Chinese semiconductor industry

gadgetcool5 said:

Get real about the Chinese semiconductor industry​

China faces competition from a massive investment surge by South Korea's Samsung Electronics and Taiwan's TSMC

On January 9, Japan’s Nikkei newspaper reported that Samsung Electronics may invest more than $30 billion in its memory and logic semiconductor businesses in 2021: “Chipmaking equipment manufacturers say the company has provided order plans for 2021 that point to a further 20% to 30% increase in spending.”

On January 14, it was reported that leading semiconductor foundry TSMC plans capital spending of between $25 billion and $28 billion this year, an increase of 45% to 63%. It is widely expected but not yet confirmed that some of this investment will be in anticipation of outsourcing from Intel.

The Information Network, a market research company focused on microelectronics, predicts that “investments in new [semiconductor] fabs or capacity expansion will exceed US$160bn in China over the coming 5-7 years; we expect this will drive an increase in China’s equipment spending to more $40 billion in 2025.”

In other words, China’s total semiconductor investments in 2025 may be less than Samsung and TSMC’s combined investments in 2021.

Please, Log in or Register to view URLs content!

Click to expand...

You are confusing money spent with result obtained. China's fast advancement already to 3nm design already discredits your theory as it is by far the fastest-moving. So what it spent on semiconductor development is irrelevant as it is unreliable (a top national priority as this will have government support that can be unannounced and classified). But the competition is not with TSMC or Samsung because that seems too easy a task. The real fight is making the lithographs. The author should get real as he's not comparing the right things and not measuring the correct units.

gadgetcool5 said:

Not only will China not catch up, it will fall further and further behind as it underinvests by tens of billions against competitors that are already far ahead. Frankly it seems like this competition is over before it started. China lost.

Click to expand...

Did falun gong teach that trick? State your wish as fact and it'll come true? LOLOL Your efforts to instill defeatism in the Chinese had failed before they began. You bet against China and you will lose every time.

gadgetcool5 said:

Get real about the Chinese semiconductor industry​

China faces competition from a massive investment surge by South Korea's Samsung Electronics and Taiwan's TSMC

On January 9, Japan’s Nikkei newspaper reported that Samsung Electronics may invest more than $30 billion in its memory and logic semiconductor businesses in 2021: “Chipmaking equipment manufacturers say the company has provided order plans for 2021 that point to a further 20% to 30% increase in spending.”

On January 14, it was reported that leading semiconductor foundry TSMC plans capital spending of between $25 billion and $28 billion this year, an increase of 45% to 63%. It is widely expected but not yet confirmed that some of this investment will be in anticipation of outsourcing from Intel.

The Information Network, a market research company focused on microelectronics, predicts that “investments in new [semiconductor] fabs or capacity expansion will exceed US$160bn in China over the coming 5-7 years; we expect this will drive an increase in China’s equipment spending to more $40 billion in 2025.”

In other words, China’s total semiconductor investments in 2025 may be less than Samsung and TSMC’s combined investments in 2021.

Please, Log in or Register to view URLs content!

Not only will China not catch up, it will fall further and further behind as it underinvests by tens of billions against competitors that are already far ahead. Frankly it seems like this competition is over before it started. China lost.

Click to expand...

Shall I do my own version of cherry picking using your exact same article?

Get real about the Chinese semiconductor industry​

The value of semiconductors made in China has risen 3.9 times since 2010 and is forecast to increase another 1.9 times to $43.2 billion in 2025. According to IC Insights, this would be at least 7.5% of the global market. And it won’t stop there.

Depending on the expansion of foreign companies’ production capacity in China, the share accounted for by Chinese companies may or may not increase. Both Samsung and SK Hynix are likely to make substantial new investments in China.

As for spending on wafer fab equipment alone (not counting the cost of new factory buildings and related facilities), SEMI, the global semiconductor equipment and materials industry association, calculates that China accounted for 26% of the total in 2020, compared with 24% for Taiwan and 23% for South Korea. China’s share has doubled over the past five years.

In this context, keep an eye on Yangtze Memory Technologies Company (YMTC), China’s homegrown maker of NAND flash memory. (NAND stands for NOT-AND; it is a logic gate which produces an output which is false only if all its inputs are true.) YMTC is investing more than $20 billion in its fab in Wuhan. It aims to double 300mm wafer processing capacity to 100,000 per month by the end of this year.

If and when yields reach an acceptable level, this could give the company a market share of more than 5%.

NAND flash is a commodity used in cell phones, memory cards, USB flash drives, solid state drives (SSDs) and other products. More competition means lower prices and more innovation. What’s not to like?

