Opinionated - China Chipping Away to Semiconductor Dominance

[j20blackdragon]

So if it's so stupid and useless then why is Trump so worried? Our friend Paul here seems to suggest we should microcontrollers instead of CPUs. What a genius Indian.

Global microcontroller market is estimated to be valued at $26.9 billion by 2020, while Intel revenue for 2018 alone was already $70.8 billion.

So of course 'they' don't want China to pursue x86.

 

[Bussard Ramjet]

Semi matters of significance 2019

①. 三维闪存芯片 长江存储量产自研Xtacking架构64层三维闪存芯片，打破了美日韩厂商的垄断。
②. 车规级碳化硅Mosfet芯片 比亚迪微电子推出自研车规级Mosfet芯片，打破了德法日厂商的垄断。
③. CMOS图像传感器 北京豪威推出自研0.8um，4800万像素级的CMOS图像传感器，与索尼之间的技术差距缩短到不到一年。
④. X86架构桌面CPU 上海兆芯研发的KX6000桌面CPU正式量产，达到英特尔第七代Core i5的水平。
⑤. GPU 上海兆芯推出国产最高水平的GPU极瑞3000。
⑥. 手机芯片 华为海思推出业界最高性能的5g手机芯片麒麟990。
⑦. 电视芯片 华为海思推出业内性能最强的智能电视芯片海思Vxxx。
⑧. DRAM 合肥长鑫正式量产8GB LPDDR4型内存芯片。
⑨. 晶圆代工 中芯国际14纳米工艺晶圆代工线正式量产。
⑩. FPGA芯片 紫光同创推出28纳米工艺制程7000万门级titan2系列高性能FPGA芯片。

View attachment 551165

Can you actually tell me more? I have responded below by points as listed earlier:

1. Yangtze river's 64 layer NAND I know about, and can potentially be game changing for China.
2. I want to know more about this BYD Microelectronics Mosfet.
3. I have never heard about this company's CMOS sensor, and now it comes and claims that it is only 1 year behind Sony? Can you give me the name of the company? Google Translate doesn't translate very well./
4. Zhaoxin has minimal market share. Why is that? How much market share does it have in China?
5. Again, what is the market share?
6. Hisilicon has been a true success
7. How do you know it is the best?
8. Hefei Changxin, any more news? I thought it was producing specialty DRAMs and not general purpose DRAM.
9. SMIC again I know.
10. How does its FPGAs compare with market leaders. What is its market share?

 

[Paul2]

Global microcontroller market is estimated to be valued at $26.9 billion by 2020, while Intel revenue for 2018 alone was already $70.8 billion.

So of course 'they' don't want China to pursue x86.

I repeat my rationale there:

Having PC CPUs supply embargoed will be nowhere as painful as being denied microcontroller — a giant portion of Chinese light industry wholly depends on imported microcontrollers, while Intel CPUs are only needed for PCs to play videogames.

China stands to loose much more loosing access to MCUs than x86, that's indisputable in my opinion.

Second, China has domestic MCU.

GD32 is a new 32-bit high performance, low power consumption universal microcontroller family powered by the ARM Cortex®-M3 RISC core
http://www.gigadevice.com/products/microcontrollers/gd32/arm-cortex-m3/

I know of Gigadevice, but not much of their MCUs. They are much bigger in their flash business. Still, they have near no market presence in MCUs. Were they did, I would've known.

Can you actually tell me more? I have responded below by points as listed earlier:

1. Yangtze river's 64 layer NAND I know about, and can potentially be game changing for China.
2. I want to know more about this BYD Microelectronics Mosfet.
3. I have never heard about this company's CMOS sensor, and now it comes and claims that it is only 1 year behind Sony? Can you give me the name of the company? Google Translate doesn't translate very well./
4. Zhaoxin has minimal market share. Why is that? How much market share does it have in China?
5. Again, what is the market share?
6. Hisilicon has been a true success
7. How do you know it is the best?
8. Hefei Changxin, any more news? I thought it was producing specialty DRAMs and not general purpose DRAM.
9. SMIC again I know.
10. How does its FPGAs compare with market leaders. What is its market share?

