A decade of fierce competition in RF chips

Yuanchuan Technology Review recently learned exclusively from industry chain insiders that SMIC Ningbo Co., Ltd. (hereinafter referred to as "SMIC Ningbo")'s fully independent intellectual property bulk acoustic wave resonator process technology platform SASFR® has supported many design companies to achieve mass production of RF front-end medium and high frequency bulk acoustic wave filters (BAW). The performance of customers' products has reached the best domestic and industry-leading levels, and can begin to be supplied to high-end 5G mobile phones.

This indicates that Huawei's 5G mobile phones, which were previously rumored to be blocked by Broadcom's BAW filters, may have a glimmer of hope.

This also indicates that after ten years of hard work, China's RF chip industry has finally made up for its last shortcoming.

Radio frequency plays an inconspicuous but respected role in the chip industry. It is like the limbs of the baseband chip and is the core executor of the mobile phone communication function.

From low to high technical difficulty, this type of chip is subdivided into: switch, which selects the signal path; LNA, which is a low-noise amplifier, which receives weak signals; PA, which is a power amplifier, which amplifies and transmits the weak signals sent by the mobile phone; filter, which receives signals in a specific frequency band and filters out interference signals. Among them, SAW filters are mainly used in medium and low frequency bands, and BAW filters are mainly used in medium and high frequency bands, especially in 5G communication scenarios.

China's RF chip industry started with the historic opportunity of the explosion of smart phones ten years ago. Under the combined effect of time, place and people, three representative companies, namely Zhuoshengwei, Weige Chuangxin and SMIC Ningbo, were born in the fields of switch, LNA, PA and filter. This article takes this as a sample and attempts to sort out the development context of China's RF chip industry.

This article takes the company as the research sample and is divided into three parts:

1. Zhuoshengwei: The Infinite War of the Starter

2. Weige Chuangxin: Prisoner’s Dilemma in Red Ocean Competition

3. SMIC Ningbo: Seizing a strategic high position

0 1

Infinite War of the Opener

The pioneer of China's RF chips is undoubtedly Zhuoshengwei. From the very beginning of entering this field to making progress every year after its listing, it can be said that among domestic RF players, Zhuoshengwei has been fighting a "magic battle" all the way, and it is an excellent business case in the history of China's chip industry.

However, Zhuoshengwei's beginning was very bleak. Founder Xu Zhihan returned to China to start a business as early as 2002. After working in Hangzhou Zhongtianwei for four years, he founded Zhuoshengwei. At the beginning of its establishment, the company chose to enter the field of terrestrial TV and mobile TV chips, but ended in failure.

During the most difficult period, the company burned through all the money it had raised from investors such as Sequoia and could not pay salaries. He and two other partners borrowed 1 million yuan a month for nearly a year until new investors joined. During this period, the three partners only received 8,000 yuan in salary a month.

After reflection, Xu Zhihan believed that the objective reason was the wrong choice, lack of understanding of the domestic market, and lack of grounding.

Later, in the trend of smart phones, Zhuoshengwei did not follow the trend in the RF chip track and choose PA, which seemed to have higher technology and higher value at the time, but instead chose the seemingly marginalized switch and LNA.

This choice was immediately proven to be correct. In 2012, Apple's mobile phone became a global hit, which took up the production capacity of traditional RF chip manufacturers such as Skyworks, Infineon, and Qorvo. Android manufacturers such as Samsung had no RF chips to use.

A radio frequency chip worth a few cents now holds the key to a mobile phone worth thousands of yuan. Compared with traditional IDMs in Europe and the United States, Zhuoshengwei is a typical Fabless company that is only responsible for design and outsources manufacturing to wafer foundries. This business model is flexible and can quickly launch products with strong current profitability. Zhuoshengwei received an order from Samsung and found TSMC to jointly develop a GPS LNA based on RF CMOS technology, which met Samsung's numerous stringent supply chain requirements, solved Samsung's urgent needs, and established a firm foothold in the strict Korean electronics supply chain. Since then, jealous Chinese companies have continued to challenge this position, but all of them have failed.

Mastering the core know-how of switch and LNA, Zhuoshengwei took advantage of the victory and entered the supply chain of Xiaomi, vivo, Huawei and OPPO. The main players in the Android camp supported Zhuoshengwei's international status in RF chips.

However, when Zhuoshengwei became the fifth in the world in the switch field, if it wanted to move up one more place, it would have to become a well-established international RF manufacturer, and the competition was too fierce for PA. Therefore, when it went public in 2019, Zhuoshengwei wisely chose to use the raised funds to make filters, RF modules and factories, and the subsequent round of private placement also continued along this line of thought.

In general, Zhuoshengwei has played the role of a leader in RF chips for quite a long time. It had a difficult start, but it has been upgrading and fighting monsters all the way, from low to high, creating a rare miracle in the history of the chip industry, which always makes people look at it with admiration. However, if it wants to continue to move up, there are still difficult battles to fight.

