Chinese Economics Thread

[FriedRiceNSpice]

Industrial automation is rarely just about the physical robotics but also the software. Specifically the machine vision software/cameras/algorithm provided by Cognex and Keyence. The moment they turn that off is the moment your factories stop working. This is an area where domestic players are even further behind. How do I know? I visited them both.

Additionally even if domestic parts can be made a lot of the CNCs and machine tools used to make these parts are still imported. How do I know? They told me.

Th critical difference is that the Chinese market is big enough to domesticate everything and have it be world class which is why there is the drive to do so. This is absolutely happening but as I’ve been very clear, this is a conversation to be had in 2025/2027.

I really don’t care how badly the Taiwanese economy functions but “at least we’re not as bad as them” or “they’ll be impacted worse off” is not a solution to your own problems.

The Taiwanese economy will be likely a zero in case of AR - the question is how the Chinese economy does in the aftermath where it will be full on decoupling from the West. That is what I’m interested in better understanding.

Software is the last thing China needs to worry about, it is the easiest and quickest hurdle to overcome. If there is an incentive, China has the software engineering talent to produce domestic equivalents within a year or two. Hardware takes much longer.

 

[tankphobia]

Please, Log in or Register to view URLs content!

Does that include military UAVs? China has been selling tons overseas, especially now with major deals in the ME $14 billion seems a bit low?

 

[broadsword]

Does that include military UAVs? China has been selling tons overseas, especially now with major deals in the ME $14 billion seems a bit low?

No.

 

[abenomics12345]

Software is the last thing China needs to worry about, it is the easiest and quickest hurdle to overcome. If there is an incentive, China has the software engineering talent to produce domestic equivalents within a year or two. Hardware takes much longer.

This is objectively not true - and I'm telling you what the engineers themselves are saying.

For example if you look at design software, the company ZW Soft is the Chinese equivalent of a mishmash of Autocad + Dassault Systems. (Which by the way they actually acquired an American company for core technology). While 2D Autocad is 'easier' to domesticate (no surprise, with the help of Huawei as a customer), the 3D CAD (Dassault Systems CATIA) is the difficult part and at current paces will take a decade to reach similar capability. This is a result of all the kernels you need to fulfill design needs. This is sort of like how domestic Chinese analog semi companies are trying to catch up with Texas Instruments/Analog Devices - the difficult is the fact that they have a manual with 100k SKUs - and there is no way to catch up unless you brute force this. The situation is no different with Glodon which supplies design software for construction related activities.

Anyone who's done analog semi design will tell you that it is very different from logic semi design and it takes years to train a new analog semi engineer.

A good analogy of looking at this is the number of dentists in China - as of 2020 there are 16 dentists per 100k population in China - you may as well argue that "its not hard to train a dentist" until you realize that it takes 8 years to train a dentist. Like sure, it's 'not hard', but that doesn't mean you're going to catch up to a developed economy (say Taiwan in 2019 at 40 dentists per 100k population) in the next 2 years.

The framing of "hardware vs software" is an incorrect way of looking at domestication difficulty.

 

[Michaelsinodef]

This is objectively not true - and I'm telling you what the engineers themselves are saying.

For example if you look at design software, the company ZW Soft is the Chinese equivalent of a mishmash of Autocad + Dassault Systems. (Which by the way they actually acquired an American company for core technology). While 2D Autocad is 'easier' to domesticate (no surprise, with the help of Huawei as a customer), the 3D CAD (Dassault Systems CATIA) is the difficult part and at current paces will take a decade to reach similar capability. This is a result of all the kernels you need to fulfill design needs. This is sort of like how domestic Chinese analog semi companies are trying to catch up with Texas Instruments/Analog Devices - the difficult is the fact that they have a manual with 100k SKUs - and there is no way to catch up unless you brute force this. The situation is no different with Glodon which supplies design software for construction related activities.

Anyone who's done analog semi design will tell you that it is very different from logic semi design and it takes years to train a new analog semi engineer.

A good analogy of looking at this is the number of dentists in China - as of 2020 there are 16 dentists per 100k population in China - you may as well argue that "its not hard to train a dentist" until you realize that it takes 8 years to train a dentist. Like sure, it's 'not hard', but that doesn't mean you're going to catch up to a developed economy (say Taiwan in 2019 at 40 dentists per 100k population) in the next 2 years.

The framing of "hardware vs software" is an incorrect way of looking at domestication difficulty.

