Opinionated - China Chipping Away to Semiconductor Dominance

[Shotgunner51]

@Shotgunner51 How long before TSMC's market share starts decreasing and when Chinese SMIC will start dominating? What will be the major goal of this particular industry in Xi Jinping's Made in China 2025 strategy?

Mainland-funded firms are represented by Tsinghua Unigroup and SMIC, well despite making fast progress I don't see either one displacing TSMC's global dominance anytime soon (TSMC is well vested/positioned/aligned across the supply chain). Either way it's not a zero sum game, the general goal (of "Made in China 2025") is to build biggest and best semiconductor industry on Mainland soil, TSMC is not considered foreign and in fact they do have increasing stake in the big picture.

I believe Tsinghua Unigroup and SMIC will continue to do their best to gain better positions, gradually. Among all elements talents are central to hi-tech, so naturally both are actively recruiting, as we see hi-profile news like recently Shih-Wei Sun (ex-CEO of UMC, Taiwan's second largest foundry, world's third) joined Tsinghua Unigroup, Chiang Shang-Yi (ex-COO of TSMC) joined SMIC. Let's not forget, SMIC was co-founded with TSMC veterans in the very first place.

Mainland funds are not sitting idle, they are actively looking for M&A opportunities. National Integrated Circuit Industry Investment Fund (NICIIF) has been eyeing on the world's second largest foundry US-based Globalfoundries. Abu Dhabi's Advanced Technology Investment (ATIC) is said to have a willingness to release its holdings, however since Globalfoundries' 14nm process is licensed by Samsung Electronics, it is believed that Samsung is unlikely to agree to the acquisition bid from any potential competitor. Moreover, US government may still prefer ownership by Emerati not Chinese. Anyway let's see how this unfold.

The name "Made in China 2025" is self-explanatory, it's gonna take a long process to achieve.

https://defence.pk/threads/semiconductor-three-way-race-taiwan-us-and-china.412251/
https://defence.pk/threads/former-tsmc-executive-to-join-chinese-rival.468162/#post-9035301
https://defence.pk/threads/tsinghua...ith-investment-totaling-us-70-billion.472281/
https://defence.pk/threads/china-se...nd-reportedly-targets-globalfoundries.395023/

 

[52051]

@Shotgunner51 How long before TSMC's market share starts decreasing and when Chinese SMIC will start dominating? What will be the major goal of this particular industry in Xi Jinping's Made in China 2025 strategy?

China is far behind TSMC in this regard, their 14nm is expected to be ready around 2020 or later.

This is partly because the Wassenaar Arrangement on semiconductor tech on China.

The key device to produce a semiconductor is called mask aligner, and the most advanced mask aligner in the world is developed by a Holland company called ASML.

Intel/TSMC hold stocks of this company, actually all top semiconductor players use ASML's mask aligner exclusively nowadays.

But due to sanction China can only buy outdated mask aligners from ASML (acutally 2 generation behind).

That's why China spend lots money on developing their own mask aligners, but it take significant amount of time and money to catch up the world leading ones there, although there are breakthroughs here and there in semiconductor manufacturing process research in China but to develop a fully functional mask aligners that is competitive to the world best is not a trivial feat, and will take quite some time and money.

 

[Huan]

China is far behind TSMC in this regard, their 14nm is expected to be ready around 2020 or later.

This is partly because the Wassenaar Arrangement on semiconduct tech on China.

The key device to produce a semiconduct is called mask aligner, and the most advanced mask aligner in the world is developed by a Holland company called ASML.

Intel/TSMC hold stocks of this company, actually all top semiconduct players use ASML's mask aligner exclusively nowadays.

But due to saction China can only buy outdated mask aligners from ASML (acutally 2 generation behind).

Thats why China spend lots money on developing their own mask aligners, but it take significant amount of time and money to catch up the world leading ones there, althrough there are breakingthroughs here and there in semiconduct manufacturing process research in China but to develop a fully functional mask aligners that being competitive to the world best is not a trivial feat, and will take quite some time and money.

