Recently, Samsung wants to build non-American device chip manufacturing plant news was once again hotly hyped by the media. In fact, the news was reported as early as the end of May, when Taiwan’s Economic Daily reported that Huawei was trying to persuade Samsung and TSMC to build an advanced process production line for non-U.S. devices, and even said samsung already had a 7nm production line using non-U.S. equipment for Huawei’s Hess.
Does Samsung really have the strength to build a chip manufacturing plant for non-U.S. devices? Behind the reflection of samsung chip industry how real strength?
Non-U.S. devices are hopeless, Samsung has no sense of presence in the field
Since it is chip manufacturing, it is the relevant chip manufacturing equipment that naturally determines the role. As we all know, IC (including chips) manufacturing usually requires lithography machines, etching machines, thin film equipment, diffusion ion injection equipment, wet equipment, process detection and other six categories of equipment.
According to VLSI Research, U.S. companies have a large monopoly on the market in PVD, CMP equipment, testing equipment and ion injection equipment. From a revenue perspective, the three largest U.S. semiconductor equipment providers, AMAT,Lam Research, Lam Research, and KLA, accounted for an absolute advantage of $13.468 billion, $9.549 billion and $4.665 billion in 2019, respectively, accounting for 35.9 percent of global total revenue and 43 percent of the top 15.
In terms of equipment market share, AMAT products cover the entire industrial chain, and many products are among the world’s leading market participants, including PVD deposition equipment (84.9%), CMP equipment (66%), and ion injection equipment (73%) in the leading sector. In addition, Panlin’s etching machine market share reached 52.7%, the market share of Theo lei’s testing equipment reached 70%.
What makes the outside world feel weak erasing is that, in addition to the three major equipment manufacturers, the U.S. semiconductor equipment industry has a large number of parts suppliers, even the industry’s well-known as asthet ASML photolithography machine, even can not be separated from the United States semiconductor componentsuppliers suppliers.
According to data compiled by Bloomberg, ASML’s lithography machines also use U.S. technology, and from ASML’s entire upstream supply chain, nine of ASML’s 17 core suppliers are from the U.S., including Brooks Automation, Sparton, II-VI Blocked, Lumentum Holdings, and ASML Litugraphy, which has been acquired by ASML. Metering technology equipment is manufactured by Hewlett-Packard in the United States. Because most components of ASML lithography opt for outsourcing, they are heavily dependent on suppliers, particularly in the United States.
So the question is, can Samsung build a semiconductor manufacturing line for non-U.S. equipment?
To that end, EE Times ranked leading equipment suppliers in 11 stages of the chip manufacturing process and concluded that while it was theoretically possible to build a semiconductor production line without U.S. equipment, equipment in Japan, Europe and even China was not a leader in many areas.
At this point we just want to say that being able to build, use and use is a completely different concept.
It should be noted that in the top 15 rankings, there is not even a South Korean chip manufacturing equipment enterprises, and according to statistics, in 2018, South Korea SEMES has entered the list, ranked top12, but due to insufficient investment, last year completely with the top 15 no chance. In fact, even in the top 15, in the field of chip manufacturing equipment, South Korea should not say that the United States, that is, compared with Japan and Europe companies are far from.
According to THE Korea International Trade Association (KITA), the top three countries in South Korea for imported chip manufacturing equipment in 2019 are Japan, the United States and the Netherlands. This means that South Korea’s minimum gaming power with the US in chip-making equipment is not, or will even be clamped down by Japan in the first place (the trade war between Japan and South Korea, which began last year, in the chip field).
Japan-South Korea trade war, Samsung semiconductor materials exposed to deadly short plate
In addition to chip manufacturing equipment, the importance of semiconductor materials to the chip industry is self-evident. If Samsung works in this area, it will, to some extent, give it the leverage to play a game against the Us when it builds a non-American device chip manufacturing plant. Unfortunately, in this key area, last year’s Japan-South Korea trade war exposed Samsung to another deadly short-cut in the chip industry chain.
Through this trade war, the industry knows that Japan is the world’s largest producer of semiconductor materials, and in semiconductor materials in the long-term to maintain an absolute advantage, the production of semiconductor chips requires about 19 kinds of necessary materials, indispensable, and most materials have very high technical barriers. Among them, Japanese enterprises in silicon wafers, synthetic semiconductor wafers, light masks, photoresist, target materials, protective coating film, lead holder, ceramic plate, plastic plate, TAB (strip auto-joint), COF (film compounding), welding wire, packaging materials and other 14 important materials accounted for more than 50% of the share.
For example, in the silicon chip sector, where the highest cost of materials is higher (more than 30%), Japan’s Shinyu Chemical leads the way, with the largest market share, followed by SumCO of Japan (Mitsubishi Sumitomo), Taiwan Global Wafer, Germany’s Siltronic, and South Korea’sSK Hynix. In 2018, the top four silicon chip suppliers accounted for 94% of the global market share, including Japan’s Shin-Yue Chemical accounted for 28% and Japan’s Mitsubishi Sumitomo 25%.
