Odd Lots - China Made a Chip Breakthrough That Shocked the World

Episode Date: December 7, 2023

Both the Trump administration and the Biden administration made moves to constrain China's ability to build out an advanced homegrown technology industry. But the country is still investing billions in its chip sector and there are signs that it's really starting to pay off. Huawei recently released the Mate 60 Pro smartphone, with capabilities that shocked the world in terms of its performance. So how is the country making such strides in face of technological trade restrictions? We spoke with Dylan Patel of SemiAnalysis and Doug O'Loughlin of Fabricated Knowledge about how much progress China is making, and the policies that are accelerating these gains.See omnystudio.com/listener for privacy information.

Thanks for listening to OddLots. Follow the show on Amazon Music for more future episodes or just ask Alexa play the podcast, OddLLLLots on Amazon Music. Hello and welcome to another episode of the Odd Lots podcast. I'm Joe Wisenthall. And I'm Tracy Allaway. Tracy, it's been too long since we've done a semiconductor episode. I know. I actually feel really behind on what's been happening in the industry. I've seen some headlines. I mean, obviously we talked about InVIDIA earlier in the year and that stock. Just explore. and recently they released their earnings. I've seen some news about additional export restrictions and things like that, but I haven't been paying close enough attention, and I feel really bad about that. Do you ever read, Tracy, or have you ever tried? Do I ever read? Yes, Joe. You read. Have you ever tried to read those like really technical semiconductor publications that talk about, oh, the new Nvidia chip is four different cores and a thing here? And if,

Have you ever seen those sites? I have. Sometimes people drop the links at our Discord, and I try to catch up to speed on what the hot new chip is, and I just, I always give up one third of the way through the pose. It's really difficult. I mean, even the ones where they talk about, like, wafer thinness and stuff like, like, I can kind of understand it, but there's so much within this specific subject. There's sort of the big picture things like geopolitical tensions and trade tensions and things

like that. And then you can really get into, I don't want to say the nuts and bolts. of the specific technology, but, you know, I don't know, the wafers and pins of specific technology. So the one thing is it seems like the lower nanometer number, the better. Yes. But even there, and I know we've done some episodes in the past with Stacey Raskin, etc., even there that only tells you so much about a chip's performance, and there are all different kinds of architectures and yields.

We've talked about that, too. Or in theory, you could have a really powerful chip, but maybe it's not economical because you lose so many chips in the process, et cetera. So wrapping one's head around chips is tough. Yes, agreed. What will we be doing on this episode, trying to wrap our heads around chips? Yeah. So, you know, there's been some news particularly related to China. And I think Huawei came out with a phone fairly recently in the last couple months. And what caught people's attention was it seemed to have performance that people didn't assume it could have, given what was known about the state of domestic Chinese semiconductor capacity.

Seven nanometers. Seven nanometers. I shouldn't whisper on a podcast. I don't know why I did that. It's not a secret. So supposedly it had a chip made by SMIC, a Chinese chip maker that was seven nanometers, and something that people thought China wasn't able to produce just yet. And yet here we are talking about the 7NM in these phones.

And I guess the question is, is this represent a major domestic, breakthrough for China's semiconductor industry. Does it mean that some technology, which wasn't supposed to get into the country, somehow got in? So there's some sort of U.S. national security implications. But it's a good time. You know, we've talked about this for years, you know, with people like Dan Wong, like, what is the state of Chinese semiconductor? Are they catching up? So I think it's a good time to take stock of the situation. I agree. And also, you didn't mention the biggest thing that people are talking about now, which is the idea that have the restrictions basically created the exact opposite result intended and maybe accelerated China's

semiconductor technology. Which is something people have warned about that ultimately, sure, maybe you set the country back a few years in its development, but if you restrict its capacity to get international technology, then it just builds faster its own homegrown. And with all kinds of questions related to chips right now that we need to catch up on. Yes. And also, can I just say that I blame? the semiconductor restrictions on me having to read the three body problem. Did you ever read that? And now these visions, I don't know how many people have read it,

but a lot of people used it as an analogy for China's technological development. But now I have nightmares about like little shriveled up dead people. And if you've read the book, if you've read the book, this makes sense. Okay. Well, let's talk chips. And we really do have two perfect guests, one of whom we've had on the show before, Doug O'Loughlin. He is the chief analyst at Fabricated Knowledge, a Semiconductor Research Service, and Dylan Patel, chief analyst at semi-analysis, a boutique semiconductor and AI research firm.

So we are going to pick their brains about the state of Chinese chip. So Doug and Dylan, thank you so much for coming on OVLOTS. Thank you for having us. Thanks for having me again. Let's just start. What was it? What's the deal with this Huawei phone that caught everyone by surprise? So Huawei has historically been a leader in technology. That wasn't a surprise to anyone. And a few years ago, of course, they got banned from, you know, many aspects of the U.S. and Western semiconductor supply chain. They got banned from utilizing TSM, the world's largest chip maker.

And all of a sudden, they come out with their own phone, right? Their new phone that has their own chip made in China. The expectations, of course, when they first announced it were whatever. But then once people got their hands on the phone, it was like, oh, my God. this is actually very good. You know, when you compare to foreign phones, right, with, say, Qualcomm chips from, like, say, Samsung, it is only, you know, a year and a half behind, right?

