Nuclear Silos in the Chinese Desert

Jeffrey Lewis 0:02

Maybe it's time to think about arms control. You are listening to the Arms Control Wonk Podcast, a podcast on arms control, disarmament, and nonproliferation. I'm Jeffrey Lewis, the founding publisher of the Arms Control Wonk blog, and a professor at the Middlebury Institute of International Studies at Monterey.

Aaron Stein 0:24

And I'm Aaron Stein, the Director of Research at the Foreign Policy Research Institute. It's been a weird week, Aaron. You were texting me at 4 a.m. West Coast time, and I think you were alive and kicking. It wasn't like I got up early.

Jeffrey 0:39

No, I had insomnia. I have, for the first time in two years, a cold, which I think is because I'm stressed out and run down because after the Iranian space launch story, we got another one, Aaron.

Aaron 0:58

So I thought you were up all night excited and giddy. I only found out you had a cold when we called this morning. But you want to tell the listeners what you got?

Jeffrey 1:06

Well, by the time this drops, they will have already read in the Washington Post that China, a country which as of September 2020, according to the Department of Defense, had about 100 ICBMs, is building a giant facility with 119 at last count ICBM silos.

Aaron 1:31

So, you know, that's like a thing. You've made news. Now, I had heard about this at 3 a.m. West Coast time, so about 7 a.m. my time, or I guess 6 a.m., 7 a.m. my time. This isn't the first reports of Chinese ICBM silos, but something tells me that you have more to add to those reports.

Jeffrey 1:54

Yeah, so we've been watching this for a while. A couple years ago, Catherine Dill noticed the Chinese building, a test silo, and Hans Kristensen at the Federation of American Scientists has done a great job documenting different silos under construction at a place called Jilantai, which is like a – it's really considered a training center. So if you look at what DOD says about it, they say it's – You know, a place where they're developing con ops, concepts of operations for their ICBM force.

So even though they had silos, it's like not necessarily the place. So, you know, like we knew that silos were a possibility for the DF 41, their new big solid propellant ICBM, but it also gets carried on, you know, vehicles and they might be looking at rail. So, you know, like silos were a possibility, but we did not expect to find 119 silos in the desert.

Aaron 2:51

So where, first of all, where are they? Cause I don't think they're at the same place.

Jeffrey 2:54

No, so they're at a place called Yumen, which is actually not very far from China's original plutonium production reactor, although it is best known today as being the site of a large wind farm.

Aaron 3:10

Well, we know wind farms aren't good, right? Yeah. What's worse, a wind farm or an ICBM silo?

Jeffrey 3:16

I don't know. I just like you have both sides of your attitude about the environment. On the one hand, a nice, clean energy source for a better future. On the other hand, many, many megatons of death and destruction.

Aaron 3:33

So should we talk about how you and the team and point out one member in particular found these silos before we talk about what it means?

Jeffrey 3:43

Yeah, there's no I in team, but there's no team in Decker. He's a podcast veteran. He's been on the show. I love Decker. Right. What happened was, listeners should know Decker Eveleth. We've had Decker on the podcast before. Decker is an undergraduate at Reed. We hired Decker last summer to do an ORBAT for China's strategic rocket forces.

He had already done his own ORBAT, but in the great tradition of things, I like giving young people money. And so we brought him on as a summer intern, paid him a little money, and had him tidy up that work. And then Decker actually was admitted to Miss. And so he was going to be an incoming student.

And in the interim, right, just talking to Decker, you know, I'd heard these rumors that like, you know, maybe silo construction was going a little faster than we realized. And maybe Hans hadn't gotten them all. And I wasn't the only one who heard that. Like I had this conversation with some friends in D.C. and they had kind of heard the same thing.

And so I suggested to Decker, like, hey, man, there are these rumors floating around. They don't strike me as being crazy. Why don't you take a look? And so he decided he'd take a look at Jilantai and look for signatures, and he'd go through China and see if he could find things that matched. And holy crap, did he find things that matched.

Aaron 5:17

We were talking about this off pod, but it's one of those incredible things where with a little Oversight, a little bit of help, you know, the proliferation of capable people to do this type of analysis is really spread far and wide. And it's quite incredible, just basically what you can do from your home computer and the right types of accounts that one can have access to to get update satellite imagery.