In 2019, when Zhuoshengwei went public, China's PA industry finally produced a participant selected by MediaTek: Vigorchip.

When it comes to Vijet, we have to mention its legendary founder Rong Xiuli. She founded Tianyu Mobile in 2002 and built it into the top domestic mobile phone brand in 5 years. But the good times didn't last long, and Tianyu was soon defeated by a group of smartphone players. Mobile phones are becoming more and more difficult to make, but Rong Xiuli smelled another opportunity - RF chips, so she founded Vijet in 2010 and plunged into the wave of PA chip design in China at that time.

With the R&D personnel inherited from the old RF chip manufacturer RFMD and a keen business sense, Visionox took the lead in the domestic PA market and started a price war with its Taiwanese competitor Airoha. The two sides were both losing money, and finally Airoha's shareholder, mobile phone SoC chip manufacturer MediaTek, could not stand it and decided to end the war. In April 2019, MediaTek made a strategic investment in Visionox and became a major shareholder, while disbanding Airoha's PA department.

This move by MediaTek basically announced the victory of Vigetronix in the PA market. For large chip companies like MediaTek, peripheral chips such as RF chips, which have a smaller market but still require a lot of investment, have a low input-output ratio. For the giants, self-development is not cost-effective. Therefore, the general way of playing is to cultivate a group of small companies like raising insects. If they die, they die, and the winners eat a little meat. In this way, the most competitive companies can be screened out and enjoy the winner-takes-all dividends. With its strong appeal to mainland mobile phone companies, MediaTek has attracted investment from the four complete machine manufacturers of Xiaomi, Huami, OV, for Vigetronix, thereby achieving a strong binding with its supply chain.

Despite this, the gross profit margin of Weige Chuangxin's main business - 4G PA is only about 20%, far lower than the 50% of Zhuoshengwei's switch and LNA.

The technical difficulty and gross profit margin are actually opposite. There are three reasons behind this:

First, there is no bargaining power with the foundry and packaging and testing ends. As a fabless company, the raw material cost disadvantage of Weige Chuangxin's products is inevitably held in the hands of the wafer foundry. PA uses GaAs (gallium arsenide) technology, and the wafers are mainly purchased from Win Semiconductors, a manufacturer in Taiwan, China, accounting for more than a quarter of the total procurement cost. Moreover, GaAs does not have the advantages of silicon-based, and the cost cannot be reduced for a while. At the same time, the packaging and testing suppliers of PA in mainland China are concentrated in a few manufacturers such as Changdian and Suzhou Suning. Weige has poor bargaining power for the two major parts of manufacturing and packaging and testing.

Second, PA modularization means a shortcoming in RF procurement. With the trend of miniaturization and thinness of mobile phones, OEMs are more willing to use highly integrated modules in mid-to-high-end products, preferably with integrated filters. However, filters are the most difficult to develop, and Weige Chuangxin's core advantage does not lie in this, so it can only be purchased externally, which means it is subject to the control of others.

Third, there are already six or seven domestic PA manufacturers, and the price war cannot be stopped.

In order to break through the gross profit margin dilemma, Weige Chuangxin has two measures:

First, move towards 5G high-end products. In the first half of 2021, the proportion of 5G PA modules in the total revenue of Visionox increased from 10% last year to 25%, and the gross profit margin of 5G PA modules exceeded 40%, which is nearly twice that of mature products 4G PA modules.

Second, improve the layout of the test end. As the proportion of high-end 5G product shipments increases, general test equipment is difficult to meet the requirements, while costs are rising and profit margins are squeezed. Therefore, Weige Chuangxin plans to invest more than half of the funds raised from the IPO into the testing business line.

After domestic companies have successively conquered switch, LNA and PA, filters have become the last hurdle for China's RF chip industry. There are many domestic manufacturers that can make mid- and low-end SAW filters, and they have even started a price war. However, BAW filters suitable for 5G high-frequency bands are rare.

In short, high-end BAW filters face three major challenges :

First, the factory needs to be very capable and able to come up with the design. But if no one in China helps to manufacture it, it will just be talk on paper. Building a wafer fab requires a lot of talent. Second, there is a patent wall. Broadcom and Qorvo, two major American companies, have monopolized this field. New entrants can hardly touch this forbidden area. If they do, they will be hit with a big stick. Third, there are no users. The 5G market has just started to develop, and BAW filters are extremely critical. As long as American companies are still supplying them, domestic equipment companies will lack the motivation to apply them.

Faced with three major difficulties, SMIC Ningbo began to try.

Here we need to introduce the background first. First of all, let’s talk about Huang He, the founder of SMIC Ningbo. He returned to China and joined SMIC as early as 2002. He was one of the early veterans who followed Zhang Rujing to start a business. Now we see that SMIC has been given the mission of pursuing process, but in fact, the team founded by Zhang Rujing has always had its own strengths, and all important chip product lines have been laid out, including the special process that Huang He is engaged in.