That is sad to hear. It sounds like another delay tactic to put a brake on China's rise in technology in this Cold War-like mindset.

 

[Shotgunner51]

That is sad to hear. It sounds like another delay tactic to put a brake on China's rise in technology in this Cold War-like mindset.

It's fine, already a decades old norm, it isn't even necessarily a bad thing. While industries in most nations in general can tap into a globalized market of tech/patents, tools, components, materials even talents, some Chinese industries do operate almost like antarkies in parallel universe, especially those can relate to defence however slightly. There are both pros and cons, say it elevates demand for internal R&D, demand for complete internal supply chain (breeds a wide range of industries), avoid potential trade deficits and current account losses.

 

[ahojunk]

It's a status quo norm. While industries in most nations in general can tap into a globalized market of tech/patents, tools, components, materials even talents, some Chinese industries do operate almost like antarkies in parallel universe, especially those can relate to defence however slightly. It isn't necessarily a bad thing, there are both pros and cons, say it elevates demand for internal R&D, demand for internal supply chain (breeds a wide range of industries), avoid potential trade deficits and current account losses.

No worries, just require changing the problem into opportunities.
IMO, in the long term, this is better way to go.
I am not sure about other countries, but China has scale, talent and money to do it.
Short term pain = long term gain.

 

[AndrewJin]

No worries, just require changing the problem into opportunities.
IMO, in the long term, this is better way to go.
I am not sure about other countries, but China has scale, talent and money to do it.
Short term pain = long term gain.

It's fine, already a decades old norm, it isn't even necessarily a bad thing. While industries in most nations in general can tap into a globalized market of tech/patents, tools, components, materials even talents, some Chinese industries do operate almost like antarkies in parallel universe, especially those can relate to defence however slightly. There are both pros and cons, say it elevates demand for internal R&D, demand for complete internal supply chain (breeds a wide range of industries), avoid potential trade deficits and current account losses.

U guys are absolutely right.
Long-term pros outweigh short-term cons significantly.

 

[Bussard Ramjet]

TSMC won 59% of Global Market Share 2016, confirms 12nm for 2017, works on 7nm for 2018

View attachment 369448

TSMC remains as world's #1 largest IC foundry, claiming 59% of global market share in 2016 as per IC Insights:

• Among the 10 largest IC foundries, 4 from Taiwan, 2 from Mainland, 1 US-based (Abu Dhabi owned), 1 from South Korea, 1 from Israel, 1 from Germany.
  
• The 4 Taiwan foundries combined market share 72.28%
• The 2 Mainland foundries combined market share 7.27%

View attachment 369441

http://www.icinsights.com/news/bull...Market-Surges-11-In-2016-To-Reach-50-Billion/
http://www.icinsights.com/data/articles/documents/945.pdf​

______________________________________________________________________________________

Taiwan Semiconductor Mfg. Co. Ltd. Confirms “12nm” Chip Technology Plans
Jan 18, 2017 at 7:40AM

As the competition for more mature chip manufacturing technologies heats up, TSMC isn't standing still.

View attachment 369442

A while back, DigiTimes reported that Taiwan Semiconductor Manufacturing Company (NYSE:TSM), a major contract chip manufacturer, planned to introduce an enhanced variant of its 16nm manufacturing technology dubbed "12nm."

On the chipmaker's most recent earnings call, an analyst asked management about this potential new technology. The company seemingly confirmed its existence, though it's not clear if the technology will, in fact, be marketed as 12nm.

Let's look at just what management had to say about the tech, and why it matters to TSMC investors.

A refinement of 16nm tech

On the call, TSMC Co-CEO C.C. Wei told analysts that its strategy is "continuously to improve every node in the performance, such as 28-nanometer." He went on to say that TSMC is "continuing to [improve] the 16-nanometers technology." Wei explained that its next revision of the 16nm technology may be worth calling 12nm because it will deliver improved "density, classical density, performance, and power consumption," according to a transcript by Seeking Alpha.

