The world is facing a global memory chip shortage, with companies scrambling to buy what they can. The crunch puts desperate U.S. firms in a bind: will they need to start purchasing chips from China?

In this episode, reporters Rachel Cheung and Noah Berman discuss the rise and global allure of China’s homegrown chipmakers. Even though the U.S. government designated two of China’s biggest suppliers as military companies and subjected them to export controls, looming chip shortages mean purchases from China aren’t off the table just yet.

We also feature an extract from our Q&A with author and historian Jeffrey Wasserstrom on the importance of understanding Chinese history.

Start your subscription to The Wire China today! We’re making it easy with your first month for just one dollar. Sign up for our newsletter here and we’ll send you the discount code to apply at checkout.

Transcript

Rachel: Hello, and welcome back to The Wire China podcast. I’m Rachel Cheung, staff writer at The Wire China, and will soon be joined by my fellow reporter Noah Berman to discuss if China could be the answer to the global crunch in memory chips.

We will also have a Q&A with Jeffrey Wasserstrom, a history professor at the University of California, Irvine, who has just published a handy primer on contemporary China called Everything You Wanted to Know About China But Were Afraid to Ask.

Before we go into memory chips, we are currently offering a discount code for new subscribers to The Wire China. To get your first month for just $1, sign up for our newsletter at the link in the podcast description, and we’ll send a discount code your way.

Great to have you on, Noah.

Noah: Thanks for having me.

Rachel: Well, let’s start with a bit of basics. Why is the world running out of memory chips, and how big of a problem is this?

Noah: Well, it’s a classic mismatch of supply and demand. In short, the answer on the demand side is AI. Demand from the companies building advanced AI, companies like Nvidia and Amazon and Google, is just through the roof.

You need memory chips, particularly an advanced version of them called high-bandwidth memory, to power AI data centers, and demand for high-bandwidth memory, or HBM, is growing in the triple digits. Supply, on the other hand, is growing at less than 20%. So the result is a shortage, in particular for the types of chips that go into everyday electronics, like laptops and smartphones, because the makers of these memory chips are pivoting to sell more profitable HBM than the types of memory chips that go into those commodity electronics.

And so what we’re seeing is that prices are growing rapidly. They grew more than seven times over in the year up to February, and that’s largely because of this mismatch.

Rachel: So we know for advanced AI chips, China is still playing catch-up with the U.S. What about memory chips? What role do Chinese companies play in this segment at the moment, and how competitive are they compared to industry leaders like Samsung or U.S. Micron?

Noah: So there are two big memory chip champions in China. Both are state-owned enterprises, and they’re still not huge players in the global market, but they are both gaining ground pretty rapidly. Those companies are YMTC and CXMT, and they don’t directly compete with one another. YMTC makes a form of memory called NAND or NAND flash, and CXMT makes a form of memory called dynamic random access memory or DRAM.

YMTC has about 11 percent of the NAND market. CXMT has about 5 percent of the DRAM market. So of course, that leaves a big piece of the pie left.

People that I spoke with for the story, experts on the industry, say that the so-called big three, those two companies that you mentioned, Samsung and Micron, as well as a third Korean company, SK Hynix, are still in the lead as far as technology goes. But these companies are pretty openly concerned about the threat posed by the Chinese champions, and Micron even specifically called out YMTC and CXMT in its most recent quarterly filing.

Rachel: You mentioned CXMT, and in China, there’s a lot of excitement about this company because it just received approval for an upcoming listing in Shanghai, and it is expected to be China’s biggest IPO since 2022.

So tell us about the momentum behind this company. And right now, its share of the market is mostly just in China, or does it also have global ambitions?

Noah: Like you mentioned, there’s enormous momentum behind CXMT right now. This IPO could be huge.

The company is looking to raise at least $4 billion. From what we know, it’s too early to say what the valuation of the company would be, but some estimates peg it at over $100 billion U.S. dollars. That would make it one of the 25 largest companies in China by market cap.

So this is really set to be a major offering. And from what we can tell from the IPO prospectus, CXMT is benefiting largely from China’s push for self-sufficiency. CXMT said it made $7.5 billion in sales in the first quarter of this year, which was eight times more than it made in the same period last year.

And its profit increased by more than 17 times. So huge increases there. And almost all of those sales, 97 percent, are coming from China and Hong Kong.

So these are Chinese companies driven by government incentives that are buying up CXMT’s capacity. As far as their global ambitions, the company’s prospectus is pretty clear there, too. It says the company is focused on the domestic market, but it adds that a gradual expansion overseas is coming.

And so we’ve seen what happens in other industries when Chinese companies go global. When it rains, it pours. And that could be why companies like Micron are voicing concern about this in their filings.

Rachel: So some have predicted that the supply crunch will persist till next year, even as far as 2030. Given this predicament, are companies like Apple and Dell turning to Chinese players? Is that a potential solution? And what do you think are the concerns or sort of drawbacks there?

Noah: There are reports that Apple and Dell are among several makers of electronics, things like computers and smartphones and TVs that are looking to potentially add these Chinese companies YMTC and CXMT to their supplier lists. And people who I spoke to for the story said there’s particular interest in CXMT.

The problem for U.S. companies is that both those Chinese champions are designated by the U.S. government as Chinese military companies. And so buying from them could have significant political costs. In the past, when reports have emerged about U.S. companies sourcing from YMTC in particular, it attracted scrutiny from politicians in the United States who expressed extreme concerns.

One of those politicians was Marco Rubio, then a senator, now the secretary of state. A bigger question, though, is does that matter right now? Someone who I spoke to for the story, who’s an executive at an electronics distributor, said companies are concerned about the political costs. But right now, the memory chip shortage is so severe that they’ll pretty much take what they can get.

On the other hand, there’s export controls. So long term, companies may be hesitant to lock in deals with CXMT and YMTC because there’s still some degree of uncertainty about whether their products will be competitive if they can’t buy the tools that they need. The U.S. has export controls on tool sales to CXMT and YMTC and many other Chinese companies.

So it’s an open question of whether those will begin to bite or if CXMT and YMTC will be able to innovate around them.

Rachel: I have one more question. We know there is a global memory crunch. Is there one in China? And if there is, does that mean these companies’ priority will be the domestic market?

Noah: You’re right to mention that, Rachel. The same dynamics that are playing out in the United States are playing out in China. And so that’s another obstacle that U.S. companies looking to expand their supplier lists will have to overcome is even getting supplies from these Chinese memory chip makers whose products are being snatched up by the Chinese AI developers, the so-called hyperscalers, companies like Alibaba and ByteDance.

And so that will be another challenge for U.S. firms.

Rachel: Great stuff. Thank you, Noah, for joining us today.

You can read his story today on thewirechina.com. Coming up next is a clip from our interview with Jeff Wasserstrom on the importance of understanding Chinese history.

Jeffrey: I think it’s crucial to both know something about the arc of Chinese history and also know something about the spin on that arc put by the Chinese Communist Party. Because I do think that, you know, the way historical narratives are told, it really matters.

They’re important in efforts to legitimate governments. And you can’t have a critical distance on them unless you have some degree of knowledge about that. Otherwise, it’s too easy to just take at face value the story that the Chinese Communist Party tells about the country’s past.

Yeah. And there’s some odd misunderstandings about Chinese history that the Chinese Communist Party embraces and opponents of the Chinese Communist Party embrace as well. Both the Communist Party and some of its direct opponents buy into this idea of 5,000 years, relatively continuous story of Chinese civilization and a kind of singular view of that, which either the Communist Party represents or the Communist Party has been bent on destroying.

So I think it is really important to break away from that whole dichotomy and think there is actually this story of Chinese history is full of various strands being pulled together and recombined in different ways by different kinds of power holders and different kinds of people challenging power holders.

Rachel: You can catch the rest of Wasserstrom’s perspectives in our upcoming issue, published Sunday evening Eastern Time on our website. Now, Kate MacKenzie from The Polycrisis podcast joins us to discuss China’s role in the emergence of what she has dubbed the Electric World Order.

You can listen to more on what’s happening to the green transition in the midst of the Iran war energy shock on new episodes of The Polycrisis podcast.

We’ll leave you with Kate’s insights, and thank you so much for listening.

Kate: China’s played a critical role in what we call the emergence of the Electric World Order, which is a kind of a quiet revolution in energy transition that’s almost happening below the radar.

It’s a little bit hard to trace in the data, but a lot of countries are actually switching their energy systems in different ways towards more and more renewable energy supply. So it’s a huge change and the volume of exports of these goods out of China is just immense. The thing is that this wasn’t really a grand plan from the Chinese central government, you know, to kind of export clean power and electrify other countries. This is almost like a side effect of China’s own economic policy, of its own industrial strategies.

So the new season, we’re calling it a bonus season. It’s actually looking at the implications of the Iran war.

This is the biggest energy shock that the world’s ever experienced, because this is the first time we’ve had a big energy shock where there was also this option of actually switching out big parts of national energy systems, you know, away from oil and gas dependencies and towards like a more electrified system. A lot of countries are now importing a lot more Chinese clean tech gear. That’s then inevitably like prompting questions within some countries about, oh, are we just swapping out one dependence for another? Are we just swapping out exposure to an oil and gas chokepoint to exposure to a solar panel chokepoint?

Move over, DeepSeek and Minimax. Investors are flocking to a new set of AI winners: the Chinese manufacturers making circuit boards and other components that keep the world’s AI models humming along.

In this episode, reporters Rachel Cheung and Savannah Billman discuss the growing importance of these so-called ‘pick and shovel’ makers and where the money is flowing within China’s AI ecosystem. The AI industry is fraught with risk for developers and investors, but the manufacturers will win no matter which LLM comes out on top.

Start your subscription to The Wire China today! We’re making it easy with your first month for just one dollar. Sign up for our newsletter here and we’ll send you the discount code to apply at checkout.

Transcript

Savannah: Hello, and welcome back to The Wire China podcast. I’m Savannah Billman, staff writer and podcast producer here at The Wire China, and will soon be joined by my fellow reporter, Rachel Cheung, to chat about the hardware manufacturers actually making money from AI in China.

But first, it’s been a busy two weeks in the China journalism world. The whole lot of us have been swept up in the hubbub around the Trump-Xi summit. We’ve seen a flood of pre-summit op-eds, minute-by-minute coverage from reporters on the ground in Beijing, and now the post-summit evaluations are rolling in.

This Sunday evening, we’ll throw our own hat in the ring. Our upcoming cover story will analyze what the summit did and didn’t mean, and where the leaders of the two largest economies in the world may take us next. As always, this story and more can be found on our website, thewirechina.com. If you sign up for our free newsletter, you can get each week’s edition right in your inbox, along with a discount code for your first month for $1.

This is a limited-time subscription offer for our listeners, so make sure you register for our newsletter at the link in the podcast description. You’ll hear more about the summit in the interview clip in the second half of this podcast, but let’s move right along to the meat of this episode. Love AI or hate it, someone’s making money off of it.

Here to talk about who that may be is Rachel Cheung.

I should also mention Rachel has been named a finalist in two categories of this year’s Society of Publishers in Asia, or SOPA, awards, Young Journalist of the Year, and Excellence in Arts and Culture Reporting. All of Rachel’s nominated articles are now free to read on our website. Rachel’s also won two previous SOPA awards, so our listeners are about to hear some truly high-quality insights.

Rachel: Thank you. Thank you for your kind words.

Savannah: So, Rachel, your story on AI is already live on our website.

Some people think of AI as living in the cloud in a land far, far away, but the reality is that AI lives in hardware, in nuts and bolts and circuit boards, what experts call picks and shovels. And as you write, Rachel, the pick and shovel makers are trending in China at the moment. Can you explain it to us?

Rachel: Globally, or at least you look at the US, hyperscalers are pouring billions and billions into AI infrastructure. Microsoft, Alphabet, Amazon, Meta, and Oracle alone have committed $600 billion in CAPEX — capital expenditure — this year alone. It’s a little similar story in China. It’s a fraction of what the US is spending, but it’s also increasing.

Alibaba, for instance, is committing 380 billion RMB into AI cloud and infrastructure over three years. With this huge amount of money flowing into the AI supply chain, a lot of investors are also looking at where the opportunities are at. And so because of this demand and how quickly it has increased, we’re now seeing some bottlenecks in the AI supply chain.

The most well-known one is memory chips. Demand for memory chips have soared and supply hasn’t really caught up, so we’re seeing shares of SanDisk, Micron, gone through the roof. My colleague, Noah, is going to have an offlede about what’s happening in memory chips this coming week.

But there are also other areas of the supply chain that are less scrutinized, such as printed circuit boards, optical modules, or the newer one, for instance, multi-layer ceramic capacitors, where there are bottlenecks, which is pushing up the prices. For instance, for optical modules, a lot of companies in the supply chain, they’ve made three-digit gains in their shares over the past year. Their revenue has gone up, and because of this demand-supply pinch, their pricing margins have also gone up.

A lot of these hardware suppliers, the most well-known ones, are Taiwanese and South Korean companies. But there are also Chinese companies, too. So in this story, we’re looking at some of the Chinese players and the global AI supply chain.

Savannah: So we have these big hyperscalers, Microsoft, Meta, Alibaba in China, just pouring in billions. And we have all of this hardware. Who are the Chinese companies that you’re watching in this area who are making this hardware?

Rachel: So one company we feature in the story is Victory Giant.

It was actually founded in 2003, and it’s based in Huizhou. So it’s a manufacturer of PCB, printed circuit boards. So it’s used in AI servers, for instance.

They count NVIDIA, Tesla, and Microsoft as its clients. It was listed in Shenzhen, and its shares in Shenzhen have gone up three times in the past year, despite a more recent pullback. It made a secondary listing in Hong Kong last month, where they raised 20 billion Hong Kong dollars, about 2 billion US dollars.

And that is Hong Kong’s largest listing so far this year. So this is one example of a Chinese company benefiting from the global AI build-out. So a lot of these components, some of them are AI accelerators, they’re AI servers, but essentially, they’re little different critical components within the AI supply chain.

And what I find interesting, besides some of these well-known and well-established manufacturers doing this, is that this component they’ve been making for years is now used in AI servers. They’re also newer companies and startups, and they have really thrived by finding pain points within the supply chain, whether it is a computing cluster is how they’re connected. And by providing a solution to this, they have really been able to sort of find their space in the market and get really good profit as well.

Savannah: So how does the business model of these pick-and-shovel companies compare to the businesses of the model developers themselves?

Rachel: This is where things get quite interesting as well. So for the pure-play AI vendors in China that are known really well, for instance, Minimax and Zhipu, they made their trading debut in Hong Kong in January, and since then, their shares have skyrocketed. This is partly because of the scarcity premium as well.

For people investing in the public markets, there are not a lot of choices for them right now. Later on, you will have StepFun and Moonshot AI that when they’re available, they’re also eyeing IPO. When they’ve also debuted, we might see that scarcity premium come down a little bit.

Another thing investors are watching is the lockup period. For cornerstones or pre-IPO investors, that period is usually six months. So after the initial six months is over, I think that is where the real test is gonna be.

But I think the question for these, especially smaller startups, is we’re seeing more volatility in their shares lately because of this growing question of how do you monetize the offerings? We’ve seen different strategies. Minimax is targeting the overseas market quite aggressively. Zhipu is looking at a lot of Chinese enterprise and government clients.

But I think this is a question people will keep asking, and it’s gonna weigh on the shares of companies like Alibaba and Tencent, these hyperscalers that have the full stack as well. In China, we’re now in the phase of showing investors the money, right? And so people are gonna look and see not just how fast your revenue is growing, but also at what point are you gonna make a profit? At what point is your investment in AI gonna pay off?

So the Chinese market is known to be very, very competitive. We’re already a long way from the early stage where we have the hundred model war.

And another point is that with the rise of AI agents, which could run autonomously overnight instead of handling queries, it’s going to do tasks and complete tasks. So they use way more tokens, and that has also helped these model companies improve their pricing. But back to the market and sort of how competitive it is, that also means we’re still seeing how strongly they could improve their pricing power.

I think there is still a question over that. And there are worries that when a new competitive model comes out, then the pricing expectations will change as well. And one more point on this is that last year, a lot of these Chinese companies were not yet competing at the frontier.

So when DeepSeek was launched last February, you might remember that it wiped billions off US stocks and US AI companies. If the same thing happens, what people call the model shocks, it is gonna pose a risk, not just to the shares of US companies, but also to the Chinese companies. So I think that is one thing investors are watching as well.

As for the hardware, the picks and shovels, the reason why they’re more attractive to some investors is that these companies will benefit regardless of which model wins. And it’s easier for investors to see what its future growth looks like because its monetization is already very steady. There are also expectations that they will have more of a technical moat compared to the software players.

For instance, to build, to make something, you need factories, you need to accumulate IPs, there are more constraints of the physical world. And to become a qualified supplier for a company like NVIDIA, it’s usually a very lengthy and rigorous process. And I think that also gives a lot of these hardware suppliers a bigger moat and sort of protection.

Savannah: Now, let’s say I have a cool million to invest somewhere in AI. I’m getting pretty convinced that investing in the hardware supply chain might be a more stable bet. And there’s no totally risk-free option, but what will happen if the AI bubble bursts in China? How is the conversation about risk in this industry different in the US and China?

Rachel: I think the topic of the AI bubble has come up over and over again in the past few years, right? And right now we’re in a moment where we’re seeing early phases of productivity gains from AI.

So there’s a bit more confidence that the investment into AI will eventually pay off. For instance, there’s a Wall Street Journal report today that Anthropic’s revenue is set to double to 10 billion in the second quarter of this year. And that means they’re going to record their operating profit for the first time.

So this is proof of the economics of AI business, right? But at the same time, they’re also worried about concentration risk. If you look at the shares, the market value of the Magnificent Seven, which is from Apple to these top seven companies, they account for almost one third of the total market cap of the S&P 500. That means it’s very concentrated.

The conversation in China is a little different. And so with the whole AI industry, I think the biggest question is still, will consumers be willing to pay for AI services? I think that is the top question. As for the hardware, a lot of the investors are betting on the bottlenecks.

So they have to look and see if demand is going to be as strong and the difference in the demand and supply. If demand is lower than expected, or if supply really catches up, or if some of the companies choose alternative solutions or technical paths that will also change the valuation of these companies. For factors that are unique to China, I think for Chinese equities, there’s always a trend where people really like to rush into a popular theme and make it very, very crowded.

And so if there is any earnings disappointment, you would see a huge correction in their market value. Another more familiar problem is involution, which is when you have a good idea, when you have a good business idea, a lot of people will rush into it. So some of the investors into this pick and travel play are also looking to see if there will be new competitors that emerge in this space and if they might compete away the margins.

Savannah: Rachel, thanks so much for joining us today. You can read her story today on thewirechina.com. To get around that pesky paywall and read more of our content, don’t forget to register for our free newsletter and get a discount code emailed straight to your inbox. Coming up next is a clip from our interview with David Daokui-Li, a Chinese economist at Tsinghua University’s School of Economics and Management on Beijing’s goals for leader-level summits with the United States.

After that, we’ll feature a new podcast we’re pleased to recommend, The Polycrisis Podcast. More on that in a moment. Here’s David Daokui-Li.

David: The Chinese side treats the meeting as part of a long-term project. The long-term project is to build up a better relationship and a mutually trusted relationship with the U.S. And this is the first step, a one-step of the long-term project. And at the current stage, China is not particularly eager to reach a certain deal with the U.S. Rather, the Chinese side wants to build up person-to-person trust between the two leaders.

Whereas from the U.S. side, from President Trump’s perspective, he wants deals. He wants short-term deals. China is not particularly eager to have short-term deals.

China wants to have a better person-to-person relationship between the two leaders. So President Trump, the Chinese president really made an effort to treat President Trump as a good friend.

Savannah: You can catch the rest of David Daokui-Li’s perspectives on what China wants in our upcoming issue, published Sunday evening Eastern time on our website.

Now, Kate MzcKenzie from The Polycrisis Podcast joins us to discuss China’s role in the emergence of what she has dubbed the electric world order. You can listen to more on what’s happening to the green transition in the midst of the Iran war energy shock on new episodes of the Polycrisis podcast. We’ll leave you with Kate’s insights. And thank you so much for listening.

Kate: China’s played a critical role in what we call the emergence of the electric world order, which is a kind of a quiet revolution in energy transition that’s almost happening below the radar. It’s a little bit hard to trace in the data, but a lot of countries are actually switching their energy systems in different ways towards more and more renewable energy supply, more and more battery storage, but also more electrification on the demand side.

That’s one of the really important components. China has been instrumental in this, of course. It is now the biggest manufacturer by far in the world of solar panels, batteries and battery components, electric vehicles.

So it’s a huge change. And the volume of exports of these goods out of China is just immense. The thing is that this wasn’t really a grand plan from the Chinese central government to kind of export clean power and electrify other countries.

This is almost like a side effect of China’s own economic policy, of its own industrial strategies.

So the new season, we’re calling it a bonus season. It’s actually looking at the implications of the Iran war.

This is the biggest energy shock that the world’s ever experienced, because this is the first time we’ve had a big energy shock where there was also this option of actually switching out big parts of national energy systems, you know, away from oil and gas dependencies and towards like a more electrified system, which then gives you like a lot more optionality in terms of where you get the actual energy from. And of course, you know, the availability of incredibly cheap solar components and much cheaper battery components than have ever been available before.

A lot of countries are now importing a lot more Chinese clean tech gear. That’s then inevitably like prompting questions within some countries about, oh, are we just swapping out one dependence for another? Are we just swapping out exposure to an oil and gas chokepoint to exposure to a solar panel chokepoint?

Transcribed by TurboScribe.

TP-Link is having a tough year. The California-based, Shenzhen-founded internet router company can’t seem to shake skepticism that moving to the United States means it is no longer Chinese.