It isn't so cut and dry either though.

If going with the dentist analog first, the productive and services of a dentist has changed over the years as tech improves and innovations are made. If we say something as a teeth filling 20 years ago maybe only lasting say 2 years vs a current new one that could last 10 years, that directly means one current day dentist can service more people than one 20 years ago (less time lost servicing same person to 'redo' a filling because it doesn't last as long).

This is a very rough analog, but it also holds true for software for CAD.
How many of those 100k SKUs are maybe duplicate code? Inefficient code? Could it be possible that it could be rewritten in a cleaner form in say 60k lines?
What about other auxiliary tools that can help writing those lines of code for CAD? Maybe a designer can write good 2k SKUs in just a week in comparison to say 2 or 3 weeks it maybe once took?

So just slapping on a conclusion saying it will take a decade to catch up, is really not something that can be 'securily' be made, although it is 100% for sure that catching up will take years possibly even a decade or more.

 

[abenomics12345]

It isn't so cut and dry either though.

If going with the dentist analog first, the productive and services of a dentist has changed over the years as tech improves and innovations are made. If we say something as a teeth filling 20 years ago maybe only lasting say 2 years vs a current new one that could last 10 years, that directly means one current day dentist can service more people than one 20 years ago (less time lost servicing same person to 'redo' a filling because it doesn't last as long).

This is a very rough analog, but it also holds true for software for CAD.
How many of those 100k SKUs are maybe duplicate code? Inefficient code? Could it be possible that it could be rewritten in a cleaner form in say 60k lines?
What about other auxiliary tools that can help writing those lines of code for CAD? Maybe a designer can write good 2k SKUs in just a week in comparison to say 2 or 3 weeks it maybe once took?

So just slapping on a conclusion saying it will take a decade to catch up, is really not something that can be 'securily' be made, although it is 100% for sure that catching up will take years possibly even a decade or more.

Agreed that its not cut and dry, but its not a walk in the park either. I would caution against the logic of "it is possible therefore it will happen". Flipping your argument, while it is "possible" it is faster, that argument also cannot be 'securily' made. The reality is somewhere in between.

However, thus far, the best guess, from the people who are doing this day in day out, are saying its going to be a decade. Let's just say that when reality goes against your personal beliefs, reality is not the one that is wrong.

Going back to the dentist analog, things haven't changed all that much for this to matter. For analog semi companies, a significant portion of their revenues are generated from products over a decade old - meaning the innovation cycle is quite slow and that products are generally 'classics' that last a long time. These are chips that sell for less than the price of a Nongfu bottle of water and are mission critical to the entire tech stack. Once spec'd in, people don't switch. It is why this is a lot harder to substitute.

 

[Michaelsinodef]

Agreed that its not cut and dry, but its not a walk in the park either. I would caution against the logic of "it is possible therefore it will happen". Flipping your argument, while it is "possible" it is faster, that argument also cannot be 'securily' made. The reality is somewhere in between.

However, thus far, the best guess, from the people who are doing this day in day out, are saying its going to be a decade. Let's just say that when reality goes against your personal beliefs, reality is not the one that is wrong.

Who are those people? Actual people working on it in China? Or from established players such as Dassault Systems CATIA?
Afterall if it's the later, when they throw it will take China 10 years to catch up, is that really such a good measurement? Maybe they don't have a good grasp of how fast or many resources are throw into this by China? Or they might underestimate (or overestimate) some parts and stuff to get that number? Not to mention we could even have cases of trying to say something for investors or to manipulate perception.

Personally I never had the belief that it would be 'easy' or just a few years to catch up, but rather that 3-5 years to get some 'base' that's a bit behind the 'leading edge' but more than useful enough, while completely catching up and basically becoming a peer competitor might very well take about ~10 years.

Going back to the dentist analog, things haven't changed all that much for this to matter. For analog semi companies, a significant portion of their revenues are generated from products over a decade old - meaning the innovation cycle is quite slow and that products are generally 'classics' that last a long time. These are chips that sell for less than the price of a Nongfu bottle of water and are mission critical to the entire tech stack. Once spec'd in, people don't switch. It is why this is a lot harder to substitute.

The dentist analog likely really isn't that good, but even then, in 20 years stuff has changed, I was to a dentist like 1-2 years ago, and was told they are now using a newer and better filling for stuffing holes in teeth, compared to what I got like ~10 years ago.