1. No likely. Memory business is all about pricing cartels, and is not that profitable with scale involved when such cartels fall apart. See, memory prices are on a huge downward swing because Micron was first to break. The upward swing itself began when Micron bought Toshiba's DRAM business and Korean "picked up the signal" and began racking up prices.
2. Nothing much exciting. Power MOSFETs are important to switch large currents and voltages. Before electronics got that powerful (or better to say power hungry,) there was no incentive to get consumer electronics rated MOSFETs up into hundreds of amps and voltages range except for niche applications (in which China was not a player.) China is there just last man to get on the boat because low volume, niche products was never its game.
3. No idea either, but I know that Chinese makers in that area grew out of manufacturers of chips for webcams and low end stuff. It's quite natural that eventually they began trying higher end stuff too as tech is not much different in between low and high end.
4. I am very doubtful of "Chinese x86" — government bureaucrats been trying to make it happen for a very long time, but I see not commercial need for it.
5. Same opinion
6. Well, it is. Huawei made a nice SoC and been a principal driver behind 5G technical standard, so here it is.
7. Nothing to add.
8. Will be nice if they can also make money, and not live on government handouts while making. With the current market situation, that's unlikely.
9. Have to see it before I believe it.
10. Likely close to 0.00000000001%

 

[j20blackdragon]

I repeat my rationale there:

Having PC CPUs supply embargoed will be nowhere as painful as being denied microcontroller — a giant portion of Chinese light industry wholly depends on imported microcontrollers, while Intel CPUs are only needed for PCs to play videogames.

China stands to loose much more loosing access to MCUs than x86, that's indisputable in my opinion.

x86 PC/laptop CPU ban is not the concern here.

What happens when there is an x86 CPU ban on servers, workstations, and supercomputers?

You realize Intel has a near monopoly here, right?

According to IDC, global server revenues were $53 billion in 2016, of which $46 billion were x86 servers and $7 billion, non-x86 servers
https://seekingalpha.com/article/4084040-large-deep-learning-data-center-market

The MCU ban is not likely to happen. Already addressed in previous post.

First, there are too many companies/countries making microcontroller (MCU) for a ban on China to be effective.

View attachment 551713

First, MCUs are low-cost, mass-produced, commodity ICs. They are all about the same.

Second, do you see any company/country that has a monopoly on the MCU market? I see a large number of companies with evenly distributed market share. Any company dumb enough to ban MCUs to China will instantly have their market share gobbled up by someone else.

I know of Gigadevice, but not much of their MCUs. They are much bigger in their flash business. Still, they have near no market presence in MCUs. Were they did, I would've known.

First you criticize China for not having MCU and breathlessly warn us of a supposed MCU embargo. When we show you the Chinese MCUs, you move the goalpost and criticize China's lack of market share.

 

[gambit]

What a long thread we have here. Semiconductors are such and industry about which 99% stuff known in the popular culture is wrong.

Very good, sir. Regarding the semicon issue in this forum, your sobering and realistic perspective will be beneficial to the forum, but also unfortunately for you, your fellow countrymen will not be so kind precisely because you have...

- Relevant industry experience
- Critical thinking skills
- Objectivity
- Honesty

Qualities sorely lacking in this sub-forum.

As for meself, am currently in process engineering in the Probe area of the wafer manufacturing process, from wafer start to wafer sales.

https://www.mjc.co.jp/en/technology/column/wafer_prober.html

MJC is just one vendor I work with. Other vendors are FormFactor and TEL. I do not engage in die extraction and packaging but am high level familiar with those processes in order to fine tune the Probe testing processes.

I will be watching your participation about this particular issue with interests. Hope you keep those qualities.

Yangtze river's 64 layer NAND I know about, and can potentially be game changing for China.

Not likely.

NAND is at the end of its technological progress. Multi-bits per cell, stacking, or controller improvements are minor and not very profitable add-ons. NAND has reached commodity status and profit margins on commodities are always thin. The NAND technology itself -- like every technology -- is the final determinant of limits. Below the 20 nm node, NAND begins to decline in possibilities for innovations.

The world demands an alternative to NAND and the criteria are the same always: non-volative data and capacity. NAND have reached its technological limits for those criteria.

 

[Han Patriot]

I repeat my rationale there:

Having PC CPUs supply embargoed will be nowhere as painful as being denied microcontroller — a giant portion of Chinese light industry wholly depends on imported microcontrollers, while Intel CPUs are only needed for PCs to play videogames.

China stands to loose much more loosing access to MCUs than x86, that's indisputable in my opinion.