This week, reporters Noah Berman and Savannah Billman return to the podcast to discuss the strange story of TP-Link. In a bid to shed its Chinese origins and sell in the U.S. market, the company’s owner Jeffrey Chao brought his entire family to the Golden State and restructured the firm. But, as Noah reports, the router company’s American Dream is caught up in the U.S. government’s suspicion of Chinese tech companies.

We’ll also have a Q&A with Jonathan Cheng, the Wall Street Journal’s China bureau chief, on his new book exploring North Korea’s founder Kim Il Sung’s Christian background.

Start your subscription to The Wire China today! We’re making it easy with your first month for just one dollar. Sign up for our newsletter here and we’ll send you the discount code to apply at checkout.

Transcript

Savannah: Hello and welcome back to The Wire China podcast. I’m Savannah Billman speaking today with my colleague Noah Berman about TP-Link, a California-based internet router company that can’t seem to shed allegations that it’s actually Chinese.

We’ll also have a Q&A with Jonathan Cheng, the Wall Street Journal’s China Bureau Chief, on his new book exploring the Kim dynasty’s Christian background. Yes, those are the Kims of North Korea. In fact, Pyongyang, today the ruling seat of a nuclear-armed hermit kingdom, was once known as the Jerusalem of the East.

We’ll get into TP-Link in a moment, but we can’t let a China podcast air this week without mentioning the ongoing Trump-Xi summit. Our most recent issue of The Wire China covers some of the more pressing issues facing the US-China relationship today, such as the need for AI safety agreements and the startling lack of American correspondence on the ground in China. We’re also offering you a discount code for your first month subscription to the Wired China, so you can stay informed during the summit.

To get your first month for just one dollar, sign up for our newsletter at the link in the podcast description and we’ll send the discount code your way. Now to TP-Link and Noah. Hi Noah, it’s been a while. Welcome back to the podcast.

Noah: Thanks for having me.

Savannah: This will be the second time you’ve written about TP-Link for The Wire China. You previously wrote about them back in January 2025. Can you walk us through TP-Link’s troubled timeline in the US and why you’re reporting on them again now?

Noah: I think it helps to start at the very beginning of the company’s history. It was founded in 1996 by two brothers, Jeffrey Chao and his brother Cliff, and over the ensuing three decades it has become the largest player in the market for internet routers.

And over the past five to seven years in particular, it’s really become one of the largest players, if not the largest player in the United States. And as a result of that, it’s begun to attract a lot of scrutiny from the US government against the backdrop of overall concerns about Chinese companies in the telecom sector.

In December 2024, there were reports of investigations by federal agencies in TP-Link. We took that as an impetus to write about it in January 2025, as you noted, looking at the interesting way that TP-Link has sought to address that scrutiny. And what they’ve done is something truly very clever. Jeffrey Chao has split his company from Cliff’s.

Cliff has a company in China for China that is unrelated to the company that Jeffrey has, which moved to Irvine, California. Cliff’s is now called TP-Link Technologies. Jeffrey’s is called TP-Link Systems.

But really, TP-Link Systems has absorbed much of the international empire of TP-Link. It has subsidiaries in a number of countries in Europe, in India, and in China. And a lot of those still fall under the purview of Jeffrey Chao’s TP-Link Systems, we learned from reviewing legal filings.

And so as a result, the move hasn’t done much to ward off scrutiny from the U.S. government. The investigations, according to people familiar with the matter, have continued even as TP-Link Systems ramps up its hiring in Irvine and has even attracted new investigations from some states.

In December of last year, Florida subpoenaed the company. In February of this year, Texas sued the company. And in March, the Federal Communications Commission, the FCC, banned all foreign-made routers, in theory, that applied to every company. But over the past couple of months, we’ve seen a number of firms founded in the United States get exceptions, including TP-Link’s main rival, Netgear, as well as Eero, which is a subsidiary of Amazon.

But TP-Link hasn’t yet gotten an exception. And that really imperils its future in the U.S. market, as its competitors will be able to continue manufacturing in cheaper places overseas, in closer proximity to their supply chains, while TP-Link, if it does not get that approval, will have to manufacture domestically in the United States, which is not something an electronics company in 2026 really wants to do.

Savannah: Seems like TP-Link is having a very rough run in America so far. There are a lot of other Chinese firms caught up in this push to get rid of Chinese tech, which, as you note, has been going on for quite a while — the most famous examples being Huawei and ZTE. So what is the reason why so many in the U.S. government are pushing to block Chinese origin firms from U.S. networks?

Noah: Yeah, like you said, this has been going on for some time.

The U.S. government first targeted or began looking at Huawei and ZTE way back in the Obama administration. You can even go back to 2012, when a House Intelligence Committee report found that those companies, Huawei in particular, could pose national security risks. So this has been going on for some time.

And it really relates to government fears of two things, espionage and sabotage, especially as the U.S.-China relationship went from this period of hardcore engagement to the more complicated position that it’s in now of competition and now whatever the Trump administration is trying to project stability, I suppose. One thing that’s often cited by the U.S. government is a law in China that passed in 2017 called the National Intelligence Law. That law says that companies have to comply, share information with the Chinese government if the Chinese government asks.

And it has really raised a lot of fears in Washington that Chinese companies could theoretically share data on Americans with the Chinese government, which the U.S. government really fears. And at the same time, there’s been an escalation in the number of hacking attempts, particularly by threat actors, by hackers, that both U.S. government agencies as well as independent researchers say are likely affiliated with the Chinese government, targeting critical infrastructure in particular, things like the grid, but as well as as federal agencies in the United States. And in December 2024, the U.S. Treasury Department said that it had been hacked and it pointed to a Chinese state-sponsored hacker as the likely culprit.

Savannah: So Huawei and ZTE are producing, you know, phones and 5G networks, but TP-Link makes internet routers. What’s different about routers themselves?

Noah: So routers are what cybersecurity researchers, people in that industry, call an edge device, meaning they’re essentially an entry point to a given network. Once you have access to a router, you can use it to go deeper into the network and look at all of the data, theoretically, that passes through it.

It’s something that governments around the world are concerned about. There’s pretty widespread acknowledgement of the cybersecurity risks as well as the rewards for hackers associated with routers. Just in April last month, intelligence and law enforcement agencies in the United States and its so-called five eyes partners, UK, Canada, Australia, New Zealand, issued an advisory warning that Chinese state-sponsored hackers had developed covert networks made up mostly of consumer routers.

So it’s not just the U.S. government that’s concerned about this. And really the fear is that once a threat actor is in one router, it could potentially own thousands or tens of thousands of devices by just exploiting one vulnerability. And these devices are really very valuable, not just because of that data that you can get from one person’s router, but because of what you can do if you have access to thousands.

So let’s say you instruct them at the exact same point to try and communicate to a certain server or service, and all that heavy traffic causes the service to be taken down or to get a lot slower, you can really flood a target with attacks. And that can be pretty damaging, particularly if it’s at a critical infrastructure or target. And of course, there’s also surveillance concerns.

You can see who someone is contacting, you can see their passwords. And a lot of the reason for this is that no one really thinks about routers after they buy them. They may not change the default password, they may not ever update the software, and they may not replace it when the router reaches what’s called the end of its service life, which is only a couple of years.

But people just don’t want to buy a new one, even though the cybersecurity protections are dramatically degraded after it reaches that end of service life point.

Savannah: This is making me feel bad that I haven’t updated my router password recently, so I’ll have to get on that. So TP-Link has tried to get around these challenges by trying to make itself not Chinese at all, which is an interesting strategy.

You already gave us a little bit of an overview, but how have they tried to prove their lack of ties to China? And has that really happened? What is the nature of their current ties, if any?

Noah: TP-Link moved its headquarters, like I mentioned, from Shenzhen to Irvine. Its CEO, Jeffrey Chao, moved to the United States with his family to California. And the company now says that all of the important decisions, including on things like information security, are being made in the United States.

It’s also, as I said, restructured its corporate structure. It’s now owned by a Delaware LLC called J.H.1 Chao Holdings LLC, for Jeffrey Chao’s Chinese name, and J being the first letter of his English name, H being the first letter of his wife’s English name, Hillary. But at the same time, it still has a major presence in China.

TP-Link had more than 13,000 employees in mainland China as of last January, according to a sustainability report that the company put together at the end of 2024. That doesn’t even include Hong Kong or Macau. And the company says that it now has around 550 personnel in the U.S., but that’s just a fraction of the number of people that it has in China.

So it’s really about the people. It has an R&D center in Shenzhen, and it has three manufacturing centers in southern China, too. So it has way more people in China, and those teams are also cooperating with one another, as you might expect for a big company that has a footprint in both countries.

In fact, a former employee told me that teams in the U.S. and teams in China were closely linked and worked on shared code bases with Chinese employees pushing updates to the code repositories. So even if the headquarters is now in California, the company still has quite a few links to mainland China.

Savannah: Have their efforts convinced anybody in the U.S. government?

Noah: No. And in fact, I think you could say that they have backfired. The Texas lawsuit against TP Link accuses the company of hiding its ties to China. I’ll quote from it.

“TP Link misrepresents to Texas consumers that its software and devices are not affiliated with China while knowing that its ownership and supply chain are Chinese.”

And regardless of whether or not that is fair, if the company had stuck in Shenzhen and just opened large operations in the United States while owning its Chinese background, it’s unlikely, I think, that states or opponents or critics of the company in the U.S. would be able to use the way that they’ve restructured as ammo against it.

Savannah: That’s an interesting misstep they’ve made there, but it seems to be one that other companies are also running into with increasing frequency.

Your reporting here shines a light on how hard it is to actually decouple from China. Both the United States and China certainly desire to separate their economies in some aspects, at least, as we saw through Trump’s tariff blitz last year aimed at China-based manufacturing and then Beijing’s more recent block of the Manus-Meta deal. So what does TP Link’s case reveal about the difficulties of making a clean break?

Noah: I think what TP Link shows is the intense skepticism that both Washington and Beijing have toward each other’s companies.

Here you have a company that was founded in China and moved to the U.S., but because it still has so many employees in China and significant operations there, Washington is very skeptical. And as you note with the Manus meta deal, when Manus employees left China for Singapore to be absorbed by a U.S. tech giant in Meta, that made Beijing very skeptical. And so it seems like these founders, Chinese founders of tech companies in particular, are really caught between a rock and a hard place.

Because if you make too big of a move, like Manus, you’ll piss off Beijing. But if you keep significant operations in mainland China, then you’re going to face scrutiny from Washington, too.

Savannah: So with all of this going on, what is next for TP Link as its case continues to evolve?

Noah: The company is focused right now on getting what’s called conditional approval, getting an exception from the FCC rule that requires firms to manufacture routers domestically or else face exclusion from the American market.

It’s a big deal because the U.S. market is really big. It accounts for more than two thirds of TP Link’s rival Netgear sales. And making routers domestically is going to be a lot more expensive than making them outside of the United States, probably in Asia.

But for the purposes of this rule, it doesn’t even matter where, as long as you don’t make them in the U.S., then you are not in compliance. Now, I should note that the rule only applies to new models of routers and not existing models. So this is not something that’s going to change right away.

But the fear, if you are TP Link, is that if you do not get that approval, then you will have to make routers in the U.S. and pay a lot more to do so and then maybe have to sell them for more and lose market share to other companies that are not subject to the same rules.

Savannah: Thanks so much, Noah. We’ll definitely be keeping up with the tangled story of TP Link here on The Wire China.

Noah’s story on TP Link’s origins and its fight to be American will be published this Sunday evening on our website, thewirechina.com. Don’t forget to sign up for our free newsletter at the link in the podcast description. We’ll send you a discount code to get your first month subscription for $1 so you can read Noah’s full story and the many other interesting articles that we’ll be publishing this weekend.

Our listeners may also be interested in a new podcast we’re pleased to recommend, The Polycrisis Podcast, hosted by Kate Mackenzie and Tim Sahay on the political economy of climate and the energy transition, in which China, of course, plays an outsized role.

Kate: China’s played a critical role in what we call the emergence of the electric world order, which is a kind of a quiet revolution in energy transition that’s almost happening below the radar. It’s a little bit hard to trace in the data, but a lot of countries are actually switching their energy systems in different ways towards more and more renewable energy supply, more and more battery storage, but also more electrification on the demand side. And that’s one of the really important components.

China has been instrumental in this, of course. It is now the biggest manufacturer by far in the world of solar panels, batteries and battery components, electric vehicles. So it’s a huge change and the volume of exports of these goods out of China is just immense.

The thing is that this wasn’t really a grand plan from the Chinese central government, you know, to kind of export clean power and electrify other countries. This is almost like a side effect of China’s own economic policy, of its own industrial strategies, which are about getting ahead in emerging industries like EVs, where it was still possible to become a leader, as opposed to old style cars where, you know, other countries already were very dominant. That’s really had this spillover effect in lots of the rest of the world.

Savannah: That was Kate Mackenzie on China’s role in the emergence of The Polycrisis Podcast dubbed the electric world order.

Kate: So the new season, we’re calling it a bonus season. It’s actually looking at the implications of the Iran war for all of these trends and changes that we identified in the first season.

This is the biggest energy shock that the world’s ever experienced. That’s not necessarily been apparent early on, but you know, the scale of it is huge. It’s really going to permanently change the composition of the energy mix around the world, because this is the first time we’ve had a big energy shock, where there was also this option of actually switching out big parts of national energy systems, you know, away from oil and gas dependencies and towards like a more electrified system, which then gives you like a lot more optionality in terms of where you get the actual energy from.

And of course, you know, the availability of incredibly cheap solar components and much cheaper battery components than have ever been available before. A lot of countries are now importing a lot more Chinese clean tech gear. That’s then inevitably like prompting questions within some countries about, oh, are we just swapping out one dependence for another? Are we swapping out exposure to an oil and gas chokepoint to exposure to a solar panel chokepoint?

Savannah: You can listen to The Polycrisis Podcast wherever you get your podcasts or visit them at thepolycrisis.org.

As promised, we’ll end this episode with a clip from our interview with Jonathan Cheng. The full Q&A with him on his new book on the Christian background of North Korea’s Kim dynasty will be published Sunday evening Eastern Standard Time. Don’t forget to sign up for our free newsletter to get that interview and all the rest of our articles delivered into your inbox along with that discount code. Here’s the clip with Jonathan Cheng and thank you for listening.

Jonathan: I want to say that to understand North Korea, you have to think of it as a pseudo-Christian state. It’s more that in the cocktail of influences that makes North Korea what it is today. I think you need to look at Kim Il-sung because he came to power at the age of just 33.

When you compare him to Stalin and compare him to Mao, he came to power and to absolute power decades earlier than Stalin and Mao did when they were each in their early 50s. Kim Il-sung took power when he was just 33. He lived a relatively long life.

He ruled North Korea for 49 years. When you look at the person of Kim Il-sung and you look at the first 20 years of his life, more than half of his life by the time he had become the leader of North Korea was spent in a deeply Christian environment. What I’m trying to posit in my book here is that to understand North Korea, you cannot overlook the Christian influence of the Jerusalem of the East, but specifically the Christian influence of Kim Il-sung.

Transcribed by TurboScribe.

Features editor Tom Mitchell and reporter Eliot Chen return to the podcast to discuss Eliot’s investigation into why the journalist population in China has struggled to rebound. While Beijing doesn’t seem to mind this turn of events, the dearth of reporters on the ground in China means the world knows much less about what happens there — a challenge we aim to overcome here at The Wire China.

Reporter Eliot Chen returns to the podcast with editor Andrew Peaple to discuss how the operation worked, and why smuggling chips to China is seemingly so hard to stop.Also, we’ll have a clip from our forthcoming Q&A with author Nicholas Niarchos on the impact that the push for clean energy is having on the countries that supply many of the key resources needed for products like batteries and electric vehicles.

Transcript

Andrew: Hello, and welcome back to The Wire China podcast, our weekly dive into the latest stories coming up on thewirechina.com. I’m Andrew Peaple, The Wire’s news editor. And this week, I’m joined by our Toronto based reporter Eliot Chen to talk about his new cover story on one of the hot topics of the moment, chip smuggling.

And later on, we’ll also have a short clip from our new Q&A with Nicholas Niarchos. He’s a journalist who’s written a fascinating book on the realities behind the rare earth industry, drawing on his travels to the Democratic Republic of Congo, Indonesia and lots of other places. Both of those stories will be published this Sunday evening on our website. And so to get them delivered to your inbox, you can sign up for our free newsletter at the link in the podcast description.

But let’s start with Eliot and his story about the San Jose based company Supermicro. Hi to you, Eliot. How are things going?

Eliot: Hello, it’s good to be back here.

Andrew: And Eliot, before we start, I should say, last time you were on the podcast, you talked about some side reporting that you’ve done that I didn’t know about. So I just wondered, have you been on any fun trips lately?

Eliot: No more unsanctioned travel!

Andrew: Well, I should think so too, because you’ve obviously been working hard on this story about Supermicro and chip smuggling.

But let’s start with the basics. Who and what is Supermicro? And what do they do? And who runs the company? Can you just give us some of the background details before we launch into this interesting case?

Eliot: So Supermicro is an American hardware company. It provides computing equipment. Supermicro right now is perhaps best known as a manufacturer of servers for data centers. So to take a step back, when we talk about the hardware needed to produce AI these days, we talk a lot about chips, chips like the ones that are made by NVIDIA.

But in order to actually run the chip, there’s a whole bunch of other stuff that you actually have to install around it. There’s a power unit, there’s a CPU. When I was a kid, one of my favorite things that I did with my dad was build computers.

And so Supermicro at the time was also known for making motherboards, which is the sort of primary component that you put in the computer that everything else plugs into. So you have to basically build the computer for the server before you can run the AI on it. And increasingly, actually building those servers is an increasingly sophisticated business.

Because not only do you have to build a running computer, but on top of that, there’s all these issues with these massive data centers these days that consume a huge amount of electricity and produce a huge amount of heat. And so an increasingly challenging task for the server makers is actually keeping all this stuff cool. And so that’s one of the areas where Supermicro has sort of stood out in the last few years.

It’s actually a data center cooling technology and being able to build these innovative, effective servers. So a little bit very quickly about the company’s history. It was founded by this guy, Charles Liang, who was born in Taiwan, but now the company is based in Silicon Valley.

He co-founded the company with his wife, Sarah Liu, and Wally Liaw. So the three of them co-founded this company in 1993. At the time, they’re known, as I mentioned earlier, for making these motherboards.

But since then, and particularly since the AI boom that started in late 2022, the server business has really become its dominant business line and where it’s making its most money.

Andrew: And it’s been quite close to NVIDIA, the big US chip giant that’s become so successful in the last few years. Is that right?

Eliot: That’s right. So NVIDIA and Supermicro have ties that go back really quite a long time. Part of it is because both of their company founders came from Taiwan. So you have Charles Liang at Supermicro, and then at NVIDIA, you have Jensen Huang, who was also born in Taiwan.

They were founded in the exact same year, in 1993. And in fact, their headquarters are now a mere 11 minutes drive, according to Google Maps, away from each other in the south San Francisco Bay area. So the origins of the company really started around the same time.

NVIDIA has also turned to Supermicro at various points in the past as a sort of trusted partner for its business. Two decades ago, in 2007, when NVIDIA actually developed its first dedicated graphics chip for data centers, they went to Supermicro as their first go-to-market partner, according to one of NVIDIA’s VPs in a presentation a couple of years ago. Today, Supermicro remains hugely reliant on NVIDIA as a supplier for its business.

So according to Supermicro’s latest annual report, its top supplier accounts for 65%, so almost two-thirds of all of its purchase spending. Supermicro doesn’t disclose the names of the suppliers, but most analysts widely regard that top spot to have been taken by NVIDIA.

Andrew: So Supermicro, a successful company in a fast-growing area, is well-connected as well. So that’s one element of the story. Could you give us the background then as well on chip smuggling? Why has that become such a big deal for the listeners who don’t know? Why has this whole issue around exporting chips to China become so important in the US in the last few years? And why has, as a corollary really, chip smuggling become such a feature of what’s going on?

Eliot: This has been an area of coverage for us that really we spent quite a lot of time writing about in the last few years because it’s so important to the US-China relationship. The abridged history is, very quickly, to go back to October 2022, the Biden administration unveiled these sweeping new export controls that month targeted at preventing China from getting access to the most powerful AI chips and also from building those AI chips on its own.

And so this was a monumental shift in US policy at the time. A lot of observers saw it as the US government sort of escalating and really going for the most important chokehold that the US controls in the US-China competition. And it’s also just worth putting in perspective that timing, which was actually quite early, because October 2022 was two months before ChatGPT came out.

We all now know what ChatGPT is. But at the time, this AI competition wasn’t even something that was on the general public’s radar. And already the US government was moving to restrict and try and prevent China from keeping pace with US and AI development.

And so since then, as this has become a more widely understood issue in the public, it’s become a massive debate whether it’s right or wrong that the US imposed these export controls. NVIDIA and founder, Jensen Huang, are really strongly opposed to the policy. And last week, there was this prominent debate in which Jensen Huang went on a podcast and really argued very forcefully that these export controls were a mistake.

And for a while, Huang and NVIDIA also had a pretty sympathetic ear in the White House in David Sacks, who is the president’s AI czar. Proponents, however, say that it’s really important that the US keeps these controls on, especially as these AI models get more sophisticated. And so people argue that as these models are able to do these increasingly sophisticated, scary things, it’s really important that China is not able to outpace the US in development in doing that.

And then amid all of this and amid this big policy debate, there is this burgeoning cottage industry of people that have been helping to get the chips to China anyway. And so far, prior to the indictment in the Supermicro case, when we looked at these cases, and we’ve written stories about them as well, we’ve mostly looked at sort of a hodgepodge group of individual operators, maybe these opportunistic actors working with small shell companies that are maybe shifting a few hundred, maybe a few thousand chips to China at a time. And so they’re not huge operations.

That being said, the US is now prosecuting four such cases, and they’re all over the country, which sort of shows how prolific this seems to be getting these cases in California, Florida, Texas, and Georgia.

