The timing to buy Tesla back? Wall Street is showing interest in FSD again.
Most of the time, the V11 works well, but the V12 is called Demo Learning (LfD), which surprised people and made them less interested. However, imitation learning, which is also called Demo Learning (LfD), is a machine learning method in which learning agents aim to imitate human behavior, and according to Tesla Engineering, it is improving five and ten times a month. Elon said the other day that a car company showed interest in Tesla FSD, but eventually it fizzled out in disbelief, but he believes that if he starts showing interest in Wall Street again, the company will come back, too. When teaching semiconductor manufacturing at school, he teaches students terms such as the eight major processes. The first thing that starts with is the manufacture of silicon wafers. As is well known, the manufacture of single crystal silicon ingots is important for silicon wafers. This ingot refers to a pure silicon cast, and its raw material is sand rich in quartz (SiO2). More precisely, sand, which is mainly quartzite, is the best, because the ingredients that make up the silicate are silicate. Silicate is mainly composed of SiO2, and you can reduce SiO2 to obtain Si from it. The reduction process does not take place in one queue. Once SiO2 + carbon powder is reacted in a high-temperature arc discharge furnace of about 1800 degrees Celsius, liquid-grade Si (metal-grade Si) with 99% purity can be obtained. This is called MG-Si in the industry. If you harden the liquid, it becomes a solid silicon mass, so it seems that you can use it right away, but it is not. This is because single crystal Si is not produced if it is hardened like that. Since polycrystalline or amorphous Si is not very good for making semiconductor devices, an additional process is required to treat MG Si. When Si is dissolved in hydrochloric acid again, two products emerge. First, SiHCl3 (Trichlorosilane (TCS) can be obtained, and the other is SiCl3, and when hydrogen gas is additionally injected thereto and reacted, SiH4 (monosilane) is produced. SiH4 can be obtained even if H2 is injected into SiHCl3 again and reacted. Anyway, this TCS is not yet high purity. This is because there is SiH4 and the like. Therefore, fractional distillation of TCS is required in order to obtain high-purity Si with maximum purity, which is a process completely similar to distilling in chemical engineering. TCS has a boiling point of 32 degrees, and accordingly, a distillation column is installed to cool the vaporized TCS in a wall or tray inside the distillation column, and when it is collected, it becomes a liquid, and pure liquid TCS can be obtained. Hydrogen gas can be injected while flowing liquid TCS to the surface of the rod while electricity flows through a high-temperature silicon rod. Then, solid high-purity silicon can be obtained by growing high-purity silicon as if ice were growing on the surface of a rod. Solid high-purity silicon thus obtained may have high purity, but it is still in a polycrystalline state, so a process of making a single crystal is required. A well-known process for this is the Chochralski process. First, solid polycrystalline silicon is dissolved in a furnace, and single crystal silicon, which will act as a seed for crystal growth, is brought into contact with the melted silicon and hardened little by little. In a small single crystal that acts as a seed here, since the silicon atom arrangement on the surface is regular, the liquid silicon atoms attached to it also self-assembly adhere to the arrangement. Of course, there is a possibility that defects may exist in the process of sticking, because before taking the path to thermodynamic energy minimization, if growth is induced too quickly, it hardens without taking the path. Therefore, in the growth of a silicon single crystal, the rate of crystal growth, temperature, and the atmosphere of the surrounding gas are very important. Each company has its own know-how in this.

As you can see above, in order to secure silicon wafers in the semiconductor manufacturing industry, it is not only necessary to have sand. As much as possible, there should be good sand with a high silicate content, and the high-temperature primary reaction, fractional distillation, high-purity liquid TCS acquisition, and silicon single crystal growth process are much more important. Of course, good sand is definitely advantageous, but what is more important than sand is process know-how.