I know of Gigadevice, but not much of their MCUs. They are much bigger in their flash business. Still, they have near no market presence in MCUs. Were they did, I would've known.

1. No likely. Memory business is all about pricing cartels, and is not that profitable with scale involved when such cartels fall apart. See, memory prices are on a huge downward swing because Micron was first to break. The upward swing itself began when Micron bought Toshiba's DRAM business and Korean "picked up the signal" and began racking up prices.
2. Nothing much exciting. Power MOSFETs are important to switch large currents and voltages. Before electronics got that powerful (or better to say power hungry,) there was no incentive to get consumer electronics rated MOSFETs up into hundreds of amps and voltages range except for niche applications (in which China was not a player.) China is there just last man to get on the boat because low volume, niche products was never its game.
3. No idea either, but I know that Chinese makers in that area grew out of manufacturers of chips for webcams and low end stuff. It's quite natural that eventually they began trying higher end stuff too as tech is not much different in between low and high end.
4. I am very doubtful of "Chinese x86" — government bureaucrats been trying to make it happen for a very long time, but I see not commercial need for it.
5. Same opinion
6. Well, it is. Huawei made a nice SoC and been a principal driver behind 5G technical standard, so here it is.
7. Nothing to add.
8. Will be nice if they can also make money, and not live on government handouts while making. With the current market situation, that's unlikely.
9. Have to see it before I believe it.
10. Likely close to 0.00000000001%

Paul, MCUs had been almost the same for the past 40 years. Go figure genius. Chinese companies do make them with some oem brand .

 

[Paul2]

What happens when there is an x86 CPU ban on servers, workstations, and supercomputers?

You realize Intel has a near monopoly here, right?

They are not important. Playing videogames is not a matter of national importance.

Existing stocks will last for quite a while, you can use lower power SoC to take the slack in a crunch, and you really don't have PCs being a big export category, unlike the entirety of light industry.

Internal consumption of PCs matter relatively little in comparison to export value of light industries and its importance for maintaining China's grip on the world.

Paul, MCUs had been almost the same for the past 40 years. Go figure genius. Chinese companies do make them with some oem brand

I figured that out long ago, and deal with it on every day basis. Chinese MCU companies are not competitive at large, otherwise we would've be buying them and not importing Atmel, NXP, and STM by tonnes.

For the few worthy contenders, I know each of them, and in few cases their C-levels personally. And the best contender, Espressif, is not even run by a Chinese citizen. Zhang Ruian is Singaporean.

First, MCUs are low-cost, mass-produced, commodity ICs. They are all about the same.

They are way, way, way more to it than you ever imagine, and even in the most commoditised parts of the MCU market, there are own intricacies. China once made those generic 8051 clones, but where they are now?

Programming model of 8051 was very old school, archaic, and inconvenient: multiple memory banks, inability to use external memory in a convenient manner, low code density and unsuitability for use with modern compilers, wildly varying instruction execution times. I can count 10 more things that put 8051 (and therefore its clones) into the scrap heap of history.

Even 5 cents MCUs that are used for "singing postcards," are now using Atmel, PIC or 32bit Arm based MCUs because there are ones that are cheap enough and easier to program than 8051.

The importance for an MCU to be easily programmable by a modern programming language was completely overlooked by domestic makers.

As for meself, am currently in process engineering in the Probe area of the wafer manufacturing process, from wafer start to wafer sales.

https://www.mjc.co.jp/en/technology/column/wafer_prober.html

MJC is just one vendor I work with. Other vendors are FormFactor and TEL. I do not engage in die extraction and packaging but am high level familiar with those processes in order to fine tune the Probe testing processes.

I will be watching your participation about this particular issue with interests. Hope you keep those qualities.

Nice to see somebody else from semiconductors industry here. Stuff like testing, process control, and metrology are not less important than the lithography itself. My own biggest suspicion about what is wrong with Chinese fabs is that those bimbos very likely continuing to dish out big money for the latest litho equipment while completely foregoing the metrology and control. My latest experience talking with Unigroup and Canyon Bridge (really just a another Unigroup front) staffers, only reinforced that opinion further.

In semiconductors, you can buy an equipment for a certain node, but can not really "buy a process." These days, you can only develop the process (in a literal sense of the world) for your own use scenarios, tuned for your business circumstances. For what Chinese industry needs, yields are paramount for many reasons, including the need for reducing the testing expense. And those guys are trying to make commodity products on an expensive, low yield multiple patterning node...

