00:12:11] Simone McCarthy: - So China has significant advantages when it comes to, yeah, as we've been talking about, its ability to build out infrastructure, but also energy and centralized top-down planning. Obviously, the chip shortage is a major issue, but at the same time, China's own national tech giants are ramping up their efforts to be able to match some of the capabilities of their US rivals. But if we're talking about those areas where China's really shining and where it has those advantages, I mean, there is a national plan which has designated eight computing hubs and 10 so-called data center clusters. So Beijing is looking at this like from a bird's eye view of where should we place these data centers? What are the hotspots that we want to have them in? [00:12:55] Ivan Watson: - As opposed to like an ad hoc. [00:12:57] Simone McCarthy: - Exactly, yeah. [00:12:58] Ivan Watson: - And when you bring up the energy question, that's a really important one. - It's a key point. - Because these things consume vast amounts of energy. I'm gonna provide some statistics here. The United Nations says that in 2025, globally, data centers consumed an estimated 448 terawatt hours of electricity. I don't know what a terawatt hour is. - It's a lot, Ivan, it's a lot. But here's a comparison. That would be, if the world's data centers were a country, it would be the 11th highest consumer of electricity in the world, right? So that's how much these facilities, electricity, they're consuming. And an advantage that China brings to the table is it has a surplus of energy. [00:13:39] Simone McCarthy: - Yeah, let me trade you back some statistics, going over to my notes here. So between 2022 and the end of 2025, China added the equivalent of the US's entire industrial power feeding into its grid. So that's just showing-- - Okay, how did it do that? What was it doing? - Well, green energy. I mean, China is the world's leading force in green energy. [00:14:00] Ivan Watson: - Wind farms and solar. [00:14:01] Simone McCarthy: - Yeah, hydro as well. But yeah, especially when it comes to tech, it's these. Now, China still uses a lot of coal. I mean, we're not gonna forget about that, obviously. [00:14:07] Ivan Watson: - I mean, it's expanding coal-powered plants, too, so it's not all green. [00:14:12] Simone McCarthy: - Exactly, and those things, in a way, come hand in hand because you need coal, in a way, to stabilize solar and wind power when you're having times when there isn't wind. And, you know, I mean, that could be a whole other podcast. But the other stat that I think is incredible is that China wasted enough green energy in the first half of this year alone that would have been enough to power Mexico for an entire year. So they have this tremendous amount of excess green energy, and they also need something to do with it. So this is the grand vision, is, hey, let's link these things up. In Ulan Qab, we had access to a data center. We were the first international media to get access into a data center that will be one of the world's largest by area when it's fully opened. And they're actually getting upwards of 80% of their power directly from a wind farm. And so it's not even going into the grid at all. [00:15:06] Kunal Sagal: That's an extraordinary amount of energy going unused, Simone. And so presumably, of course, there's a huge opportunity there. But there is a catch, isn't it? Because a lot of that energy that is being generated in China's Western regions, while the biggest centers of business and demand are further east, how is China trying to bridge that gap? [00:15:25] Simone McCarthy: Yeah, I'm really glad you brought that up, Kunal, because China's overarching strategy, which it released in 2022, was this Eastern data, Western compute. And that strategy basically was, OK, we're going to have these data centers located out in the West, far away from our populous eastern seaboard, but where all of that ample excess energy is, because that's where... [00:15:46] Ivan Watson: And land. [00:15:47] Simone McCarthy: Exactly. There's cheap land. There's space for sprawling wind farms for solar panels. There's also a lot of hydropower in that part of the country because it's mountainous. So that was the plan. But at the same time, a lot of this AI, you can't have seconds and seconds and seconds in between sending the data from those centers. [00:16:05] Ivan Watson: You don't want a long delay in your transmission. [00:16:07] Simone McCarthy: You don't want a long delay in your transmission, right, Kunal? So, I mean, think about like an autonomous vehicle, for example, or a factory that's using AI. You definitely are going to need to have that much more instantaneous. So I think it's definitely fair to say that that strategy is undergoing some transformation. I wouldn't say it's been a complete failure, but it's like, OK, we need to tailor this to make sure that, yeah, there's some things that you can do further away, but you can't do everything. [00:16:33] Ivan Watson: But let me bring up another side to the AI data center boom, right? And that is the environmental impact here. And it's not small, right? It's not just the energy consumption, which has its own wear