And in some respects, it's actually just as good. And certain specifications, you know, depends on how nitty, gritty you want to get into it. But at worst, 1.5 years behind, at best, on par, right? And so that was a big, big shocker for everyone. Just real quickly on those specifications, like when you say the phone is really good, Because I don't know, all phones sort of seem pretty good, and I don't really notice advances in phones.

So when you say that the phone surprised people by its capacity, could you just be specific about what shocked people when they picked up this device? Sure. So stage one is sort of the network performance, right? So it can download and upload data just as fast as any foreign phone, which is the sort of on-par thing, right? In fact, it was better than the current iPhone-based phones. And so then that's like one specification. Another specification is your CPUs and GPUs and AI aspects of the phone, right? And there, they were not only using, you know, they're, again, using domestic manufacturing capabilities.

They were also using domestic design capabilities and were able to match, again, what folks had done a year and a half ago, or in some cases, even sooner, or more recently, right? So this is sort of on a performance basis, whether it's in gaming, whether it's in, you know, uploading and downloading videos, whether it's in camera. every aspect of this phone was was on par, if you will. I think that people just probably don't appreciate how impressive that is, given the fact that how little technology they have access to. So they did this without EUV, which was the big October restrictions in 2022. And I think the thing that's most impressive about this is like, it's a really good chip with both hands tied behind their back.

And I think that like if the restrictions weren't there, the implication is that SMIC could probably ship a leading edge phone as good as TSM. and maybe even better than Intel or domestic production. And I think that that's probably the biggest takeaway that I think people need to understand. And there's definitely some D.V tools that snuck in and we can talk about all the mechanics of the cross, how slippery the restrictions have been

and how maybe poorly enforced it's been. But they did an amazing phone with both hands tied behind their back. And I think that in the conversation that we've been having about semiconductors in China, for a long time, it's always been like, well, they're really far behind. they'll never catch up. And I think that this is the first time where you can say, like, if they had what we had,

they've caught up. Like, I think that's Dylan, do you agree or disagree with that? I would say, you know, if you compare just what's shipping in the market, yeah, of course, TSM's shipping better. Samsung has shipped better. And everyone talks about this Intel turnaround that may or may not happen. Intel's best chips that they ship today are 7 nanometer. The same is what China has today, right?

So it's on par. Now, of course, Intel's close to releasing their 4 nanometer. and you could caveat that in 100 ways. But if you look at what's in the market today, the densest, right? Again, as you mentioned earlier, lower nanometer chip that you can buy today from an American manufacturer is from Intel and the dentist is from Smick and Huawei, and it's the same, right? It's a similar capability. Okay.

So not since the invention of Pringles have people been so excited about a single chip. But you sort of alluded to this, Doug, but can you maybe walk us through what is needed to produce a 7NM. And what, to your point about China basically doing this with a hand or two tied behind its back, what was actually available to them? So this is a really hard question. I'm going to have to ask Dylan a lot about this as well. But the thing that I think probably differentiates it from, let's say, TSM's process, is that they did not have access to EV. That's clearly the big delineation. But if you remember the original- This is the extreme ultraviolet. Yes, EUV, extreme ultraviolet.

That's the latest and greatest from ASML, and they cost like, you know, 300 million plus a pop. They're extremely advanced technology to make tiny, tiny weight lengths of light. But they managed to get around this with something called quad patterning, self-aligned quad pattering, which is like we're not going to go into the details of that, extremely technical, but an extremely hard thing that Intel got caught up on. So they managed to ship the 7-nometer chip much quicker than Intel managed to get through all of their problems, doing the same quad-patterning DV process. And I think that that's a big deal.

Like, it shows that there's a lot of technological umph underneath the restrictions alone. And then on top of that, the restrictions have been extremely poorly enforced. That's why they had some, you know, some restatements this year about specific entities. But what happens is, like, a good example is SMIC has a leading edge and a lagging edge factory, right? And they're both related entities. But my understanding is the lagging edge entity can go by extremely advanced deposition and etch tools, ones that are on the restrictions for the October 2022. do, and then they can just kind of shuttle those tools into the fab of the leading edge and

then effectively be able to use it. That doesn't seem like it's doing, you know, the restriction is doing what it was intended to do in that case. Yeah, not at all. So far, pretty much what's happened. And this has happened every single time we've had restrictions on American semi-cap companies, is that the restrictions come out. All the companies say, oh, this is going to impact us.

And then they slowly find ways for loopholes to be pushed through. An example is applied material. I had a South Korea in factory that probably has. had, I think they are right now under investigation. And I think the thing is, it's pretty clear that some of the leading edge tools, X the EUV stuff, is getting into China. And they're able to use clever engineering to make a better chip than we thought was possible. Yeah, I would say that while in spirit, the regulations pretty much said, hey, you can't have less than 14 nanometers. The specifics of what was actually banned were quite a bit more varied, right? The tool that you can