 

[gambit]

...completely foregoing the metrology and control.

The in-line param process is crucial...

http://www.lricks.com/wlrt.htm

The WLR paradigm offers several advantages over a sample burn-in process monitor. The most obvious and immediate advantage of WLR testing is its ability to supply real-time information. This is a result of using an automatic parametric tester (APT) to collect in-line data. Feedback on process results and their potential impact on reliability are immediately available.

In-line parametric testing is essentially measurement of the physical structures of whatever design it is. Measurement is done via the standard criteria of voltage, current, and resistance. Things we learn in high school electronics, except doing it at the nanometer level. If a certain voltage input produced a higher out of bounds resistance, a wall maybe too thick, lower and the wall is too thin. Either way, performance and long term reliability -- wafer level reliability (WLR) -- will be affected.

In-line parametric testing is when all the dies on the wafer are tested while the wafer is still under fabrication, not yet ready for full functional testing. Every time there is a piece of new equipment or new recipe (process) or new gas/chemical, a sample percentage should be in-line parametric tested. Even if there is a new deionized (DI) water vendor, and DI water is a common component in wafer processing, a sample percentage of the product line should be in-line parametric tested. At least all the major semicon manufacturers are this strict even with matured products.

The problem with commodity product manufacturing, like NAND or DRAM or NOR or SRAM, is that it is easy to get overconfident that our processes do not need periodic verification that everything is within specs. Plus, because the profit margin on commodity products are already thin and the cyclical market nature of the semicon industry, the pressure to reduce production time is high among managers at all levels going up to the executive officers themselves. Machines sometimes skips their periodic maintenance inspection (PMI) and if the pressure to produce is great enough, especially at fiscal quarter end, decisions to 'pencil whip' the records can be made at the line level, out of sight of the managers.

When I used to work for Micron Technology and helped set up their Shanghai facility, we had a lot of conflicts between Micron's ways of doing things vs the 'Chinese' ways as many local hires balked at how often we sample inspect, measure, verify, and they especially objected to destructive testing, which Micron's US sites often does. We will take wafers and literally high voltage/current functional test them to 'full dead' and analyze the data. For nearly two yrs from start up, most of Micron's Shanghai's products were sold to second tier clients. Micron's US sites usually need only six months to qualify for tier one customers. A major problem with manufacturing in mainland China was power bumps or 'dirty' electricity, although it is less of a problem today, and every time we had a power bump, wafer scraps were inevitable, and to make sure all the machines were not affected, metrology measurement had to be done and done at a higher quantity of wafers and frequency between line starts, which did not sit well with the Chinese executives.

In semiconductors, you can buy an equipment for a certain node, buy can not really "buy a process." These days, you can only develop process (in a literal sense of the world) for your own use scenarios, tuned for your business circumstances. For what Chinese industry needs yields are paramount for many reasons, including the need for reducing testing expense. And those guys are trying to make commodity products on an expensive multiple patterning node...

Absolutely -- the highlighted. Intel have their 'Copy Smart' process intended to transfer knowledge and skills among sites.

http://www.chinadaily.com.cn/world/2006-05/30/content_603764.htm

Engineers strive to duplicate even the subtlest of manufacturing variables, from the color of a worker's gloves to the type of fluorescent lights in the building. Employees from around the world spend more than a year at a development lab in Oregon learning their small piece of the new recipe so they can bring it back to their home factory.

But even so, Intel has to allow for small variations from site to site.

 

[qwerrty]

First you criticize China for not having MCU and breathlessly warn us of a supposed MCU embargo. When we show you the Chinese MCUs, you move the goalpost and criticize China's lack of market share.

mcu is simple to design and make. a lot of companies already doing it and selling at very cheap price. that's why chinese just ignoring that buisiness. it's bad, but they can get into that buisiness easily if banned.... i think a ban is actually good for china, but too bad the US don't have monopoly in that industry.


new startup releasing high perfomance general-purpose mcu based on risc-v
http://www.takungpao.com.hk/mainland/text/2019/0401/270629.html
aec-q 100 grade mcu from navinfo for auto in mass production
http://finance.jrj.com.cn/2018/12/20152726772696.shtml

.

 

[Paul2]

Plus, because the profit margin on commodity products are already thin and the cyclical market nature of the semicon industry, the pressure to reduce production time is high among managers at all levels going up to the executive officers themselves.

