Looking back on a series of bans imposed by the United States on Huawei in the past: China lacks the core, and the United States restricts le neck through chips; China gets the chip, and the United States will stop it through chip manufacturing; China will manufacture chips, while the United States will control them by manufacturing equipment, materials and software. Therefore, China needs courage and rationality to get out of the chip dilemma.
So where is China's neck in the semiconductor industry? To this end, we need to sort out the chip industry chain.
(1) Upstream of the industrial chain-chip design is self-controllable.
Looking back at the development of chips in China, we will find that we are better at catching up in the field of chip design than in the field of chip technology. For example, 20 years ago, China's wireless modem chip was still blank, but today Huawei has been able to release the 5G modem chip with leading performance in the world. Analyzing this problem from the economic and social point of view, this is because the chip design can adopt the business model of no wafer factory, so that the chip design company does not have to bear the risk and huge expenses of building a wafer factory, so the amount of funds needed by the chip design company is relatively low, and the assets are mainly virtual assets such as ip design. In other words, a small amount of funds can incite the chip design company to get a lot of income and bring returns. Therefore, chip design companies can do well without a large amount of state capital, and the cause of chip design can also be done well by social forces alone. In the field of R&D and mobile phone chip design, China has formed a pattern of "1+N". To put it simply, its composition is that Huawei Hisilicon is the big brother, and there are a number of younger brothers such as Will, Sheng Bang, Zhuo Shengwei and Zhao Yi Innovation as substitutes around Huawei's industrial chain.
Huawei Hisilicon, which focuses on the core SoC and baseband chips of mobile phones, has stood at the commanding heights in the field of fabless design in the world. Huawei's market share of SoC chips in China has reached 43.9% this year, even surpassing Qualcomm Snapdragon supported by Xiaomi, OV and several second-tier manufacturers, ranking first in China. The Kirin 990 5G running score released last year is the same as that of Qualcomm Snapdragon 855plus and similar to that of Snapdragon 865 released at the end of the year.
Baseband chip can be said to be one of the core components in mobile phones. Better than Apple, in the face of Qualcomm's baseband patent, we have to pay the protection fee. And the only thing that can compete with Qualcomm is Hess's Ba Long series baseband chips. At the beginning of last year, Huawei released Ba Long 5000, which became the world's first single-chip multimode 5G baseband chip, and supported SA independent networking and NSA non-independent networking for the first time.
On top of this, Huawei also integrated Ba Long 5000 into Kirin 990 5G processor. On the chip, stronger integration means smaller area and lower power consumption. However, after Snapdragon 865 released Kirin 990 5G, it chose to plug in 5G baseband in order to go public as soon as possible. No wonder in China, where the shipment of 5G mobile phones has reached 40%, Qualcomm's market share will be surpassed by Huawei.
(2) The middle reaches of the industrial chain-chip foundry still needs to be cultivated in China.
For Huawei, both outsourced and self-developed chips are mostly produced by wafer foundries and then packaged and tested by OSAT factory. Without the support of OEM, even if Huawei designs high-tech chips that make Qualcomm tremble and Intel cry, they may become waste paper. This piece is concentrated in Taiwan Province Province, China, especially TSMC and Riyueguang, which are responsible for the manufacturing and testing of most Huawei chips. The foundry is mainly engaged in semiconductor manufacturing, and its core technology is wafer manufacturing and wafer processing, which is much more difficult than the later sealing and testing, and the equipment investment here is very huge. In the total investment of the factory, the investment in semiconductor equipment accounts for 67%, and in this 67% investment, wafer manufacturing equipment accounts for more than 70%. Therefore, the OEM mode has a large investment scale and heavy assets, and the cost of maintaining the normal operation of the production line is high; It needs continuous investment to maintain the technical level, and once it falls behind, it will be difficult to catch up. So in general, the investment in chip manufacturing can be described as massive. Some people compare it to building two latest 12-inch fabs, which is equivalent to a Three Gorges Dam. As a result, chip foundry has also entered the knockout stage, forming a pattern of "one super and many strong": TSMC in Taiwan Province, China, with a market share of 5 1%, is a well-deserved number one player; South Korea's Samsung share 19% is struggling to catch up; The third and fourth UMC and Grofonde have surrendered, indicating that they will no longer develop processes below 14 nm. SMIC, ranked fifth, is the largest and most advanced foundry in Chinese mainland, but its market share is less than 6%. In the past, due to TSMC's leading technology and Huawei's initiative to expand production, Huawei Hisilicon chips have been manufactured by TSMC, and the order amount accounts for 14% of TSMC's revenue, second only to Apple's 23%. The importance of interdependence between Huawei and TSMC is obvious. However, this "precise strike" by the United States against Huawei means that TSMC's supply to Huawei will be hindered, so we will ask whether there are any companies in the mainland that can get ahead of us. The answer is SMIC, but there is still a long way to go in improving technology and investing in R&D capital. On the one hand, SMIC lags behind TSMC's second-generation process, on the other hand, it needs a lot of capital expenditure to ensure production capacity. However, from 20 15 to 20 19, SMIC's total capital expenditure was about 10 billion US dollars, less than TSMC's annual expenditure. At this point, SMIC is not an opponent of TSMC, but at least it has completed the self-cultivation of excellent spare tires. However, due to the new regulations in the United States, SMIC is worried that if it wants to contract chips for Huawei, it will also need to obtain a license from the United States. Some people think that SMIC is a local enterprise in China, so it can ignore the US policy. However, once the United States thinks that SMIC is illegal, it may force the United States to remotely lock SMIC equipment, such as Applied Materials Company and Marco Lin Semiconductor Company, and turn it into "scrap iron" that cannot work.