use for 28 nanometer. Well, you just use many more of them for seven, right? I mean, obviously, there's a lot more complications there. And so the government sort of handed out export licenses like candy to companies like Smic saying, hey, yeah, you can expand your 28 nanometer all you want, right? That's not the spirit of the regulation. Applied materials, go ahead and ship whatever tools you want to the 28 nanometer fab in Beijing for Smic. And then Smic then gets those tools shipped to Shanghai, right, where they are making 14 nanometer, where they're making sense. 7 nanometer. And this applies not only to applied materials, but also applies to ASML and every other equipment company, right? So when you say like, hey, yes, they're banned from having less than 14

nanometer on one hand. On the other hand, every single tool that they used for their 7 nanometer was the equivalent of what TSM had when they made their 7 nanometer or an upgraded version of it. Right? So it's not like any specific tools were banned that were required for 7 nanometer. And so it's kind of like, the regulation and the implementation were so far away from each other. And that's sort of what these recent regulations hopefully are going to try and help, but there's still maybe some holes there. The news doesn't stop on the weekends. Context changes constantly. And now Bloomberg is the place to stay on top of it all. Hi, I'm David Gurra. Join us every Saturday and Sunday for the new Bloomberg this weekend. I'm Christina Ruffini. We'll bring you the latest headlines, in-depth analysis,

and big interviews, all the stories that hit home on your days off. And I'm Lisa Mateo. Watch and listen to Bloomberg this weekend for thoughtful, enlightening conversations about business, lifestyle, people, and culture. On Saturday mornings, we put the past week's events into context, examining what happened in the markets and the world. That on Sundays we speak with journalists, columnists, and key political figures to prepare you for the week ahead. Join us as soon as you wake up and bring us with you wherever your weekend plans take you. Watch us on Bloomberg Television. Listen on Bloomberg Radio.

stream the show live on the Bloomberg business app or listen to the podcast. That's Bloomberg this weekend. Saturdays and Sundays starting at 7 a.m. Eastern. Make us part of your weekend routine on Bloomberg television, radio, and wherever you get your podcasts. So actually explain this in an abstract sense. Why is it hard to align the implementation of the law with the letter of the law or the spirit of the law? I think it has a lot to do with the fact that the U.S. doesn't want to wholesale ban chip production in China, period, right? While China's automotive, you know, sort of chip manufacturing is

exploding BYDs of a vertically integrated monster that is like really taking over the world with amazing vehicles and chips for those vehicles, I don't think the U.S. has really an intention to block those. But the tools that are required there, it turns out all the advances in tools that have happened over the last, you know, decade, many of them would still also be applied to that, say, 28 nanometer chip, just the same as it would be applied to that 7 nanometer chip. It's just the differences, you know, throughput versus, you know, accuracy, if you will, right, at a simplified sense. And then the other problem is this is so incredibly technical, right? Like, Doug basically told me 10 times, don't say, you know, like a list of words that are too complicated for

the audience, because it's, it's like, the problem is the government is mostly talking to, hey, like, who's a lithography expert? Well, they all work at ASML. And what is their, what is their incentive, right? It is to ship as many tools, period, as possible, and including to China, right? Like, what does ASML care that, you know, X, Y, Z is happening? They want to be the monopoly and lithography and continue to ship. And if they don't ship, then, you know, there will be a Chinese company eventually one day, right? So their incentive is for the government to have as weak of a control around lithography as possible. And they want to be like, well, hey, like, this tool's used heavily here.

You can't just blanket ban it. No, no, no, no, just ban it for that fab. And we promise we'll make sure that we won't ship it to that fab. But if the customer decides to move it from Fab A to Fab B, then, you know, oh, no. And also I think there is like an implication that China is going to play. extremely fair with the regulations. And that's just clearly like, I feel like we keep making these regulations and they're really cute. And it's like, well, the spirit of the law and stuff. And then like, meanwhile, you look at what China is doing for their domestic semiconductor production. And they're

like, they don't care. Like, they do not care. And what they're doing is probably one of the most aggressive industrial policies ever to ramp their leading edge semiconductor production. Like I think- it looks like the US in the 1930s. Yeah. It's totally different. Like, you know, the Chips Act, I'm sure everyone is like heard of and understand the Chips Act, right? It's like $52 billion and then plus like another 20-something in tax credits. That's peanuts. Like we are talking a completely different game and how meaningful the incentives are there. Like we could go on about this forever, but like essentially every step along the way has massive cost and taxes and R&D credits and rent reductions and, you know, the big fund one, two and subsequently three are about to launch.

And all of these things together is probably putting hundreds of. of billions of dollars in subsidies to encourage the semiconductor industry to figure it out. And I think that, you know, with that much push from the top, you know, who cares if a tool is not being used in the spirit of the wall, right? Like China's government isn't going to be like, well, darn you for using this, for this, using this deposition tool for not its intended purpose. It's pretty clear that they're weaponizing the split in the east and west semiconductor supply chains. And they're trying to do it as fast as possible. And I think that that's something that the West just continues to underestimate how much gumption they have toward that.

