Chinese semiconductor industry


AndrewS

Major
Registered Member
Agreed.

iOS and Android are already deeply entrenched in the mobile space.

So Huawei, wisely, decided instead of attacking them head on on their advantageous position, it would first start taking over the new industies which are wildly growing lately, IoT (IoE for Huawei).

After 2-3 years with more 5G applications coming up on the IoT space and more advances with AI, the Huawei would dominate this huge industry and then start leveraging this user base of its HarmonyOS to finally attack the mobile space.

IMO in 2-3 years we can draw some preliminary conclusion on how successful HarmonyOS has become.

This is a winning strategy for Huawei

As expected. I wrote something similar 2 years ago here
 

voyager1

Captain
Registered Member
Expanding on my previous post, the IoT market in 2024 for China is estimated to reach $300 Billion!

Thats just for China!

And around ~$270 Billion each for EU and US separately.

So Thats a combined ~$900 billion global market (actually more, thats just the 3 biggest economies...)

If Huawei (monetizing its OS ala Apple's app store style) can mooch "just" 10% from this thats $90 billion of "free" money flowing to Huawei.

Make that 40% and it is $360 billion !!

Ofc there will be sanctions, restrictions etc, but there is crazy crazy money to made on IoT.

Please, Log in or Register to view URLs content!
 
Last edited:

jfcarli

New Member
Registered Member
If China keeps manufacturing and exporting air conditioners, refrigerators, lamps, etc... which operate with Harmony OS, sooner or later even outside China people will discover that Harmony OS on their phones will be indispensable to control all the other equipment in their houses/offices/cars.
 

krautmeister

Junior Member
Registered Member
Besides lithography though, is actual chip design. Everything i have research on kupeng, loongson , zhaoxin, and alibaba risc-v boards is that they are pretty weak :(
Loongson ISA is just getting started and Alibaba's RISC-V is already leading all RISC-V processors. I think there's a lot to be excited about when it comes to RISC-V. Once LoongsonArch ISA has finished removing the remnants of MIPS and replacing it with RISC-V, China will have 2 major RISC-V projects, along with a handful of RISC-V support ICs from the likes of Huawei. The question is when will there be a competitive SoC and/or CPU for RISC-V?

RISC-V design philosophy minimizes the ISA to bare minimum in order to serve as a monolithic ISA for everything. This is similar to the design philosophy of microkernel based operating systems. RISC-V ISA can be thought of as the "hardware" microkernel upon which IC designs would need to add their own customizations on top whether it be for general purpose CPU, SoC, ASIC, microcontroller, etc. So, in a very real way, adoption of RISC-V complements the spread of Harmony OS with further and further integration of the software stack and hardware which China can use to break foreign technology monopolies.
 

gelgoog

Captain
Registered Member
RISC-V is a mess because the available architecture isn't 100% adequate for doing a decently performing design. You need to add your own hardware instructions which make the processors incompatible with each other.

I think China is still better off with an ARM architectural license and making their own CPU designs.
 

gelgoog

Captain
Registered Member

GlobalFoundries is in deep shit.
Please, Log in or Register to view URLs content!

"IBM is suing GlobalFoundries for breach of contract and is asking for $2.5 billion in damages.
...
Back in 2014, IBM was trying to get out of the commercial semiconductor production business. It was shopping some of its aging fabs, but failed to find any buyers, and ended up paying GlobalFoundries $1.5 billion essentially to take them. The two agreed GF would complete the 14 nm IC production process IBM had been working on (which GF did). GlobalFoundries would then supply IBM with 14-nm chips (and so it did), and GF would simultaneously work on the next node. The next node was supposed to be 10nm, but given the competitive situation in the IC manufacturing business, GF decided to skip 10nm and go directly to 7nm. GF says IBM agreed with this decision.

GlobalFoundries was already badly lagging TSMC and Samsung on 7nm development, and in 2018 it announced it would not complete work on the process. GF argues it would have been financially ruinous to complete development on 7nm.
The crux of IBM’s breach-of-contract accusation is GF’s failure to successfully develop a 7nm IC production process.