Andrew: That’s a fabulous explanation of the background to this case then, Eliot. Thank you for that.

But let’s get into the details. Let’s get into the meaty, meaty story here. How did Supermicro get involved in this case about smuggling chips to China? Can you tell us about one detail in particular? How was a hairdryer involved in what they were up to?

Eliot: So the Supermicro case, I know earlier we mentioned that there are all these smaller cases.

The Supermicro case is different. It’s different for two reasons. I mean, one, because this was allegedly an inside job.

You have a top executive, a co-founder at a major company allegedly in on the smuggling. But also two, because of the scale of the operation, which is a magnitude, an order of magnitude larger than anything else that we’ve seen. And so to overview what prosecutors say happened, and so to emphasize, this case has not yet been tried in court.

These details are based on the indictment that the prosecutors have published, but has not yet been tested in a legal setting. But allegedly what happened was that Liaw and his co-conspirators, they identified a front company. This is a company that prosecutors haven’t named.

They’ve only said that it’s a company based in Southeast Asia. And it’s a real company. It’s not just a shell.

It has actual operations, supposedly. And they found an executive at that company who was willing to work with them to place humongous orders for AI servers to then be sent to Southeast Asia. Once the servers were sent abroad, the real servers, the ones that had been sent from Supermicro, were repackaged in warehouses there into unmarked boxes, and then allegedly shipped onwards to China.

Meanwhile, Liaw and his co-conspirators at Supermicro worked to obfuscate this whole thing and to dupe Supermicro’s own compliance team. And they did that by allegedly fabricating, for example, lease agreements for data centers where these servers were allegedly being set up. They also worked to dupe inspectors and auditors when they actually went to these sites.

And so this is where the hairdryer comes in. When the real servers were already on a ship or en route to China, the defendants then allegedly, in preparation for Supermicro’s auditors and inspectors from the U.S. government coming to the site where the servers were set up, what they did was they removed the labels and NVIDIA’s serial numbers from the real chips. And they did so by essentially sticking a hairdryer to the sticker and peeling them off.

And the U.S. government actually caught the defendants doing so on CCTV camera. And so they removed the labels. They put them on dummy fake servers that the inspectors then inspect.

And so in doing so, apparently managed to dupe, well, eventually not the government, the government caught on, but Supermicro’s compliance team and kept this operation going. And so all in all, between 2024 and December 2025, Supermicro sold, allegedly, according to the document, $2.5 billion worth of servers to this front company that then went on to China. And that is an astonishing number.

It is 7% of Supermicro’s total revenue over this two-year period. More than one in every $20 paid to Supermicro was a purchase that was then an export control violation. And so that’s why this case is entirely different and in a different ballpark from previous ones that we’ve seen.

Andrew: You’re right. Those are stunning numbers. And the scale of this is quite remarkable.

In some ways, though, I guess this whole scheme, even though there’s some complications to it, it looks quite simple, really. Ship stuff to an in-between company and then ship it on to China. It’s not the most sophisticated plot.

I guess it begs the question, is there a broader issue here for the US that actually stopping people from getting chips or products to China that the US doesn’t want to get to China is actually pretty hard? And it can easily be got around. Does that kind of have some broader implications, do you think, for how America should approach its tech competition and its export controls to China?

Eliot: Yes and no. I mean, I think to take a step back, Andrew, I feel like a lot of the stories that we do these days or the stories that I pitch you about China and the US are all around this theme of one or the other country wanting to stop trade of a certain type of good for whatever reason, but then not really succeeding in doing so.

Whether it’s AI chips, goods from Xinjiang to the US, small value goods from Temu or Shein, the flow of commerce just wants to happen anyway. And so trade officials and the government face this Sisyphean battle of trying to stop lots and lots of goods from moving around the world at a time when globalization has us connected, even when some people don’t want us to be. And so in that context, I think it seems difficult to stop chips from going to China if the government, if Chinese companies are determined to get them.

On the other hand, I think it’s important to emphasize that in this case, this is also a special case because of the, again, the inside man role in this story. While at once we talk about how it’s difficult to identify smuggling, in this case, what’s interesting is that Supermicro’s compliance team was actually in some ways successful at doing their jobs. At one point, their compliance team actually looked at the numbers and were like, huh, this Southeast Asian company is buying a lot of chips from us.

Maybe we should pause and take a look at where they’re all going. And so they actually froze the shipments to the front company pending an audit in late 2024.

Then in response, Liaw and his co-defendants, according to the indictment, supposedly, quote unquote, took steps to remove the hold and undermine the audit. And there’s even one email that’s mentioned in the indictment in which Liaw apparently urges the compliance team to speed the whole process up.

Yes, identifying chip smuggling is hard to begin with, but it’s even harder when you’re apparently getting overruled and pressured by a VP and director at your company who is in on the smuggling, allegedly to push forward with the sales anyway. Fascinating.

Andrew: Thanks. Thanks for those thoughts, Eliot. I just wondered then, what have been some of the consequences of this case and some of the other chip smuggling stories that you’ve looked at over the last year or so? What sort of action is being taken in the US by the authorities or by lawmakers?

Eliot: Since last August, when the Justice Department began announcing some of these cases prosecuting chip smuggling, I think there’s been this steady drumbeat of, you know, growing awareness that this is an issue.

This last case, though, I think was sort of the final straw and really catalyzed a bunch of action and activity in Congress. So for one, you know, shortly after the indictment was released, Senator Tom Cotton, who’s a pretty vocal China hawk, wrote to the Commerce Secretary calling on him to sort of as soon as possible enact provisions of this bill that hasn’t even been passed yet called the Chip Security Act. And so this is a bill that we’ve actually written about in the past.

It proposes sort of enacting tracking measures so that NVIDIA can actually sort of geolocate where their chips are in the world, which will give us a better idea of how many of them are ending up in China. The company really strongly opposes this, I mean, in part because China absolutely detests the idea that, you know, a US government can mandate that suppliers send chips to China that sort of track where they’re going. The House Foreign Affairs Committee advanced that bill out of committee for the first time a couple of weeks ago, which means that it’s actually now going to go to a full vote in the House of Representatives.

And then also just this past week, on Wednesday, the Foreign Affairs Committee also voted to advance another bill called the Stop Stealing Our Chips Act, which essentially establishes a whistleblower protection and incentive program. And so we can see that even though the White House has been pretty quiet when it comes to, you know, legislation targeting this issue, Congress isn’t really waiting around any longer. And there are bills that are sort of getting moving as a result of all this detention smuggling. So some real impacts from this story.

Andrew: Thanks so much, Eliot, for talking us through this story. It’s a really great read.

I’ll encourage all of our listeners to take a look at Eliot’s full investigation into Supermicro this weekend, which is available on our website. It will be available on Sunday as of 5pm Eastern Time in the US.

And there you’ll find his other reporting on another smuggling ring involving Nvidia chips, which was busted late last year. You can also sign up for our free newsletter at the link in the description of this episode. We’ll leave you this time with a clip from that Q&A that I mentioned at the top with Nicholas Niarchos. Thanks so much for listening and do take a look at thewirechina.com for Eliot’s stories and much more besides.

Nicholas: Companies like Apple or Tesla, they also use the Chinese as a kind of insulating layer. For them, it’s very convenient that there are Chinese companies out there mining cobalt and whatever, because they say, look, you know, we can’t do anything.

It’s China. You know, this is where the minerals come from. And, you know, we couldn’t get minerals from anywhere else. We have no way of really checking on Chinese mining practices.

This allowed for Apple for many years to buy from one of the worst, I mean, I guess you would say abusers of human rights and one of the companies that does buy off miners and child miners and so on, which is to buy your cobalt. The focus has always been on the Chinese miners and much less on the companies that are buying from the Chinese miners and the Chinese miners give these companies a convenient way of just saying, look, it’s somebody else’s problem.

Transcribed by TurboScribe.

In this episode, we are joined by a new guest: journalist Grady McGregor, who discusses his upcoming piece on crypto leverage and its Chinese origins with features editor Tom Mitchell. His cover article will be published this weekend on our website, thewirechina.com.

Transcript

Tom: Hi, and welcome to The Wire China podcast. I’m Tom Mitchell, The Wire’s feature editor, speaking to you from Singapore.

In the podcast, we aim to take you behind the scenes of the stories we cover in our online magazine. This week, I’m joined by Grady McGregor, also in Singapore. He writes about the rise and rise of leverage-fueled investment in the global cryptocurrency industry.

It is a phenomenon that has its origins in China, and also a phenomenon that the Chinese government was very wary of. It was a factor leading to Beijing’s decision to ban the industry, after which the country’s most successful crypto entrepreneurs, many of them self-made billionaires, to camp for Hong Kong, Singapore, or other more friendly jurisdictions. As ever, you can read these stories and much more on our website, thewirechina.com. But first, Grady, welcome to your first of many Wire China podcasts.

Grady: Thank you so much, Tom. I’m excited to make my debut on the new podcast here. Excellent.

Tom: Today, we’re discussing your upcoming cover article on the increasing amount of borrowing in the cryptocurrency industry, its origins, and the potential risks it poses. That article will go online this Sunday, April 19th, around 5 p.m. New York time. To get this story and the rest of our issue in your inbox this weekend, you can sign up for our free newsletter.

So, Grady, what is the current scale of leverage or borrowing by investors in the crypto industry?

Grady: One key gauge for leverage in crypto is Bitcoin open interest. It’s not a perfect measure for leverage, but it’s pretty good. It hit its all-time high of over $100 billion or so last October, and there was a big crypto crash.

It’s down, I think, around $70 or $80 billion now. But the big point here is that it’s probably up three or four times from two years ago, and so this ultimately points to a new phenomenon, that there’s just been this massive buildup in leverage and derivatives trading in crypto in the last few years.

Tom: So, why is this a bad thing or a good thing? The people you spoke to were divided on this point.

Grady: Exactly. Yeah. So, if you talk to the crypto companies, a lot of people in the industry, this is a good thing for the most part.

The idea is that the growth in derivatives trading, and this is options and futures, it’s a sign of maturing markets that look much more like traditional equities. But there is certainly some concern about building up too much leverage, even in crypto circles. Last October, there was this massive crash, and a lot of it was fueled by these derivative bets.

These sort of instantly wiped out hundreds of billions of dollars in crypto over this Trump tweet that we write about in the story. But then if you talk to a lot of scholars or crypto skeptics, they see this buildup in leverage as completely terrifying. They view it really as a house of cards, that you are making a ton of bets, highly leveraged bets on things that they don’t see as having any sort of underlying value.

I think one thing that stuck out was somebody saying, at least in 2008, houses have value, and it’s not clear that this does. And this fear is really increasing as crypto becomes more mainstream. Wall Street is involved, big banks are involved, legislation in the US will further mainstream it, so now people’s pensions could be impacted.

And so the fear here is that a crash in crypto maybe used to be able to be contained within the crypto universe, but now it could spread everywhere.

Tom: Okay, let’s just explain this as simply as possible. So let’s say I’m a crypto investor with $1 in my pocket, and you’re running a crypto exchange.

The formula is here, I invest my $1 in crypto on your exchange, and you loan me an additional $100 to also invest in crypto. Even though you don’t know whether or not I’m going to be able to pay it back, because for all you know, that $1 in my pocket was the last dollar I had in my pocket. And if the tide rises, everyone’s a winner, but then what happens if the $100 I’ve invested against my $1, the price of the assets start to go down? What do you do as the exchange owner?

Grady: Yeah, so this is kind of exactly what we saw last October.

These small price movements within crypto triggered this much larger crisis. And that’s because you have so many bets, you had so many bets, these leverage bets were exactly right up to 100 times leverage. And so if they’re small price movements, especially going down, these positions can be liquidated immediately, and then that triggers further price movements down.

And so they call this kind of liquidation cascades, and that leads to these sort of wild swings in the market. And so that’s what we saw last October. And that’s kind of a broader fear about introducing that sort of volatility, not just within the crypto ecosystem, which is known for its volatility, but more broadly.

Tom: Okay, right. Because as the exchange owner and the prices go, the prices are going down, you realize, boy, it was a really dumb idea to give Tom $100. So I’d better try and get back what I can before the prices fall even further.

So I’d have to sell my position to maybe get you back $80, $60, whatever it is. This trend of increased borrowing in the crypto industry, as you write, and the reason we’re running this piece in The Wire China, is that it really began in China about a decade ago, yes?

Grady: I mean, the story, the idea for this story came from two central things that I was seeing. Last October, covering the token conference in Singapore, the biggest crypto conference in the world, it was just so evident how dominant Chinese founded exchanges have over the entire ecosystem.

A few weeks after that conference, there was this massive crash, and it was so evident how core leverage is to crypto now. And so through this reporting, I found that it’s not that these phenomena are just linked, but understanding leverage is also key to understanding how these Chinese founded exchanges really grew and developed. And so in the beginning, in the early 2010s, when there was the kind of start of this crypto scene, it started in WeChat trading groups.

And these traders loved margin, they loved risk, and would kind of egg each other on. And this was happening as exchanges like OKX, at the time known as OKCoin, Huobi, which is now known as HTX, were competing for customers. And to get customers, they competed on leverage.

And so it was a thing where they would offer spot trading, which is just trading the assets for free. But the leverage trading, where you’d be trading on margin up to borrowing up to one or five times at the time, they would charge fees. And so they would explicitly try to court as many of these leverage traders as possible.

And that’s how they made their money and grew, especially through the mid 2010s in China.

Tom: So at that time, the Chinese government is watching these Chinese exchanges lend money to Chinese investors to invest in these currencies that don’t really exist.

So what does the Chinese government think about that as it develops and what does it do?

Grady: Yeah, I mean, I think it’s a pretty common story for how China treats new technology.

Martin Trozempa from PIIE, who I talked to, kind of described this as a wait and see approach. And so while the industry was small, sort of seeing what the innovation was around it, seeing what the technology was, they did have some very core concerns. In 2013, they banned Bitcoin from being used as a currency.

Towards the mid 2010s, as these exchanges grew, we started to see much more concerns from the Chinese government, from the People’s Bank of China about speculation. And this period also corresponded to when Xi Jinping was talking much more about sort of the real economy and avoiding speculation in general. The big crackdown happened in 2017.

And this was around these initial coin offerings, but extended broadly to the entire sort of crypto ecosystem. All of the major Chinese exchanges, OKX, Binance, which was I think founded in 2016, moved offshore. So from the people I talked to, their central concern, probably the top priority for the Chinese government, was capital flight, which China sees as a threat to its financial stability, stopping Chinese people from being able to get money out of China.

But they also took elements of the technology that they liked, the blockchain technology specifically that they could control. So obviously, they developed the E-yuan, the centrally backed digital currency. And this was a way to take elements of crypto that they liked and get rid of the things that they don’t like, like speculation, capital flight.

Tom: Now we’re at the point where China’s cracked down on its leading crypto startups. So what do they do? They all basically decamped for either Singapore, Dubai, some went to Hong Kong, and then left for Singapore and Dubai. But they were all essentially kicked out of China.

We have a very Chinese story for The Wire China. But then a new character enters, and it’s an American entrepreneur, Arthur Hayes, who’d been a Citibank trader in Hong Kong. What does he do?

Grady: Yeah, so he’s this really interesting character.

He was laid off from Citibank. I think he was an equities trader around 2014. He noticed there was an arbitrage in crypto prices in China versus the world.

And so he could buy crypto in Hong Kong, literally take the train across the border, sell it and take bags of RMB back across. But as he got more into crypto, he wanted these more sophisticated financial instruments that he had used as an equities trader in crypto. At the time, companies like OKX, Huobi had been kind of growing in the derivatives markets a bit.

And this was mostly in futures, options trading, that sort of thing. He couldn’t really compete with those exchanges on liquidity. But he developed this financial innovation that he called, I forget his specific name for it, maybe perpetual contract.

It eventually became known as just perps. And what this was, essentially, is a futures contract that doesn’t have an expiration date. And so if you have a normal futures contract, you are betting, oh, Bitcoin will go up and we’ll settle that a month from now.

If it’s up 10%, I get X amount of money. If it’s down 10%, I give you this amount of money. With perpetual futures contracts, these contracts are settled at a regular cadence.

So every eight hours or so, I think, is kind of standard. And so you are always kind of, you can bet on these things indefinitely. And what he also did was introduce the leverage of up to 100 times.

And this was quite extreme, I think, even for crypto. But what was really interesting was he came up with this innovation. The Chinese founded exchanges caught on super quickly and saw their value.

And somebody told me, basically, he pioneered this, but these perpetual futures contracts were aggressively scaled by Binance and OKX. And now they really dominate the whole ecosystem. Something like 70% of all crypto trading last year was in these perpetual futures contracts.

And so it really is how most trades now happen in crypto.

Tom: Hayes is, he refines or I guess really simplifies what was a Chinese innovation. And then his Chinese rivals embrace this refinement, the simplification of their innovation, and they run away with it.

It’s amazing that 70% of all the trades now are through these derivatives. So what are the Chinese founded companies now doing with this refinement of their innovation?

Grady: You know, they’ve scaled it aggressively within crypto. And now it’s kind of, the idea is kind of taking it everywhere.

And so we are seeing Binance now just in the last few months has released a new platform where you can trade perpetual futures contracts on everything from silver, gold to now traditional stocks like Tesla, oil. And so Binance is not the only company doing this. A bunch of them are.

This is kind of part of this broader push towards tokenized securities, which will introduce a lot of these crypto native concepts to the traditional financial system, which some people see as quite dangerous. Others see it as, you know, I guess, bringing innovation to the financial system. But a lot of Wall Street is also embracing it.

And so we saw ICE, a kind of parent of the New York Stock Exchange, announced this big new strategic investment and relationship into OKX earlier this year. And they want to launch these U.S. regulated perpetual futures contracts. And so it really does seem like this kind of trading is now being brought into the mainstream.

And so I think something I wanted to point out in this article is it’s not necessarily clear if this is going to be dangerous or not yet, but it’s running this massive experiment on quite a new financial technology that was confined to crypto. And we’re going to see what happens now that it’s being introduced everywhere.

Tom: OK, well, let’s let’s wrap up on this point. I’m in a charitable mood tonight. So let’s give the crypto industry the benefit of the doubt, something I’m not normally inclined to do. And let’s assume the worst case scenario that many fear, a quote unquote subprime 2.0 bursting of crypto lending doesn’t come to pass.

Sure, there will be booms and busts, but nothing on the scale of the global financial crisis 20 years ago or the crash of 1929. In that event, do you think it’s possible the Chinese government might rethink its policies towards the industry?

Grady: It’s really hard for me to see them changing their stance towards a lot of these products. I think Chinese policymakers do recognize that they might be losing out on some financial innovation.

People do talk about how they are kind of essentially a seeding ground and letting the U.S. become the center of the crypto world. But I think China’s central concerns about crypto won’t really go away. They are very concerned about capital flight and they’re very concerned about control.

One of the sources told me these exchanges weren’t banned because of crypto or innovation, but they were banned because the state really only cares about control over the yuan and control over the country’s financial infrastructure. As long as I guess crypto’s central ethos of being this decentralized currency that doesn’t have state control, as long as those principles are core to how it operates, it’s really hard to see China letting in these products or letting in these services because it will ultimately still pose a threat to sort of China’s economic and financial security, which is its top concern.

Tom: Yeah, I think it’s Napoleon who’s credited with the observation that you never interrupt your enemy when he’s making a mistake.

So if the U.S. is being silly and embracing this and there’s going to be some bad consequences down the line, it would make sense for China just to sit back and see how that all plays out.

Grady: Well, exactly. Yeah, because I think, I mean, the U.S., it’s really running an experiment and running an experiment of what does it mean to more closely integrate crypto with the financial system and to integrate these products? And maybe it works out for the U.S. and maybe it doesn’t.

And I think China is quite happy to sit on the sidelines while the U.S. figures that out.

Tom: Thanks so much, Grady. For more on this story and others, please do go to thewirechina.com. Also in this week’s issue, we have an interview with Eyck Freymann, author of the recently published book Defending Taiwan.

In addition, our staff writer Rachel Cheung investigates the row over Chinese scientists attending a major international AI conference, something that the U.S. has tried to stop. Savannah Billman writes about how Chinese EV makers are benefiting and not benefiting from Donald Trump’s war on Iran. And finally, we have an op-ed from China security expert Peter Mattis.

All these stories and more can be found on our website, thewirechina.com. Thank you for listening.

Transcribed by TurboScribe.

But, as reporters Eliot Chen and Savannah Billman discuss, security risks haven’t disappeared as the tariffs have. In this episode, Eliot explains why the North American neighbors fell out over EV policy, and traces the close ties between one Chinese car company and a sanctioned surveillance firm.

If you missed it, Part I of our Hainan spy plane incident oral history is now free on our website. Read eyewitness accounts of the crisis here.

The Wire China has been covering Chinese electric vehicles’ slow creep into North America for some time now. Catch up with Eliot’s previous reporting on why the Ottawa government is considering dropping 100 percent tariffs on imports of Chinese EVs and why it’s now trying to attract China’s automaking giants.

Transcript

Savannah: Hello and welcome back to The Wire China podcast. I'm Savannah Billman calling in from New York City and on the other end of this call is my colleague Eliot Chen based in Toronto.

But before we get going on this week's episode, I want to make sure all our listeners had a chance to read through our big story from the last two weeks, an oral history of the Hainan spy plane crisis of 2001. We've just made part one of this epic free to read on our website.

25 years later, it's not a breaking story anymore, but it's no less important. Planes and ships transiting the South China Sea and airspace still have close and risky encounters with Chinese military craft. The U.S. government still springs into action to rescue stranded Americans after crashes in foreign airspace, as we saw in Iran just a few days before recording this episode.

You can listen to the previous two episodes of this podcast for a preview, but we've also made part one of the oral history free on our website. Part two is still exclusive for our subscribers and we hope you'll consider becoming one.