Yeah, I remember Dan Wong, another one of our favorite Oddlots guests, he called it China's Sputnik moment when the tech restrictions started coming into play because, like, basically, it was a huge wake-up call for China that it could no longer count on the U.S. to supply its technology and that it would have to basically encourage its own domestic alternatives. So on that note, I take the point about industrial. policy and the scale of the way China is doing it here. And certainly China has a lot of experience in both industrial policy and just generally a centrally planned economy. But what sense do we have of how efficient their semiconductor manufacturing process is so far? I've seen bits and pieces

about this. You know, I've seen people talk about there is a huge profit drop in SMIC earnings for the third quarter. I think something like 80%. So a lot of people, are going like, yeah, okay, they're producing these new chips, the 7 nanometers, but maybe they're spending an insane amount of money to do it. And also, I think there was an inventory shortage of the phones. So maybe there's a sense that they're not able to produce these at scales just yet. So the thing about the manufacturing here is that, you know, yeah, their volumes are limited, right? Only about 7 million phones will ship this year, looking like they're going to ship maybe 40 million next year versus the, you know, 1.4 billion or now 1.2 billion phones.

that ship a year. This is, you know, kind of like an initial drop in the bucket, but the scale of the ramp is huge. And what's more important, I think, that's hard to recognize for most people is that semiconductor manufacturing is literally the most complex manufacturing supply chain in the world, bar none. There's more process steps. There's more complex. There's more R&D in this field than any other field. It is the most complex supply chain period. And so when you're talking about thousands and thousands of process steps, each step has 100 different knobs that you could turn on each tool. Getting good yield. getting the number of chips you try to make versus how many actually work in the end is very,

very difficult. But the way you learn is by ramping, right, by producing, and then, hey, if I'm running a thousand experiments in flight and each one has one knob slightly turned differently, and then I see, oh, which one worked? Awesome. Now I know that I leave that knob turned, and now I move on to another piece, right? Sort of there's this, like, complex, like, you know, constant tweaking of the process, right? This is what's made TSMC, you know, so amazing versus, say, a company like Intel is that TSMC, their engineers are constantly doing this even on 15-year-old process technology, still getting the yield better, still getting the performance better on, say, 90 nanometer, whereas Intel would move on to work on the newest technology every single generation.

And so SMIC by, you know, yes, they are not profitable, especially if you strip away the subsidies, right? The SMIC, Shanghai joint venture, the subsidies they're getting from Beijing, these are massive. If you strip those out, they'd be losing billions a year, right? But, you know, that's, that's, the thing is like, hey, let's skip forward. Now, is there 28 nanometer profitable? Probably, yeah. It's their 7 nanometer.

That's not. Okay, but we skip forward a couple years and it's like 7 nanometer will be very profitable. And guess what? Over 85% of the world's chip value is not under 7 nanometer today, right? And so, like, there is a long tale of like, hey, my car has 07 nanometer chips, right? But there are a ton of chips that are made by Texas instruments and analog devices and microchip and you go down the list. And it's like, well, China's going to.

to compete with these companies and they're going to compete with them very strongly. Yeah. And I think that also just looking at in a pure, you know, profitability perspective, maybe loses some of the context of history. Taiwan, Korea did this exact same thing where you can look at the early days of TSM and they lost a lot of money along the way. But the important thing is that by pushing their self-sufficiency, what they're going to be able to do is they're going to be able to write up the experience curves on every single technology as they're pushing for full self-sufficiency. And I think that this story continues to have, evolve. And I think that like that's one of the reasons why we're here to be excited to talk about it is like

CXMT today actually. I think is literally today or yesterday. Which company? CXMT. It's their DRAM company. So they have a they have a NAND company, which is YMTC, which for context is shipping the most dense memory in the world. I think that's correct, Dylan. Like the most dense memory. Maybe the yields aren't great. But they've been effectively banned from tooling like properly, like no loopholes. And so they're they've been, they've been stumbling. But CXMT company founded in 2016. Yeah. And they are, they're looking to spend about $7 to $8 billion on equipment next year. Yeah. Which is more than Micron, by the way. Yeah. And so, so like they're shipping these memories that are like maybe two to four years old.

But like this conversation, if we had this five years ago, it would seem insane. Like truly, I think the rate of change is something that people continue to miss. And I think that as this continues, it's going to be, you know, after they have their own domestic phone or domestic 5G modem, the domestic CPU for the phone or like, whatever, the NPU, then they have their own domestic NAN, then they have their own domestic DRAM, and then all of a sudden you just like look at it, and that's the entire semiconductor tool chain. Like, you know, so this is, and so yeah, is it capital efficient? Hell no. Like Big Fund, too, for example, has allegations of massive amounts of fraud, and that definitely is happening.

And there's hundreds and hundreds of companies that have started probably this year to take advantage for the subsidies for semiconductors. And yes, there are lots of misses. But in terms of being able to get the talent there and then also bring them all up the learning experience, I think that that's a huge thing. And I think that's also a conversation that is completely missed about this as well is that like Dylan and I, we go to, you know, trade shows in Japan or Taiwan or Korea. And like the amount of young people there versus the amount of young people in the West that are making chips is just drastically different. And I think that you fast forward that same equation. And yes, maybe they don't have EV tools.

but there are a lot of ways to make better phones using packaging or other clever engineering tricks. And so what's going to happen is that not only do they have the technological know-how, they also have the talent too. And we could be talking about in, you know, 10, 20 years that America's chip dominance is very, you know, backwards looking. And I think people need to watch this closer. The rate of change has just constantly shocked me at least, and I'm sure Dylan, but like the progress is staggering. You can get the news whenever you want it with Bloomberg News Now.