YMTC was spun off from the Tsinghua Unigroup in 2016 after two years of NAND flash development work (Tsinghua Unigroup maintains a controlling stake of 51%). It introduced its unique Xtacking® architecture and began mass production of China’s first 3D NAND flash memory in 2018. This year it should begin trial production of 192-layer chips, which could close the technology gap (although not the economies-of-scale gap) with its more experienced rivals.

Then there is SMIC, the largest Chinese-owned semiconductor foundry, which the U.S. sanctioned as part of its effort to stifle Huawei. For SMIC to catch up with TSMC now looks like nearly impossible. YMTC also supplies Huawei, but so far seems to have escaped the attention of America’s global economic police.

In his essay “How America Can Shore Up Asian Order,” published in the January 12 issue of Foreign Affairs, Kurt Campbell wrote that “… it is neither practical nor profitable to exclude Beijing from Asia’s vibrant future.” As President-elect Joe Biden’s incoming coordinator for Indo-Pacific affairs, Mr. Campbell was presumably telegraphing a shift to constructive engagement in Washington, DC.

Constructive engagement in the areas of IP protection and market access is definitely called for, but U.S. policy makers might better focus on rebuilding their own country rather than trying to keep other countries down.

gadgetcool5 said:

Not only will China not catch up, it will fall further and further behind as it underinvests by tens of billions against competitors that are already far ahead. Frankly it seems like this competition is over before it started. China lost.

Click to expand...

Can't grow an industry by dumping simply machines worth billions into it.

That is not organic growth, if you buy thousands of machining centres, assembly machines and buildings and hire few thousand farmers the end result won't be a car manufacturing plant, but a huge pile of wasted money.

gadgetcool5 said:

Not only will China not catch up, it will fall further and further behind as it underinvests by tens of billions against competitors that are already far ahead. Frankly it seems like this competition is over before it started. China lost.

Click to expand...

It will take at most 10 years for China to develop a competitive EUV lithography machine.
Other wafer fab equipment will take a lot less time.

And as soon as domestic equipment becomes competitive for a process node, there's going to be a massive programme in new wafer fabs .

Given that we're now seeing a wartime level of industrial mobilisation in the semiconductor field, I expect that in 3-10 year's time, we will have Chinese companies buying more fab equipment than the rest of the world combined.

That is based on today's China consuming or manufacturing more microchips than the rest of the world combined.

WTAN said:

Just a misunderstanding, no worries.
The CETC 45nm node (65nm resolution) DUVL is less advanced than the SMEE 28nm node (38nm resolution) DUVL.
But CETC is doing alot of R&D and will eventually catch up with SMEE. They aim to eventually produce their own 28nm node DUVL very soon.
CETC and SMEE are systems integrators and as long as the subsystems are available, they will produce more advanced Lithographs.

Just heard that ICRD is doing EUV Lithograph R&D using the Light Source from the Shanghai Synchrotron. They have a prototype EUV equipment of some sort & are working on it. Not much else information available.

Click to expand...

I am certain that CETC has a 40nm node internally for two years now --- the node used only for military --- if they are producing NRIET Huarui-2 octacore DSP for military radars since 2018.

NRIET Huarui-3 is probably 28nm or lower. This has been in development since 2018 or earlier.

Oldschool said:

@WTAN
siom.cas.cn Shanghai been doing lpp-euv since 2012 , how's that coming along?

Beijing and Shanghai competition heating up.
CETC Beijing has ion impl, cmp, and now litho
Naura Beijing has pvd, ald,

Shanghai has imec etch, smee litho

Click to expand...

The Shanghai Institute of Optics was the partner of Changchun Institute when they jointly developed the 28nm - 32nm EUVL Prototype back in 2017.

Shanghai Inst has a lot of experience and i believe their LPP-EUV is a front runner in Light sources.
The other party of course is the Harbin Institute with their DPP-EUV.
There may be a competition going on between the 2 types of Light Sources.
They have not yet decided which Light Source to use.
Yes i think there is a Beijing (CETC) and Shanghai (SMEE) competition in both DUVL and EUVL.
A SMEE insider confirmed that Changchun Institute will not be the Integrator of the EUVL. It will just develop the Optics.
So i think the integrator of the new EUVL will be CETC or SMEE. Maybe even 2 integrators.

Tam said:

I am certain that CETC has a 40nm node internally for two years now --- the node used only for military --- if they are producing NRIET Huarui-2 octacore DSP for military radars since 2018.

NRIET Huarui-3 is probably 28nm or lower. This has been in development since 2018 or earlier.

Click to expand...