Now back to the main topic of today, Chinese EVs in Canada. In January, Canada announced a deal with China that would involve opening its market to Chinese cars. After so many years of Canada and the U.S. presenting a united front against Chinese EVs, some were surprised by the move, but certainly not our unofficial Canadian car correspondent, Eliot Chen.

Eliot, last year you reported that Canada was mulling over allowing Chinese EVs on its roads. Now obviously your words at the time have come to pass. Can you give us a very high level overview of Canada's divergence from its southern neighbor on the EV issue?

Eliot: So I think in order to understand what's going on in Canada with Chinese EVs right now, you have to go back to at least two years ago. So around the time of 2024, the Biden administration at the time and the liberal government under Justin Trudeau in Canada were very much aligned on many things on foreign policy, particularly on China. And at the time, when the U.S. was seeing the emergence of the Chinese electric vehicle industry and its growing exports abroad, the U.S. imposed 100% tariffs on Chinese electric vehicles. And very quickly after that, Canada basically moved in lockstep and did so as well. Then everything changed.

Of course, Donald Trump was elected and started threatening to invade and annex Canada. And with that, we saw a growing divergence between Canada and the United States on trade issues, but also foreign relations issues. So in January, now Prime Minister Mark Carney visited China. At the time, he made this famous pronouncement while he was there that we are in a new world order. And of course, shortly after that gave his viral Davos speech that sort of laid out Canada's new position vis-a-vis U.S. and the world. And one of the big announcements as well to come out of that Beijing visit was that Canada said that it would drop that 100% tariff on Chinese EVs and also allow in, starting this year, a limited number of EVs into Canada imported from China, starting with 49,000 cars this year. And that number will slowly tick up by 6.5% every year.

At the same time, Canada is also actively courting Chinese automakers to set up factories in this country. BYD has said that it's looking, actively considering it.And just last week, we also learned that Leapmotor, which is another Chinese EV company partially owned by Stellantis, is also in active discussions about setting up in a mothballed Stellantis factory here. So that's where we stand. The U.S. right now doesn't allow Chinese EVs in, but Canada is very much taking a different path.

Savannah: What has happened to all the pushback against Chinese EVs that was there in the first place, protecting domestic industry, for example, or the need for alignment with the United States? Are these things still concerns for Canadian policymakers?

Eliot: So Savannah, I think this is a good time for me to make a confession. This is something our editors don't actually know about. And so I'm going to use this opportunity to share a little of a story. So in the spring of 2023, I was still a relatively new reporter at The Wire. And that week in March, President Biden was visiting Canada for his first state visit at the time. And I snuck into that summit.

I had just bought my first car. I had a surprisingly light week at work. And so I got an email from Global Affairs Canada, which is sort of the State Department over here that was saying, like, do you want a media pass? And so I signed up. I mean, it was in part that I wanted to go see the President and Prime Minister. I also wanted to check out Air Force One for myself. And so I drove to Ottawa. And without telling our editors, I ended up sort of joining the media pack at that visit.

So that may be news for them. But I will add that in retrospect, sneaking off and doing that was actually a really valuable experience because I got to sort of witness the U.S.-Canada relationship that once was. So at the time, President Biden gave this address to the House of Commons in Canada. And he gave the speech that I actually thought was really lovely. He described the relationship between the U.S.-Canada. I'll just sort of quote a line from that speech in which he said, no two nations on earth are bound by such close ties. Friendship, family, commerce, and culture. Our labor unions cross borders. So do our sports leagues, baseball, basketball, hockey. And then he adds, Americans and Canadians are two people, two countries, in my view, sharing one heart.

So that's all gone now. Thanks to overwhelmingly President Trump's repeated threats to annex Canada. Over here, where I live now, and I've lived for five years, there's this real feeling of betrayal. The number of U.S. trips, the number of Canadian trips to the U.S. are down. The country is really quickly trying to diversify its trading partners. And the feeling of betrayal is just especially acute in the automotive sector, thanks to cases like Stellantis shuttering a factory in Ontario in order to double down on investment in the U.S. And so there's this feeling that in this industry, there's this zero-sum game when it comes to production. And I think that's all just really important context to understand why it is that Canada has shifted tack on Chinese EVs and why it is that some of the things that aligned Canada so closely with the U.S. even two, three years ago are no longer priorities.

I think I would also just add to that that while protecting domestic industry is something that's important over here, I think one of the factors that made it easier for Canada to embrace Chinese electric vehicles is the fact that for the most part, there actually isn't a domestic auto industry in the same way that there is in the U.S. Of course, there are factories, but most of those factories are producing cars on behalf of American brands or foreign brands. Canada doesn't have a homegrown automaker. Just all of those factors together, I think are important to understanding why it is that we're seeing this divergence when it comes to autos today.

Savannah: Thank you, Eliot, for your entrepreneurial reporting and sitting on that anecdote until just the right time. Which we're very lucky is this podcast. So let's zero in a little bit on the security concerns. I mean, U.S.-Canada relational breakdown aside, a lot of the U.S. actions have been motivated by data collection technology from Chinese EVs and security concerns around that. You'll be covering this in our upcoming issue, of course, but let's give a little bit of a preview now. You've identified one company that could prove particularly problematic for Canada. Can you give us an overview of that company and why it's raising alarms?

Eliot: Early on in this podcast, I mentioned the name Leapmotor. This is a Chinese EV company. It's based in Zhejiang province. In Chinese, its name is Lingpao. Most people probably haven't heard of Leapmotor to the extent that people have heard of Chinese EV brands. Really, it's companies like Geely or BYD that have really dominated the headlines. But Leapmotor is no slouch.It made 600,000 cars last year, but it's also controversial because it has had longstanding ties to a controversial Chinese video surveillance company called Dahua Technology.

Dahua has been blacklisted in the U.S. and the European Union, as well as a number of other countries which have accused it of aiding repression and surveillance in China, particularly of the Uyghurs and Muslim minorities in Xinjiang. In the U.S., Dahua has been added to the 1260H list, which is the Defense Department's list of Chinese companies that are alleged to be contributing to China's military-civil fusion strategy.

Leapmotor was born out of this company in 2018. It was basically a spin-out. Its largest shareholders were Dahua and Dahua's two co-founders.And so at a time when the U.S. and also a number of other Western countries are growing concerned about the growing data and cybersecurity implications of an increasing number of Chinese connected vehicles rolling out on Western roads, it's notable that here we have an automaker with extensive connections to this highly blacklisted surveillance firm that is also actively considering setting up in Canada.

Savannah: So how close are these connections between Leapmotor and Dahua?

Eliot: When Leapmotor was founded in 2015, as I mentioned, its largest shareholders were Dahua and its two co-founders, Fu Liquan and Zhu Jiangming. In addition to that, Dahua has sort of extensive commercial relationships with Leapmotor. In 2023, when Stellantis bought a 21% stake in Leapmotor, Dahua sold its direct stake in the company, but at the same time has maintained personnel and commercial relationships with the company. So for example, several of Leapmotor's vice presidents previously worked at Dahua. One of Dahua's co-founders is now chairman of Leapmotor. And the two companies together jointly developed an AI chip that is used for autonomous driving capabilities called Lingxin 01. And that chip is embedded in many of Leapmotor's cars.

Savannah: For those wanting to know more on what this could imply for Canada, Eliot's story will be published on thewirechina.com this Sunday evening, Eastern Standard Time. We'll leave the rest of your reporting for our readers on Sunday. But one final question from me, Eliot. The context of all of this is, of course, Chinese EVs' seemingly unstoppable global rise, as our readers and listeners are no doubt aware.We've covered the major players and expansion into markets from Southeast Asia to Latin America for some time now, but North America was always a nut that China's car giants could never crack, at least until now. Is the world just in an unstoppable wave of Chinese automobile technology that we can no longer avoid? Was Canada's decision to allow Chinese EVs inevitable?

Eliot: So this is a story that is continuing to play out in real time. And I think what's interesting and exciting about it is that it's also just changing really quickly. On the question of whether this was inevitable, and whether these cars might be coming to other countries in North America, I think I would just add that one interesting dynamic that we've also reported about in recent articles is how the U.S. is going to respond to the prospect of an increasing number of these Chinese vehicles, but also Chinese auto factories coming to North America. One of the things that seem to have captured President Trump's attention, for example, is the fact that some of these automakers want to build here. And we heard from him in January at the Detroit Auto Club saying, let them come in, let them build a factory in the U.S. And that's created a really interesting dynamic here, because even though that hasn't happened yet, I think a lot of people in the industry are waiting to see whether a deal will come out of his visit to Beijing in a couple of weeks. But also it could create a sort of competition, a bit of a bidding war for these factories in North America that could really play into the hands of the Chinese automakers.

There's a lot of complicated dynamics that could really tip the balance. In Canada, for example, the unions have a really important voice when it comes to how factories are set up and also the extent to which the cars will be fully built here. Whereas in the U.S., President Trump may not be so picky. He could offer up, for example, as one analyst noted to me, just a plot of tax-free subsidized land in Alabama, where historically unionization rates are lower and automaking has really thrived in the last few years. And so if Trump goes to Beijing and announces this deal, we could see the entire landscape change all over again. And this talk about where these EVs are going to show up, there could yet change.

Savannah: We might not all be driving BYDs or Leapmotors next week, but there might be a day where the tradition of the Great American Road Trip is undertaken behind the wheel of a Chinese car sooner than we think. Thank you so much, Eliot. Eliot's reporting on Leapmotor will be published on thewirechina.com this weekend, along with an op-ed on Xi Jinping's perplexing vision for the Chinese economy by Yanmei Xie, an essay adapted from Eyck Freymann’s new book, Defending Taiwan, and more. To get full access to all of our story archives and read Eliot's reporting this week in more detail, you can visit us at thewirechina.com/subscriptions. Or if you prefer to first dip your toe in the water, you can sign up for our free newsletter on Substack or thewirechina.com/newsletter.

As usual, we'll end this episode with a preview from our upcoming Q&A, which is a weekly column where we chat with some of the keenest minds on China today. Our Q&A this week is with Martin Thorley, an academic who has written about the so-called golden era of relations between China and the UK in the 2010s. In this clip, Thorley explains how it was that a network of policymakers and business leaders in the UK became so enamored of China during that time. And after this clip, we'll conclude with a recommendation for a podcast that we think is well worth your time. Thank you for listening. See you next episode.

Martin: From my time on the ground in China, there was this gulf and I think there was this hard edge of the Chinese party state, sort of political economy. This is the ultimate zero-sum economy, the legacy of decades of battle, a very different beast than a liberal democracy. And it was, I was very fortunate to have experience working with various individuals really in effectively a domestic business environment in China.So you see it close up. And then on the other hand, here you have individuals who have grown up in a liberal democracy coming over to engage, maybe to do business deals. And I think that for me, the lesson there was that in many cases, they were very well-meaning, good people, but totally lacking the tools and lacking the awareness.

I mean, I think it took me maybe a couple of years in that particular position to really understand the domestic political economy. So in terms of why there's a group who's willing to engage in this way, I think there is that lack of appreciation of the fact, or lack understanding of the real Chinese political economy, unfortunately. I think that what you see is not only in the group just mentioned, but also in policymakers who would go over and visit China.

It's almost like they learn, they understand China through the back of a chauffeur-driven car and through an air-conditioned conference room. And I think that is also one of the factors that enabled this approach.

In our latest episode, editor Andrew Peaple and reporter Noah Berman discuss his upcoming reporting on Chinese solar’s booming sales to the continent. Solar proponents spy opportunity to improve electrical grids across African nations with Chinese imports, but are widening trade deficits and China’s anti-involutionary policies going to slow the rush?

Last episode, we dove deep into eyewitness testimony from the Hainan spy plane crisis 25 years ago with audio of an erratic Chinese pilot from a few months before the fatal crash, the U.S. pilot who steered the disintegrating plane to land on Hainan, and the U.S. ambassador to Beijing during the crisis.

In the second part of this episode, we look at how the U.S. intelligence community viewed the crisis through the eyes of Dennis Wilder, who was then CIA China Division Chief.

Our oral history of the incident continues this week on our website, thewirechina.com, where Part II will be released this Sunday evening.

Transcript:

Andrew: Hello and welcome back to The Wire China podcast with me, Andrew Peaple, The Wire’s news editor, hosting this week from the UK, where spring is in the air and the sun is out.

And talking of the sun, in a moment I’m going to be speaking with my colleague Noah Berman about a piece he’s written this week for our latest edition. It’s on China’s burgeoning exports to African countries of solar panels. Before that, just to say that our main cover story this week will be the second part of our oral history of the Hainan spy plane incident 25 years on from that event, which rocked relations between the U.S. under then President George W. Bush, and China, which was then led by Jiang Zemin. We’ll also have an opinion piece on the state of China’s overseas development and one of our Q&A’s with the author Michael Luo about his book on Chinese immigrants into the U.S. from the 19th century up to the present day. Plenty to look forward to in this week’s edition. But let’s turn now to Noah, who’s over in The Wire’s head office in New York. Noah, great to speak to you. How are things there? Noah: Warm here as well, Andrew. Spring has sprung.

Andrew: Excellent. Well, a sunny part of the world is obviously Africa, but it’s becoming increasingly reliant on solar panels from China. That’s what your latest piece is about this week, those growing trade ties between China and Africa, and in particular in the solar industry. So just to kick off, can you talk about just how dramatic the rise in solar exports from China to Africa has been and why this really matters?

Noah: So China’s exports of solar panels to Africa grew nearly 50% last year when measured in gigawatts. That’s a big gain in any year, but it’s particularly dramatic relative to recent history. In 2017, China exported nearly a negligible amount of solar panels to Africa. And this matters because for Africa, it’s a sunny continent, but it generates very little power from solar energy.

So people in the energy sector view it as a region with a lot of growth potential for solar. And at the same time, it matters for China because its exports of solar panels to other regions are slowing. So Chinese manufacturers are actively looking for new markets. And as a representative for the African solar industry told me, that means they’re, quote, betting a lot on Africa right now. So if their bets pay off and if countries continue to buy at the same pace, that will make China a major supplier of Africa’s energy infrastructure in the not so distant future.

Andrew: So what are some of the other main reasons behind this surge in solar exports to Africa just at this time?

Noah: Well, as with anything, there’s supply and demand factors. Africa has a big electrification gap. According to the International Energy Agency, 600 million people, almost double the population of the United States, still don’t have electricity in Africa. So solar companies around the world may view that as a major untapped market. But many African countries don’t have as much money as countries in Europe or other parts of the wealthy world to invest in solar. And for a long time they were priced out of solar panels. But recently, those panels have gotten much cheaper, cheap enough to be competitive with other forms of energy, like fossil fuels. And some estimates suggest that panel prices fell by half in 2023 alone, by a quarter more in 2024. And they’ve really crossed the Rubicon now into affordability that’s helped unlock new markets that may have previously been priced out of solar energy.

Andrew: And I guess solar also helps some of these African countries move more towards sort of climate-friendly electric sources away from fossil fuels. Is that a factor as well?

Noah: Exactly. And you know, solar doesn’t produce the same greenhouse gases in the same way that oil and gas does. And so there’s a big push, or there has been a big push in Africa to drive down greenhouse gas emissions, and solar could be one way to help out that effort.

Andrew: Do you think this is something that’s been driven by the Chinese government sort of telling Chinese companies, solar companies to export more to Africa, or is it something that’s coming from the companies themselves looking to sell more into Africa?

Noah: I don’t know if the Chinese government is explicitly telling solar companies to export more to Africa, but it is certainly happy that companies in general are diversifying their exports. In fact, in the most recent five-year plan, which was released last month, it’s said that efforts to diversify exports delivered notable results. And of course, this comes at a time when exports are a bright spot of China’s economy, which is suffering some from some other problems like high unemployment. And for solar companies in particular, they’re dealing with slowing growth in developed markets like Europe and the United States. And as we mentioned, Africa is a bright spot. So they have their own incentives to sell on to Africa, which are potentially aligned with Beijing’s.

Andrew: Is there any risk here? Do you think that the more African countries import solar panels from China, the more dependent they become on the Chinese solar industry in the same way that countries can become dependent on other nations for their oil and gas? Or is that not really a factor here?

Noah: So I think that it differs from oil and gas in that local companies are heavily involved in the distribution of electricity. So that the potential dependence here is on the ability to buy uh the infrastructure, cheap enough panels or the components used to make them. In that way, African countries may be dependent still on China in the same way that many uh countries around the world are dependent for all sorts of manufactured goods. It seems likely that if solar does take off in Africa in the way that proponents of the industry hope, countries on the continent will be dependent on buying cheap enough panels from China. But whether they worry about that probably depends on the country. You’ve a wide range of economies of different sizes in Africa. Some of them are smaller and they don’t have local manufacturing of solar panels, they may be content to keep buying panels for as cheap a price as possible, regardless of where they come from.

Andrew: Yeah, I see. I see. You did some calculations, right? And at the current growth rate, if it carries on the way it is, solar from China is going to be a pretty key part of Africa’s energy mix in not a very long amount of time.

Noah: Yes, absolutely. If sales continue to grow at the same pace, then probably by the end of the decade, China could become a significant player in Africa’s energy infrastructure. A lot of that infrastructure could be imported from China.

Andrew: So, what about the kind of broader macro risk here? We’ve seen in the past years that the US, Europe, other places pushed back against Chinese solar and Chinese solar panel imports to varying degrees because they’re worried that essentially China is dumping the solar panels into their markets and reducing the competition from local players, you know, getting an unfair trade advantage. Do you think there’s a possibility that African countries will come to push back? That at the moment they’re sort of welcoming this export surge or this import surge from China, but in the future they might come to see it as something that’s undesirable because it’s just increasing their trade deficit with China?

Noah: I think it’s helpful to think about the macro trading relationship between China and Africa before zooming in a bit. The trade imbalance is growing. Last year, Africa’s trade deficit with China reached a record high. It was 65% higher than in 2024. And certainly in some of Africa’s larger economies, the South Africa’s, the Kenya’s, maybe Nigeria, frustration could begin to bubble or may already be bubbling. They may worry about local manufacturers’ ability to compete, including in industries like solar, with echoes of what we’re seeing and hearing in Europe and the United States, where there’s high tariffs that effectively keep solar panels made in China out of the market.

China is making some efforts to balance things. Next month it’s going to, it says that it’s going to drop tariffs on 53 African countries to zero. But the trading relationship in terms of what China buys from Africa and what Africa buys from China is pretty uneven. Some analysts are skeptical that a move like dropping tariffs to zero will have a big impact. China primarily buys natural resources from Africa, sends back finished products like solar panels that are worth much more. And to top it off, Chinese companies are pretty active in extracting natural resources from Africa. So a good chunk of those exports from Africa to China could be between Chinese companies already. And Africa may have more leverage if its countries bound together, but the African Union is pretty widely regarded as weak.

And as we mentioned, countries have loads of different incentives. So even if the larger ones that may be at risk of losing manufacturing jobs, for example, are not in agreement with smaller ones, then the African Union may be unlikely to push back much on China. And even those larger economies are just a fraction of the size of China. So all of that points to, I would say, an unlikely chance of Africa pushing back in the same way that the US and Europe have on Chinese solar panels.

Andrew: In Europe, to an extent we have the same problem, which is that often it’s not in the interests of individual European countries to push back against China, but it can work if the countries come together and work through the European Union. So it’s pretty interesting. Thanks for that, Noah. It’s a really fascinating situation. I think this whole relationship between China and Africa in trade and investment is one that uh we’re going to be keeping more of a close eye on. So thank you for that.

Just while I have you, Noah, I wondered if you could give us a quick word on the second part of the Hainan cover story. Absolutely fantastic story that you and the other reporters put together last week. And I think you’ve got more coverage of that week. Can you just run us through some of the highlights that we might have in the second part of the big spy plane incident that happened 25 years ago?

Noah: So this week we’ll have interviews with more people involved in the Hainan Island incident, including two of the detainees who were in different roles in the US Navy at the time, who were on the U.S. Navy spy plane, a couple of more U.S. political officials, and one so-called freelancer who flew from Hawaii to Hainan on his own to try to help resolve the incident and bring both sides to terms.

Andrew: Well, up next we have a recording from one of the interviews that we did for the latest episode of our Hainan uh cover story. It’s with Dennis Wilder. Dennis was chief of the China division at the CIA at the time of the Hainan spy plane incident. And after that, we’ll share a recommendation for another podcast which we think is well worth your time.

Dennis: Wang Wei was a “hot dog” pilot and he was well known to be a “hot dog” pilot. He had even put his email address in the window of his cockpit at one point and shown it to the people on other EP-3 planes.

The maneuver he was doing is a standard military maneuver that pilots like to do to each other that is highly dangerous and not recommended, which is called the thumping maneuver.

And what you do is you bring your aircraft up under the other plane and you lift quickly in front of the plane. What that does is create an air pocket, and then everybody else on the other plane gets all shaken up.

Vietnam-era pilots used to do it to each other in the US military all the time. It was a game. Okay?

The difficulty this time with the game is the F-8 aircraft is a dog. It’s a terribly unsophisticated aircraft. It’s a knockoff of the MiG-21. And the problem is that when you fly that aircraft at low speed, it becomes unstable. And in order to fly at the same speed as an EP-3 turbo, you have to fly slow. And when you do that, your stability is not great and he just miscalculated the distance and consequently came up and you know, took off the nose cone of the EP-3 and then killed himself, basically, by crashing into the ocean.

One of the things the White House was asking us was this some sort of test of the new president? People may not remember Bush is a governor. He wasn’t a foreign policy expert. He was the governor of Texas. And he had just come into office. The team was barely in place at the White House at that point in time. And thus they had to wonder whether this crisis was not something that the Chinese desired to test the mettle of a new president.

Jiang Zemin, he had to deal with his public who were angry but with the foreign policy necessity. So what we felt he did was in the initial period, he decided that he had to make a big deal of mourning the pilot. They did this massive search at sea for the aircraft and the pilot. The United States offered to assist and they said no way.