TSMC has, to date, announced several 16nm variants. The first was vanilla 16nm, which didn't seem to gain much traction as a performance-enhanced version of the technology quickly replaced it, branded 16nm FinFET Plus.

After introducing 16nm FinFFET Plus, TSMC rolled out yet another version of the technology, called 16FFC (with the 'C' standing for "compact") that allowed chipmakers to build smaller, more cost-effective chips.

The upcoming "12nm" technology looks like TSMC taking an additional step in its efforts to try to maintain technology leadership against competing 14/16-nanometer technologies, particularly as competition in those technologies heats up in the coming years.

Why it matters to TSMC investors

There are few chip manufacturing companies that can bring leading-edge technologies to market. However, over time, the weaker chip manufacturers bring out products that can compete with those the stronger companies debuted several years earlier.

Since contract chip manufacturers tend to generate significant revenues from older-generation manufacturing technologies (TSMC's 28-nanometer technology, first introduced in late 2011, accounted for 26% of the company's revenue in 2016), it is important for TSMC to remain cost/performance/feature competitive with these older technologies.

United Microelectronics (NYSE:UMC), for example, has said that it expects to begin "commercial production" of a 14-nanometer technology "by the second half of 2017," per EETimes.

China-based Semiconductor Manufacturing International (NYSE:SMI), too, is planning to introduce a "14nm" technology at some point.

By continuing to enhance its 16nm technology, TSMC should be poised to defend its market share position against upcoming competitors while at the same time keeping its cost structure competitive and its average revenue per wafer as high as possible.

All that should ultimately translate into robust revenues and profitability on this technology.

Looking out to 2017, TSMC management appears confident that it will be able to maintain its strong market share position in the contract chip manufacturing space. "I will say that we certainly do not think we will lose market share," Chang told analysts. "We're not going to grow less than foundry," he said, referring to the contract chip manufacturing, or semiconductor foundry, market.

http://www.fool.com/investing/2017/01/18/taiwan-semiconductor-mfg-co-ltd-confirms-12nm-tech.aspx

______________________________________________________________________________________

TSMC Already Working On 7nm Chips For 2018
Filip January 6, 2017 20:14 CST

View attachment 369443

10nm chipset from both TSMC and Samsung are about to hit the market. Samsung is producing Qualcomm's Snapdragon 835 chipset, while TSMC is said to be the sole supplier of the Apple A10 Fusion and A11 chips alongside MediaTek's Helio x30. To get ready for 2018, TSMC is working on the first 7nm chips.

TSMC will tackle 7nm manufacturing process in the second half of the year, and the process is expected to enter full production in early 2018. The company is currently working on the Tape Out phase, the last step in the design process for the prototype chip before commencing mass production. More information about the 7nm process will be revealed on January 15.

http://www.nextpowerup.com/news/32899/tsmc-already-working-on-7nm-chips-for-2018/

The article and the stats you pointed out are only for "Pure Play" Foundries.

Intel produces almost all of its chips in house as well, and were its production to be counted, it would easily be in top 3.

 

[TaiShang]

Either way it's not a zero sum game, the general goal (of "Made in China 2025") is to build biggest and best semiconductor industry on Mainland soil, TSMC is not considered foreign and in fact they do have increasing stake in the big picture.

Among all elements talents are central to hi-tech, so naturally both are actively recruiting, as we see hi-profile news like recently Shih-Wei Sun (ex-CEO of UMC, Taiwan's second largest foundry, world's third) joined Tsinghua Unigroup, Chiang Shang-Yi (ex-COO of TSMC) joined SMIC. Let's not forget, SMIC was co-founded with TSMC veterans in the very first place.

Together, stronger.

Hsinchu Science Park pools immense talent to be recruited back and forth between mainland and Taiwan.

 

[Shotgunner51]

The article and the stats you pointed out are only for "Pure Play" Foundries.