Jeffrey 5:41

Yeah. And, you know, Decker is in pretty regular contact with Scott LaFoy and with me. And so, you know, while I like to think that Scott and I have Helped Decker grow, we should just be upfront that there's just a lot of native talent there. And that if you replace Scott and I with less charming versions, Decker would probably be equally successful.

But he's smart and he's really dedicated. He's very diligent. Because this stuff can be kind of tedious sometimes. Extremely tedious. And like, I, I think once we were talking about his search method and he like gridded out inner Mongolia at one point and just was like powering through it block by block. So that's, that's one piece of it is just talent and dedication.

The second piece is that I got Decker a planet account. Friend of the show. And this is, this is a really important thing because. You know, some people don't really see the value in planets, moderate resolution imagery, the three meter stuff.

Aaron 6:45

Yeah.

Jeffrey 6:46

But if you ever look at one of their images at three meters and one of their high resolution images, you know, a high resolution image is just going to really be like, You know, 10 square kilometers, right? It's going to be like a really tiny space. So if you think about it, high resolution images are great because you can see stuff in great detail, but you are staring at the earth through a straw.

Yeah, definitely. And you have to know where to look. Right. And that's great when you know where to look, like I want to count cars in a parking lot, but it is not great if you are doing wide area search.

Aaron 7:21

Which looking for 119 ICBM silos requires.

Jeffrey 7:25

Right. And so the way the silos are constructed, and we can talk about this a little bit, but they have these large inflatable structures on them.

Aaron 7:33

Inflatable?

Jeffrey 7:35

I think they're inflatable. I mean, they look inflatable to me. I mean, maybe I'm wrong. But they, I mean, I call them the, I call each of them the bouncy house of death.

Aaron 7:44

I was about to say the moon bounce of death, but okay. Yeah.

Jeffrey 7:46

Yeah. Moon bounce of death. That's good.

Aaron 7:48

Yeah.

Jeffrey 7:48

And so those are big enough that you can see them in three meter imagery and But you can also get a three-meter image of all of China every couple weeks, right? That's the value, yeah. Right? Because you're getting an image every day of almost every place. Sometimes it's cloudy. Sometimes you miss some. But... So Decker was able to go through the three-meter image pretty efficiently, spotted the structure, and was like, that looks like the thing we see at Jilantai.

And then I went ahead and went to plan it and was like—

Aaron 8:25

Please, sir, take some high resolution pictures.

Jeffrey 8:29

And, and boy, did they deliver, right? So, so, you know, what we end up with is these stunning high resolution images of these really important parts of the facility. But we couldn't have done that unless we'd had access to the three meter imagery. And so like, it's these two parts, it's you have to have a decker, But you have to give Decker the tools he needs to succeed.

Like if you had turned Decker loose on this problem 20 years ago, he was hosed.

Aaron 9:00

Right. Or even with just basic Google Earth. Right.

Jeffrey 9:03

Oh, yeah, because Google Earth doesn't update enough. I mean, one of the big things, I guess the first thing we'll say about this site is all of the construction has occurred after February. Oh, really? So it's happened really fast. And honestly, the facility is not done. It's still in what I would call a fairly early stage of construction.

But waiting for Google Earth to update it, it could have been a year or two. Totally. Because this thing is in the middle of nowhere. This is not a place that is a priority for Google Earth to update.

Aaron 9:34

Right. And I assumed it was in the middle of nowhere. I'm just trying to think what... I mean, 119 silos is a lot. I mean, it's a lot more than the training facilities that we're seeing. And so, I mean, do you think that this explains perhaps some of the hand-wringing we were seeing towards the end of the Trump administration about the growing Chinese arsenals or some of that stuff going on?

Or that's like a disaggregated thing that was going on beyond the ICBM silo construction?

Jeffrey 10:05

I don't know. And the reason I say I don't know is because all the construction happened after February. So this thing was probably in planning, right? At the end of the Trump administration. But did they know that? Did they have access to those plans? In which case, it would explain it. Or maybe it didn't, right? Maybe the hand-wringing was just happening, and then now the 119 silos have shown up.

But one thing I would say, we should probably talk about the signatures that we're seeing. Yeah, definitely. One thing I do want to say is that It is not clear that 119 silos means 119 missiles. And that's a really important clarification that goes to the heart of something Scott loves, which is MX Basing. Yeah, he's so upset he's not on this podcast, by the way.