I'm Amy Morris. And I'm Karen Moscow here to tell you about our new on-demand news report delivered right to your podcast feed. Bloomberg News Now is a short five-minute audio report on the day's top stories. Episodes are published throughout the day with the latest information and data to keep you informed. Yes, there are other products like this from a variety of news organizations. But they usually rerun their radio newscast. throughout the day. That's not what we do. We create customized episodes that can only be heard on Bloomberg News Now. And we don't wait an hour to publish breaking news. When news breaks, we'll have an episode up in your podcast feed within minutes.

So you're always getting the latest stories and developments. Get the reporting and the context from Bloomberg's 3,000 journalists and analysts. We're all over the world. Listen to the latest from Bloomberg News Now on Apple, Spotify, or anywhere you listen. This kind of gets me to, I don't know if it's a strange question or theoretical question, but when it comes to semiconductors, you hear a lot about the importance of the overall supply chain. So, you know, people will say that NVIDIA is really good at managing its own supply chain and that's been part of its success story. If China is now manufacturing more and more of its own chips on a domestic basis, I guess like maybe it'll take them long. to develop some things, but does that in the end kind of lead to a more resilient or reliable

supply chain for that technology? How do you see those two things interacting? You know, it depends in capital intensive industries. What you tend to see is a very big sort of winner-takes-all vibe, which is why in prior, you know, sort of technologies, if you will, right? Solar, for example, hey, Germany in the U.S. invented all of it. And they were great. until China dumped way more capital in a very capital inefficient way, put way more people at work at it, right? Smart engineers working on it. And all of a sudden, you know, hey, 90 plus percent of solar cells come from China, right?

And likewise, you know, U.S. companies were the first to do electric vehicles, right? Like GM initially than Tesla, but, hey, China makes like three times as many electric vehicles as the West does now. And so, you know, sort of you see these technology curves, and there is a significant amount of winner takes all, right? If you look at Latin American markets, I mean, even Europe had to put up tariffs recently. Chinese EVs are killing it, right? The U.S. has had to put up massive tariffs and then has had the sort of the Inflation Reduction Act with the huge subsidies for battery production and cathodes

and anodes and all these sorts of things. There is a big winner-takes-all sort of element of capital-intensive industries, and the semiconductor industry is not different. In fact, it is even more tilted that way. Every single vertical you look in, and there's hundreds of verticals, people think, oh, it's such a complex supply chain. Well, every single spot where you zone in, there's two, maybe three competitors, right, in any specific technology, in any specific chip, and many times one. And everyone makes good profit, and it's a very, like, you know, strong industry. But what happens when someone comes with breakthrough innovation and now all of a sudden,

you know, it's capital intensive and those players start to fall off or they exit markets, that's something that happens, right? And the winner does take all. And so it's not just necessarily a, hey, it's a domestic supply chain. it's hey well what about all of the vehicles that are being shipped out of china into lat am southeast asia and you know maybe even europe if they don't have tariffs or you know the u.s already blocks chinese vehicles mostly or or how about you know solar right like battery you know solar inverters to convert the power from sort of the solar panels to what's acceptable for your home or for the grid that's going to

become a chinese supply chain why wouldn't it right they already have the solar panels so it's not just like hey domestic versus not it's actually there there's a very strong element of hey this is going to be a supply chain or this is going to be a Japanese supply chain or this is going to be a Korean supply chain, which is already the case in sort of semiconductors, right, or a U.S. supply chain. Yeah. And I think there's also like an important thing to note there, at least I don't think this will happen in the United States, but Europe is where this is happening the hardest. So we've been talking about the leading edge, which they put out this new Huawei phone, really amazing technology, but there's problems with ramping, there's problems with yield.

We have some cutoffs. But like, you know, China at the same time is also pressing their foot to the gas in the lagging edge. And we've been talking about automotive, and I think that that's probably where it's scariest, frankly. So if you are a European automotive OEM and you are trying to sell more EVs, as everyone is, and you're buying your semiconductors from other semiconductor companies, and then you're buying the battery,

and then you're putting it all together and you're trying to have this margin. You know, it's a single digit or, you know, maybe double digit margin. It's a very low dollar value added industry. And then you're trying to compete against BYD, and BYD in this conversation, is making their own batteries. They're buying their own ships to ship the cars from China to

yeah. They have their own fab. Yeah, they have their own fab. They make all the semiconductors. And like the semiconductors are because they're becoming larger and larger parts of cars and they are the most profitable. They are the largest gross dollar profit pool within cars. Huh. And so if you just do it. We talk about batteries all the time, but the real profits are in the chips. Well, well, batteries as well. Batteries and batteries and semiconductors. And they're doing both completely vertically integrated. So, what they're going to do is they're going to collapse all that margin that the suppliers make and then they're going to take it out of the price of the car and they'll make the full stack margin.

But if you're a European or American company competing against this, you will never compete on price. And I think that this is... I mean, you see it with Tesla, right? Tesla has the highest gross margins in the industry. You can say what you want about their cars, right? But they have the highest gross margins. And especially after the seventh price cuts these year, they're cheap. Right? And BYD is the same except even better in many regards and even more vertically integrated. Yep. And it's happening a lot quicker than people realize.