The trick here is that you maximise control and test before the wafer leaves the fab, so your fab yield does not influence yields downstream, and reduce the need for all forms of "after fab testing."

The rationale is that there is very little what can happen with the chip after fabrication, but a messed up chip not caught by test means scrapping them after expensive packaging (yes, in cheap chips, package some times costs more than the silicon,) or on PCBs of expensive devices, if not finished ones. Or imagine one such chip getting into some heavy industrial machinery and failing...

When I used to work for Micron Technology and helped set up their Shanghai facility, we had a lot of conflicts between Micron's ways of doing things vs the 'Chinese' ways as many local hires balked at how often we sample inspect, measure, verify, and they especially objected to destructive testing

Well, most of China's bimbo managers are American schooled, Stanford, Harward etc. Some time I feel that we have more Stanford grads in big Chinese corporations than American MNCs. What those guys objecting to destructive testing were doing was as good as scrapping the whole wafer for a single chip.

A major problem with manufacturing in mainland China was power bumps or 'dirty' electricity, although it is less of a problem today, and every time we had a power bump, wafer scraps were inevitable,

A lot of industrial companies here own a following equipment: giant flywheels that are powered by AC motor from the grid on one end, and DC generators on the other end. Never seen those things outside of China.

mcu is simple to design. a lot of companies already doing it and selling at very cheap price. that's why chinese just ignoring that buisiness. it's bad, but they can get into that buisiness easily if banned... i think a ban is actually good for china, but too bad the US don't have monopoly in that industry.


new startup releasing high perfomance general-purpose mcu based on risc-v
http://www.takungpao.com.hk/mainland/text/2019/0401/270629.html
aec-q 100 grade mcu from navinfo for auto in mass production
http://finance.jrj.com.cn/2018/12/20152726772696.shtml

.

Yes, they are easy to do and make, but making a viable business with them is terribly hard.

I don't care so much for individual MCU's performance as the performance of companies making them, and particularly supply reliability and engineering support.

 

[gambit]

The trick here is that you maximise control and test before the wafer leaves the fab, so your fab yield does not influence yields downstream, and reduce the need for all forms of "after fab testing."

The rationale is that there is very little what can happen with the chip after fabrication, but a messed up chip not caught by test means scrapping them after expensive packaging (yes, in cheap chips, package some times costs more than the silicon,) or on PCBs of expensive devices, if not finished ones. Or imagine one such chip getting into some heavy industrial machinery and failing...

Regarding the highlighted, Micron does not sample test, even with commodity product lines. Micron literally test %100 wafers and packaged dies.

I started with Micron at the Boise, Idaho, fab, specifically in the Burn-In dept. In Burn-In, we %100 tested extracted and packaged dies.

https://caeonline.com/buy/burn-in-systems/micron-technologies-ambyx/9013033
http://www.semidice.com/active-die-components/micron-technology/

...data from Micron’s AMBYX™ burn-in system provides detailed failure information, which is correlated to abundant wafer parametric and functional data.

The AMBYX system was essentially a man-size oven that have variable temperatures that depends on the specific product. The packaged dies are literally baked at food cooking temperature and while under heat stress, the dies are injected with data and read cycling. At the end of the Burn-In period, the dies are categorized into 'bins'.

Later, I moved into the Probe department. In Probe, we literally test %100 wafers. Micron considers Probe to be under Fab, other companies consider Probe to be under Packaging. I think Micron have the better way. In Probe, the wafers are tested for functionality and final metrology (structural dimensions). Since Probe is part of the fab, any deviation discovered by the Functional and Parametric teams are quicker to feedback to the fab process engineers. I also worked on the military line as well but will leave that out of this discussion.

Well, most of China's bimbo managers are American schooled, Stanford, Harward etc. Some time I feel that we have more Stanford grads in big Chinese corporations than American MNCs.

Schools can only teach theoretical knowledge and although that is important, it is the working environment that truly conditions you to be an engineer.

What those guys objecting to destructive testing were doing was as good as scrapping the whole wafer for a single chip.

Destructive testing sets the extreme. From that line, we can better extrapolate the longevity and reliability of the design with less the need for 'real world' feedback. For example, regarding NAND, the average consumer of SSD drives will run Windows 20-something before the SSD fail, so if we have to rely only on users that have heavy read/write actions, like the New York Stock Exchange (NYSE) or Facebook or Boeing, it would take at least 2 yrs before we have any reliable 'real world' data. If a die takes longer to fail than other dies -- on the same wafer -- that die would be examined by Yield Analysis and Enhancement under the electron microscope to find out why. Many useful information from that examination were feedback to the fab for process improvements.