(3) The downstream of the industrial chain-the bottom needs to be broken.
The importance of semiconductor equipment was mentioned earlier. American equipment enterprises accounted for 49% of the global semiconductor equipment industry in 20 19, among which the applied materials ranked first in the industry (applied materials, Fanlin), while China's localization in mask aligner, ion implanter, measuring equipment, SOC tester, PECVD and other equipment fields was low, and its dependence on American equipment brands was 47%. The latest US policy towards Huawei is largely based on semiconductor equipment. Even if most of Huawei's chips can be self-developed, they should be handed over to the foundry. Whether TSMC or SMIC, a considerable part of the semiconductor equipment they use comes from the United States. Therefore, if you want to use these devices to produce chips for Huawei, according to the United States, you must first obtain approval. Such "long arm jurisdiction" is very lethal.
Similarly, in the upstream link, semiconductor software (mainly including EDA design software and lithography computing software such as OPC and SMO) is absolutely monopolized by the United States, and local brands are still in the incubation period. In the field of semiconductor materials, Japan is in a leading position, and American companies are in a leading position in many sub-sectors. As the localization of semiconductor materials is in the initial stage, the United States and Wassenaar have upgraded their technical control (on February 24th, 2020, Japantime media reported that 42 members of Wassenaar, including the United States and Japan, agreed to expand the scope of export control on February 20th19, including semiconductor substrate manufacturing technology for military purposes and military software for cyber attacks, which will still restrict the overall development of China's semiconductor industry.
For semiconductor software, EDA software still has a solution. On the one hand, some technical measures can be taken to continue to use the original software; In addition, even if foreign advanced software is completely in short supply, local products can be used, but the overall level is definitely lower. If used, the overall performance of the chip is likely to be compromised in a few years. However, this is inevitable.
(4) Break the game and deduct points in the future
The crux is semiconductor equipment, which directly affects whether the foundry can supply Huawei. In the short term, semiconductor equipment can't be controlled by itself, so China may introduce countermeasures, but it may be restrained, and it will gradually increase depending on the US reaction. It is irrational to play all the cards you can at once. It is expected that there will be several round trips between the two sides. There may be no results in the short term.
There are opportunities in the crisis. The upcoming supply cut-off policy in the United States will further promote the domestic substitution of Huawei's supply chain, which is a once-in-a-decade opportunity for the domestic semiconductor industry chain and will accelerate the domestic substitution opportunities in manufacturing, equipment and materials. This new round of control agreement is also a great opportunity for domestic semiconductor manufacturing and semiconductor equipment. It is understood that Huawei's goal for localization is to adopt 100% for domestic suppliers whose performance meets the requirements.
Based on this, from the perspective of American enterprises, the previous ban in the United States has prevented some American chip manufacturers and chip manufacturers containing 25% American technology from supplying Huawei. Now the new regulations force Huawei to use its own research chips. So doesn't this mean that the United States has been busy for half a day, soliciting business for other chip manufacturers in Japan, South Korea, Europe and China? Therefore, Xin Zhixun believes that the United States may release many American chip manufacturers that can be replaced by domestic chips, issue licenses to them, let these manufacturers do business with Huawei, and urge Huawei to increase the use of American chips. In this way, Huawei can continue to operate, but it is equivalent to abolishing Huawei's martial arts (unable to use self-developed chips), and then making it rely on American chips (referring to those chips that can be replaced but still have certain advantages compared with domestic chips), thus making its products lose their leading position in technology and market.
Cited documents:
1, who will help Huawei: the rise and dilemma of the three major tracks, 2020
2. Variables of Huawei chip industry chain in 2020
3. The ultimate deduction of Huawei incident after upgrading control measures in the United States, 2020.
4. Comment on BOC Securities-Semiconductor Equipment, Materials and Software, Yang, 2020.