Latest Post

Crafting Magic: Exciting Materials for Moms to Create Memorable Moments with their Children

Failed to mass-produce DRAM, and only NAND released prototypes.
Most of China’s semiconductor machinery is overpacked.
Future declines in market share and sales should be prepared.


Global electronics company A has won the victory of a new system semiconductor factory in Zhuhai, China. It was entrusted to a Korean consulting firm from equipment selection and line configuration of the plant to the operation of the manufacturing plant. The CEO is Choi Jin-suk, who has been recognized for a long time at Samsung Electronics and SK Hynix. Choi, who is staying in Shenzhen to select equipment suppliers, told an interview with the Korea Economic Daily on October 15 to shed light on the problems of the Korean semiconductor industry and move forward. Move his words as they are. Some of his arguments may be different from ours.

I think it is one of the most accurate people in the Korean semiconductor industry. By the end of last year, he worked in Taiwan for three years and six months and closely watched the progress of Chinese semiconductors. Taiwanese, who participate in the Chinese semiconductor business, as well as Chinese company officials come to seek advice. As such, the reality of China’s ‘semiconductor rolling’ is understood in detail.

China’s semiconductor operations reported in Korea are either exaggerated or different. In conclusion, there is no China in memory semiconductor area. Despite various rumors, there is no Chinese company producing memory semiconductors. Samsung will remain the number 1 memory leader for at least 30 years.

China established two DRAM makers and one NAND maker in 2016 to manufacture memory semiconductors. Hefei Changxin, Fujian Jinda DRAM and Tsinghua Uni are NAND makers.

DRAM makers copied Micron’s production line in Taiwan. Hefei Changshin moved to Micron’s 23nm factory ‘100S’ and Fujian Jinhua’s 28nm ’90S’ process to produce DRAM. To this end, Hefei Changxin recruited up to 400 Taiwanese engineers. Fujian has teamed up with UMC, a Taiwan-based foundry company that recruits micron process engineers.

Micron invested 3 trillion to 4 trillion won to develop new process lines such as 100S and 90S, respectively. Of course, it’s hard to see Chinese companies copying the process without paying a royalty without a license agreement. Micron, who only grasps the current situation, took action last year when Fujian Jinhua announced that it had completed the process development.

Based on the accumulated data, he filed a patent infringement lawsuit with the US court and based on this, he requested the US Department of Commerce to prevent US semiconductor equipment companies from selling the equipment to Fujian Jinhua. Fujian Jinhua, which is still in the equipment setting stage, could not be produced without the addition of US equipment such as Applied Materials and Ram Research. As the Commerce Department accepted Micron’s request, preparations for Fujian Jin’s DRAM production were all-stop. In the future, DRAM production is unlikely.

Electronic integrated circuit chip

The situation in Hefei is not much different. The construction of the process line, which has worked hard for the past three years, is difficult to avoid the patent issue with Micron. As a result, it has declared that it will develop its own process line. The recruited Taiwanese technicians have expertise in setting up and managing existing lines and have no experience developing new processes.

Hefei Changxin will introduce a large number of equipment for new process development in August this year. Process development has not yet taken place, but purchasing equipment now is relatively cheap. Even if process development goes well, DRAM production will be possible at the end of 2021, two years later. However, Hefei Changxin’s own capacity is unlikely to develop new processes.

What will be the future of Fujian Jinhua and Hefei Changxin, which have already invested enormous amounts of money? Fujian is expected to be acquired by Micron. Because it is a line made using Micron’s process technology, there is no one who can operate the plant to avoid legal issues unless it is Micron. As China’s inevitable choice is on the defensive, Micron is very likely to take over a new DRAM line at a low price. Micron can easily increase DRAM production, which is bad news for Samsung Electronics and SK Hynix. It shows how bad news and good news can be divided among competitors depending on how the Chinese government uses semiconductor drives.

Hefei Changshin is hopeless for now. From an expert point of view, the development and production of proprietary process technologies is impossible. There is a good chance that they will have to choose to collaborate with a Korean or US DRAM company or hand over a plant within a few years. Where to take the plant will also affect the market share fluctuations between Samsung Electronics, SK Hynix and Micron.

NAND is relatively better than DRAM. In 2017, Tsinghua Uni completed development of 32-layer 3D NAND. Compared to DRAM, NAND’s production difficulty was significantly lower. If the level of technology required for DRAM production is 100, NAND will be available at around 30.

But Tsinghua Uni hasn’t produced NAND yet. With the global NAND market shifting to 64D or more 3D NAND, there is no demand source for 32 NAND. Accordingly, Tsinghua Uni is also hastening to develop and produce 64D 3D NAND.

After finishing the development of the process at the end of last year, we introduced the equipment in May to produce 20,000 sheets of 64-layer 3D NAND on a wafer basis. But equipment setup and production will take at least a year. At the same time, there is a lot of doubt about the reliability of the developed process.

After all, until the first half of next year, there are no Chinese memory semiconductors on the market, whether DRAM or NAND. Some reports that China’s semiconductor industry has made the memory semiconductor market difficult or that Korean memory companies are in trouble are far too far away.