Intel produces almost all of its chips in house as well, and were its production to be counted, it would easily be in top 3.

If you say like Samsung then I don't know yet, their foundry is also sizable, but Intel? You can forget that. Even their foundry chief Sunit Rikhi disclosed on May 2015 "the business was on a path to ramping volume to over a billion dollars of revenue run rate", now even if that "run rate" was fully achieved in 2016, $1B does not get anywhere close to Taiwan's UMC or Emerati-owned Globalfoundries, let alone TSMC which was $30B. Looking forward, Intel does not possess any capabilities to challenge TSMC's technological/overall advantages, not even close. Read this summary from Barron's (September 27 2016, that's even before TSMC releases further details on next gen 7nm process info):

Intel a Year Behind TSM in Foundry Capabilities
Citigroup
"We conclude that it is premature for Intel to become a meaningful player to challenge TSMC in the foundry business. We believe TSMC’s foundry leading position on technology, ARM process capability, foundry capacity, cost structure, production flexibility, balance sheet and valuations are well secured against Intel. While Intel is better in microprocessor technology and manufacturing, its foundry manufacturing lags its microprocessor manufacturing by at least two years, putting it a year behind TSMC. We do not expect Intel to become an imminent threat to TSMC in the foundry business anytime soon. TSMC: Maintains foundry technology leadership throughout 2018E – While TSMC’s 10nm and 7nm gate scaling density is behind Intel’s 10nm and 7nm on an apples-to-apples comparison, TSMC is closing the gap by migrating to the finer geometries 1-2 years ahead of Intel’s foundry from 10nm. With its proprietary InFO package technology, we expect TSMC to dominate 10nm and 7nm foundry share in 2017 and 2018, respectively. We expect limited competition to TSMC for 10nm and 7nm in the market."
http://blogs.barrons.com/techtrader...behind-tsm-in-foundry-capabilities-says-citi/​

 

[terranMarine]

TSMC , good job to our Chinese brothers in Taiwan

 

[cirr]

China’s Tsinghua Unigroup to Build $30 Billion Memory-Chip Factory in Nanjing

China's Tsinghua Unigroup to build $30 billion Nanjing chip plant

 

[onebyone]

Tsinghua Unigroup to build US$30b Nanjing chip plant

China's top state chip maker plans to boost local production capacity targeting a monthly capacity of 100,000 wafers in phase one at the new factory

By REUTERS JANUARY 20, 2017 8:55 AM (UTC+8)

Tsinghua Unigroup Ltd, China’s top state chip manufacturer, revealed plans on Thursday to build a US$30 billion memory chip factory as the government seeks to boost local production capacity.

The firm is targeting a monthly capacity of 100,000 wafers in phase one of the development which will cost US$10 billion and is located in Nanjing.

In a statement on its website, Tsinghua Unigroup said the project is part of China’s efforts to build a world-leading chip industry, and it hopes it will create a siphoning effect to attract more development.

y capacity of 100,000 wafers in phase one at the new factory
By REUTERS JANUARY 20, 2017 8:55 AM (UTC+8)

Tsinghua Unigroup Ltd, China’s top state chip manufacturer, revealed plans on Thursday to build a US$30 billion memory chip factory as the government seeks to boost local production capacity.

The firm is targeting a monthly capacity of 100,000 wafers in phase one of the development which will cost US$10 billion and is located in Nanjing.

In a statement on its website, Tsinghua Unigroup said the project is part of China’s efforts to build a world-leading chip industry, and it hopes it will create a siphoning effect to attract more development.

Tsinghua Unigroup announced plans for a separate US$24 billion chip factory based in the Chinese city of Wuhan in March last year.

The development comes as Chinese memory chip firms face increasing regulatory resistance to acquiring overseas technology.

In December US President Barack Obama issued an executive order barring the acquisition of the US business of German semiconductor equipment maker Aixtron by a Chinese-backed chip fund over security concerns.

