According to the Worldwide semiconductor Trade Statistics Society (WSTS), global semiconductor sales in 2019 are expected to be $412 billion, a year-on-year decrease of 12.1%. From 1999 to 2019, the global semiconductor compound growth rate was only 5.2%. If you only look at the growth rate, the semiconductor industry is not a high-growth market. In addition, from the perspective of annual growth rate, the growth rate of the semiconductor industry is unstable. From the perspective of scale growth and stability, the semiconductor industry is not attractive for investment.
The semiconductor industry in the United States, Japan, South Korea, and Europe developed far earlier than in China, and the overall industry has entered a mature and stable stage. In addition, semiconductors are the basis of information technology, and the end user is computer automobile communication, which is highly correlated with the macro economy. Therefore, from a global perspective, it is reasonable that the growth rate of the semiconductor industry is highly correlated with GDP. From the perspective of the semiconductor industry, due to the increase in mergers and acquisitions, the number of chip manufacturers is decreasing, the ratio of capital expenditure to revenue is stable, and the correlation between global GDP and the growth rate of the semiconductor market is getting higher and higher.
High income growth is unsustainable
As discussed above, the overall growth rate of the semiconductor industry is slow, so the growth rate of individuals in the industry is also relatively low. Taking the 34 chip design companies whose US stocks are more than 1 billion US dollars as an example, the 5-year compound growth rate of the two giants Intel and AMD computer CPU is only 6.1% and 4.1%. Analog chip leader TI Texas Instruments, the compound growth rate over the past five years is only 4.9%. CPU and analog chips are the two areas with the highest technical barriers, and the leading growth rate is basically the same as the industry growth rate. As of 2018, there were 9 companies with a growth rate of more than 10% for two consecutive years, accounting for 26%; only 4 companies with a growth rate of more than 10% for three consecutive years, accounting for 12%. There are three growth drivers: First, mergers and acquisitions, such as Broadcom, Microchip, and Cypress.
1) Broadcom, which has the highest 5-year compound growth rate, is mainly driven by acquisitions. For example, it acquired PLX, Emulex, former Broadcom, Brocade Communications Systems, and CA Technologies during the period from 2014 to 2018.
2) Cypress, which has a 5-year compound growth rate of 28%, is mainly driven by the acquisition of spansion in 2014, and its revenue growth rate in 2018 is only 6.7%.
Microchip Technology, with a 5-year compound growth rate of 22.6%, acquired Atmel in 2016 with a revenue growth of 56.8%, and acquired Microsemi in 2018 with a revenue growth of 34.4%.
3) ADI (ADI) with a 5-year compound growth rate of 18.7% mainly relies on Linear acquired in 2017 and Symeo acquired in 2018, and its revenue growth rate from 2012 to 2016 is very unstable, there are 2 years Still falling.
The second is the short-term explosion of demand. From 2016 to 2018, mining chips drove Nvidia’s revenue growth, and from 2016 to 2017, mobile phone storage drove the growth of Micron Technology.
The third is to launch new products, which are recognized by the market. For example, AMD launched Radeo graphics cards and Ryzen processors in 2017. By 2018, AMD continued to rely on upgraded Ryzen, Snapdragon, and Radeon graphics products to drive revenue growth of 23.3%.
Sales volume growth cannot hedge against unit price declines in the long term
Although chips are products with the highest technical content in hardware, chips with low technical barriers are not necessarily a very profitable business. Most chip companies have small shipments and low unit prices of chip products, making it difficult to expand their revenue scale. Unlike consumer products, except for memory and processor chips in special periods, the price of general chips is gradually decreasing. If chip design companies want to increase the scale of their performance, they can only rely on increasing sales and use the increase in sales to offset the decline in unit prices. However, chip sales cannot grow indefinitely. All kinds of smart terminals are composed of many chips, and it is impossible for a certain chip to increase in sales alone. After all, in the long run, the compound growth rate of the semiconductor industry in the past 20 years is only 5.2%.
Case 1: Espressif Technology, mainly produces wifi and bluetooth chips, wifi and bluetooth technology solutions. Products are used in mobile devices, household appliances, industrial equipment and application scenarios that require high safety performance. In 2018, Espressif’s chip revenue was 320 million yuan, and the chip sales volume was 75.97 million. The unit price of the chip showed a downward trend. To expand the scale, only by increasing sales.
However, due to the low technical barriers of Bluetooth chips and fierce competition, the manufacturers engaged in TWS Bluetooth chips include not only foreign companies such as Qualcomm and Cypress, but also Yuanxiang, Luoda and Realtek from Taiwan, and Hengxuan, Huawei, Juxin, Ziguang Zhanrui, Jerry and Zhongke Lanxun and other manufacturers. If you want to increase sales, you have to cut prices. In the medium and long term, sales growth cannot offset the decline in unit prices.