But there was huge effort and huge publicity. They took the wife and child and had them go out to sea and put wreaths in the water. And then they had a huge ceremony in the Great Hall of the People making him a martyr, you know, with Jiang Zemin there. They needed to mourn him in order to assuage the Chinese people so that the Chinese people didn’t attack America, but also had a chance to say how terrible what we did was and to make this guy a hero.

The frustration at the White House, I will tell you, was that we had to live through the mourning period. And consequently, I was under great pressure to tell them how long this mourning period was going to be. And I couldn’t tell them how long that was going to be. I didn’t know. I mean, we felt it would be, you know, in the range of a week based on other things from the past, but we couldn’t guarantee anything.

The key moment for us actually came when Jiang Zemin went to South America during the crisis period. That to us said he doesn’t want a crisis. If he were really worried about a crisis, he wouldn’t be going on his six nation tour of Latin America. And so he left on 4 April for Latin America, which we said to the White House is a pretty clear sign that that’s not what he wants.

In this special episode, The Wire China kicks off our two-part oral history series on the Hainan spy plane crisis, a defining diplomatic test for both nations at the beginning of the so-called “Asian Century.” Editor Tom Mitchell, reporter Rachel Cheung, and The Wire China team bring together the voices of the officials, pilots, diplomats, and military personnel who were there to understand how they navigated the crisis — and if such an incident might occur again.

The full oral histories will release in two parts on thewirechina.com over the next two weeks.

Cover image: Members of the detained EP-3 crew salute as they board a chartered aircraft which will fly them from Haikou, China, to Guam. Credit: DoD

Transcript

Tom: Hi, I’m Tom Mitchell, The Wire China’s features editor. This week we are publishing the first of a two-part oral history series to mark the 25th anniversary of the Hainan spy plane crisis.

On April 1, 2001, a Chinese jet fighter and an American EP-3 reconnaissance or spy plane collided over the South China Sea near the tropical island province of Hainan. The jet fighter plunged into the sea and its pilot, Wang Wei, was never found. The spy plane almost crashed as well, with 24 crew on board, but the pilot, Shane Osborn, was able to pull it out of its near-fatal dive and make an emergency landing at the same People’s Liberation Army airbase that Wang flew out of.

George W. Bush’s presidency was supposed to be the first American presidency of the Asian century, and this crisis occurring less than three months into his administration seemed to confirm that. It would be a tense 11 days on Hainan before Beijing and Washington negotiated a resolution and the PLA released the American crew. But memories of the high drama over Hainan, an incident that appeared to confirm that China and America’s emerging geopolitical rivalry would be the defining feature of the new century, faded fast after the September 11th attacks on New York and Washington. Rather than a China-U.S. military showdown over Taiwan or some other hotspot, the Bush, Obama, Biden, and Trump administrations have instead presided over a quarter century of U.S. wars in the Middle East that continue right up to this very day.

I should also confess here to a deep personal interest in this incident. At the time, I was the Guangzhou correspondent for the South China Morning Post. The scene at the hotel in Haikou, Hainan’s capital, where an international press pack and a six-member U.S. diplomatic team stayed was chaotic when I arrived. When I opened the door to what I thought was my room, I found it was already occupied by Duncan Hewitt, the BBC Shanghai correspondent. I made my excuses and went back to the hotel front desk where a harried staff member accidentally gave me a room on the same floor as the U.S. diplomatic team. The floor was supposed to be reserved for that team. There were no other journalists on it. And the Chinese guards posted on that floor didn’t seem happy about my appearance, but nobody insisted that I move. What followed was for me a fascinating front row seat on the crisis.

Our first interview is with the American pilot Shane Osborn by our staff writer Rachel Cheung. I will also be talking with Rachel later in this episode about the memoir written by Ruan Guoqin, the widow of the Chinese pilot Wang Wei. At the time, the couple had a young son who graduated from university two years ago and joined the Chinese Navy.

Before the April Fool’s Day collision in 2001, there had been other close calls between U.S. spy planes and the Chinese jets that flew out to challenge and monitor them. The following audio clip is from one such encounter a few months earlier in January of that same year, when a Chinese J-8 fighter plane veered close to a U.S. spy plane.

And next from our interviews, the crash, as described by Shane Osborn.

Shane Osborn: When they got really close, I knew it, but usually they’d come out, they’d intercept us, they’d break away, and then they’d leave. Okay. So they come out for a while, stay off our wing, and leave. In this case, they all of a sudden started coming back again. And we were already heading home, and I was like calling up the navigator going, confirm we’re in international airspace. Because that was my first concern is that we’d wandered off. You know, this airplane back in that day did not have the most sophisticated system. The only GPS they had was back at the nav station, so the pilots had no access to that. That was one of my concerns because I couldn’t believe how aggressive they were, you know, they’ve become a little more aggressive lately, but nothing like this. He came in and he came in too fast. And so he tried to slow his aircraft down by pitching the nose up and kind of making it you ought to kind of slow it down. Well, when you pitch your nose up, what happens? The aircraft goes up. Well, he started out below my wing and quickly rammed into my wing. And that was you know, right where his tail beats his fuselage is where my prop, my far left engine, literally cut him in two. His tail continued up, tore through my aileron, and then his nose came apart. We were flipping inverted because of the impact on my left wing. As we rolled his nose, we had a separate impact. His nose hit my nose and tore my nose off. And so next thing I know, I’m in this big lumbering aircraft upside down, looking up at the South China Sea, and I thought we were dead for sure.

Tom: Osborn then pulled off his miraculous landing. His radio requests for permission to land were never answered by the Chinese military. He also remembers that a second Chinese jet fighter that was not involved in the collision took up position behind his crippled plane and requested permission to shoot it down. Permission to do so was not granted.

President Bush: That I’ve just talked with Brigadier General Sealock, who earlier today met with our 24 men and women in China. The general tells me they are in good health, they suffered no injuries, and they have not been mistreated. This is an unusual situation in which an American military aircraft had to make an emergency landing on Chinese soil. Our approach has been to keep this accident from becoming an international incident. We have allowed the Chinese government time to do the right thing. But now it is time for our servicemen and women to return home. And it is time for the Chinese government to return our plane.

Tom: Our next excerpt is from the U.S. ambassador to Beijing during the crisis, Joseph Prueher. Prueher was also a former U.S. Navy admiral and pilot and led the U.S. negotiating team. He told our staff writer Noah Berman the following.

Joseph Prueher: But their immediate five demands were or accusations were you invaded our airspace, you rammed our airplane, uh, you landed without permission, we demand an apology, and we demand reparation. And I responded that I disagreed with every point. We spent a lot of time the first couple of days talking about whose fault it was. You know, they said your airplane rammed our airplane. Well, I had an advantage here that I didn’t usually have is that I knew what I was talking about. Yeah, I was a pilot, and did test work for three years. I’d rendezvoused on Russian bears and badgers myself and been in command of other squadrons and stuff. I knew a lot about it, and I knew that the big airplane didn’t go ramp to other airplanes. And uh I could throw language at them that they were unfamiliar with about aviation. They just backed away from making that point. And also, we wasted two days talking about whose fault it was. Most of the rhetoric on it afterwards just caused a collision. And it was clear that their airplane hit ours. And we also had some films of the pilot, the Chinese pilot, whose name was Wang Wei, but of him making rendezvous previously on other airplanes that were barely under control and uh and doing some things that where he could be recognized, and they didn’t want those films to come out, and we never used them publicly at all. So we took whose fault it was out of the discussion. We had hard information on where the collision occurred, and it wasn’t international airspace. That left the landing without permission. Well, our guys had called for permission, but they did not get a response. And so that left the issue of apologies, and I said, you know, we don’t have anything to apologize for. This went on for a while. Finally, I said, We regret that you lost the airplane, and we went through a long bit of linguistic nuances about the difference between regret and being sorry. Finally, to get an agreement, I agreed to say we were sorry that their pilot had been lost, and we were sorry that they had not heard us call for landing.

Tom: Of course, one critical voice missing from any history of this event is that of Wang Wei, the Chinese pilot.

Chen Ci: Over 800 fishing vehicles from Hainan, Guangdong, Guangxi, Hong Kong, Macau, and Taiwan joined the rescue. We will do whatever we could, no matter what it will cause, to search to search for and rescue the pilot.

Tom: That was Chen Ci, a Hainan provincial government official, speaking to journalists on April 8th, 2001. At this point, the search for Wang was continuing. Two years ago, Wang’s widow, Ruan Guoqin, wrote an often-touching memoir about her husband and their relationship. Miss Ruan’s publisher did not respond to our request to speak with her. Wang’s parachute was deployed in the accident, suggesting that he may have survived the collision and landed alive in the water in difficult sea conditions of forced five to six winds and three-meter waves. Joining me now is Rachel. So, Rachel, tell us a bit about Ruan’s memoir.

Rachel: So, the bulk of the book is personal letters between Ruan and Wang, when Wang was in military academy and later became a pilot. They were high school sweethearts, um, but after they got together, they were separate most of the time because Wang was enrolled in the military academy since he was 18. Based on these letters, Wang was clearly very passionate about flying. In one letter to Ruan, he wrote that learning to fly is very romantic, but it is mostly hard work. I’m not afraid of the hardship. What I fear is having no emotional anchor. I’m happy because I’m loved by you.

Tom: Now the US always maintained that Wang was a reckless flyer and caused the accident. China says the opposite. He was a seasoned professional with a great safety record. What light, if any, does the book shed on his flying?

Rachel: So some of the letters were written while he was a cadet, um, spending five years at academy training to become a pilot. He was one of the best pilots at the time. Um he wrote about his training, how he found solo aerobatics very thrilling. While some of his peers were told to fly further away and try to become more bold, he was actually asked to remain in the inner airspace, so he’s not trying out some of these guns that he was pulling off at the time. But in one letter he wrote to Ruan, he talked about how he performed 18 barrel rolls. The binoculars he wrote at the control tower were fixed on me, and the instructors were a bit shocked. Consequently, I’ve become quite famous for having a strong physique. He would also go on to fly different kinds of planes to train and adapt and learn how to fly different planes. According to Chinese state media, according to the memoir, they also emphasize Wang’s subsequent safety track record. By 2001, when the accident happened, he had flown 2,000 sorties and accumulated over 1,000 accident-free hours.

Tom: And how was it that Wang ended up pursuing a military career in the first place?

Rachel: According to the memoir, he has wanted to join the military since he was a child. A lot of his cousins were in the military. His aunt and uncle fought the Japanese and were a part of the long march. The opportunity for him presented itself when the Air Force came to Huzhou, where he lived and grew up, to recruit in 1986. He was 18 at the time and he went behind his parents to apply on his own, probably knowing that his parents would not agree with the decision. The Air Force made them go back, made him go back to their parents and get their permissions. He actually went to their unit, set there until they would give their approval, and that’s how he became a cadet.

Tom: So, Rachel, this memoir by Wang Wei’s widow, why did Ruan write it? And how is Wang Wei remembered today?

Rachel: So it is actually published last year. According to the epilogue, when she introduced a book, she got the idea of writing this book because of the 100th anniversary of the Chinese Communist Party. And she spent a couple of years compiling the letters and sort of going over everything and writing this book. And I think he wants Wang Wei to be remembered and also to be an example for others, including her son, who has gone on to become a naval officer to pass on Wang’s legacy.

Tom: Thank you, Rachel. And to our listeners, please do check out the first of our two-part series, which will go online this Sunday evening, U.S. East Coast Time. This week’s issue also features a review of the open claw AI craze in China, written by our very busy Rachel Cheung, profile of XPeng, and an interview with Neil Shearing, author of The Fractured Age. We’ll leave you with a recommendation for Face Off, an excellent podcast by fellow journalist Jane Perlez. Thank you for listening.

In our eighth episode, reporters Rachel Cheung and Savannah Billman discuss Rachel’s upcoming story on where the United States and China score across the stack of technologies and capabilities that make AI possible.

We also have an excerpt from our upcoming Q&A with veteran diplomat Sarah Beran on the role of U.S.-China diplomacy in an evolving international environment. You can read these stories and more at thewirechina.com.

As she discusses in the podcast, Rachel has written several excellent deep dives into how AI works in China across different industries. Read more of her work here:

* Could Chinese AI Videos Kill The Hollywood Star?

* China is Churning Out a Growing Army of Service Robots

* Manus, Meta and a Deal That Goes to the Heart of the U.S.-China AI Rivalry

* China’s Labor Market Braces for an AI Shock

* China’s Search for an AI Magic Cure

Transcript

Savannah: Hello and welcome to The Wire China Podcast. I’m Savannah Billman, usually behind the scenes as the producer of this podcast, but today speaking with my colleague Rachel Cheung about her upcoming scorecard on the U.S.-China AI race.

We’ll also have a Q&A with Sarah Beran, former Deputy Chief of Mission at the U.S. Embassy in Beijing, with a behind-the-scenes look at U.S.-China diplomacy and her predictions for the future of the relationship.

These days, AI is inescapable. It’s hard to imagine a time before it was the hottest thing in U.S.-China competition, before the DeepSeek moment last year when a little-known Chinese AI startup catapulted Chinese AI into the international spotlight.

It’s about time that we tallied everything up and figured out where exactly the U.S. and China stand. Rachel, so glad you’re here to guide us through this scorecard.

Rachel: Hi, really happy to be here.

Savannah: An AI scorecard is a fairly straightforward format. What can we learn from ranking an entire industry like this, and what are the limits to this format?

Rachel: I’m really glad you asked about the limits. Perhaps I could start by telling you why we’re doing this story now, and it’s not just because it’s one year right after deep-seek, but more that for the longest time when we talk about U.S.-China AI race, it’s about who’s building the smartest model.

So it started with OpenAI’s ChatGPT, and then there were expectations that China would be behind, and then with deep-seek, it seems like China is able to be a very fast follower and really catch up. But I think we’re at a point where we’ve moved from just talking about the smartest model to delivery of AI as a product. So it’s not just about chatting with AI, but it’s also integrating it into your regular workflow and making something out of it.

And that means it’s no longer just about the smartest model, but also people and users and companies are also looking at the subscription prices, the speed, the reliability, is there going to be any limits to how they’re going to use it. And when it comes to delivery of AI as a product, it requires reliable inputs in every layer of what the industry usually calls the AI stack. And in this story, we classify it according to the way Jensen Huang of NVIDIA likes to classify it.

So at the bottom layer, you have energy, and then you have infrastructure, which is the data centers, you have chips, you have models, and the final layer on top is the application that we see. In terms of limits to this format, we try to find and source data to best capture where each country is in each of these layers. These metrics, we’re trying to get a picture that is as close and as accurate as possible, but they are not perfect metrics.

I think that is sort of the first caveat there. And so there are also a lot of additional points to really flesh out what each country is doing in each layer, the strategies, how are they different? Secondly, this is just a snapshot of where we are. Things change very quickly in this industry.

It’s very dynamic. It’s far too early to declare who is the winner, whether it’s a country or it’s a company. And so this is just where we are today.

And the last thing, when I speak to people, you hear about the China-U.S. AI race a lot, but what’s the definition of victory? It’s different for everybody. What does it mean to win? Is it just capturing the biggest market share possible? Is it building the best model? Is it the money they’re making from this or the productivity? So there are different ways of calculating it. And so I think those are sort of the limits we have to this format.

Savannah: As you mentioned, the scorecard doesn’t just examine different models. The market has moved past that by now. You’re examining the stack of technologies and capabilities that make the entire AI industry possible.

There are five levels, energy, infrastructure, chips and computing power, models, and applications. So in which one of these five categories does China have the biggest advantage?

Rachel: There’s no doubt China’s advantage is in energy, the bottom layer. We have seen, well, the argument we’re hearing is that because the marginal cost of AI is just the cost of electricity, China has abundant power.

So it has the upper hand. I think that is partially true, but there are also some caveats to this argument. China has been really ramping up its energy capacity to meet its growing demand.

And so the demand that comes from data centers is actually smaller than, for instance, manufacturing or just air conditioning by regular residents. And so in order to prepare for that demand, it has been ramping up like crazy, building renewables and so forth. By comparison, U.S. energy capacity has remained largely stagnant in the past decade.

We see executives like Jensen Huang, Sam Altman of OpenAI really petitioning the U.S. to do more. There’s a lot of red tape. And I think the U.S. is trying to work on that.

So it’s true that China has abundant power, but there are also some unknowns or question marks over here. Ideally, you have electricity from all kinds of sources, thermal plants, renewables all go into a large pool and everybody can draw from it when they need it. That’s how you distribute power most efficiently.

Right. But in China, the way it works is by a lot of fixed contracts. And so there are still a lot of institutional bottlenecks that determine who can use what kind of power or how it is being delivered to households and consumers and factories and so forth.

And so even after adding all these capacities, there’s the question of how it’s being distributed. And I think these bottlenecks would mean electricity, at least the price of electricity might not be as low as what some might expect, especially compared to Europe and the U.S. And so how big of an advantage China has in here, I think there’s also a question mark over there. And the other part is ideally, you would have renewable energy to supply data centers at some point, or at least part of the demand.

But renewable energy is intermittent by nature. So I also spoke to a scholar who visited some of these data centers in China that built themselves as green projects, but they are still at the moment relying on coal power. So how do you bundle renewable with thermal power at a point so that it’s stable enough to supply these data centers? I think there’s still quite a bit that needs to be done to sort out in order for that to happen.

Savannah: What about the U.S.? Where does America’s advantage lay on this scorecard?

Rachel: So the U.S. has a lot of advantages for the ecosystem, from the way it works. But from just sort of this layer by layer evaluation, it’s clear that the U.S.’ biggest advantage comes from its export controls on chips and related equipment to China. And so I spend a lot of time figuring out if we should discuss data center or chip first, because those two layers are very correlated.

Well, if you build all the data centers and you don’t actually have the chips, they’re just empty buildings, right? And so because the U.S. is able to restrict the supply of advanced chips for AI to China, and China is facing this, it’s essentially a scarcity in computing power. There is a global shortage right now.

But I think the situation in China is far more acute compared to the U.S. We’ve seen companies from AI model companies to even robotaxi companies talking about how this is impacting their operations. We’ve seen Zhipu or ByteDance struggling to deliver their services and making sure they’re stable because they don’t have enough computing power. And earlier this year, we also saw Chinese top executives come out in a conference and really sort of petition the government to say, we are running out of computing power.

Just the amount of computing power we have from the chips we have, we could use. It’s just enough to deliver the services, which is AI inference. And so it’s not enough to train a model.

We also see these petitions. And so there’s the question of whether NVIDIA would be able to sell some chips to China, whether that could change the situation. And I think that remains unseen.

Savannah: The technology is obviously very exciting, but these companies still have to make money. That’s covered in the final level of the scorecard application. So how are AI companies making money?

Rachel: They are making money. Revenues are soaring. And as adoption continues to grow and as they have better pricing powers, we’ve just seen Alibaba clap to increase their pricing. I think the revenue is going to go up.

The revenues for Chinese companies, still a fraction of the U.S., but they are growing rapidly. The question is, will this be enough to cover the amount of investments they’ve made? In other words, is this revenue sustainable or is their business model sustainable? And I think that’s a question both in China and the U.S.. In the U.S., we have seen these worries about an AI bubble rock the market.

In China, we’ve seen these AI companies rushing to go public because their investors are sort of weighing on their shoulders. And so they’re in the public capital markets where we could see their numbers. We’ve seen sort of their IPO earlier this year, and the question that was being raised over and over again is, great, you have raised money, your revenue is growing, but we also see the amount you’re burning in R&D, and when is that going to cover that?

Savannah: We won’t go layer by layer. We’ll leave that for when your article is published this weekend, and everyone can go on and read your detailed insights into each layer of that stack. But let’s take a look at China’s just approved 15th five-year plan.

AI is a major component of the high-level vision for the country over the next five years. What does Beijing want to see happen with AI? And what would an ideal high score look like for China?

Rachel: I’m actually really curious about how Chinese authorities internally are doing their own scoring. But I would say that, for instance, if we look at the five-year plan, AI is mentioned 52 times.

That’s compared to six times in the last one from 2021. I think that shows AI is definitely a priority. As Martin Chorzempa of Peterson Institute for International Economics recently wrote, he analyzed the paper and sort of made some comparisons between China and the U.S. AI’s plan, and there are actually a lot of similarities.

For instance, both really emphasize defusing AI, essentially promoting AI across sectors. So they’re adopting AI to really boost productivity growth. In China, the biggest difference is that the government plays a bigger role to ensure companies will have reliable inputs, whether it is chips or data centers.

There is, for instance, a national-scale compute planning project going on. People have also petitioned the government to do more kind of planning to direct and organize the data centers in such a way that people will have the services they need.

I think China recognizes the strategic value of AI, and the key point is not just to build a competitive supply chain, but to make sure we have a domestic supply chain that is reliable. So at no point of the supply chain, whether that is chips or other parts, will there be choke points where another country, the U.S., could stop or halt its progress.

And the other, I think, would be something that Beijing want to see happen with AI is for it to really lift other parts of the economy, to see it brings real value to the economy, and not just be this little tiny sector and the whole economy that is doing very well, but really able to sort of solve some problems they’re facing, whether that’s in healthcare or other sectors.

Savannah: So Rachel, you yourself are no stranger to reporting on AI in China, and in just the last few months have written some amazing pieces on AI in China’s healthcare system, labor market, service robots, and both on the ground and flying cars, all of which are, of course, available to read on thewirechina.com. So let’s score up your own coverage.

What have you learned about China’s AI industry by looking at it from so many different vantage points?

Rachel: That’s a really tough question, but sort of the way I see it is when I approach a subject, I’m trying to separate the hype from reality, especially with a topic or an industry like AI, where you do have a lot of enthusiasm, some of them very well-oriented, but when AI is actually deployed, it really runs into some problems with each of the industries. It runs into problems within companies. It is interacting with a lot of forces.

