Taipei, Thursday, Apr 25, 2017, 01:06

News

Foreign Fabs in China to Compel Country’s Domestic Foundries to Expedite Their 28nm Production

Published: Apr 10,2017

According to the global market intelligence firm TrendForce, Semiconductor Manufacturing International Corporation (SMIC) and Shanghai Huali Microelectronics Corporation (HMLC) are the two leaders among Chinese foundries in the development of the 28nm processing. The existence of foreign-owned fabs in China will pressure these two domestic manufacturers to further accelerate their timetables for this technology.

More on This

IR Sensor Modules for Mobile to Reach US$145M in Scale This Year, Says TrendForce

Among the possible new technologies to be incorporated in this year’s refresh of the iPhone series, 3D sensing has generated a lot of interests in terms of applications...

TrendForce Anticipates Global Server DRAM ASP in 2Q17 to Rise Over 10%

DRAMeXchange, a division of TrendForce, reports that demand continues to outpace supply in the server DRAM market. In this first quarter...

TrendForce’s research finds that the most advanced node currently available for China’s foundry industry is the 28nm. SMIC’s wafer capacity for 28nm devices represented less than 1% of the global total for 2016. The world’s largest foundry TSMC accounted for 66.7% of the global 28nm capacity for the same year, while the second-largest GlobalFoundries had 16.1% and the third-place UMC had 8.4%.

Data from SMIC’s quarterly earnings report reveals that the share of 28nm devices in the company’s total revenue in the fourth quarter grew to 3.5%. At a road show held on February 15, SMIC’s CEO Dr. Tzu-Yin Chiu stated that the 28nm node will be a major revenue growth generator for 2017.

Chiu expects that 7~9% of the company’s total revenue by the end of this year will come from products made with 28nm or more advanced technologies. SMIC’s 2016 annual results announcement also projects that 28nm products will make up nearly 10% of the total revenue at the conclusion of 2017.

HLMC is the other major domestic foundry that is working on the 28nm processing. Though the company announced that its collaboration with MediaTek has led to successful tape-out of 28nm mobile chips, there is still some progress to be made before mass production can be realized.

In terms of capacity planning, HLMC started the construction of its second fab at the end of 2016. This facility, which will be initially responsible for 28nm production, is scheduled for completion and trial runs in 2018. HLMC is also considering to have its second fab transition to more advanced nodes (the 20nm and the 14nm) and adopt FD-SOI processing in the future.

TrendForce notes that currently China’s IC industry is in its high-growth period. With support from various levels of government and local funds, domestic IC manufacturers are making huge investments to expand the scales of their operations. The take-off growth of China’s IC market is also drawing foreign companies and capital. Major international semiconductor enterprises want to establish themselves in the country during this critical time so that they will be in a more competitive position later on.

From the aspect of technology transfer, foreign semiconductor companies that have set up shop in China generally have licensed technologies that are at least a generation behind from what they currently have in their headquartering countries. This practice puts domestic manufacturers at a disadvantage because they are competing in the home market with products made from similar technologies.

Within the foundry sector, international companies so far have undertaken several fab projects in China to compete against the production capabilities developed by domestic manufacturers. TSMC’s Nanjing fab, which will begin to operate in 2018, is assigned to produce devices on a 16nm FinFET process. This will put pressure on SMIC’s timetable for its 14nm FinFET process that is expected to be ready for mass production at the start of 2019.

In Chengdu, GlobalFoundries is also expediting a fab project that will be completed in 2019 and feature 22nm FD-SOI manufacturing. This plant can be seen as attempt to overtake HLMC’s development schedule for the same technology.

More recently, UMC announced that it will license its 28nm technology to its China-based subsidiary United Semiconductor Xiamen (USCXM). This move followed UMC’s success in mass production for its 14nm FinFET technology. USCXM was at the early stage of adopting 50/40nm technology when it began operation at the end of last year.

The subsidiary foundry will now undergo a dramatic upgrade and at the earliest will begin making devices on a 28nm Poly/SiON process this second quarter. At end of 2017, USCXM monthly wafer capacity is estimated to reach around 16,000 pieces, of which 10,000 pieces are expected to come from its 28nm process. This would make USXCM the leading 28nm producer that is located in China.

SMIC, which is also currently producing on a 28nm technology, will be impacted by UMC’s plan. SMIC has been making low-end and mid-range IC products using the 28nm Poly/SiON process. The company is inexperienced with HKMG, which is the more advanced manufacturing solution.

Additionally, UMC is ahead of SMIC by two years in the mass production on the 28nm technology. USCXM will thus gain access to a more mature 28nm technology that can offer better and stable yields. The licensing deal will ultimately affect SMIC’s plan for its 28nm manufacturing.

In sum, TrendForce believes China’s foundry industry will face a very heated and direct competition for technical personnel, resources and market shares once foreign-own fabs located within the country start operating. Domestic IC manufacturers will now have to really fast track their technology roadmaps so that they are prepared for the coming struggle.

CTIMES loves to interact with the global technology related companies and individuals, you can deliver your products information or share industrial intelligence. Please email us to en@ctimes.com.tw

697 viewed

comments powered by Disqus