Given the characteristics of the memory semiconductor industry, which takes five to 10 years to develop processes and products, install factory production lines, and accumulate know-how for profitable operation, we are confident that Samsung Electronics will continue to dominate for at least 30 years. SK Hynix will remain the No. 2 memory player for at least 15 years.

Japanese companies, which beat the United States in the early 1980s and occupied the throne of the global semiconductor industry, began to fall behind Samsung Electronics in the mid-1990s. The first Mercury period is only about 15 years. By comparison, South Korea will dominate the memory semiconductor industry for quite some time.

However, this does not mean that Korean memory companies are only walking along the ‘flower road’. China’s memory semiconductor production itself cannot be prevented, and if Chinese semiconductors enter the market, a drop in semiconductor prices and a decrease in Korea’s market share are inevitable. For the Korean semiconductor industry as a whole, it is time to prepare for this.

Choi has rarely heard his name among those who have worked in the semiconductor industry for more than a decade. Samsung joined the company in 1984 when Samsung Electronics just started the memory business and moved to SK Hynix in 2001, where he served as CTO (Chief Technology Officer) and left the company in 2010. Rarely, he is involved in the entire process from semiconductor development to manufacturing and processing. It received three times the Samsung Technology Grand Prize, which is referred to as ‘executive promotion check’ once received. It is the result of recognition for the development of 12-inch wafer processing technology, development of 256 mega and 16 mega DRAM. At SK Hynix, the company raised yields and increased production with existing production facilities without funding from creditors. In 2006, the company recorded the lowest manufacturing cost and the highest production volume in the memory semiconductor industry. This led to Samsung’s chairman Lee Kun-hee in 2007, saying, “How can Samsun

g fall behind Hynix?”
Since then, he has worked as CEO at STX Solar, Hanwha Q CELLS, and other solar industries with similar manufacturing processes. In 2015, he founded a production and factory consulting firm and consulted on improving productivity at Taiwan and Japan plants in Micron, USA. Mr. Choi has attracted the attention of the global semiconductor industry by improving productivity by 20% without additional facility investment. It is because of these results that A has entrusted the factory.














Dラムに比べるとNAND型は、比較的事情がよい。 2017年清華ユニは32段3D NAND型の開発を完了したからである。 Dラムに比べてNAND型の生産難易度が大きく低下可能だった。 Dラム生産のために必要な技術レベルが100であれば、NAND型は、30位に可能である。

Electronic integrated circuit chip

しかし、清華ユニはまだNAND型を生産していない。世界のNAND市場が64段以上の3D NAND型中心に変わった状況で32段NAND型を必要する需要先自体が残っていないからだ。これにより、清華ユニも64段3D NAND型の開発と生産を急いでいる。

昨年末の関連プロセスの開発を終えたのに続き、64段3D NAND型をウエハ基準に月2万枚生産することができる装置を、5月にかけてきた。しかし、機器のセットアップと生産まで、少なくとも1年かかる見込みだ。それさえも開発したプロセスの信頼性に疑問がかかる部分が多く、収益を出すことができるかは未知数だ。


プロセスと製品開発、工場の生産ライン設置、収益性の高い操作のためのノウハウの蓄積など、5〜10年かかるメモリ半導体産業の特性を考えると、今後、少なくとも30年は、サムスン電子の独走が続くものと確信している。 SKハイニックスも、少なくとも15年以上のメモリ2位の座を維持する。

1980年代初め、米国を破って世界の半導体業界の王座を占めた日本企業が、1990年代半ばサムスン電子に押さ始めた。 1位水性期間は15年程度に過ぎないものである。ここに比べれば、韓国はかなり長い期間メモリー半導体業界を号令することになる。


◆チェジンソク社長は、半導体業界で10年以上働いたこれらの間に彼の名前を聞いたことのない人はほとんどいない。サムスン電子が膜メモリー事業を開始した1984年の研究で入社し、2001年にSKハイニックスに席を移しCTO(最高技術責任者)などを歴任し、2010年に会社を出た。業界でまれに半導体開発から製造・工程までの全過程を渉猟している人物である。一度だけ受けても「役員昇進保証小切手」と話しているサムスンの技術対象を三回受賞した。 12インチウエハー加工技術の開発、256メガ、16メガDラム開発の功労を認められた結果だ。 SKハイニックスは債権団の資金支援がない中で、既存の生産設備での歩留まりを上げ、生産量を増やした。 2006年のメモリ半導体業界最低製造コスト、最高の生産量の拡大などの記録を出した。これはイ・ゴンヒサムスン会長が2007年の「どのようにサムスン電子がハイニックスにおくれることがあるのか​​」と、半導体の幹部を強く非難することでつながった。
以降、半導体や製造工程が似たような太陽電池業界のSTXソーラー、ハンファキューセルなどCEOとして働いた。 2015年の生産や工場コンサルティング会社を建て、アメリカミクロンの台湾と日本の工場の生産性向上をコンサルティングした。ここで、崔社長は追加の設備投資なしに20%の生産性向上を成し遂げ、世界の半導体業界の注目を浴びた。 A社の推進する工場を任せたのは、このような実績からである。

Leave a Reply