Companies do competitor research all the time. We bought finished products from Korean, JPNese, and Chinese companies. With the scanning electron microscope (SEM), we have a good idea on how others manufacture the same concepts. For my personal use, I will buy SSDs first from Micron or Intel, then from Samsung, but not from any Chinese foundry fab if I know where the SSD's source.

 

[j20blackdragon]

If USA, or Taiwan will embargo an Nvidia GPU, or Intel CPU, nobody but videogame addicts will complain,

Having PC CPUs supply embargoed will be nowhere as painful as being denied microcontroller — a giant portion of Chinese light industry wholly depends on imported microcontrollers, while Intel CPUs are only needed for PCs to play videogames.

They are not important. Playing videogames is not a matter of national importance.

We now have 3 separate instances of you making the absurb claim that x86 CPUs are only useful for videogames.

Repeating a falsehood over and over again like a broken record does not make the claim true.

How come you won't address Intel's near monopoly for x86 servers, workstations, and supercomputers?

Furthermore, explain how China benefits from an Intel monopoly in the first place.

 

[Paul2]

We now have 3 separate instances of you making the absurb claim that x86 CPUs are only useful for videogames.

Repeating a falsehood over and over again like a broken record does not make the claim true.

How come you won't address Intel's near monopoly for x86 servers, workstations, and supercomputers?

Furthermore, explain how China benefits from an Intel monopoly in the first place.

First point, no Chinese PCs, nor servers, nor supercomputers are as important for the economy as our light industries and exports.

Second, nothing prevents you from using non-x86 hardware to make the aforementioned stuff. You address this issue by not using Intel hardware, and using something else for that instead.

I think I restated that point a few times already as well.

Why to use 5000 RMB Intel Xeons, if you can simply make higher power versions of existing Chinese consumer electronics chips for less than 50 a pop. Yes, it will probably be not as efficient, but nowhere near the crippling level.

Those in Chinese industry bodies who kept complaining about that for the last 2 decades, and been wasting fantastic amounts of state funds trying to make a Chinese x86 over and over again don't get that simple idea.

You can say that individuals really have no choice, but to keep buying expensive x86 hardware, as making microchips is outside of what individuals can afford. But when you talk about businesses with tenths of billions of RMBs and a national level industry body having access to a bloody 1 trillion yuan fund, not taking note of that it is plain stupidity in first case, and a criminal dereliction of duty in the second.

 

[j20blackdragon]

First point, no Chinese PCs, nor servers, nor supercomputers are as important for the economy as our light industries and exports.

There isn't going to be a MCU ban. Stop trying to beat that dead horse.

Second, nothing prevents you from using non-x86 hardware to make the aforementioned stuff. You address this issue by not using Intel hardware, and using something else for that instead.

China already tried that years ago with the Loongson/Godson CPUs based on the MIPS architecture. MIPS-based processors had a near-zero market share back then, and has a near-zero market share today.

The moment you remove yourself from x86, you also remove yourself from the x86 software and OS environment that the rest of the world has been using for the last several DECADES.

Why are you so eager to put China on an island?

wasting fantastic amounts of state funds trying to make a Chinese x86 over and over again

They are not trying. The Chinese x86 CPUs are already in production.

buying expensive x86 hardware

x86 is expensive because Intel has a monopoly. How come you won't address this?

 

[Paul2]

remove yourself from the x86 software and OS environment that the rest of the world has been using for the last several DECADES.

And what so? Just recompile your software, no problems.

China already tried that years ago with the Loongson/Godson CPUs based on the MIPS architecture. MIPS-based processors had a near-zero market share back then, and has a near-zero market share today.

I think your own point can be turned around in the exactly same manner here.

Industry bodies been wasting billions over the last 20 years, trying to make useless x86 clones without any commercial success, and the situation remains the same to this day. Those people never got to stop and think why they keep failing every time.

Yes, this time they got expensive licensed versions of gimped Via and AMD chips, just as was with Cyrix 20 years ago. To the effect, they are still the same relabelled versions of low-end offerings, for which we still pay a premium price. And yet again, I bet, it will be dead on arrival.