Last year, Chinese chip makers withdrew a record volume of overseas deals.

In 2015, Tsinghua Unigroup tried unsuccessfully to acquire US chip group Micron Technology.

Tsinghua’s new plant will produce DRAM and 3D-NAND flash chips which are used in a range of devices including smartphones and personal computers.

The firm also announced it would invest roughly another 30 billion yuan (US$4.37 billion) in building an “international city” facility including apartments and international schools for foreign employees.
http://www.atimes.com/article/tsinghua-unigroup-build-us30b-nanjing-chip-plant/

 

[onebyone]

Huaxintong Semiconductor Technology is developing a server chip based on ARM architecture

By Agam Shah


U.S. Correspondent, IDG News Service | JAN 30, 2017 1:36 PM PT

• 
  

Credit: Qualcomm
The number of powerful chips coming out of China keeps growing as a war of words on semiconductors with the U.S. escalates.

A joint venture between Qualcomm and China's Guizhou province, called Huaxintong Semiconductor Technology, has started the development of a new server chip based on ARM technology.

The joint venture is "now busy developing a customized server CPU product based on our technology and designs for the China market," said Derek Aberle, president at Qualcomm, according to a Seeking Alpha transcript of an earnings call last week.

Other companies are also developing custom chips for the Chinese server market.

developing a CPU based on IBM's Power architecture, though the venture has raised security concerns. AMD has also created a joint venture to create Chinese x86 server chips.

Chipmakers are making a run at the Chinese market, which is considered a big opportunity for data center technologies. Like Facebook and Google in the U.S., Chinese companies like Alibaba and Tencent are establishing mega data centers for cloud and machine-learning services.

But the Chinese market has its quirks because companies there prefer to buy hardware from local vendors. It's partly because servers made by Chinese companies are cheaper and potentially come with fewer national security risks.

China's long-term goal is to be self reliant in the hardware market, with a majority of devices in the country running on homegrown components. The country already has the world's fastest supercomputer, TaihuLight.

has accused China of rigging the semiconductor market by giving an unfair advantage to Chinese chip companies.

Though Intel rules the China server chip market, Qualcomm is trying to push its chips, based on ARM architecture.

Qualcomm late last year announced its first 48-core server chip, the Centriq 2400. It is considered the best ARM server chip yet. But Intel has more than a 90 percent server chip market share, while ARM servers are virtually nonexistent and are still being tested. China represents a big opportunity to Qualcomm and ARM architecture to grow in the server market.

The Guizhou province is building up a reputation as a hub for big data, with many cloud server and telecom companies establishing data centers there. Making homegrown chips and servers will boost the region's economy and keep more workers employed.

Huaxintong Semiconductor Technology is a separate company from Qualcomm and developing its own CPU technology, a Qualcomm spokeswoman said in an email.

What the joint venture company is developing is unclear, but the technology could be based on the Falkor CPU core used in Centriq 2400.

Huaxintong Semiconductor Technology could take Qualcomm's technology and customize the CPU for local customers. It could strip out or add I/O and throughput technologies to directly address customer needs, Brookwood said.

The chips could drive cloud installations and target Intel's Xeon E3 and E5 chips.

Alternately, Huaxintong Semiconductor could create an ARM chip for high-performance computing by cramming many ARM CPU cores together, Brookwood said. That could allow the joint venture to create a chip to compete with Intel's Xeon Phi.

The server chip design will also depend on the software a customer uses. For now, the only proven software stack for ARM architecture is LAMP (Linux, Apache, MySQL, PHP) for web serving, but new usage models in areas like deep learning and high-performance computing are emerging every day.

http://www.computerworld.com/articl...he-way-as-country-boosts-home-grown-tech.html

 

[onebyone]

Good Work

 

[艹艹艹]

http://www.spreadtrum.com/en/show_news.html?id=fe766282-e9cc-4ffd-b211-e3d19675d3f7
Spreadtrum launches 14nm 8-core 64-bit mid- and high-end LTE SoC platform
2017-02-27

Spreadtrum delivers an industry leading smartphone SoC platform, manufactured on Intel’s advanced 14nm process, featuring a high performance 2GHz Intel® processor, 5 mode CAT 7 LTE modem, Ultra HD display, and advanced 26 Mpixel imaging technology.