And I think that that's something that it's another example of where the industrial policy is just going as fast as possible with these hundreds of billions of dollars to subsidies and people are not really paying attention to the story that's happening right in front of us. I'm glad you brought up the car chips because it does seem as though the disruptions to car chips that we saw in 2020 is sort of what brought the semiconductor. supply chain at the forefront. It's what sort of got everyone's consciousness. Oh, we can't make cars because we can't get the chips. But those are, you know, the lagging edge chips. So we sort of, it feels like there is this sort of incoherence that we maybe tell about the story here where it says,

okay, we need to invest in chips, but then we're talking about we need to invest in the leading edge, whereas actually the real disruption was at the lagging edge. Can you talk a little bit more what you said about how the U.S. companies gave up on the lagging edge, whereas you say if TSM and others continue to improve their yields on the lagging edge. We don't have that here with, say, Intel. So in some cases, right, with companies like analog devices who used to vertically manufacture all their chips, they're moving more and more to having TSM make for them. But in other cases, Texas Instruments, right, both of these companies are over $100 billion companies, but Texas instruments, they are actually investing hugely in making more fabs and having more

than 80% of their chips manufactured by themselves. And, you know, I think, you know, just a little bit of a chime in on the whole term lagging edges. I think a bit of a misnomer because it's not like they're selling the same chip that they designed in the 1990s, right? Well, they are, but they're also selling many, many specialized chips that just aren't low nanometers, but they have some material innovations or, you know, they have some different properties with them that aren't necessarily the smallest possible, but definitely specialized. And there's a humongous variety of these chips, right? So it's not like, oh, there's one chip, right? It's, you know, phone chips are quite easy to understand it's a single chip that it has a tens of billions of dollars market right

AI chips same thing tens of billions of dollars are market right so we can point to that one invidia chip and be like wow look at it but when you look at Texas Instruments catalogs the the catalog is like so thick it's like it's incredible of how many different chips they sell it's thousands of products it's thousands and thousands of products and I think um Texas instruments will say this that like half of their products have been from 1990 and earlier um so So there's a lot of lagging edge chips, but the difference here is that Texas Instruments is making like a 65, 70% gross margin. The business is managed, extremely like a mature semiconductor company it is.

But then you have the new entrance in China that are just saying, let's like completely screw up the industry structure. And that's something that's very new. MCUs, microcontroller units in China that is like the hottest topic of disruption and what people are trying to do. And MCUs, yeah, it might be a 28 nanometer chip or maybe a. It's like a, you know, 14-anameter chip or something like that. It's not the sexiest thing. But that is a, you know, that's a tens of billions of dollar market there. And I think for the reason why it's maybe not as viewed as strategically important is let's put this in the context of AI, right?

It is not a leading edge AI chip that, you know, and the original restrictions that we put on China was to stop AI progress, right? So that's not exactly, you know, strategically, quote, unquote, important. But in terms of for the businesses, for economics, it's a big deal. So what they've done is they've kind of refocused on the lagging edge where they can, and they're throwing as much money into there and is going to probably create crazy price. Going back to this sort of idea of that, hey, this catalog is thousands of chips. This is why no one competes with each other, right? Like Texas Instruments, they're the only player in many markets, right?

Or analog devices. And maybe the chip is only $0.8 to sell. But it maybe only costs them two cents to manufacture. The thing about China is they've literally sciop their entire generation into wanting to work in semiconductors. It's literally the coolest job. There are two different dramas that I know of. I was about to ask, like, are there any TV characters that are working in semiconductors? Because that's when you know.

Yeah, there's one of, like, this kid who's in college and he's lame, but then he, like, goes and works in the semiconductor industry and now he's super cool. And then there's another one where it's like a love story. And they both work in a fab and they, like, fall in love. And it's like a drama, right? They're like gazing at each other through the glass and stuff like that. That'd be pretty funny. Yeah.

Both wearing bunny suits. Wait, I don't know if anyone heard, but I think the sound earlier was both Joe and I scrambling to pull up the Texas Instruments catalog. Literally we both, how did you know what I was typing? We just know each other that we both Googled Texas Instruments catalog. But I can tell you, I'm looking at the page just for MCUs, and it looks like it's more than a thousand two hundred. That is just one category product. Yeah. Yeah, 1,238 microcontrollers and processors.

Can we just order them online? I guess you can. I mean, click the box. See what happens. Each of those has a couple hundred pages of PDFs associated with how to design them into a product. And it's like there's a reason no one redesigns this stuff. But now when you have thousands, hundreds of thousands of people coming into the supply chain and like, hey, we can't order this from TXN anymore. We need to order this.

We need to do this domestically. It's like, and then you also have subsidies that say, hey, every time you design a chip of, you know, certain ages, you get pure tax. credits right you get you get you get just straight up a payout local governments essentially each tape out will give you like a million bucks tape out as a design of a yeah yeah a tape out all the way to design so if you make one of these it doesn't matter what it is you just get money and you know if you're engaged in doing anything related to a semiconductor you have five years for every company 10 years for leading edge companies tax free like 200% R&D

credits meaning that like for every ten dollars you spend or let's say a hundred dollars you spend in R&D you get two hundred and twenty dollars back for certain leading edge categories Like the subsidies here, you know, they want you to work in it. They want to give you as much money as possible to, you know, be moderately successful. You don't have to be that great at it, honestly. If you're just taping out bad chips, you can probably make a living doing this. Now, that probably creates some pretty messed up incentives there.