How do we adopt AI in a way that is responsible, is safe, and that actually brings about real impact? That is sort of the hardest part. That is harder than training a very smart model. Why that is important is, when we see adoption of these AI tools and technology, you usually hear from companies, and they’re very, very optimistic.

There are a lot of claims about what these AI technologies can do, a lot of imagination about how they can solve a problem. And in reality, when they’re adopted on the ground, there are often challenges. There are often caveats, whether it’s about safety or concerns, privacy.

So there are a lot of things that have to be sorted in reality. And the other thing that really stays on my mind is the rise of AI anxiety I’m seeing both in the U.S. and China. In the U.S., in just the past few weeks, we’ve seen reports of Meta, Oracle, Amazon laying off quite a number of staff.

We’ve seen Jack Dorsey fired 4,000 employees from his FinTech company because AI has changed the way companies are organized these days. In China, we’re not seeing sort of similar waves of layoffs yet, because I don’t think AI is at a point where it could replace labor in China, especially given the low labor cost. It’s also harder for a Chinese company, because of their social responsibility, to just lay off that amount of people.

That being said, and generally, when we look at surveys, Chinese users, Chinese residents, they’re more optimistic than their foreign counterparts about the adoption of AI in the tool. That being said, I still think there is an increasing fear about being left out. People are facing a very tough job market.

You see, for instance, we see a lot of AI adoption in major cities. But when we talk about the diffusion of AI, it’s also about the smaller cities, the fourth-tier city that people might not have heard of, about whether those people could employ these tools eventually. There is a question mark here about the path for the technology to travel from the frontier labs, from these companies, to these ordinary users.

I think there is worry about the disruptive impact that AI would bring, and that anxiety is also on the rise in China.

Savannah: Thank you so much, Rachel. The complete U.S.-China AI scorecard will be published in our next issue of The Wire China on Sunday evening, Eastern Standard Time. Please do read her excellent work at thewirechina.com.

We’ll also be featuring the full Q&A with Sarah Beran, an excerpt of which will be on in a moment. Sarah has a diplomatic career spanning more than 20 years, which has taken her from President Biden’s National Security Council to the State Department, and finally to the U.S. Embassy in Beijing. I’ll leave you with her thoughts on the evolving but still important role of high-level diplomacy in U.S.-China relations.

Sarah: Least likely, in my view, is wide-ranging agreement on economic, political, militantism.

Fundamentally, I think that’s unlikely because of the depth of mistrust and the lack of real channels other than those that lead.

The second possible scenario is rapid deterioration. This could be precipitated by a crisis of some metal-on-metal incident where Americans or Chinese soldiers die in Taiwan Strait or in the South China Sea.

An airplane goes down, as it did with the EP-3 crisis 25 years ago, and someone dies. Or a massive cyber attack that hits domestic critical infrastructure, civilian critical infrastructure. All of those, I think, could precipitate a real break in relations where both sides escalate with sanctions, with export controls, and potentially shut down channels.

I think that’s unlikely because both sides see a need for stability right now, and I think we’ll work to try to find off-ramps, but it is a possibility.

The most likely scenario, and this has been consistent, I would say, for the last eight years, in my view, is just continued competition, but stabilization. An effort to buy time, to de-risk and diversify away from reliance on each other, to strategically decouple in certain areas.

Not all. Their trade and investment continues, but at least in national security or sensitive sectors, much of that trade probably stops. And body communication between the two sides to just slow the decline.

That’s a pretty dark view, I know, of the future, but I think that’s what’s ahead of us for the next, if we’re lucky, five to ten years.

Transcribed by TurboScribe.

Coming up this week, we have an extract from our latest Q&A with the economist Eswar Prasad about his new book, The Doom Loop: Why the World Economic Order is Spiraling into Disorder, and a guide to the other articles you can read in this week’s edition, published every Sunday evening EST at thewirechina.com

Transcript

Eliot: Hello and welcome to The Wire China podcast. I’m Elliot Chen, one of the reporters at The Wire. And this week I’ll be hosting from my home in Canada, bringing you the highlights from this week’s edition of our magazine.

And today I have the pleasure of chatting with two of my colleagues, Rachel Cheung and Savannah Billman, about their fascinating recent reporting on developments in China’s AI sector, particularly with video generation. Later on, we’re going to have a clip from our new Q&A with the economist Eswar Prasad, the former IMF China head, now a professor at Cornell University.

Prasad talked to my editor, Andrew Peaple, about his new book, The Doom Loop: Why The World Economic Order Is Spiraling Into Disorder and his views on China’s current economic outlook. But first, we’ll turn to Rachel and her story this week on ByteDance’s AI video generator called Seedance 2.0 that’s generating lots of hype. Later, we’ll speak to Savannah about her recent big picture on the challenges facing Chinese AI models when it comes to adoption.

Rachel, to start with you, can we talk a little bit about your story and maybe begin with what exactly ByteDance’s new tool does? Is it as good as people are saying it is?

Rachel: ByteDance dropped a new video generation tool called Seedance 2.0 last month. I wanted to try it out for myself, but there’s a long waiting time on the app right now because of traffic.

But we’ve seen a lot of videos generated with this tool that have gone viral on Chinese social media, on X and on other platforms. You might have seen a video of Kanye West singing in a Tang jacket in Imperial China. There’s a whole music video of it.

And I think what stood out about this tool is that we’ve had video generation tools for a while and they have improved incrementally. They’re getting to a point where you could just give it a text prompt and it will generate a video based on it. But in the past, you could usually tell this is an AI generated video because there’s distortions, the fingers might be crooked, the sound is off, the physics are inconsistent.

Seedance 2.0 gets to the point where it is very believable. One common yardstick you might have seen on the Internet is called “Will Smith eating spaghetti.” With Seedance 2.0, people have generated a video where Will Smith doesn’t just look like Will Smith, eats spaghetti, there’s sauce on the fork.

It’s the sound of people eating, but that he also laughs and talks like Will Smith. And it’s very believable. And I think this is why people found it very striking.

You might have also seen the video of Brad Pitt facing off with Tom Cruise. It’s also generated with a text prompt, and it looks like it came straight out of a big budget movie.

Eliot: And so what’s the controversy around this tool and why are some people concerned about it?

Rachel: So the first worry is just that we now have AI-generated video that is virtually indistinguishable from real video.

It’s very believable. And the other thing is with the earlier version of Seedance 2.0, when it first dropped, there weren’t the kind of copyright filters you usually find in American tools such as Google’s Nano Banana. So one thing, a caveat here is all these copyright filters and safeguards, they’re not perfect.

You could probably generate a copyright material with a lot of tools. But I think earlier on, Seedance 2.0 is a little bit more lax about it. And so people could generate all these things.

There are spoofs and parodies of Hollywood films, of Hollywood actors, of Disney characters and all these iconic characters. As you might understand, film studios from Disney, Netflix and Warner Bros are not happy about it. And so they have issued cease and desist letters to ByteDance and in response, ByteDance has said that they would put some of these safeguards in place.

I’ve checked it earlier. And right now there are some of these safeguards in place. But again, these are not perfect.

Eliot: I suppose it’s understandable why these big name actors are concerned that technology is getting to the point where it can replicate them. But given these safeguards that have been put in place, in your view, is Hollywood’s reaction justified? Is there an overreaction, perhaps? I mean, you’ve mentioned that there’s still this long wait list that you have to get on before you can access the tool.

Rachel: So I would first say that there are still limitations to these tools, whether you’re using it to generate images or video.

There are still very short clips. They’re repetitive. There might be a lot of motions, but it doesn’t necessarily move the story forward.

There’s also a question of is this actually original footage or is it basically a video that is generated or it’s by a model that’s trained on existing visuals? And then there’s also the question of is this original? Is this something that would truly stand out? Or is it just something that is really eye-catching at the moment? Or in the long run, the novelty will wear off very quickly. But it changes the cost structure of film production. It means instead of having an actor, you could tweak or instead of having prosthetics, you could tweak the facial features of someone with AI.

You could generate background very quickly. So I think in the long run, there’s this larger issue of what AI means for the future of the entertainment industry and film production. And I think it’s right to think that the impact is going to be quite disruptive, especially as this technology develops.

The additional layer of complication that I explore in this story is that these are Chinese companies. To go after them, to hold them accountable, to enforce laws across jurisdictions has proven to be quite difficult. It is already challenging in the US. We can see court cases dragging on for a long, long time.

But when you’re doing it across jurisdiction, there is that extra layer of politics. And I think that will complicate things.

Eliot: I think one of the interesting things as well that you touch on in the story is how the reaction to the emergence of this new model has really been pretty different in the US versus in China.

Could you tell us a little bit more about how people in China are reacting to Seedance 2.0? How is that different from these concerns that we’ve been talking about?

Rachel: So Hollywood is up in arms, but in China, we do see an AI generated content industry in China emerge. For instance, I spoke to a 25 year old AI filmmaker who called himself Xiao Kai and he started dabbling in these tools when he was in university. At the time, AI video generation was at a point where it couldn’t even generate a PowerPoint.

But over the years, it has become much better. They’ve become much more realistic. And so he made a film that won an AI film competition in China.

And that has led to some commercial opportunities. And by last year, he had his own studio. He’s making films and now he’s leading his whole team in Hangzhou.

Their schedule is packed till mid this year. They had more requests than they could take on. They’re doing AI dramas.

They’re doing shorts. They’re doing ads. This is just one small example.

There are many more AI studios like this. Some of the most prominent filmmakers in China and creators are also jumping on this trend and pushing out AI dramas. And at least they’re very keen to explore this and find out what’s the utility of this.

Eliot: So I have to admit that when I think about how I use AI, making videos is not really the first thing that I would want to do. I’m curious, though, you mentioned that this could be huge for the industry, for filmmakers. Are there broader implications for China’s AI sector? How big of a deal is this going to be for the Chinese industry in general?

Rachel: So I think what we’ve seen with the Chinese industry is that the big question is, where is the money going to come from? Right.

We’re now at the phase where it’s not just about building the most capable AI system, but how do we monetize it? Where is that revenue going to come from? The consumer side of the business is difficult because consumers don’t want to pay or they don’t have that habit of paying for AI software. Enterprises are also difficult and the economy is not great. There is this urgency of recouping the costs or the investment investments they’ve made into AI.

And so one investor and entrepreneur suggested to me that this emerging AI generated content industry could provide a new revenue stream for Chinese AI companies. And the way they look at it and envision it is to a point where you could have AI avatars playing sports. Instead of watching real athletes, you could have AI sports.

So I think in a lot of places where we might have more reservations or more hesitancy to explore what AI can do in China, because there is not that legacy of an entertainment industry, I think people are more willing to adopt new things. But when I spoke to that AI filmmaker that I just mentioned, he wasn’t all that optimistic as well, to my surprise. And that’s because he also envisioned there could be a crackdown at some point when the government sees sort of the industries, the output, and it will move to regulate it.

And I think we already see some small signs of that. Before the new year, Chinese government, the cyberspace administration removed 540,000 pieces of illicit AI content on Douyin, on WeChat and all these social media platforms. So I do think that the government is watching very carefully.

And so for the Chinese industry, or at least for the AI generated content industry, the backlash might not just come from IP rights holders, the traditional entertainment industry, but from authorities.

Eliot: So it does sound like there are a lot of headwinds facing these AI companies in China on that topic. My other colleague, Savannah, recently wrote a big picture — that’s our graphics focused reporting — about the challenges Chinese companies face precisely on this topic when it comes to diffusing their AI models and getting consumers to actually start using them. So I want to bring in Savannah here. Could you tell us a little bit about that story and what are the challenges facing these AI companies in China?

Savannah: One thing that stood out over the Lunar New Year holiday was the really aggressive advertising campaigns from a lot of these leading AI companies.

We have ByteDance, Alibaba, Tencent. These companies are putting out really popular models in China. Their advertising campaigns over the new year were incredibly expensive.

They were literally giving users money and paying people to download and use their tools. But their regular user base is actually quite low considering the technical capabilities of their AI models. Gemini has over 750 million monthly active users, whereas DeepSeek only has 200 million. Qwen, Alibaba’s open source model, has just 100 million active users.

Now, it’s really hard to measure the real world reach of AI, especially in China, for several reasons. A lot of Chinese AI models are open source, including Qwen and DeepSeek.

So that means people can just download these models and use them and edit them, re-upload them for their own individual purposes. And it becomes really hard to track how many people are using AI tools that are built on these open source models after that initial download. Another challenge in measuring this is that because of these AI companies, you know, they have their separate platforms.

Alibaba not only has Qwen, it also has Alipay. It also has Alibaba, the shopping platform, Taobao. And it can deploy its AI models across all of these different platforms, similar to the way Google just kind of puts Gemini everywhere.

So it’s really hard to measure if users are seeking out AI intentionally or if active user numbers are mistakenly inflated. But there are several metrics that point to this real world adoption of AI in China being lower than expected for the visibility that it gets.

Eliot: That’s fascinating.

Well, thank you very much both. On the topic of diffusion, our Q&A this week with Eswar Prasad also talks about a global expansion, albeit of a very different kind. In Prasad’s new book, suddenly titled The Doom Loop, he looks at how economics, domestic politics and geopolitics have become intertwined and are stuck in this negative feedback loop that brings out the worst in each other, setting off a spiral of instability.

Here’s an extract from that interview where the two of them talked about the role of businesses and financial flows as stabilizing and destabilizing agents in a changing geopolitical landscape.

Eswar: No matter what the rivalries were between the US and China, American businesses, American financial institutions that had a very strong interest in using China as part of their supply chains of selling into the Chinese market when it comes to goods or financial services that, you know, increasing affluent Chinese consumers might demand. There was a very strong interest in maintaining stable ties.

Likewise, Chinese businesses saw America’s worthwhile investment in destination, again, as a production base and also in terms of enabling them to get easier access to certain types of technology. So business flows in both directions of both goods and services and also financial flows, you know, acted as a stabilizing force. Businesses are now retreating back to their home countries, especially in the US-China context.

While they may think they’re de-risking, what they’re doing is removing themselves as a stabilizing force in the relationship between the two countries. So now any time there is a provocation between the two countries, there is no longer a very strong constituency and saying, let’s calm things down because we would really like to keep the relationship on a stable footing. With that balancing force gone, the internal political dynamics in both countries are leading to a ratcheting up of tensions between the two countries.

The once powerful force of businesses operating across national borders has become substantially weakened simply because businesses are rationally trying to reduce risk. But paradoxically, their behavior is actually increasing risk.

Eliot: You can read the full edited transcript in The Wire China from this Sunday evening.

And just to preview the rest of our issue, our cover story this week is from Brent Crane about Chinese investments in Latin America. It’s a timely piece, just as the Trump administration is re-invoking the Monroe Doctrine to advance its interests in that hemisphere. And we have an op-ed from the Asia Society Policy Institute’s Neil Thomas on what to expect from China’s climate policy when Beijing unveils its next five year plan.

Lastly, before we wrap up, in case you’re looking for something to listen to after this ends, one podcast recommendation from our team is Face Off by the Pulitzer Prize winning journalist Jane Perlez. It’s launching its new season this week on the U.S.-China relationship. And here’s a bit more detail from that podcast: https://pod.link/1734890307

Transcribed by TurboScribe.ai.

In our fifth episode, The Wire China reporter Eliot Chen ventures into the criminal underworld with news editor Andrew Peaple. They discuss how Eliot mapped out an international smuggling ring busted late last year for sending advanced chips to Chinese customers. It's a case that reads more like a thriller — and it shows just how far companies will go for an edge in the increasingly competitive race for tech dominance.We also have an extract from our latest Q&A with former National Security Advisor Jake Sullivan on how successive presidential administrations navigated U.S.-China rare earth competition, and a guide to the other articles you can read in this week's edition, published every Sunday evening EST at thewirechina.com

Transcript:

Andrew: Hello, and welcome to The Wire China podcast. I’m Andrew Peaple, the news editor at The Wire, hosting this week from my home in the UK, here to bring you the highlights from the latest edition of our magazine.

In a moment, I’ll be joined by my colleague, one of our reporters, Eliot Chen, who I think is buried somewhere in the snow in Toronto still. Later on, we’ll have a clip from our new Q&A with former US National Security Advisor, Jake Sullivan. Jake talked to Bob Davis, both about his dealings with Beijing during his four years in office, and where he’s worried about the direction things have taken under President Trump.

But first, let’s turn to Eliot and his terrific new story about the case of a chip smuggling ring in the United States, through which advanced chips used in AI and made by NVIDIA were ending up in the hands of Chinese customers. Eliot, hello to you.

Eliot: Hello, good to be back.

Andrew: Great to have you with us. Eliot, I always love all of your stories, of course, but I particularly love it when you get really into the nitty gritty of a good court case and a good kind of step-by-step story. So, really looking forward to talking to you about this.

Just to start with, I think it’s fair to say that the timing of the announcement of this smuggling ring that the US authorities had busted was in itself pretty interesting, right?

Eliot: Yes, that’s right. So, the announcement that the US government had broken up the smuggling ring came the exact same day last year that President Trump announced that the US government would allow NVIDIA to sell its H200 chips to China. So, you may remember that the debate over whether to allow those chips to be sold, it was a pretty fierce one.

And the announcement itself drew a lot of criticism from China hardliners and Democrats. And in part because of that, this smuggling ring announcement sort of became a footnote to that story, which I think actually probably contributed to it getting less attention at the time.

Andrew: Yeah, it’s quite weird on the very same day that they’re busting this smuggling ring, or at least telling everyone about it.

They’re at the same time actually making it legal to export these chips. But anyway, there’s quite a bit of detail in your story that’s coming out in our latest issue. But can you give us the bare bones of how this chip smuggling ring really operated?

Eliot: The part that made this story somewhat difficult to report out just for starters is that a lot of the court documents that were released, anonymized the names of the key companies involved.

And so, I basically spent a couple weeks digging around trying to figure out the identities of these companies based on clues in the documents.

But very simply, the ring was a US-based smuggling operation that brought together a sort of ragtag group of people spanning from Texas to New York and New Jersey to Virginia, which began with a guy in Texas named Alan Hao Xu, starting in October 2024. Xu was basically the money guy.

He took orders for chips from customers in China, who wired money to the account of his company, Hao Global. At the same time, Hao Global negotiated with a large American server company in order to buy Nvidia chips from them. In Hao’s plea agreement, because he ended up pleading guilty, the prosecutors only described that company as a Hong Kong-based global technology company with headquarters in Morrisville, North Carolina.

But it’s pretty clear from that fairly detailed description that this company was in fact Lenovo, a very important and trusted Nvidia partner that helps Nvidia sell its AI chips to businesses.

Xu basically lied to Lenovo, telling Lenovo that he was buying chips for either domestic customers or customers in countries that weren’t subject to export controls like Thailand or Taiwan. And he bought the chips from them, in total agreeing to buy almost 7,000 chips. In the end, he only purchased a fraction of that, but that was the original agreement.

And then sent the chips onwards to intermediaries who relabeled the chips, hiding the Nvidia brand names, hiding the Nvidia labels and barcodes and replacing them with a generic fake brand name, and then shipped those chips on to China, sometimes through further intermediary countries, including Singapore and Canada.

Andrew: We should say at this point that obviously in the story, we reached out for comment to both Lenovo and Nvidia, and you can see their responses in the story when you read the full account when it comes out this weekend.

We should also say, of course, obviously these chips are pretty crucial at the moment in the ongoing AI race. They are used to power the data centres that lie behind the AI revolution, and that’s why they’re so important. Eliot, you’ve done quite a lot of digging as well, looking into who it was in China who was actually buying these chips. Can you tell us more about that?

Eliot: So that’s the big question here, ultimately, right? Which is, who’s the mystery buyer in China who’s trying to acquire all these highly coveted AI chips? I basically spent a whole week digging at this, trying to identify the buyer, and ended up on quite a journey. Bearing in mind that the DOJ anonymised all the names in this story, their description of the Chinese buyer was that the buyer was a Hong Kong company that was quote-unquote affiliated with a Shenzhen technology company. They also write that certain public financial records describe the Hong Kong logistics company as part of the corporate network of Shenzhen technology company.

So this is weird, right? Because normally, if you have a Hong Kong company that’s linked to a mainland one, you just call it a subsidiary, or just point to the ownership ties that link them together. And so that struck me as unusual. And the other interesting thing that the investigators did was they actually made one omission, unclear whether it’s a mistake or not, in the indictment, in that they left the Chinese characters of a group chat title, which included the Hong Kong logistics company’s name, in the indictment.

And so I was actually able to, from that name, then sort of backtrace and figure out who this company was. So the Chinese characters that they left were the word Fuyun. The document also mentions that one of the players in this story was brought in because his company is best friends with Fuyun.

Now, we went to that company, the company that was brought in, which had a parent company that was in Beijing that was publicly listed. And so all their financials were public. And there was a financial document that mentioned that their so-called best friend is, in fact, a supplier to the publicly listed company.

And so from there, we were able to get the full name for Fuyun, which was a company whose English name is Fortune Shengda Supply Chain Management, later renamed Fortune Global Service Technology Co. This company is based in Shenzhen. It is a Shenzhen technology company.

There was also online a Chinese financial news article that was questioning Shengda, this Shenzhen technologies company’s weird relationship with a Hong Kong logistics company. And so in fact, the Chinese media were looking at this for separate reasons unrelated to smuggling, but they had written up that there was this relationship. And so from there, I was able to find the Hong Kong logistics company’s name.

The challenge then became, OK, well, we have a media article that’s claiming that they’re related. Is there anything else that we can use to show that they’re tied? The ownership records didn’t provide anything. They were really careful in separating them by ownership.

Even the directors, the key people, the leadership at the two companies were fully separate. But Fortune Global, the Hong Kong company, made one mistake. When they registered the company’s online web domain, they used the email for Shengda, the Shenzhen company, to set up that website, therefore meaning that through the website registration information, we could actually find a business relationship between the two companies.