China, which calls for the rise of semiconductors, has continued to invest heavily in all areas of the semiconductor industry over the past decade. The semiconductor industry is one of the key industries penetrating the 12th, 13th, and 14th economic development plans, and China, in particular, is strongly pursuing policies to self-sufficiency in the entire cycle of the semiconductor industry from A to Z. It is also focusing its investment on the production of silicon wafers, which are the most basic, but now exports to Korea as well as its own domestic production. As of 2022, Korea’s two major exporters of silicon wafers were Japan and China, with approximately $900 million worth of imports from Japan and approximately $800 million from China. Japanese wafers are still overwhelming in quality, but Chinese wafers are also gradually improving. In fact, the number of wafer producers in China, which was insignificant 10 years ago, is increasing. Companies such as TCL Zhongshan, Yoyenkui, Hukui, Zanluihong Micro, Liang Micro, Shanghai Silicon, Zhong Jingkouji, and Eswin are Chinese wafer companies that have recently grown rapidly to cope with the exploding semiconductor industry in China, and they have a strategy to first gain experience in low-cost wafer production and then gradually enter the high-quality, high-purity wafer market occupied by Japanese, Korean, Taiwanese, and German companies. In other semiconductor industries, China’s strategy is to target from low prices, gain know-how, gain mass production experience, and move to high value-added areas. This strategy has not changed much in wafers.

In fact, silicon wafers themselves are important as a starting point for the semiconductor manufacturing process, but the added value of the entire industry is not very significant. The market size is also less than $2 billion as of 2022, even if we consider the global scale. Less than 5% of the capital invested by top global semiconductor manufacturers such as Samsung and TSMC in CAPEX per year. Nevertheless, the fact that China continues to increase its share of silicon wafer production in China gives two things to think about. First, if China takes the base of silicon wafer production, it will be more advantageous for China to produce semiconductors than to disrupt the global semiconductor supply chain caused by wafer monopoly. This will be one of the factors that will increase the manufacturing cost competitiveness of Chinese semiconductors in earnest if the current US semiconductor industry regulations on China weaken. The other is that silicon wafers are not only used in semiconductors. Silicon wafers are used not only in semiconductors, but also in displays and solar panels. Silicon wafers of a certain quality that are mass-produced are the driving force behind China’s current dominance in the global silicon solar cell market. In the case of compound semiconductors, it is very difficult to make a high-purity single crystal wafer using a standard process mass-produced as much as silicon, because it is not made by making and dicing a high-purity ingot like silicon in the first place. For example, a compound semiconductor wafer such as GaAs is made by growing a thin film, but the production speed is very slow compared to a Si wafer, so the price is much higher. However, if Chinese wafer makers build up enough power in the silicon field, they are likely to target compound semiconductor wafers as the next market, so the possibility of becoming China’s strategic point in the next-generation semiconductor industry increases.

In one book attached below, sand is the main cause, concluding that it is impossible for China to make cutting-edge semiconductors such as Korea or Taiwan. Of course, sand must be recognized for its importance as a silicon wafer addressing material, but as mentioned earlier, the process of making high-purity silicon into a liquid, solid, or single crystal is much more important than sand itself. The process of converting sand into a silicon wafer is not the main battleground for supremacy in semiconductors as we understand it. Also, China is not currently at a disadvantage in its ‘battlefield’.

We should avoid overinflating certain areas of the semiconductor industry in order to properly face the ever-expanding global, especially the war of semiconductor hegemony between the U.S. and China. If silicon wafers are really important in the hegemony competition, especially securing sand is important, it is China that is rather advantageous. This is of course not the case, and silicon wafers are not an important priority even in the US public regulation. Several key technologies that act as bottlenecks in major semiconductor manufacturing processes are the really important competition for supremacy. Although much attention is already being paid to lithography, heterojunction, which has become important in anisotropic etching process such as HARC or post-processing, NCF bonding and filling, W2W, W2D, and D2D bonding processes, etc. These processes are not areas in which technology is secured overnight, nor are technical information easily disclosed through papers or patents. Only a few companies make specialized equipment for exclusive use, and each equipment is now excessively expensive, so it is only a handful of manufacturers that can purchase it.

Even when promoting a book, it is necessary to give accurate meaning based on proper information. The publisher’s intention and sincerity are fully understood, but this kind of interpretation or promotion can be a point of some doubt about the quality of the book as a whole. It is not necessary to denigrate the importance of silicon wafers and even sand that makes silicon wafers, but we must also be wary of committing a mistake that is overrepresented or exaggerated.