But the focus and the desire to, you know, to get to Chinese domestic by 2030 or whatever. And that number, they're meaningfully below that. But this is a long journey and it's pretty clear how they feel about. it, right? Like, this is, this is, like, splitting the internet. One of the most interesting, like, observations of this sort of, like, warped incentive structure is that the company that makes rail cars in China has made it their, like, national goal to, like, because they make profit from that, right? Which company? CRRC. Yeah, yeah, CRRC. Yeah, so one interesting thing about this warped incentive structure

is that state-on enterprises or very successful companies that are doing really well in some some market are expanding to places that make no sense, like, you know, logically, right? Because they have profits and they're like, as Doug mentioned, right, you don't have taxes for a certain amount of time. So, hey, let me take all the profits from this business and throw them into another sector. And so CRRC is a national railway car company. And they're plowing all of their money into making their national goal to basically disrupt Infinien, which is Germany's largest chipmaker, right, who makes chips for power. Nothing sexy, right? But hey, yeah, converting from one form of power to another is very important job for chips. And that's what

infinity inch chips do, right? And this is what CRR sees sort of goal is to do, is to do that, which is nothing at all related to railway cars. But hey, I might as well do it because now if I make any money on the semiconductor business, I don't get taxed. And what I was getting taxed on my profitable, stable business is now being funneled into this new business. This is the irony of Xi Jinping's crackdown on disorderly capital in like e-com. and consumer tech. And now there's just like money kind of flowing indiscriminately in other sectors. You know, Joe mentioned in the intro that it's been a while since we've done a semiconductor

episode. And I think the last one that we did was actually on Invidio. Is that right? Yeah. I think that's right. And since then, the stock has absolutely exploded and there's been this frenzy over AI. InVio is just the absolute leader here. Obviously, China wants to develop its own AI models.

It has its own AI models, its own large language models, its own open AI competitor, etc. Where is it in terms of its own access to the quality of chips that it needs to make cutting edge AI models? So that's probably the most effectively enforced part of the China restrictions so far, right, in 2022, now 2023, is cutting back on, hey, you can't get the chips that NVIDIA makes, has been the most successful part, basically. And InVidio's tried to circumvent this by, you know, releasing new China-specific versions. And China's able to get some. But really, this is where China's ecosystem looks the most interesting for breakthrough innovation, right? Which is, hey, you know, you guys are going in this route, which is awesome. You know, we can't really go that route, but we have more people working on this.

So why don't we try and do different things that will actually generate outcomes that are in the same vein, but not in the same path? Right. So Huawei, of course, is making chips. They're making this mobile chip on seven nanometer, but they also have this AI chip on seven nanometer domestically made. Invidia's AI chips are on seven nanometers. So only one generation behind on process technology. But more importantly, what Huawei and Smicker are doing is that they're investing heavily in technologies that are sort of a few generations out for everyone else because they're sort of not necessary. So things like bringing optical fibers directly to the AI chip, right? So it's called co-package optics is the technology. Other things are, called like hybrid bonding, which is like stacking chips on top of each other. They're doing really interesting things there to enable breakthrough innovation, to enable performance that is on par with the U.S. chips like from Nvidia or AMD, right? And they aren't there yet, but they're going to be there. And one interesting like sort of phenomenon is that like, you know, because they can't order all these Nvidia chips, there's really only two places in the world that will let you build data centers and have cheap power, right? East Asia can't do it because they have to

import all their power. Europe hates, you know, natural gas and natural gas is basically how you have to power data centers. And so the U.S. and China are really the only place that can build data centers. And China's been blocked off. So now China has like these companies flooding into Malaysia and Indonesia trying to build data centers that they can install foreign chips in. And meanwhile, they're also building these chips trying to do this breakthrough innovation domestically. Yeah. And I believe there was also a loophole that might have been priorly closed talking about how you can essentially rent a chip. So for example, a Chinese company could rent something, you know, in AWS, you can rent a GPU that's in a cloud and, say,

Signaport or something like that. But I think the conversation about hybrid bonding and copackage optics is really interesting and kind of where we, you know, at the beginning, we were talking about how by cutting them off, we would effectively force them on their own roadmap. And that's kind of what's already happened. The SMIC roadmap is completely different than what is sort of the industry standard roadmap for leading edge. And as we continue to cut them off, they're going to have to kind of create a new roadmap for semiconductors going forward. And especially co-packaged optics, which is always five years ahead. And the reason why it's never adopted in the United States or in the Western semiconductor world is because it's just too expensive.

But if you don't have an option, and this is your only option for scaling out chips, they'll pursue the expensive option. So there's going to be know-how created because of the necessity of this. Tracy, I think when we released this episode, we should publish a glossary. too, and all these terms with like little like definitions. I think that would be very helpful. Okay, have fun doing that, Joe. Well, actually, speaking of new terms that will no doubt need to be added to the glossary, who makes the best chips for Q-Star?