I went with all this information to the U.S. Attorney’s Office last week and asked if they would confirm these names. The press office declined to comment. But helpfully, this week, in filings that the U.S. government submitted as part of this court case, they then named Shengda and Fortune Global as the Hong Kong and Shenzhen companies that were the buyers.

And so from there, we got confirmation that indeed, this was the mystery buyer in the smuggling.

Andrew: Great work, Eliot. It just shows how much it pays to have somebody who can really read Chinese and also somebody who knows how to really follow all of these links.

So that’s great work. And obviously, they probably disclosed these names because you were asking questions about it all. What do we know then about how the U.S. authorities actually busted this ring in the end?

Eliot: The way that U.S. authorities, in this case, it was investigators with the Commerce Department’s Bureau of Industry and Security.

So these are the guys that are in charge of enforcing U.S. export control rules. The way that they figured out that this ring was operating was that they actually got an anonymous tip-off that there were a bunch of these chips sitting in a warehouse in New Jersey. And the story of how they went from the tip-off to ultimately confiscating the goods, it’s its own saga involving an undercover agent and a ransom payment and confusion about the chips being stolen by a rival buyer.

I’ll sort of put a pin in that for now and invite our listeners to get the full details in our story. In the end, it was through this anonymous tip that the U.S. confiscated the goods and also were able to secure a whole bunch of intelligence about the key players involved, thanks to a botched rescue mission.

Andrew: Yeah, I’d encourage anyone listening to this to read the details of the story. It’s kind of like a scene out of a movie, like Donnie Brasco, one of those old movies.

Obviously, quite a lot of questions arise from this all. Even though these H200 chips are now, it’s now legal for NVIDIA and any other company to export those to China, it still begs quite a lot of questions as to whether this is just the tip of the iceberg, is this a one-off, or are there lots of other smuggling rings going on that are still smuggling chips to China that the U.S. has actually banned from export? What sense did you get about that from talking to experts around this case?

Eliot: So the big thing that stands out about this whole scheme is really how crudely simple the whole thing is.

Really, it’s just a bunch of guys with import-export companies that seem to spy an opportunity to acquire these lucrative chips and send them on to China, where they could fetch a fairly high premium given how much they’re in demand. You don’t really need a whole lot of specialized knowledge in chips or AI to replicate such an operation, which then sort of raises the question of, well, how many other people might have also spied the same opportunity and are doing the same thing? In fact, there’s another case that’s being prosecuted in Florida that was announced in November, also by the DOJ, that is very similar to this one. And so we know that there’s at least one other smuggling ring that is unrelated, that the government is also aware of.

And it then also raises questions then of, well, if there are all these opportunistic actors out there, of course, what are the sellers, what are the legal vendors’ responsibilities to make sure that all of this is being detected and preventing it from happening?

Andrew: Yeah, absolutely. And as I said, we’ve talked to Lenovo and NVIDIA about both of those questions. I guess the other thing is that a lot of people might say, well, this is the problem with export controls, trying to stop things like chips getting from the US, getting from NVIDIA into the hands of Chinese customers is always going to be tricky.

It was always going to be the case that Chinese interests might get their hands on these chips anyway. Having said that, do we think, or do the people that we spoke to think that the US government, whether it’s BIS, whether it’s other authorities, are doing enough to enforce the chip controls that they still have in place?

Eliot: The elephant in the room in this question about what the US government should be doing is, of course, the fact that since December, the government has created a legal route for some of these chips to go to China. I should note, though, that while the Trump administration made that announcement in December, we know as of yesterday, which is when NVIDIA published its annual report, that it hasn’t actually sent any of the chips legally to China.

In the annual report, it says that to date, we have not generated any revenue under the H200 licensing program and do not yet know when any imports will be allowed into China. So the legal route hasn’t yet been opened. But even if and when it does, there are, of course, other chips that the US government will continue to have to monitor and make sure are not being sold to China.

The big question here then is what can be done to better alert BIS and the Commerce Department to some of these smuggling operations? On that, there’s currently a bipartisan bill in Congress that was introduced last year that would incentivize people to blow the whistle about potential export control violations and sort of notify the government of these smuggling rings. We know from this case that basically BIS caught wind of what happened through Good Samaritan. But what the bill would do is sort of take the example of what the Securities and Exchange Commission does, which is potentially reward whistleblowers with a share of the fines that are levied on export control violators.

And so that’s one of the things that people are floating as a way to potentially improve the government’s ability to detect smuggling.

Andrew: Well, it’s going to be fascinating to see what happens next in this space. And fascinating to read your story, Eliot.

Just to remind listeners, you can read more about this in Eliot’s story in the worldwirechina.com in our latest edition that’s coming out this weekend. Thanks so much for talking to me about the story, Eliot. It was great fun editing it as well, I must admit.

Export controls are actually one of the topics that Jake Sullivan discussed during his interview with Bob Davis that I mentioned earlier. That interview was pretty wide ranging. It also touched on what Sullivan thinks Xi Jinping actually intends over Taiwan and what it was like working with Joe Biden. Obviously, that’s a controversial area.

So do read that interview. In fact, we’ve got an extract from it where Jake and Bob talk about rare earths and what steps the Biden administration was taking on rare earths and securing America’s supply to those.

So you can read the full edited transcript in thewirechina.com from Sunday evening. But here’s an extract from that interview now.

Jake: There’s no doubt China has leverage, including on rare earths against the United States where they can inflict some measure of pain.

And there’s no doubt the United States has leverage on China where they can inflict some measure of pain. In my view, there is absolutely no reason to telegraph so persistently that this is a fatal weakness of the United States, which A, on substance, I don’t believe it is. I think we could have managed our way through this without just throwing up our hands.

And B, even to the extent it is a vulnerability, just going out so publicly and assertively and saying so and repeating so, and they’re continuing to do that to this day, has basically led Beijing to the very strong conclusion that it holds the high cards, and therefore, that it can use this lever, use this tool at any point along the way to make sure that the United States does not do things it doesn’t like and maybe even to try to cause the things that are problematic from our perspective that China is looking for. So in the Biden administration, we saw the rare earths challenge. We took a series of steps to try to reduce the vulnerability.

Many of those steps the Trump administration is now building on. It is patently clear that we did not move fast enough or ambitiously enough. And there were a number of hurdles involved in why we have market failures to actually get processing capacity for rare China.

But, you know, we didn’t get the job done. Frankly, I think right now the Trump administration has taken some good steps but still isn’t moving with the alacrity or the efficiency that’s required to close that particular vulnerability. But there is no reason for us to operate as though this one thing is kind of the easy button in the relationship for China, just for some undetermined length of time.

The United States has to play a more sophisticated game.

Andrew: Well, that’s all that we’ve got time for this week. Just to mention the other stories coming up in this week’s edition.

We’ve already put out this week Savannah Billman’s latest big picture. It’s a great story. It’s a great look at what some of China’s AI companies were doing over the Chinese New Year holidays to really promote their AI models.

And what Savannah has done is look into why it is that actually AI models aren’t getting that much adoption in China, at least relative to other major markets, even though we hear so much about how well the Chinese AI models are doing. So do take a look at that. That’s out already.

Earlier this week, we published a piece on the visit of German Chancellor Friedrich Merz to Beijing this week. That’s by Jörg Wuttke. Jörg’s very well placed to comment on this.

He was the President of the European Chamber of Commerce in China for many, many years, a well-known expert in this area. So do read his thoughts on German relations with China. Pretty interesting phase that they’re going through at the moment.

We’ll have a new op-ed, as I say, this Sunday. That’s by Alicia Garcia Herrero, an economist. She’s writing about China’s growth, China’s profitless growth, I should say, the fact that so many companies in China still aren’t making any profit, even though the economy is still still growing at about 5% a year.

And we’ll have a new cover story from Paddy Stephens about China and what it’s doing in the space area. Pretty interesting topic as well. So do look out for all of those stories on thewirechina.com. For now, thanks very much for listening and goodbye.

Transcribed by TurboScribe.ai.

Though no concrete deals ever materialized, the Epstein files demonstrate how power, connections, and opportunities flowed among the elite as China’s star continued to rise. In this episode, our reporters discuss how they sorted through thousands of files to build the network of Epstein associates with stakes in the China game.The Wire China will have no new issue this week as we take off for the Chinese New Year. Read the Epstein and China story on our website, thewirechina.com.

Transcript

Tom: Happy Chinese New Year and welcome to The Wire China Podcast. I’m Tom Mitchell, Features Editor, speaking to you from Singapore and joined by our staff writers, Elliot Chen in Toronto and Noah Berman in New York. We span the globe here. Elliot, Noah, happy Chinese New Year.

Eliot: Happy Chinese New Year.

Tom: This week’s podcast is going to be a little bit different from our previous ones. For one thing, we don’t have an issue to preview.

We are not publishing this week because of the Chinese New Year. We will, however, have a brief clip from a Q&A conversation that we will be posting this week with Frank Decoder, the China historian and prolific book author. First, I am speaking with Elliot and Noah about their most recent cover story with their fellow staff writer, Rachel Cheung in Hong Kong on Jeffrey Epstein and China. Rachel cannot join us tonight because she has family Chinese New Year dinner obligations.

One of the things I like about our cover article on Jeffrey Epstein is that it is a classic example of how journalism often works, that is by last minute accident. I had been planning to run a different cover for last week’s issue.

And while I was aware that you were working on something related to the files, we thought that was going to be running at a different slot. But it came together very well and was very interesting. So we swapped it at the last moment.

Elliot, first off, can you explain to us how you, Noah and Rachel, approach this story? Because the Epstein files are essentially the world’s biggest document haystack. So how do you even begin to search for needles in it?

Eliot: We started off sort of just casually searching the files for mentions of China just out of curiosity. And we noticed during that search process that there are actually a bunch of pretty interesting documents and names in there that piqued our interest.

We were curious, for example, why was there a pitch deck for Boyu Capital, the private equity firm, sort of best known for being co-founded by the grandson of former top leader Jiang Zemin? Or why is Desmond Shum, the well-known author of the book Red Roulette, mentioned in the files? And we’d also seen individual social media posts on Twitter and on Reddit that were making reference to some of these documents. So we knew that other people were looking at this too. So then we asked ourselves, well, if these files are in there, can we sort of piece together a timeline of what’s going on here and what’s the extent to which China is being discussed in Epstein’s emails? And in the end, between the three of us reporters, we probably reviewed between 4,000 to 5,000 documents within the space of a week.

Many of them were duplicates. And we also had a lot of help using a program called Notebook LM, which is a Google-made AI research and note taking tool to help piece together the timeline. So at The Wire, we never use AI for writing, and we cross-check all of the output from Notebook because AI still makes a lot of mistakes.

But in this case, the tool is really useful because what it could do is read a large number of these PDF files and then sort them into order. And so we ended up with a spreadsheet of roughly 2,000 rows with the emails sorted chronologically, from which point we can gradually get a sense of what was going on, but also what are the limits of the interactions and conversations being had between these people and Epstein and the extent to which business deals were actually being discussed.

Tom: I want to spend the rest of this podcast focusing on three of the people who Epstein corresponded really the most with about China.

These are David Stern, Peter Mandelson, and Desmond Shum. Also want to touch on some of the J.P. Morgan executives in Europe who were in communication about possible China deals with Epstein and his main associate, David Stern. With regards to David Stern, Noah, tell us a little bit about who he is.

Noah: So David Stern is a young guy who met Epstein, we think, in 2008. That’s when their first email communications that we could find were when he pitches Epstein on investing in a private equity fund that he’s raising to invest in China. But we actually know a little bit more about his background because at one point Epstein asks him via email for his resume and he sends it.

So Stern studied Chinese law at the SOAS University of London and then briefly worked, we think, while he was a student for Deutsche Bank and for Siemens separately in China, and then founded this UK registered advisory firm called Asia Gateway Limited. A couple of years later, he founded another firm called Asia Gateway China, which is a health care IT business that he later sells to a British company. And over the course of their years of correspondence from 2008 until 2019, when Epstein died in his jail cell while awaiting trial on federal sex trafficking charges, Stern pitches Epstein on dozens of business deals, including several involving China and many in which he references potential partnerships with so-called princelings, the sons of the political elite top officials in the Chinese Communist Party.

Tom: Now, Stern also has connections to Prince Andrew, and we’re recording this shortly after the news that Prince Andrew has been arrested in the UK because of his association with Jeffrey Epstein. How did that happen? How is it that David Stern knows both Epstein and Prince Andrew?

Noah: So we have some insight into this, again, because of correspondences between Stern and Epstein and Sarah Ferguson, who is Prince Andrew’s ex-wife. And in February 2010, an Epstein contact, whose name appears only as Sarah in the documents, emails Epstein suggesting that David Stern help someone named A set up a wealth fund in China.

She says that Stern has been helping her a good amount and that he has a great Rolodex for China. And then immediately afterwards, Stern follows up with an email to A, writing, Your Royal Highness, the Duchess has asked me to get in contact with you. It’d be a pleasure to discuss.

And then from then on, Stern refers a number of times to his firm, Asia Gateway, as one that manages private P.A. deals, P.A. probably being short for Prince Andrew, and suggests numerous other ventures that include involvement from Prince Andrew or P.A., as they call him. Right. So David Stern is talking to Epstein about possible deals in China, none of which ever materialize that we’re aware of.

Tom: He also has this connection to Prince Andrew and seems to be using Prince Andrew’s name to try and drum up other China-related business, none of which ever materialize that we know of. Is that correct?

Noah: We did not find evidence that any of the deals that they discussed came to fruition.

Tom: However, David Stern is not all hat and no cattle because he did do at least one deal that we’re aware of with a princeling. And this was Li Botan. Can you tell us a little bit about who that princeling was and what was the investment that the two of them did together?

Noah: Li is the son-in-law of Jia Qinglin, who was once number four on China’s Politburo. And Stern and Li invested together at least once in 2017 in an electric vehicle company called Evelozcity, which was later renamed Canoo.

We know from the emails and from a person who we spoke with that Li put at least $100 million into the venture. Things were looking good for the company, which was later renamed Canoo. In 2022, it won a contract to even supply vehicles to NASA, but it went bankrupt in 2025.

Tom: Okay. So there’s a rare deal involving Stern, not Prince Andrew or Epstein in this case, but not a good deal, one that didn’t work out very well. Elliot, let’s talk a little bit about Desmond Shum, who was introduced to Epstein by Peter Mandelson.

We’ll come back to Noah and Peter Mandelson in just a second, but let’s talk about Desmond Shum. I’m a big fan of the book he wrote, Red Roulette, about his business dealings with the family of Wen Jiabao. I think it’s one of the most important books written about China this century, if not since the period of reform and opening.

What was the gist of their correspondence, Epstein and Desmond Shum?

Eliot: Shum, of course, like you mentioned, is best known today as the author of Red Roulette. At the time in which a lot of the correspondences that we reviewed happened, though, he was a businessman and a real estate developer who was based in Beijing and married to Duan Weihong, who was an entrepreneur who, according to a New York Times investigation in 2012, actually helped the family of then-premier Wen Jiabao conceal billions of dollars worth of shares in Ping An Insurance. So when Epstein first meets Shum, none of this is known yet, however.

So this is in early 2010. In early 2010, Peter Mandelson, then still a UK cabinet minister, reached out to Epstein, offering to introduce him to his quote-unquote friend, Desmond Shum. Now, before I go on, I should add that Shum told us in a statement that he had no business transaction or money transaction of any kind with Mr. Jeffrey Epstein.

And he said after our article was released on X over the weekend that his conversations with Epstein were just talking shop. But over the years, the emails that we reviewed do show substantial evidence that Shum, in fact, did discuss deals with Epstein. In 2011, for example, the two of them talk about creating an offshore banking business in China, and Shum tells Epstein that he had discussed the idea with top decision makers in China, including the chairman of state-owned investment firm CITIC, the vice mayor of Shanghai, and the top decision maker at the State Administration for Foreign Exchange.

Two years later, Shum sends Epstein a fundraising deck for his company Great Ocean Group, in which he’s looking to raise money for a plan to develop infrastructure near airports in China. And Shum and Epstein end up staying in contact for seven years, scheduling catch-ups as late as 2017. There’s no evidence from the emails that they closed any deals together.

And indeed, in some cases, Epstein actually seems skeptical of Shum’s ideas.

Tom: He’s quite critical, isn’t he, of Shum’s ideas?

Eliot: Yeah, yeah. In one case, after Shum emails a plan for a private bank to Peter Mandelson, Mandelson forwards that email to Epstein, and Epstein responds with the word childish, to sort of dismiss Shum’s plans.

Noah: I’ll just add one point on what may have impressed Desmond Shum about Epstein, and what about Epstein may have impressed Desmond Shum, which is that in one email, Epstein claims access to more than $15 billion in capital. And Shum responds, you always talk in numbers that put me in awe. We don’t know that Epstein truly did have access to that $15 billion, but he at least said so to Desmond Shum, and Desmond Shum responded in a way that conveyed that he was impressed.

Tom: I thought the correspondence between Epstein and Shum was really interesting because I thought it said so much about Epstein. I never really followed this sad story in much detail until it’s become such a big scandal of late. One of the few things that stuck in my head about Epstein is criticism of him as basically a con man who convinced a couple very rich associates, friends, that he would manage their money.

He made money managing their money, but he was nothing in the way of a real businessman or entrepreneur. He was just good at networking and convincing people to let him manage their money. And if you think about what a guy like Epstein is facing when he tries to go into a place like China, he knows nothing.

He is woefully unprepared. And the type of guy he needs to work with is someone like Shum, who has a real track record and experience. And his book is fascinating when it comes out years later.

So Desmond was a real operator, and yet Epstein doesn’t seem capable of recognizing that. Now, it was Mandelson who brought Desmond Shum and Epstein together. Tell us a little bit, Noah, about Mandelson and his relationship to China and Epstein.

Noah: On Mandelson, we know that, as Elliot said, Mandelson introduced Desmond Shum to Epstein while he was still a UK cabinet minister. He was a business secretary in the administration of Gordon Brown, then the Labour prime minister. And over the years, after leaving government, he starts a lobbying firm called Global Council and begins to drum up ideas for how to win business in China, including with state-backed investment firms.

And he talks about those ideas with Epstein and asks for advice about them. And when Epstein is talking with Shum and Mandelson, David Stern doesn’t appear to be involved in these discussions. Am I correct about that? It mostly appears that Epstein corresponded with Shum and Mandelson independent of David Stern, though there were some emails that suggested that they were aware of each other.

For example, Epstein emails Mandelson early on in either 2009 or 2010 saying, I have someone in China who I want to introduce you to, and it’s David Stern. And at another point when he is meeting Desmond Shum for the first time, David Stern sends Epstein background information that Stern says he’s collected about Desmond Shum. So it’s clear that Epstein is discussing his arrangements with Mandelson and Shum, with Stern on the side.

It’s not clear how big a part of those discussions Stern ever became alongside Epstein. And again, with both Mandelson and Shum and their discussions with Epstein, no deals that were China deals that we’re aware of ever materialized.

Tom: I just want to wrap this up by talking about another area in which Stern and Epstein were very much in contact.

And this was with certain J.P. Morgan executives in Europe, and they were talking about possible China related deals. Elliot, can you walk us briefly through what was going on there?

Eliot: With the caveat that I think this is certainly an area in which we only have a partial glimpse of what was going on through the emails, because a lot of the conversations happened over the phone. And we know that because there are emails in which Epstein is saying, call me. Stern is asking him, when can I call you?

Essentially, what we can see from the emails about what was going on here was that Epstein and Stern were talking about setting up a sort of arrangement under which Stern could work in China with J.P. Morgan. We know for some background that J.P. Morgan, until at least or up till 2013, had close relationship with Jeffrey Epstein, thanks in large part to Epstein’s friendship with Jess Staley, who was the head of the private bank at the time. So we know that there is this connection between Epstein and J.P. Morgan.

And we know that Stern and Epstein are talking about using that J.P. Morgan relationship in some way to advance their work in China. At one point in time, Staley introduces Stern to top executives at J.P. Morgan. And one of those executives then for several months corresponds with Stern about using Stern’s relationships in China to try and secure some kind of state investment from Chinese investors in Deutsche Bahn, the German railway company.

So at this time, the J.P. Morgan banker, who’s based in Europe, is working with Deutsche Bahn on a partial privatization deal in which Deutsche Bahn was looking for external investors. And it seems that he was approaching Stern to try and see if Stern could get investors who Stern later mentions include CIC, China Investment Corp., the Sovereign Wealth Fund, as well as China Life, the Chinese state-owned life insurance giant, to invest. Now, nothing ends up happening with this deal.

In the end, Deutsche Bahn reverses course or the German government, which owns Deutsche Bahn, reverses course and opts against privatizing the company. But it’s an interesting look at the extent to which Epstein’s associates work with J.P. Morgan and also a look at how J.P. Morgan thought that they could use this person who claimed to be highly connected in China in order to advance their own business interests.

Tom: Thanks. Well, this is all really fascinating. And if any of our listeners haven’t seen that story, please do check it out on our website. Thank you for listening to today’s podcast.

We’re going to leave you with a clip of our interview with Frank Dikötter, the China historian and prolific author on some of the greatest tragedies of Maoist China. So with that, this is Tom Mitchell. Thank you very much and have a good holiday.

Frank: I picked up a book by a journalist called John Powell, my 25 years in China. So he spent a quarter of a century in China. I thought, let me read this.

This is an interesting primary source. And let’s see if it triggers my interest, if I can develop a passion for this period in which I had spent the first 15, 20 years of my career, by the way.

So in there is a chapter in which John Powell, with other journalists, stands on a boat in Manchuria and watches a Sino-Soviet war in the summer of 1929, in which hundreds of thousands of soldiers fight each other with aircraft, boats, towns destroyed, massacres of civilians, a full-fledged war.

And I thought, well, that’s interesting because I hadn’t really heard of the Sino-Soviet war. So I googled it. This is 2021.