 

[abenomics12345]

Who are those people? Actual people working on it in China? Or from established players such as Dassault Systems CATIA?

My information is based on my conversations with the companies I mentioned. I.E I spoke to people from half a dozen analog semi companies in China (include two founders) and those two Chinese software companies I mentioned.

So, the former.

The dentist analog likely really isn't that good, but even then, in 20 years stuff has changed, I was to a dentist like 1-2 years ago, and was told they are now using a newer and better filling for stuffing holes in teeth, compared to what I got like ~10 years ago.

Dentistry may change, but that is not going to make up for the fact that the shortage of dentists in China is not going to change for the next 5 years. Again, this is from regulators/industry participants in China - and if you look at the number of dentists in schools today. It's like how Moutai's production volumes today are irrelevant to how much Moutai can be sold this year - because of 5 year minimum aging.

Where your logic is correct is in dental technology - for example, Angelalign's products is very competitive (some would say even better) to Invisalign - these two companies have about 80% of the market share in China. and Angelalign is busy looking to export to global markets. In another case, Chinese CBCT machines made by Fussen and Meiya Optoelectronics are catching up and displacing Korean makers like VA Tech. I fully expect them to displace Dentsply Sirona over the coming years.

But the 'gap' here is easier to catch up - and this logic is not replicable in sectors like analog semis, or in the case of dentists, the number/shortage.

The issue here is that your argument is that it is conceptual in nature (and 'could' be right) - and that my argument is based on evidence of what's happening on the ground.

 

[Michaelsinodef]

My information is based on my conversations with the companies I mentioned. I.E I spoke to people from half a dozen analog semi companies in China (include two founders) and those two Chinese software companies I mentioned.

So, the former.

I c, so earlier when you wrote:

the 3D CAD (Dassault Systems CATIA) is the difficult part and at current paces will take a decade to reach similar capability.

Is that meant as just catching up to current capability or to actually catch up to Dassault Systems CATIA?

Also, it kind of doesn't matter, since it could be the case that they are lying or just being conservative in their estimate (not to mention, are you working for a foreign or a chinese company?) and it could always be the case that they just aren't properly able to make such estimates (maybe with more experience/work put into it, they can become slightly faster? Like if put in numbers, it might take them 100 days to write 10k lines right now, but maybe in 3 years time, they can do 14k lines in 100 day? Such a development could be the case. New tech / AI etc. might come out that could help fasten things).

Dentistry may change, but that is not going to make up for the fact that the shortage of dentists in China is not going to change for the next 5 years. Again, this is from regulators/industry participants in China - and if you look at the number of dentists in schools today. It's like how Moutai's production volumes today are irrelevant to how much Moutai can be sold this year - because of 5 year minimum aging.

I haven't disputed that though, just pointed out that new tech or developments might results in more productive / effectivity, whether that's for dentists but also for software designers and the likes.

Where your logic is correct is in dental technology - for example, Angelalign's products is very competitive (some would say even better) to Invisalign - these two companies have about 80% of the market share in China. and Angelalign is busy looking to export to global markets. In another case, Chinese CBCT machines made by Fussen and Meiya Optoelectronics are catching up and displacing Korean makers like VA Tech. I fully expect them to displace Dentsply Sirona over the coming years.

But the 'gap' here is easier to catch up - and this logic is not replicable in sectors like analog semis, or in the case of dentists, the number/shortage.

The issue here is that your argument is that it is conceptual in nature (and 'could' be right) - and that my argument is based on evidence of what's happening on the ground.

But evidence on the ground isn't the be all and end all or necessarily the truth, especially when it comes to predicting the future.
Although if one got actual good evidence on the ground, it can serve as a good rough prediction (so 9-11 years) on the condition that there aren't any new disruptive tech/developments which really can't be ruled out.

 

[tphuang]

I think we need to differentiate a few things here. China's software industry in general may be very well developed and have plenty of talent. We've seen with EV/AVs, the European automakers just can't develop anything. China also has the only alternatives to American tech companies when it comes to e-commerce, social media apps and search engines. What Huawei has done in the past few years is quite impressive. But for certain sectors, it could very well be the case that Chinese software makers are not putting out competitive products.

My personal experience dealing with software engineers from China was about 6 to 7 years ago when I interacted with Shanghai Gold exchange & CFFEX. I wasn't too impressed with what they did. But that could just be because high level of bureaucracy with those organizations.