In 2016, SMIC Ningbo was established. Huang He moved to Ningbo with dozens of core personnel, focusing on high-voltage analog, optoelectronic integration and other products. The 8-inch N1 production line was put into production as early as November 2018. Located in the Yangtze River Delta, SMIC Ningbo's main products have broad market prospects. In 2019, facing the suppression of the United States, Huawei began to save itself, and a large number of domestic chip companies ushered in market opportunities. Huang He keenly discovered the opportunity and decided to launch BAW RF chip products.

First of all, for a team that has been fighting in the domestic foundry industry for more than a decade, plant construction and technology are consistent advantages. Since 2019, SMIC Ningbo has quickly covered the entire process of BAW filters from wafer manufacturing to packaging and testing, and the yield and reliability have been rapidly improved.

The real difficulty lies in two points: first, how to make a clean chip without patent disputes; second, how to enter the 5G market in large quantities.

In response to the patent issue, SMIC Ningbo took a unique approach, avoiding the traditional technical route, and innovated in the core difficulty of BAW filters, the thin film process structure and resonator device design, and developed a SASFR process technology platform with fully independent intellectual property rights. During this period, the company's R&D investment increased from less than 100 million to nearly 400 million this year, and the total number of patent applications reached 700. Now the company has about 800 employees, which is almost one patent per person.

In response to the market expansion issue, SMIC Ningbo has chosen a wise path: cultivating an ecosystem. Generally speaking, there is still a long distance between wafer foundries and downstream terminal manufacturers, and many things cannot be done by themselves, but the foundation of the domestic BAW filter design industry is weak.

Therefore, SMIC Ningbo has built a unique business model in China - special wafer foundry plus design services, and vigorously supported emerging chip design companies, including Xinsheng Semiconductor, which was established this year and focuses on BAW filter design. The two parties have cooperated closely and quickly achieved the mass production milestone in a short period of time.

So far, SMIC Ningbo has built two core barriers around the almost blank BAW filter in China: one is a breakthrough in the entire manufacturing process, and the other is to strictly adhere to the boundaries of foundry and use design services to attract new players to expand the ecosystem.

It can be said that at the current point in time, any domestic chip company that achieves independent mass production of BAW filters will be destined to occupy an extremely high strategic position in the RF chip industry by tackling the hardest challenges right from the start. Product verification by the OEM is only a matter of time. Once the volume is increased, its revenue curve will be very steep.

Ten years ago, China's RF chip industry was almost blank. The era of smartphone explosion attracted a group of domestic players. The past decade is the "Cambrian" era of domestic RF species explosion. Although they had a rough time in the early days, many players have achieved success today.

Summarizing the past decade, China's RF chips can be divided into three stages:

First, find a market. Ten years ago, domestic mobile phone manufacturers would never dare to use domestic RF chips, because this small item directly determines whether the mobile phone communication function is normal or not. Using the products of new players may be fatal to the person in charge of procurement, and no one dares to bet their official hat. There is no way, no one wants domestic chips, and domestic RF players have found a ticket to enter the market from Samsung. However, at this stage, they only sell the cheapest categories - LNA and switch.

Second, domestic substitution. The Huawei incident sounded the alarm. China's most powerful OEM began large-scale domestic chip verification. Xiaomi, OV followed behind and became bolder. During this stage, more categories of RF chips completed domestic substitution, and a large number of PA companies emerged.

Third, solve the bottleneck component problem and complete the overall layout. At this stage, filters related to the promotion of 5G have become a battleground, especially the BAW filter that is firmly controlled by Broadcom, which has become the Achilles heel of Huawei's 5G mobile phones. Not only that, as the integration of RF chips becomes higher and higher, whether it can be integrated into the most difficult RF device - the filter, has become the biggest problem. Whoever can solve the filter will stand at the commanding heights of the RF industry.

After three stages, China's RF chip industry has gradually formed a climate. But there is a hidden worry that needs attention:

Talent poaching leads to a bank-run competition.

At present, most domestic RF manufacturers are designers, and none of them can form a full range of RF products. Moreover, they almost share a very limited number of foundry and packaging and testing manufacturers, and the most core foundry manufacturers are all outside the mainland.

Compared with several fully integrated IDMs abroad, such an industrial ecosystem is very weak in competition and is even less able to withstand the war for talent.

Because the key to the success of the RF chip industry lies in the continuous know-how of hundreds of engineers working closely together. There are three key points: 1. First, there are hundreds of people; 2. Then there is close collaboration between the upstream and downstream industry chains, and everyone is doing research and development together; 3. Long-term continuous research and development, and uninterrupted know-how. If there are only many small companies with less than 100 R&D personnel, talents are scattered everywhere, and the average length of service of engineers in each company is very short, it will definitely not be able to succeed.

In this context, China needs more responsible RF players who can take on the responsibility of building manufacturing capabilities and establishing an ecosystem.