BARCELONA, Spain--Feb. 27, 2017 -- Spreadtrum Communications (“Spreadtrum”), a leading fabless semiconductor provider in China with advanced technology in 2G, 3G and 4G wireless communications standards, launched its 14nm 8-core 64-bit LTE SoC platform, SC9861G-IA, at the 2017 Mobile World Congress ("MWC"). Built on Intel's 14nm foundry platform, SC9861G-IA is targeting the global mid-level and premium smartphone market, and features Intel Airmont architecture with powerful mobile computing performance. SC9861G-IA has a very efficient power-management design and will deliver a delightful experience to users worldwide.

SC9861G-IA platform is a highly integrated LTE chip solution that delivers outstanding power-saving performance. It supports 5-mode (TDD-LTE / FDD-LTE / TD-SCDMA / WCDMA / EGG) full-band LTE Category 7 (Cat 7) communication, as well as Carrier Aggregation and TDD/FDD hybrid networking. The platform enables peak target data transmission rate of 300 Mbps downlink and 100 Mbps uplink, and the superior multimedia configurations by supporting dual cameras up to 26 megapixels with real-time rear/front camera capture/recording, refocusing, image fusion and real 3D shooting.

The platform achieves the ultra-HD 4K2K video recording and playback with HEVC hardware encoding and decoding technology, and supports high-resolution WQXGA (2560 x 1600) displays. It features an integrated Sensor Hub for better user perception and to enable new applications. This platform enhances overall smartphone user experiences.

SC9861G-IA is based on 64-bit 8-core 2.0 GHz Intel Airmont architecture, Imagination PowerVR GT7200 GPU, and developed on Intel's advanced 14nm process technology via its comprehensive foundry services. It uses Intel® Virtualization Technology to support a multi-domain security system architecture and provide security for smart devices.

"The successful launch of SC9861G-IA, Spreadtrum's first high-end LTE chip platform based on an Intel architecture and designed on Intel Custom Foundry's 14nm technology platform, demonstrates Spreadtrum's prowess in technology and the breadth of its technology portfolio," said Dr. Leo Li, Chairman and CEO of Spreadtrum. "The platform's efficient computing performance and leading technology give our customers more possibilities when it comes to end products. Looking ahead, Spreadtrum plans to continue innovating and creating more high-end and differentiated products and services."

"Intel and Spreadtrum are bringing together our core technology strengths and assets to deliver competitive mobile SoCs to the marketplace, addressing the demanding requirements of the latest devices. This launch is a significant milestone in our partnership with Spreadtrum. We look forward to working with Spreadtrum on additional mobile platforms,”said Brian Krzanich, CEO of Intel.

Spreadtrum SC9861G-IA is scheduled to go into mass production in the second quarter of 2017.

About Spreadtrum Communications
As an affiliate of Tsinghua Unigroup, Ltd, Spreadtrum Communications is a fabless semiconductor company that develops mobile chipset platforms for smartphones, feature phones and other consumer electronics products, supporting 2G, 3G and 4G wireless communications standards. Spreadtrum's solutions combine its highly integrated, power-efficient chipsets with customizable software and reference designs in a complete turnkey platform, enabling customers to achieve faster design cycles with a lower development cost. Spreadtrum's customers include global and China-based manufacturers developing mobile products for consumers in China and emerging markets around the world. For more information, visit www.spreadtrum.com.

more read：
Spreadtrum LTE Chipset is Adopted for Samsung GALAXY Tab A 7.0’’
http://www.spreadtrum.com/en/show_news.html?id=f326e80f-40ef-4c5c-9179-334f29972c04