We'll do two, because we have a kind of emerging hypothesis about this facility, which is... Just a hypothesis. We actually really need to see them build some more stuff before we get firm opinions about it. and so I think we'll come back and revisit, but but just before before we get there, I I think we should probably talk about what I think we should probably talk about what we see. Yeah, so what were the signatures that tipped both you and Decker off to this site as you were watching it?

Right. So the main signature that Decker used was the Bouncy House of Death, this large inflatable structure that we had seen at Jilantai, which was, you know, it's a missile base. And there are images there that show that before the structure goes up, they've started a circular excavation that looks like it's for a silo. And then that, you know, boom, up goes the Bouncy House.

So We were pretty confident that that was a good signature. And then when Decker found it, we were still pretty careful. Because, you know, it's right next to a wind farm. And while it doesn't look like a wind farm, people do bad OSINT all the time. And so we wanted to be super careful. So we looked at the security features at the site.

Like there's some funny roads that are clearly designed to make it hard to go. They built a military base kind of in the center of it, which we think is like a headquarters support area. The... Bouncy houses, which hold the silos, are themselves in triangles three kilometers apart. So you get this kind of triangle or diamond-shaped grid pattern, which is unusual.

I mean, I'm jokingly calling it a bouncy house. But if it were a bouncy house, you wouldn't have 119 of them.

Aaron 13:06

Unless it's one hell of a kid's party, right?

Jeffrey 13:08

Right. Well, and then you wouldn't put them three kilometers apart, right? Right. It has all the signatures of a certain kind of ICBM field. And then the part I really like is there's at least one facility that's exposed that's like a cut and cover bunker. And it looks like it's going to be for like a launch control facility.

And it's got 10 trenches for cables leading out of it radially, each going to a different silo launch site.

Aaron 13:42

Right.

Jeffrey 13:43

So, I mean, that really is like a very classic signature. You know, you have this large repetitive pattern of things that we got some pictures with the bubbles off. Decker and I made sure that we tasked against that. And so you can see the circular excavation for the silo before the bubble goes up. You have the bubbles connected to the launch site at Jilantai.

You have the pattern. And then you have the fact that each of those launch sites is connected in groups of 10 to what looks like it's going to be an underground control facility. I mean, this is not a theme park.

Aaron 14:24

No. No.

Jeffrey 14:25

So fun.

Aaron 14:27

And does that field remind you of, of say like, like Soviet or Russian fields, or is it unique to China? Does you see similar ones in the US? Well, so this is where, this is where things get weird.

Jeffrey 14:39

It looks almost exactly like the layout that the US was examining for the MX missile when they were going to do, they called it mobile, what was it? It was mobile protective shelters, the shell game. Right. Where they were going to build 23 silos for every missile.

Aaron 15:01

Right. I think we have a podcast on this.

Jeffrey 15:05

Yeah, they're going to shuttle the missile from initially it was they ended up choosing horizontal shelters. So this is like sometimes hard to explain, but they did consider silos as well. Bill Perry, actually former Secretary of Defense Bill Perry, back when he was a slightly lower ranking Defense Department official in the 70s, calls them vertical shelters and then clarifies like a vertical shelter is a silo.

Right. So the pattern of that layout in China is so evocative. Because if you look at how China used to lay its ICBM silos out, they were very alone. Right? Huge distance between them. These are like packed together. I mean, they're spaced far enough apart that you couldn't kill two with one nuclear warhead. But they're... They're still sort of packed together.

And like the reason you would do that is if you were shuffling the missiles between the silos.

Aaron 16:10

And you should explain to listeners why that would, at least why the U.S. is perhaps, because I know that this is the hypothesis, why that this would be advantageous.

Jeffrey 16:19

Right. So missiles are expensive. And vehicles that carry missiles are expensive and silos while they're also expensive, they're not quite as expensive. And so if you're going to build 200 missiles you can get a big jump in survivability if you build a much larger number of silos, because your opponent, if you do it right, your opponent doesn't know which silos are armed.

Exactly. And there are questions about how you maintain that fiction. Do you really want to shuttle them? Do you just want to fake your maintenance patterns so no one ever knows? But the idea was to present the Soviets with more than, what was it? It was like 4,000, 4,600, I think, to give the Soviets 4,600 targets. And that if they wanted to kill the 200 missiles, they would have to expend everything going after those shelters.