In October 2019, the price of the TWS Bluetooth headset chips of Zhuhai Jerry and Lanxun has reached 1.6 yuan, while the chips of other manufacturers are still at the stage of 1.6 US dollars. The monthly shipment of chips is as high as 50KK and 70KK
From a macro industry perspective, it is estimated that 5.4 billion Bluetooth devices will be shipped by 2023. Assuming that each device uses 1 Bluetooth chip, a total of 5.4 billion Bluetooth chips will be shipped. Assuming that the price of Bluetooth chips will reach 3 yuan in the future, the market size Only 16.2 billion yuan.
Case 2: Shengbang Co., Ltd. mainly produces analog chips. There are more than 1,200 products in 16 categories. The average unit price of each chip is only about 0.3 yuan, and it shows a downward trend. In 2018, 2.02 billion chips were sold, with an average of 1.69 million chips per chip.
Case 3: Goodix Technology, which mainly produces fingerprint recognition chips and capacitive touch chips. In 2018, the sales volume of capacitive touch chips was 1.07 billion, and the sales volume of fingerprint recognition chips was 1.2 billion, of which 83% came from mobile phones. The sales volume of mobile phones determines the market size of fingerprint identification chips. In 2018, the global smartphone sales volume was 1.4 billion. From this perspective, the market saturation of fingerprint identification chips reached 86% in 2018. The ceiling of the market size of fingerprint identification chips is the ceiling of mobile phone sales.
From the unit price point of view, the overall chip unit price of Goodix Technology dropped to 6.33 yuan in 2018. The company has not announced the price of fingerprint identification chips in the past three years. From the prospectus, it was found that the price of the company’s fingerprint identification chips dropped from 61.1 yuan in 2014 to 22.8 yuan in 2015, a drop of 63%.
Profit margins are difficult to improve
Products determine profit margins. As long as the types of products remain the same, it is difficult for chip design companies to increase their profit margins. It is good to be able to maintain them at a certain level. Although Intel is the leader in PC processors and Qualcomm is the leader in mobile phone processors, the gross profit margin of the two also showed a downward trend from 2010 to 2018 (Qualcomm removed the patent fee, and the gross profit margin also declined). In the past 9 years, the gross profit margin of most chip design companies can only be maintained or decreased. The reasons are as follows:
First, the competition is fierce. For the top processor chips, MediaTek, Spreadtrum, etc. compete with Qualcomm, AMD and Intel, and other chips with lower technical barriers compete more intensely.
Second, the processors of Intel and Qualcomm are digital chips, which have high process requirements, and advanced processes bring high manufacturing costs.
The third is that the shipment of single chip products is small and the market is fragmented. A single design company cannot form bargaining power over downstream foundries, and cannot increase gross profit margins from the cost side. For example, the gross profit margin of Espressif’s wifi and Bluetooth chips dropped from 52.5% in the first half of 2018 to 47.6% in the first half of 2019.
Only a few companies have a steady upward breakthrough in gross profit margin, such as Texas Instruments, NVIDIA, NXP, Micron Technology after 2015, Maxim, STMicroelectronics, Skyworks. Companies with rising gross margins have the following characteristics:
One is the surge in short-term old demand, such as Micron Technology, because the surge in mobile phone storage demand has led to a rise in the price of memory chips.
The second is the emergence of short-term new application requirements, such as NVIDIA’s GPU for Bitcoin mining and artificial intelligence computing.
The third is analog chip design companies such as Texas Instruments and NXP. Although the operating income has increased slightly, the gross profit margin has increased.
The reasons are:
1) Driven by the three major demands of power management, signal conversion, and automotive electronics, ICinsights predicts that the growth rate of analog chips exceeds the overall growth rate of the chip market by 6.9%;
2) Digital chips such as processors and memories need to be frequently upgraded, such as 10nm to 7nm, resulting in high manufacturing costs.However, the process update speed of analog chips is slow, and the cost will gradually increase with the depreciation of the production line.
Here we can draw another conclusion: having its own fab is a necessary condition for the growth of analog chip design manufacturers. From the perspective of business model, high-quality analog products require close integration of design and process, and sufficient communication between fabs and designers can develop distinctive and competitive products.
On the one hand, due to the slow update of the analog chip process, the use time of the fab exceeds the depreciation time. This IDM model of self-built fabs is conducive to reducing costs and increasing gross profit margins in the future. On the other hand, analog chip factories have diverse needs and a high degree of customization in design and manufacturing. Without their own fabs, they cannot guarantee product quality, performance, and even production capacity. Foreign manufacturers generally define and develop new products according to application requirements. Design, process, and application form a stable triangle of product definition. Without the support of foundries in China, it is difficult to form an opportunity to combine design and technology.