Tam, you are probably right. They most likely already have a 45nm DUVL in small scale production.
The factory was built for mass production.

The CETC 45nm node DUVL is a Dry Arf Lithograph.
All CETC has to do is add a Liquid Immersion System to the Lithograph and they would have created a 28nm node Immersion DUVL similar to the latest SMEE 28nm DUVL (ASML 1980i/2000i)

WTAN said:

Tam, you are probably right. They most likely already have a 45nm DUVL in small scale production.
The factory was built for mass production.

The CETC 45nm node DUVL is a Dry Arf Lithograph.
All CETC has to do is add a Liquid Immersion System to the Lithograph and they would have created a 28nm node Immersion DUVL similar to the latest SMEE 28nm DUVL (ASML 1980i/2000i)

Click to expand...

I don't think photo resist for the 28nm is ready for mass production yet. Still in testing

via Mr unknown



China Debuts First 7nm Data Center GPU To Rival Nvidia, AMD
By Zhiye Liu
First Published 18 hours ago

Ready for prime time

Shanghai Tianshu Zhixin Semiconductor Co., Ltd., commonly known as Tianshu Zhixin, announced that its Big Island (BI) GPGPU has come to life. The BI is touted as China's first domestic GPGPU that's tailored towards AI and HPC applications and other industries, such as education, medicine, and security.

The BI packs 24 billion transistors, and it's based on a home-made GPU architecture. The chip is built with the cutting-edge 7nm process node and 2.5D CoWoS (chip-on-wafer-on-substrate) packaging. Tianshu Zhixin doesn't explicitly reveal the foundry that's responsible for producing the BI. However, the description of the node coincides with one of TSMC's manufacturing processes.

Tianshu Zhixin commmenced development on its BI chip in 2018. The company finalized the tapeout for BI back in May 2020 and should have already underwent mass production if Tianshu Zhixin wants to meets its goal of commercializing the chip this year.

The BI solution reportedly provides twice the performance of existing mainstream products on the market, while also offering a very appealing performance-to-cost ratio. BI supports a plethora of floating point formats, including FP32, FP16, BF16, INT32, INT16, and INT8, just to mention a few.

Tianshu Zhixin is keeping a tight lip on the BI's performance, but the company has teased FP16 performance up to 147 TFLOPS. For comparison, the Nvidia A100 and AMD Instinct MI100 deliver FP16 performance figures up to 77.97 TFLOPS and 184.6 TFLOPS, respectively. Note however that Nvidia's A100 also has Tensor cores that can do 312 FP16 TFLOPS (624 TFLOPS with sparsity).

Hendrik_2000 said:

via Mr unknown

Please, Log in or Register to view URLs content!

China Debuts First 7nm Data Center GPU To Rival Nvidia, AMD
By Zhiye Liu
First Published 18 hours ago

Ready for prime time

Shanghai Tianshu Zhixin Semiconductor Co., Ltd., commonly known as Tianshu Zhixin, announced that its Big Island (BI) GPGPU has come to life. The BI is touted as China's first domestic GPGPU that's tailored towards AI and HPC applications and other industries, such as education, medicine, and security.

The BI packs 24 billion transistors, and it's based on a home-made GPU architecture. The chip is built with the cutting-edge 7nm process node and 2.5D CoWoS (chip-on-wafer-on-substrate) packaging. Tianshu Zhixin doesn't explicitly reveal the foundry that's responsible for producing the BI. However, the description of the node coincides with one of TSMC's manufacturing processes.

Tianshu Zhixin commmenced development on its BI chip in 2018. The company finalized the tapeout for BI back in May 2020 and should have already underwent mass production if Tianshu Zhixin wants to meets its goal of commercializing the chip this year.

The BI solution reportedly provides twice the performance of existing mainstream products on the market, while also offering a very appealing performance-to-cost ratio. BI supports a plethora of floating point formats, including FP32, FP16, BF16, INT32, INT16, and INT8, just to mention a few.

Tianshu Zhixin is keeping a tight lip on the BI's performance, but the company has teased FP16 performance up to 147 TFLOPS. For comparison, the Nvidia A100 and AMD Instinct MI100 deliver FP16 performance figures up to 77.97 TFLOPS and 184.6 TFLOPS, respectively. Note however that Nvidia's A100 also has Tensor cores that can do 312 FP16 TFLOPS (624 TFLOPS with sparsity).

Click to expand...

@Hendrik_2000

Huawei also has a 7nm GPU, but without TSMC FAB it was place on a backburner. Hope SMIC 7nm N+2 come on line this year, they have a lot of customer lining up to use its node.