And even then, they wouldn't get them all. So, I mean, that makes a lot of sense for China, too. Where construction is cheap.

Aaron 17:29

Definitely. And where we know, despite sort of the reports and despite this finding that you have, is that they still have a comparatively smaller number of nuclear weapons than we do.

Jeffrey 17:40

Yeah. And, you know, I don't know what the cost trade-off for the Chinese is. But, you know, if you can buy two silos for every missile, that's a pretty efficient trade, right?

Aaron 17:54

Yeah.

Jeffrey 17:54

You can have... You could either have half again as many missiles or twice the number of targets. Right. So you it it makes a lot of sense. You know, it was unpopular in the US because four thousand six hundred was excessive. And that was going to be a lot of land in the West. And strangely, you know, the people who it really made angry were like the Clive and Bundy types.

Aaron 18:21

Right, exactly.

Jeffrey 18:23

It was like ranchers who were like big Reagan people and they were like, yeah, we're going to eminent domain the fuck out of your ranch. And by the way, we're going to use all the water.

Aaron 18:34

But you're a good patriotic American, so you're okay with it, right? And they were like, no, we're not. And who wants their backyard to be a name point, right?

Jeffrey 18:43

But in China, you don't have pesky environmentalists or reactionary ranchers. I mean, China did have reactionary farmers, but that problem was solved, you know, some time ago. And yeah, I guess we shouldn't make too much light of the repression of the Chinese state apparatus. But the point is, the thing that killed it in the United States is not a problem the Chinese have.

So it does make sense for them. And, you know, we know they considered it. John Lewis and Hua Di wrote an article on China's missiles about... 25 years ago. And Huadi actually went to prison for this article. The Chinese government was so happy with the disclosures he made. And one of the things he disclosed is that they knew the DF-5 silos were very vulnerable, and they examined very closely the U.S. plan for MX.

And thought that there was something to the idea of deceptive basing, but then chose a sort of easier, less expensive solution of just building some decoy silos.

Aaron 19:45

What do you make of the hundred? I mean, so the assumption is, it's just hypothesis, but let's call it an assumption for now. That doesn't necessarily mean that we see a big change in the configuration of the nuclear forces, does it? I mean, I'm not really an expert on this. You wrote the book on this. I think it's called Paper Tigers.

Jeffrey 20:06

Yeah, yes and no is my answer. One thing I always try to caution people against is, and I'll be honest, I struggle with this balance myself. We always want to explain China's nuclear posture in terms that are familiar to Americans.

Aaron 20:20

Yeah.

Jeffrey 20:21

And sometimes, and I'll admit this happens when I'm doing it, when we're describing China's nuclear posture, they sound like heroes of minimum deterrence. And they're not really heroes of minimum deterrence. They have their own strategic concepts that are different than like the nuclear force I would design. And so... What China's doing now is definitely inconsistent with the kind of idea that, you know, the Chinese force is, you know, this extreme version of minimum deterrence.

On the other hand, you know, for a force that is, I would argue, fundamentally retaliatory in nature— And intended to limit U.S. nuclear coercion more so than engage in its own coercion. This doesn't really change that. And if you were to deploy 12 missiles in 120 silos or 120 missiles in 120 silos, I don't think it would dramatically change it.

You know, the Chinese numbers are still in the kind of hundreds, right? Few hundreds. They're not in the thousands. And, you know, a force that is vulnerable is necessarily going to push— You to feel pressure to use those weapons. And so if China has a no first use policy, then building a large number of deceptive a large number of, of, of, of silos, right.

And, and, and engaging the shell game is a way to reduce that vulnerability. So I, I don't think it represents a dramatic change in their posture. I think it's sort of consistent with what they've done in the past, but with better technology, larger numbers, and more money.

Aaron 22:29

And this may be unrelated. By the way, I'm just asking questions because I'm fascinated by this. We talked in a couple of, God knows how many episodes ago, but we talked about the hypothesis that they were MIRVing missiles as well. You know, do you think this fits in here? I mean, like, I really am just asking questions because I'm fascinated by all this.