That's been in the news. We don't know what it is yet, but I'm going to go ahead and ask. Someone tried to do and kind of explain it to me last night, actually. I don't even know what my favorite take on this, and this is Dylan's take is that Q-Star is a sci-op. It's a, yeah. So I'm in the Bay a lot. A lot of my friends are researchers, and we sort of had this idea that anytime OpenEI tweets about something that's related to research, they're actually just trying to sigh up everyone

into wasting their time on a path of ML research that is actually not like, yeah, not going to result in a better model. Because there's limited amounts of GPU, limited amounts of time. So if you're wasting your time doing that, instead of actually working on, you know, what's the correct path, at least what they think, because they've done all the experiments, sort of ahead of everyone else, it's easier to innovate what's already been innovated than it is to innovate something completely new.

And so sort of the thought is, hey, they're doing these sciops and Q-Stars is one of those. We'll see. There's a million more questions we could probably ask, but we've got to wrap up. So we'll have to have you back. Doug O'Laughlin and Dylan Patel, thank you so much for coming on to Odd Law.

It's amazing conversation. Thanks for having us. Always love to chat with you guys and always love to chat chips, for sure. Yeah. Yeah, this was a blast. I didn't realize you two are even more funny when you're sitting in the same room.

Thank you. When we're pointing at each other and making faces. Yeah. All right. Thank you. That was so much fun. That was amazing. Thank you so much.

Tracy, I thought that was an amazing conversation. I guess it freaked me out a little bit. Yeah. You know what's bothering me enormously now is I've realized if you step away from semiconductors for like a month, there's some new breakthrough. Yeah. And there's like five new technologies and probably like more companies. with like four letter acronyms involved somehow.

Is that like a rule? I remember having the same thought about stepping away. Like every time I would ignore crypto for six months, there'd be some new scheme. Except in the end, none of that mattered because it was all, it didn't matter. This stuff actually matters. And so you do, it does feel like you really have to pay attention constantly to know what's going on.

There was so much in there. You don't want thing just that struck me when I was thinking about U.S. First China industrial policy. And here we freak out politically do it away. So we're still talking about, say, Cylindra. So one company that was doing tech. And it ended up being scandal that for years cast aspersions on the idea of the government investing. And it sounds like listening to Dylan and Doug that, yeah, they probably have their equivalent of plenty of salindras in China. There's probably plenty of fraud, plenty of waste, plenty of disasters, plenty of endeavors that don't have the chance. It just seems that they have the appetite to continue investing. in those areas even with some level of fraud. But also, like, they are so explicit about what they're doing. And I can't remember, I'm going to have to go back and look who said this. But someone was basically, like, the equivalent of follow the money.

Like, China is telling you where they're going to put the money in semiconductors. This would have been around 2021 when they first started cracking down on, like, for-profit education and the e-commerce platforms and things like that. And this person, I remember them saying, like, China's telling you, what it wants you to do as a domestic investor or someone who's working in tech, they don't want you building, you know, video games or something like that. They want you to make chips that might power those video games or something much more important. And here it's like, yes, we have messaging, but it's never quite that way for obvious reasons. I think there's like still a lot of discomfort

here about industrial policy in general. But of course, since COVID, that seems to be changed. changing a little bit. So many other details. We need to find one of those shows, the love story of the two people who meet in the semiconductor fab. I also think it's just really interesting, this idea of a rail car company. Yeah. Building, say, hey, here's an opportunity to build a chip that we have to buy from a German

company and because of the tax structure. It makes sense for us to invest and see if we could build it domestically. This idea that, you know, we talked about BYD before on the show a couple months ago with Corey Cantor. but that it's also a chip powerhouse and that there's a lot of margin there. So many, it's too many interesting things to go over. Yeah. I got to say, I went on one of CRRC's, the high speed trains.

I think it must have been them between Beijing and Hong Kong. That was a nice train. I really enjoyed that. I want to do one of those drives. So I look forward to seeing the semiconductors too. I'm sure they'll also be nice, maybe. Shall we leave it there? Let's leave it there.

Okay. This has been another episode of the Oddlots podcast. I'm Tracy Alloway. You can follow me at Tracy Alloway. And I'm Joe Wisenthall. You can follow me at the stalwart. Follow Dylan Patel. He's at Dylan 522P. Follow Doug O'Loughlin at underscore Fab Knowledge underscore. Follow our producers. Carmen Rodriguez at Carmen Armin, Dashel Bennett at Dashbot and Kell Brooks at Kel Brooks. And thank you to our producer, Moses Ondom.

For more Oddlots content, go to Bloomberg.com slash Oddlots, where we have transcripts, a blog, and a newsletter, and you can chat 24-7 with fellow listeners in the Discord. Discord.g.g. slash oddlots. There's even a room there just for semiconductors where people are posting links to highly technical things that I don't understand but seem very fascinating. Go check it up. And if you enjoy oddlots, if you want us to get back into semiconductors on a regular basis, then please leave us a positive review on your favorite podcast platform.

Thanks for listening. I'm Francie Lacquan, an award-winning journalist. and I've got a new podcast, Leaders with Francine Lacqua from Bloomberg Podcasts. I've interviewed everyone from heads of state to fashion icons about the news of the moment. But I've always been curious who are these people as leaders. I don't think there's one right way to be a leader. Make decisions. A poor decision is always better than no decision. Listen to new episodes every other Monday.

Follow leaders with Francine Lacois wherever you get your podcasts.