Again, IBM has yet to file its documents in court, but it provided EE Times with the following statement regarding GlobalFoundries’ countersuit:
“This lawsuit is yet another attempt by GlobalFoundries to cover up its fraud and deliberate breaches of contract in failing to fulfill its legal obligations to IBM, including the development and supply of high-performance semiconductor chips. IBM contributed $1.5 billion to GlobalFoundries to supply the next generation of chips, and GlobalFoundries utterly abandoned IBM as soon as the final payment was received and sold off assets from the deal for its own enrichment. IBM welcomes the opportunity to seek the recovery of the substantial damages it is due.”"


This lawsuit is going to make any IPO they want to make a nightmare to pull through. Pissing off your clients is not exactly a great way to run a business.
 
Last edited:

HybridHypothesis

Junior Member
Registered Member
GlobalFoundries is in deep shit.
Please, Log in or Register to view URLs content!

"IBM is suing GlobalFoundries for breach of contract and is asking for $2.5 billion in damages.
...
Back in 2014, IBM was trying to get out of the commercial semiconductor production business. It was shopping some of its aging fabs, but failed to find any buyers, and ended up paying GlobalFoundries $1.5 billion essentially to take them. The two agreed GF would complete the 14 nm IC production process IBM had been working on (which GF did). GlobalFoundries would then supply IBM with 14-nm chips (and so it did), and GF would simultaneously work on the next node. The next node was supposed to be 10nm, but given the competitive situation in the IC manufacturing business, GF decided to skip 10nm and go directly to 7nm. GF says IBM agreed with this decision.

GlobalFoundries was already badly lagging TSMC and Samsung on 7nm development, and in 2018 it announced it would not complete work on the process. GF argues it would have been financially ruinous to complete development on 7nm.
The crux of IBM’s breach-of-contract accusation is GF’s failure to successfully develop a 7nm IC production process.

Again, IBM has yet to file its documents in court, but it provided EE Times with the following statement regarding GlobalFoundries’ countersuit:
“This lawsuit is yet another attempt by GlobalFoundries to cover up its fraud and deliberate breaches of contract in failing to fulfill its legal obligations to IBM, including the development and supply of high-performance semiconductor chips. IBM contributed $1.5 billion to GlobalFoundries to supply the next generation of chips, and GlobalFoundries utterly abandoned IBM as soon as the final payment was received and sold off assets from the deal for its own enrichment. IBM welcomes the opportunity to seek the recovery of the substantial damages it is due.”"


This lawsuit is going to make any IPO they want to make a nightmare to pull through. Pissing off your clients is not exactly a great way to run a business.

kekw, can't wait to short it then
 

gelgoog

Captain
Registered Member
Please, Log in or Register to view URLs content!

SiFive are the main developers of RISC-V core designs. They are behind most of the standards. They have also been taking patents on how to efficiently implement RISC-V in hardware.

If Intel does buy SiFive they will most likely bungle it up. A lot of people working on RISC-V already have suspicions that SiFive will eventually try to use their patents to make RISC-V essentially non-free. Intel most likely wants to purchase SiFive to use RISC-V on their own products and take down ARM a notch or two. It might be good or bad for RISC-V in the long run. But the whole patent situation around RISC-V is a bit of a mess because of cases like this.
 

vincent

Senior Member
Please, Log in or Register to view URLs content!

SiFive are the main developers of RISC-V core designs. They are behind most of the standards. They have also been taking patents on how to efficiently implement RISC-V in hardware.

If Intel does buy SiFive they will most likely bungle it up. A lot of people working on RISC-V already have suspicions that SiFive will eventually try to use their patents to make RISC-V essentially non-free. Intel most likely wants to purchase SiFive to use RISC-V on their own products and take down ARM a notch or two. It might be good or bad for RISC-V in the long run. But the whole patent situation around RISC-V is a bit of a mess because of cases like this.
Just create a fork like any other open source project

The patents will be on the implementations, not the instruction set. There are more than one way to skin a cat
 

Top