Jeffrey 22:49

I mean, yes, the DoD thinks the DF-41 is going to be MIRVed. And there's no reason that it wouldn't be. It should be bigger. We don't have great open source information on it. So I kind of sat down and did a back of the envelope calculation. So here's all the data we have on MIRVing. The usual suspects say 10 warheads, which seems unlikely because it's not as big as peacekeeper and peacekeeper MX had, had 10 warheads, but whatever they say, 10 the usual suspects claim there are classified intelligence reports that say six to 10 warheads.

That also still seems high to me, but it's certainly possible that that's the real range. The actual tests that have been reported, the leaked intelligence reports say two warheads in the tests. So there's a lot of distance between two and 10. Yep. You know, they might be testing them with two because they don't want us to know whether it's three or four or five.

So, you know, I'm not saying that because they tested them with two, it is two. I just, I got some people scream 10. Some people say six to 10. The number tested is two. And what we can do is we can do sort of back of the envelope calculations. There are some open comments from Chinese people who probably know about the approximate weight of the thing, which is like 60 metric tons.

So if you take a US MX slash Peacekeeper and scale it down to 60 tons and figure out how much throw weight that should be about, you get a number. I try not to do math openly on a pod. But you have a number for how much you can carry. But then you have to divide it by how heavy do you think Chinese warheads are.

Aaron 24:40

Right.

Jeffrey 24:42

So if you divide it by how heavy... What is the mass of the last warhead that they tested in the 90s? That leaked, and that's like 500 kilograms. 470, I think. And so that's only enough for three or four warheads. If you think like, well, that was 25 years ago, and they're willing to use computer models, and they could maybe deploy something as slim as a US warhead, because kids these days don't need tests.

They're happy to live with computers. Then that number jumps up a little bit, right? So then that number kind of gets up into that like six neighborhood.

Aaron 25:28

We could see why there might even be a range because of the number of possibilities, and we just don't really know. Right. So I mean, I, the number could be as low as two.

Jeffrey 25:39

I don't think it's much above five or six. I'd be I'd be surprised. But you know, it could be that like the our estimate of how big the missile is is wrong. Yeah, okay. So I actually I pulled up my I pulled up my spreadsheet. It's If you just treat it like a mini peacekeeper, it's three to four with a Chinese warhead and five to eight with a modern US warhead.

That makes sense. So, you know, meh. I'm guessing it's probably not eight and it's probably not two. So that changes it, right? Because let's say, if you put your thumb on one side of the scale and you say, it's 120 missiles in 120 silos with 10 warheads, that's 1,200 warheads. Or if you say, well, it's probably three warheads on 12 missiles.

That's 36. That's a big delta. Right. Definitely.

Aaron 26:48

The last thing that makes me think about is, Whether or not there's thought over there about whether their mobile missiles are vulnerable to conventional counterforce. And so the silos become a way to protect themselves from conventional counterforce. You see this within the context of the arguments for the collapse of the INF, the return, the eventual return of some sort of missiles, conventional missiles to Asia on the U.S. side, and whether silos are a way to protect against that.

Jeffrey 27:21

Yeah, it has a very American quality to it. You know, we constantly hear how important it is to have these ICBMs as warhead sinks.

Aaron 27:28

Yeah.

Jeffrey 27:31

Like, well. You know, if you are worried about either numerical advantages in terms of nuclear forces or growing conventional forces, then it does make sense to say, like, well, one problem you have to deal with is going after the submarines. Another problem you have to deal with is going after the mobile missiles. And a third problem you have to deal with is...

The simply mechanical task of covering all these targets out in the desert.

Aaron 27:56

Right. It's just adding aim points, right? And adding aim points, which in some respects, you know, going back, I think, to the two pods ago, right? Which crosses the Lewis and Sagan rule where these targets probably can't be destroyed with conventional weapons.

Jeffrey 28:16

Yeah, you know, I don't know. I've been thinking about that. One thing I did, not very usefully, is I looked at the distance between the shelters, and they're a little too far apart. And the only reason you'd put them that far apart is if you were worried that they weren't very hard.

Aaron 28:38

Oh, yeah, good point. Is... And if the MX basic model hypothesis proves to be accurate.

Jeffrey 28:47

Well, spacing is spacing. I mean, even if they're all alive, you don't want them so close that a single warhead could take them both out. No, totally.

Aaron 28:53

Yeah.