In the field of photoresist, Japan’s JSR, Tokyo Chemical Industry, Sumitomo Chemical, Dow, Fuji Electronics and other enterprises monopoly;
By contrast, Samsung and even the entire Korean semiconductor industry chain have made little difference in semiconductor materials and have led to a high reliance on Japanese semiconductor materials.
Last year, for example, during last year’s japan-South Korea trade war, high-purity hydrogen fluoride, which was listed by Japan for export restrictions, was mainly used for wet etching of semiconductors. Although almost 99.99% purity of hydrogen fluoride produced by major plants around the world, only japanese manufacturers can achieve a purity of 99.999 percent.
At a time when the 7nm process is beginning to mass produce and semiconductors are becoming more sophisticated, wet etching is due to the same characteristics in all its ways, which makes the chip manufacturers’ requirements for the purity of hydrogen fluoride rise. At present, high-purity hydrogen fluoride imports from Japan account for 43.9% of South Korea’s total imports.
Also such as the restricted photoresist, with today’s various chips have entered the micro-nano-level, photoresist corresponding wavelength is correspondingly decreased. At present, the more mainstream photoresist wavelengths are g-line (436nm), i-line (365nm), KrF (248nm), ArF (193nm) and the latest EUV (13.5nm). The most commercially available KrF and ArF are more japanese, with more than 91.9% of South Korea’s photoresist imported from Japan, according to the Korea International Trade Association.
Although Samsung has since actively sought alternatives and announced a huge investment plan to invest 7.8 trillion won ($6.5 billion, or 44.9 billion yuan) over the next seven years to develop home-grown semiconductor materials and equipment to reduce its dependence on Japan. But it will be some time before Japan has the ability to play games like Japan’s killers in key semiconductor materials.
The core level of self-research is declining, and the edge of chip design is being lost
As we all know, the semiconductor production process is mainly divided into three major links of design, manufacturing, sealing testing, in the latter two links, the need for key equipment and materials, they are also to ensure the smooth production of the upstream cornerstone of the chip. But as we said earlier, Samsung has little competitive advantage in the last two key semiconductor devices and materials, leaving only the design link. Even so, Samsung is in danger of losing.
On March 21 this year, a petition submitted by Samsung users on Change.org website caused a stir in the industry. The user petition aims to put pressure on Samsung to stop carrying its own Exynos 990 on its flagship Galaxy S20 Plus. As for that, the model that carries the chip has performed less well than the Galaxy S20 Plus, which is equipped with qualsm SnapSnapdragon 865.
The Exynos 990 does have a difference in performance, power consumption, and Qualcomm SnapSnapdragon 865, according to tests bymedia Android Authority.
For example, in a configuration with FHD plus resolution and a refresh rate of 60Hz, the SnapSnapdragon 865 outperforms exynos 990 in CPU, the Exynos 990 lags behind The SnapSnapdragon 865 in terms of CPU, and the SnapSnapdragon 865 is even better in terms of system performance.
The single-core performance of the two CPUs is nearly consistent in the FHD plus, refresh-up of 120Hz, but the SnapSnapdragon 865 has a clear advantage on the multicore side; on the GPU, the gap is much smaller than in the 60Hz configuration, although the Dragon Snap 865 is still ahead; and in terms of system performance, the SnapSnapdragon 865 is better.
The industry knows that Samsung’s Galaxy S/Note “one-engine dual-core” approach has been around for a long time, with two different sets of SnapDragon and Exynos processors for different markets, while components such as basebands and CMOS are always slightly different. Thanks to the fine tuning of the Samsung LSI team, the difference in performance between the two flagship SoCs is small, and even Exynos is slightly ahead on some of its children.
With the release of the Exynos 990, Samsung’s self-developed high-performance CPU core has advanced to the fifth generation, with a $17 billion investment, but the results are not ideal, especially as ARM’s public version performance continues to improve, Samsung’s self-chip core iteration speed has been behind.
So from the point of view of technology or commercial value, Samsung’s self-research chip core has no practical significance, which is why Samsung disbanded its entire CPU research and development team in Austin, Texas, last November, and will fully shift to ARM’s public version architecture in the future, which is bound to weaken Samsung in the chip field, with Apple, Qualcomm and other chip core enterprises with their own chip-level differentiation competitiveness, and reflected in the smartphone terminal market.
To sum up, whether samsung, or other enterprises, to build non-American device chip manufacturing plant can be successful, the ultimate consideration is whether the enterprise in the chip industry chain has a game of killers, only for Samsung, obviously in the core area of the chip industry chain does not have, or even lack of such killers, while exposing Samsung itself in the chip industry “big but not strong.”
And this is also a reminder to China’s chip industry, in view of the future chip market competition complexity, especially the increase of non-market competition factors, game capability will be crucial. Therefore, in the future development, we should clear the chip industry chain which is the core of the killer technology, the focus of breakthrough, avoid the pursuit of the surface of the big and complete, but also strong, only in this way, China’s chip industry can really not be subject to people, in the game occupies the upper hand.