And indeed in 2017, a man called Michael Walker had published a book called The 1929 Sino-Soviet War. So the subtitle is very interesting. The subtitle is The War Nobody Knew.

And he says in the introduction that very few China scholars have mentioned this war at all, which I thought was surprising. So that triggered my interest.

Even as wars in Ukraine, the Middle East and Africa raged on in 2024, newly released data for that year shows that China’s arms exports decreased by ten percent as Xi Jinping’s anti-corruption investigations roiled the country’s defense industry. Meanwhile, despite trade tensions, the door for Chinese EVs to enter U.S. markets is not entirely sealed shut, but President Trump’s seeming openness to the idea puts him at odds with national and state officials alike.

We also have an extract from our latest Q&A with author Yi-Ling Liu on her new book “The Wall Dancers: Searching for Freedom and Connection on the Chinese Internet,” and a guide to the other articles you can read in this week’s edition, published every Sunday evening EST at thewirechina.com

Transcript

Tom: Hello everyone from Singapore, happy Chinese New Year and welcome to The Wire China Podcast.

I’m Tom Mitchell, The Wire’s Features Editor. In the podcast, we aim to take you behind the scenes of the stories we cover in our magazine. This week, I’m joined by one of our staff reporters, Noah Berman.

He writes this week about a rare downturn in China’s arms exports. We’ll be discussing that with him as well as his recent article on why and how Chinese electric vehicles could appear on American roads, despite longstanding efforts by US politicians to prevent this. Later on, we’ll have an extract from our Q&A conversation with Yi-Ling Liu, author of The Wall Dancers.

It’s a fascinating new book about cyberspace culture in China. In the Q&A, Liu speaks with The Wire China’s Rachel Cheung about her reporting on an online environment that, as she says, swings wildly between freedom and control. As ever, you can read these stories and much more on our website, thewirechina.com. First, to Noah in New York.

Noah, I want to talk to you about your two most recent articles, starting with this week’s piece on China recording a rare year-on-year decline in arms exports in 2024. Data for that year has just become available. So what was the decline and what is the source for these statistics?

Noah: So there’s a nonprofit in Sweden called the Stockholm International Peace Research Institute, which tracks sales of arms.

And in 2024, eight of the Chinese defense companies that it tracks recorded a combined 10 percent decline in their annual revenue, while companies in the rest of the world saw big increases as wars raged around the world.

Tom: I don’t remember 2024 as being a particularly peaceful year. I certainly wouldn’t have predicted then a coming decline in China’s arms exports, not with the war in Ukraine dragging on and on, not with the interminable and seemingly innumerable conflicts raging across the Middle East and Africa that year. So what happened?

Noah: Well, first, we should note that 2024 is just one year, but there is some evidence that this trend is going to persist. For example, between 2020 and 2024, during that five-year period, China’s total weapons exports was still lower than it was between 2015 and 2019 by more than 5 percent. And the decline was big between 2023 and 2024.

It brought the 2024 level to the lowest since 2009, except for 2020, which was a bad year overall for arms sales around the world.

What could explain this? Well, it’s important to note that during this time, China is in the process of purging top military officials, both in the government and at defense companies. And some analysts tell me that that’s creating an environment where there is very little appetite for risk, including arms sales.

Another factor is that China is heavily reliant on one customer. Between 2020 and 2024, Pakistan accounted for almost two-thirds of China’s total weapons sales, and sales to Pakistan dropped significantly in 2024.

Tom: The corruption angle is fascinating, especially as that all came to a head late last month. And yeah, I guess Pakistan is, if nothing else, a country and a customer with perennial cash flow problems. But at least India will be happy. Let’s shift a bit now from the military to the civilian, to an industry where China’s exports have been surging, not declining, the electric vehicle industry.

As impressive as Chinese EV exports around the world have been, however, it seems like sales in the U.S. were and remain politically impossible. But in your piece in our last issue on this subject, you explained why that might not be the case. What factors could bring Chinese EVs to America?

Noah: The most important factor is the president of the United States.

Donald Trump doesn’t view Chinese electric vehicles the same way that many hawks in Washington do. He said many times on the campaign trail, as he ran for president, that he would welcome Chinese electric vehicle manufacturing in the United States. And he reiterated that at a speech in Detroit last month, when he said, let China come in, with reference to auto manufacturing.

And at the same time, there’s a growing recognition in the U.S. auto industry that Chinese cars are just really good. And they also could be much cheaper than the cars that are currently available in the United States. An average new car here can cost up to $50,000, according to the most recent Kelly Blue Book estimates.

And so young people and in general, people who are looking to buy cars might be swayed by the opportunity to pay less for a better product.

Tom: Okay, that’s really interesting. But as you noted, the China hawks in Washington have long been adamant that Chinese EVs are an intolerable national security risk.

Each one allegedly, according to them, is essentially a mobile surveillance device. And similar concerns have led to restrictions in China on where Teslas can be driven there. So how can the U.S. eliminate this risk or is that simply impossible?

Noah: Hawks will tell you that if it’s not impossible, it’s pretty close to it.

But in reality, there’s a number of ways to at least reduce risk associated with the way that cars collect data now. As you mentioned, China found a way to do this with Tesla, notably a U.S. company and one tied to a defense contractor here via Elon Musk’s ownership of SpaceX as well. One way that it could happen is through joint ventures, through rules that basically make it impossible for Chinese companies to enter the U.S. market without doing joint ventures, in which the U.S. company has control over the data, has majority ownership over the structure, and that could assuage some of the concerns that hawks have about data.

And notably, the Financial Times reported last month that Ford is considering exactly this with the Chinese electronics giant, Xiaomi, though both companies denied that deal.

Tom: Okay, and Canada recently lowered its barriers on Chinese EVs in return for China doing the same with regard to Chinese agricultural commodities and other products. Do you see the China-Canada agreement as a likely template for a similar deal between Beijing and Washington?

Noah: So the deal between China and Canada focuses on dropping tariffs from a restrictive prohibitive level of 100% to 6%, which is totally manageable for a very small set of Chinese electric vehicles, less than 50,000.

That’s by some estimates around 3% of the Canadian market. I think that it’s more likely that if Chinese cars are going to enter the U.S., it’s going to be through rules or pressure to get them to manufacture here. Tariffs remain something that President Trump talks a lot about.

And so I think that he will likely retain those, particularly as he keeps tariffs on China over 30%, still higher than they are on much of the rest of the world.

Tom: There’s this apparent tension between Trump and American consumers who wouldn’t mind being able to buy cheaper cars and China hawks in Washington who are so focused on national security concerns. What do you think is the perspective of U.S. governors and U.S. labor unions on the entry of Chinese EV manufacturers? Is this something they would support, do you think?

Noah: On a local level, it’s interesting. The Governor of Michigan, Gretchen Whitmer, has shown skepticism. Michigan, of course, being important as a hub of the U.S. auto industry, has shown skepticism to Chinese electric vehicles. Senators in other auto manufacturing states have expressed extreme opposition, even after President Trump’s comments to letting Chinese auto manufacturers into the U.S. So that will be another obstacle for Chinese EV manufacturers to overcome as they seek to enter the U.S. because that local opposition could be strong.

Tom: That’s really interesting. Thanks, Noah. For more on these stories, please do go to thewirechina.com. You’ll also find there our latest long interview, which this week is with Yi-Ling Liu, author of the recently published book, The Wall Dancers, about China’s internet culture.

Liu is a Hong Kong-born and raised journalist who previously worked for the Associated Press and has also written for the New York Times Magazine and The New Yorker. She spoke with our own Hong Kong-born, raised and based Rachel Cheung.

Yi-Ling: I think one of the most alarming things that I’ve noticed is collaboration between political power and the tech elite to kind of push for their own particular agenda as opposed to that of its users.

I found that reminded me a lot of just the kind of relationship of interdependency and patronage between the Chinese government and Chinese tech companies. It’s alarming because it reveals that technology is increasingly in the hands of people in power and they’re able to kind of carve out their own turf and take over what we once saw as like the open space of the web.

Tom: You can read more of that interview on thewirechina.com. Also on our site this week, you can read our other regular features, including our stories sourced from the Epstein files and Wirescreen corporate records on Jeffrey Epstein’s China business plans and partners, and an adaption from Rebecca Fannin on her new book, The New Tech Titans of China about the evolution of the country’s venture capital industry.

Finally, it would be remiss of me not to note Monday’s sentencing of former Apple Daily publisher Jimmy Lai to 20 years in prison in Hong Kong for alleged violations of the Beijing-imposed national security law. Jimmy, the territory’s best known dissident, is 78 years old, so the sentence is effectively a death sentence. If you missed it, please do read Noah’s cover story in our first issue of 2026 about the Chinese Communist Party’s years-long campaign to get Jimmy.

So on that grim note, Happy New Year and a special 万事如意 wàn shì rú yì to Jimmy Lai and his family. Thank you for listening.

In this episode, reporter Eliot Chen walks us through a true crime thriller with news editor Andrew Peaple. A mild-mannered Chinese software engineer working for Google in California schemes to make it big as a tech entrepreneur by setting up his own company back in China. Instead, he winds up in the middle of the U.S.-China tech war, accused of stealing Google’s intellectual property for the Chinese government.

We also have an extract from our latest Q&A, with Ali Wyne of the International Crisis Group, and a guide to the other articles you can read in this week’s edition, published every Sunday evening EST at thewirechina.com

Follow us on social media:

https://x.com/thewirechina

https://bsky.app/profile/thewirechina.bsky.social

https://www.linkedin.com/company/thewirechina

https://www.instagram.com/thewirechina/?hl=en

Transcript

Andrew: Hello and welcome to The Wire China Podcast. I’m Andrew Peaple, the Wire’s news editor, coming to you from the UK where it’s pretty sunny today as signs that spring is finally on the way.

Well in this podcast we aim to take you behind the scenes of the stories we cover in our magazine each week and this time I’m joined by our reporter Eliot Chen. His cover story this week recounts the extraordinary tale of a would-be tech entrepreneur from China who lived a double life as an employee of U.S. tech giant Google. Later on we’ll have an extract from our new Q&A which this week is with Ali Wyne, an expert who argues that during his second term President Trump has ruptured the consensus on China policy in Washington, creating an opportunity now for a major rethink.

As ever, you can read these stories and much more on our website thewirechina.com. But first let’s talk to Eliot, coming to us from Toronto where I expect it’s still much colder than here in the UK. Hi Eliot.

Eliot: Hi Andrew, great to be here.

Andrew: Well I think you’ve escaped Toronto recently for a while and you’ve been to San Francisco to report on the trial of Ding Linwei, this ex-Google employee. It’s pretty remarkable stuff. Can you explain what he was alleged to have done?

Eliot: Ding Linwei was a former employee at Google. He was a software engineer who worked in particular on data center infrastructure there and so what that means is that Ding was working on the advanced chips that are used by Google to train AI models and these are used not just by Google but Google actually also rents out these chips to other AI companies that you’ve probably heard of including Anthropic and Microsoft.

The thing is AI models these days they’re not just trained on a single chip but often thousands and thousands of them and so what Ding worked on in particular at Google was this technology that allowed it to essentially wire up these thousands and thousands of chips in such a way that could run these huge AI models.

At the same time what Ding got in trouble for was that while he was working on this technology there he is alleged by the government to have copied more than 1,200 documents totaling over 14,000 pages from Google’s internal Wikipedia into his work computer and then onto a personal Gmail account and in particular the way that he did so was he was essentially screenshotting and copying key extracts from these internal confidential documents, pasting them into his laptop’s sort of Apple Notes document and then uploading them as PDFs into his own personal Google account.

So of those 1,200 documents the government alleges that 105 of them were trade secrets meaning that they contained confidential proprietary information that only Google had. And then what’s he alleged to have been doing back in China or while all this was going along? So at the same time that Ding was uploading all these documents he was actually putting together sort of a double life back in China which is where he grew up and where he went to university. In essence what Ding managed to do was he evaded Google’s oversight for months between 2022 and 2023 in order to spend large amounts of time in China trying to build a competing business to Google.

Starting in late 2022 he first did so with a business partner at a Beijing company called Rongshu Lianzhi and then later on he broke out on his own, started his own firm and applied to join this prestigious startup incubator called MiraclePlus that actually was formerly known as Y Combinator China which was the Chinese division of the sort of storied Silicon Valley startup incubator program.

Andrew: And so you’ve been at the trial of Ding after he was arrested finally for stealing these secrets allegedly from Google. What sort of person was he? How did he come across in court and how was it that he was managing to lead this double life effectively?

Eliot: I was sort of observing the proceedings from the public gallery. So Ding showed up at court every day in a dark grey suit wearing thick rimmed black glasses. By all impressions he was a pretty quiet, unassuming guy, unfailingly polite, really a software engineer’s software engineer. In order to do some of these things that he’s alleged to have done in China I think we can also assume that Ding is also a pretty pretty gutsy guy.

So in order to get into that startup program in China, MiraclePlus, he had to dupe the government alleges both MiraclePlus and Google at the same time in order to get away with it. So to get into the incubator he actually listed the name of a vice president at Google as a professional reference on his application and then went so far as to create a fake email account for that Google vice president in order to pretend to be him. And then after that even downloaded this AI voice modification app in order to impersonate the vice president on a call with a MiraclePlus employee.

At the same time while he was doing all this he was also tricking Google into thinking that he was working at the office in Sunnyvale, California where Google’s headquartered. And how he did that was he gave his employee badge to his Google intern at the time and asked her to basically swipe him into work every few days. So there’s an extraordinary amount of planning that went into this on his part.

Andrew: As I understand it from your story Eliot, his defense though rested in part on this idea that actually in Silicon Valley it’s fairly common for people who work at these big companies like Google that they eventually look to leave and set up their own business and the only difference with Ding was that he was just planning to do this in China. Is there any sort of fairness to that argument? How did that argument go across in the case?

Eliot: It’s actually a pretty well trodden path particularly in Silicon Valley for employees to be working on their own projects, maybe even planning their own startup while continuing to be employed at a big tech company like Google. And as we’ve seen from the news in the last few months with the big AI companies in the US, poaching and defection really happens all the time.

So all of that is pretty par for the course for this industry. Ding’s defense team argued that Ding was essentially being vilified then for doing what thousands of people like him have done before, except in this case he was Chinese. And they say that the trade secrets he was alleged to have stolen were really just internal notes that he was taking for himself in order to keep a record of the work that he had done at Google.

Now the government disputes all this and even in the opening arguments, the prosecutor actually addressed this issue head on. And so I’ll actually just quote a little bit from that opening statement from the prosecutor in which he said, it is not a crime to leave a big company like Google and start a business. And it is not a crime to leave a company and start your own business in China. It’s not a crime to to help China succeed.

What you can’t do, however, is take proprietary technology that doesn’t belong to you, that you didn’t create, and build a business and offer to help a foreign government build the very same technology you stole. So the clash that emerged in this trial and what the jury was asked to evaluate was basically, do you believe that what Ding was taking were merely notes or did those notes contain trade secrets? And if they were trade secrets, was Ding stealing them in order to help a foreign government, in this case the Chinese government?

Andrew: Absolutely fascinating, Eliot. One thing that seems odd to me about this case is that a big tech company like Google wouldn’t have systems in place to prevent this kind of theft of their intellectual property. What has Google said about the case?

Eliot: So one of the really interesting things for me about this case was it was also actually a pretty interesting look into how Google’s internal security systems work. The government at the trial called a number of witnesses from Google who were then asked to explain in great detail essentially how the company keeps track of their employees and what they can see their employees doing on company laptops.

And what came across pretty clearly at the end of it was that Google is a company that puts a lot of trust in their employees, but when that trust is broken, there is a huge, huge amount of data that the company has on you that they can use to audit your activity after the fact. To give one example, Google has a program called Santa that tracks everything on a Google computer. An engineer at Google sort of unprompted at the trial explained that they called it Santa because it tells you whether an employee’s computer is being naughty or nice.

So Matt Linton, a member of the Google security team that investigated the thing, explained at the trial that Google logs every day about three trillion data points of activity on its employees. And so it is 100% impossible for a human security team to keep up with and proactively monitor that amount of information. And so what the company does is that they use computer programs to aid with detecting anomalous activity.

For example, it can flag when confidential documents are being uploaded to a personal account, like what Ding did. But Ding was able to evade a lot of these automated systems because of the way that he copied and pasted the information, which was to remove a lot of its identifying features, whether it was names or sort of the confidential markings on the document that would otherwise trigger the program.

Andrew: Eliot, as well as this being a great story at a sort of human level, the context of all of this is obviously the ongoing race between the US and China over developing AI.

Now that you’ve reported on the case and written the story, how do you reflect on its broader sort of significance and what it says about this competition between these two great countries?

Eliot: Yeah, so I think there are two sort of big takeaways for me from the story. One, sort of when it comes to economic espionage and that problem in the US. And secondly, with regard to the broader AI competition.

On the first one, the FBI has been warning for quite some time now about the problem of economic espionage from China. During Christopher Wray, the previous FBI director’s tenure, he was talking about this almost all the time. After the China initiative, prosecuting these cases became increasingly fraught.

So the China initiative was this initiative under the first Trump administration, this big push to crack down on Chinese economic espionage. But what it wound up being was mostly a series of failed prosecutions of US academics of Chinese origin for things like improperly filling out a disclosure form. So when the Biden administration came into office in 2022, they ended the China initiative and emphasized that they would go after Chinese economic espionage in other ways.

The Ding case is one of those other ways. Basically the exact kind of alleged brazen theft involving a critical technology at a critical US company that the Justice Department wanted to focus on rather than just academics. And so in many ways, this case you can think of as sort of the poster child for that new way of doing things.

In terms of the implications for broader AI competition though, I think one thing that I will say, this case is actually a really interesting view into how the talent and venture capital system in China functions. I write about this a lot more in the story and go into more detail, but I think for now, we’ll put it this way. Ding actually really overstated his capabilities in order to secure funding in China.

A lot of the things that he claimed he could do, he actually wasn’t fully qualified to. And so at one point he goes on this roadshow in China and meets with all the top VC companies, as well as state backers looking for investment, and practically all of the top private VC firms turn him down. The entities that do buy his pitch are all state-backed.

There’s a few SOEs, there’s Sugon, the Chinese state-controlled supercomputing company, a few state-backed innovation parks, and a few think tanks. And so I think that actually says something pretty interesting about who is able to discern in the Chinese system between the charlatans and the true talents when it comes to AI, and who can get away with something like what Ding got away with until he got caught.

Andrew: Well, thank you so much, Eliot. It’s a fascinating story. And you can obviously read Eliot’s full story and find out the result of the trial, of course, in his piece, which will be on thewirechina.com as of Sunday, this coming Sunday. You’ll also find there our latest long interview, which this week is with Ali Wyne, a China expert at the International Crisis Group, who says that President Trump’s foreign policy in his second term has not only upset the consensus in Washington on China, it’s given Xi Jinping the upper hand ahead of the pair’s likely talks in Beijing this coming April.

Wine goes on to argue that the result of this is now is really a good time for a full rethink on how the US approaches its relationship with China.

Ali: One of the great ironies of the discourse on China policy, and one of the great ironies of the policy itself in the way that it’s evolving, is that the second Trump administration is unwinding the alleged consensus that the first Trump administration introduced. President Trump is not only far more central to formulating China policy in his current term than he was in his first term, he’s also far more dominant within the Republican Party.

So his views, which are quite iconoclastic by DC standards, carry far more weight. President Trump does not subscribe to the view, the quite prevalent view, that strategic competition between the United States and China is inexorably zero-sum, fundamentally ideological, and potentially existential. He seems to take a far narrower view of strategic competition with few evident grievances beyond the size of China’s trade surplus and China’s export of fentanyl precursors.

It isn’t clear that he subscribes to a more, to a grander view, or to a grander conception of strategic competition. He also seems to have an affinity for his counterpart in China, Xi Jinping.

Andrew: You can read more of that interview with our colleague Noah Berman on thewirechina.com this weekend.

On our site this week, you can also read our other regular features, which this week includes our other graphics-led big picture. We’ll also have a news analysis about the extraordinary growth of the service robot industry and how that’s affecting markets in the US and Europe. And we’ll have our normal opinion column as well.

And Eliot, just before we go, I wanted to just highlight as well a piece that you wrote last year. This year in China, one of the biggest stories so far that people will have probably read about is the firing by Xi Jinping of two of the country’s top generals, two very senior figures in the People’s Liberation Army. This is a topic that you actually wrote about last year in one of your other cover stories, All the President’s Generals.

Just give us a quick update on what’s happened since you wrote that story.

Eliot: When we wrote this story last year, at the time, the story was inspired by the fact that one of the central military commission’s vice chairs, this man named He Weidong, had disappeared. And we wanted to look into the trend that had already picked up at the time of Xi Jinping’s top generals disappearing for a while and then getting prosecuted for corruption.

The story also looks into and goes back to just ask, why is it that the PLA is so corrupt? And how did it end up that way? Going back to the history of PLA Inc. and its engagement with the commercial sector going back to the 1980s. Of course, since then, since we wrote the story, much more has happened, including the investigation of two other members of the Central Military Commission.

But I think a lot of the points about the origin of corruption and crucially what it means for Taiwan continue to remain pretty relevant.

Andrew: Well, if you haven’t read that story before, I recommend going back and looking that up on our website. It’s called All the President’s Generals.

You can read it on thewirechina.com, as I say. Thank you so much for that, Eliot. I think that piece really gives a lot of good background to what we’ve seen in the last few weeks coming out Plenty to read again on our website.

Do take a look. And in the meantime, for now, goodbye.

In this episode, news editor Andrew Peaple talks to reporter Rachel Cheung about her latest cover story on China's fast-growing robotaxi industry, and whether a recent spate of accidents involving self-driving cars will drive the sector off course. 

We also have an extract from our latest Q&A, with Chinese economist David Daokui Li, and a guide to the other articles you can read in this week's edition, which you can read at thewirechina.com