and tear on the globe. It's the vast amount of water that you need to cool off these facilities, right? I'm going to go back and throw out a statistic. The UN University Institute for Water Environment and Health, it warns that by 2030, data centers will consume the equivalent amount of water that you would need for all 1.3 billion residents of sub-Saharan Africa. In 2025, data centers had a water footprint of 4.5 trillion liters of water. I'm told that's enough to fill 1.8 million Olympic-sized swimming pools. That water has to come from somewhere. And if China's going to embark on a boom of constructing these things, it needs water to do it, right? [00:17:36] Simone McCarthy: Yeah, I mean, so this is a question that I really tried to dig into when I was preparing to go on this trip to Inner Mongolia, because obviously, as we said before, this is on the edge of the Gobi Desert. Of course, it's going to be a very dry area with water scarcity. So I rang up a bunch of experts talking about this. One of the things about this specific area, as well as several others that are in this overarching plan that China has, is that these are just cool areas. I mean, most of the year, you don't really need a lot of cooling because it's just really cold up there. But the other part of this is that newer technology, as far as I understand it from the experts that I was speaking with, allows you to not use nearly as much water as a lot of these older data centers that are really water guzzlers. And we see this issue in the United States coming up again and again and again. So, I mean, this facility that we went to, that we visited in Oolongkab, which is from the green technology company Envision, the way that they did it was they are using sort of an industrial alcohol instead of water for the cooling system. And they were telling me that each building that they have, which off the top of my head is 300 megawatts of computing power, is only using about a bathtub of water. And that, it sounded like, was more from the bathrooms and the kitchen and things like that, not the actual cooling facility. They're calling it a zero water cooling system. So, the technology is there. It's just a matter of whether or not it's being implemented. And I can't speak to what all of the data centers in that region are doing, but it does seem like there's a lot of focus on that highest technology. [00:19:07] Ivan Watson: You spoke to an executive from this company. Let's hear what he told you. [00:19:10] Speaker 4: This campus is our pilot project. We try to demonstrate all the technologies so that we can replicate to the rest of the world, maybe Latin America, South Africa, all the places which requires for AI and energy shortage. We can all bring this kind of solutions to them. [00:19:29] Simone McCarthy: And one of the things that you hear in China, too, which I think is really interesting and relevant to this conversation is that a lot of Chinese business people, pundits will say, you know, we have the technology that could solve a lot of the problems in the U.S. in terms of these energy constraints, but we're not able to export that. And of course, that's, you know, coming from a perspective of people who want to be exporting that into the United States. But it is part of the conversation here. [00:19:55] Kunal Sagal: I think what's really interesting about this story is that we tend to think of the U.S.-China AI race as a competition between technology companies and algorithms and the most advanced chips. But what we're seeing is like sort of a different side to it is that some of the things that could potentially ultimately determine who wins this race are much more basic, which is land, electricity, water, and the ability to actually build things. The U.S. obviously has significant at the moment, at least, technological advance. But it's discovering that having the best technology and the chips doesn't necessarily mean that you can build the infrastructure to support it quickly. And China, meanwhile, like you said, Simone, is betting that its ability to move at scale with the backing of the state can help it close that gap. [00:20:43] Simone McCarthy: The other point is the U.S. still is putting a lot more money into this than China is. So we've got chips, we've got money, we've got energy on the other side, we've got the ability to build infrastructure. We're gonna have to see in the coming years how this all plays out because the strengths and the weaknesses of each side are totally different. So it'll be an interesting one to watch. Maybe we can check back in in a year from now and see what's happening. [00:21:12] Ivan Watson: - Great, well, Simone, I'm so glad you were able to come and explain the fruits of your labor and your visit there. We're gonna start wrapping up here. But before we close, I think it's really important to kind of send a message to our colleague over there across the way of happy birthday. [00:21:30] Simone McCarthy: - Happy birthday! - Happy birthday! [00:21:32] Ivan Watson: - Our colleague Deepak there. Come on, pick it up. Show us the cake. Make a wish. - Nobody wants to jump in the shot. - Happy birthday, Kunal! [00:21:45] Kunal Sagal: - Thank you, guys.