Jeffrey 28:54

And so I don't quite understand. Three kilometers is like a little much. Now, having said that, like, well, I mean, what does it really cost them to spread this stuff out in the Chinese desert? Probably not that much. Maybe they're just being extra cautious. I don't know.

Aaron 29:12

Well, so the question is, what do we do about this thing, right? Yeah. I mean, that's usually how we end it, but it's concrete in the desert. What are you supposed to do? Well, maybe it's time to think about arms control. So I was going to ask you about that, but they just don't seem to have any willingness to engage in arms control.

When I read it, again, somebody who really doesn't grapple with these issues, is that they're just not ready for it. Well, we could ask why that is.

Jeffrey 29:45

This is my regular plea. I think the Chinese have been extremely clear for the past 30 years that they are not interested in discussing limits on their offenses unless there are corresponding limits on defenses. Which to me makes perfect sense. In the 1990s, the Chinese said it in the conference on disarmament all the time. They had a whole bunch of discussions about how there had to be some kind of limitations on defenses.

It was the argument they gave for why they wouldn't agree to a fissile material cutoff treaty. It's like, well, we don't know how much fissile material we're going to need because we don't know how big your defenses are going to be. They kept it up through the 2000s and the 2010s. And I think they still say it.

And I am not saying that it would be easy. Like, I think if we showed up saying, hey, we're finally willing to talk about defenses as part of this equation, I think there'd be a lot of people in China who would, you know, react with the same frustration that like Trump administration officials would have reacted if the Chinese had agreed to their proposals, which were not made in good faith.

Right. So I'm not saying that everyone in China would be delighted if we took their proposals to be in good faith, but I think ultimately... If we want to get a handle on this arms race, the answer now is the same answer that we that stared us in the face in the late 1960s, which is you have to limit defenses.

And I understand that it's impossible and it's dead on arrival in the Senate and Republicans will never go for it. But if you really believe that, then this is like school shootings. Right. We know what the answer is. We know what we're supposed to do about the problem. We've just decided that it's too hard. And so like some kids are going to die.

Except for the difference is, you know, the risk is like now it's not some kids are going to die. Like it's the risk is we all die.

Aaron 31:47

So we die faster than we normally would have. Yeah.

Jeffrey 31:50

Like to me, this is the arms race that we are in the middle of is a predictable consequence of leaving the ABM treaty and refusing to discuss defenses. All of these unwelcome developments in Russia and China are a consequence of that. And if we want to address them, I mean, even if you don't think they're exactly a consequence of that, your capacity to talk to them about limits is Is zero if you're not willing to talk about defenses, and I would be, but they also said that they want our numbers to come down as well too. Oh, I'm sure they have lots of lots of things, but I just think when you say my my read is the Chinese aren't interested at a fundamental level at my my my observation on that is because the thing they care about most of all is not on the table for them.

And so until that thing is on the table, they're going to just build their way to security, which, you know, I think is stupid because I don't think you can really build your way to security. But that's an arms race for you.

Aaron 32:55

You think we're in the middle of one. I think we're at the beginning of one. Well, I think we both think we're in the early stages. Early stages.

Jeffrey 33:01

Yeah. But, you know, here's one other thing. I mean, I'm like a broken record, but. You know, our numbers today are quite comparable to those during the Cuban Missile Crisis. Russian forces are bigger than during the Cuban Missile Crisis. So yeah, numbers are not what they were during the height of the Cold War. But if the Cuban Missile Crisis was scary to you, there's plenty of reason to be worried today.

Aaron 33:25

That's a good point. That's a good way to frame it because I don't think about it like that, but you're right. By the time this episode drops, this will have already been in the news, but in case people were stuck under a rock where would they find it? The Washington Post and your op-ed, which will be linked on your Twitter feed.

That's the plan. All right. So with that, Jeffrey, I know you got another call. I have another call. So thanks for doing this. If you like this, I'm sure the Slack channel will be eagerly involved in this, debating hypothesis about the MX missile hypothesis, as well as the open source imagery. And Decker is in the Slack channel, and he's the man, and he will be another proud Miss Alumni like me.

So head on over to patreon.com slash ACW podcast, $3 if you're a student, $5 if you're a normal person, and you can join that conversation. So with that, have a good one. I got to bounce. Thanks, Aaron.