01

Investment value of domestic gallium nitride track
          1. Investment value of gallium nitride trackThe material properties of third-generation semiconductors will gradually replace part of the market for silicon-based power devices. This has been continuously verified through market applications and has basically become a consensus. Compared with silicon carbide, the current main disadvantages of silicon-based gallium nitride include low voltage rating and lack of reliability verification data. In addition, due to the low overall output value of silicon-based gallium nitride, scale effects have not yet formed, resulting in no cost advantage compared to silicon carbide. However, silicon-based gallium nitride manufacturers are currently promoting the evolution of products towards high withstand voltage (currently 1200v products) and high reliability (currently have ideal reliability data for high-voltage devices) through epitaxy, device structure, and drive control circuits. Moreover, the selling price of gallium nitride devices has approached that of silicon MOS, and product performance, quality, and cost continue to approach the sweet spot. In addition, compared to silicon carbide,The silicon-based gallium nitride material itself can provide higher power efficiency and lower cost. In the 600V to 1200V range, silicon-based gallium nitride will become a very competitive technology direction.In addition, with the continuous application of silicon carbide in the automotive and photovoltaic markets, billions or even tens of billions of investments are being made at home and abroad. It is intuitively felt that the market size of silicon carbide is very large, while the market size of gallium nitride used in small charging heads feels much smaller. According to data released by Yole, the market size of silicon carbide will be US$2.5 billion in 2025, and the market size of gallium nitride is forecast to be US$1 billion in 2026. At first glance, the overall output value of silicon carbide is indeed several times that of gallium nitride.    However, the opportunity output value of silicon carbide and silicon-based gallium nitride for most domestic enterprises is actually equivalent:   One is to consider the output value of the substrate. The cost structure is different, and the output value of silicon carbide companies without substrate links will be greatly reduced.The cost of silicon carbide substrate is about 50%, the cost of epitaxial wafer is about 20%, and the cost of device manufacturing, packaging and testing is 30%. The substrate cost of gallium nitride is basically ignored, epitaxy is 50%, and the remaining device output value ratio is higher.    The second is to consider the output value of the vehicle specifications. Market demands are different. Silicon carbide focuses on automotive specifications, while gallium nitride focuses on consumption.According to data released by Yole, electric vehicle market demand is expected to account for approximately 61% in 2025. According to data from Yole, consumer grade accounts for approximately 70% of the market demand for gallium nitride. At present, domestic silicon carbide manufacturers still have a long way to go to achieve large-scale applications in the automotive market.  The third is to consider the output value of MOS. Domestic silicon carbide companies are currently mainly concentrated in the SBD market, and MOS is still rarely shipped.According to data released by Yole, MOS's output value will account for more than half after 2022. As the later market is mainly driven by electric vehicles, the proportion of MOS will continue to increase. However, due to the reliability and process source issues of silicon carbide MOS itself, it is still difficult for domestic manufacturers to confidently and boldly carry out large-scale promotion and application. To sum up, if domestic silicon carbide companies exclude the output value corresponding to substrate materials, automotive market, and non-automotive MOS, their opportunity output value is actually not much different from that of silicon-based gallium nitride. Superimposed on the 30% and 70% market growth rates of gallium nitride and silicon carbide, silicon-based gallium nitride is a good track from the perspective of industry scale. The focus of Muscle Man Silicon Carbide and Flash Gallium Nitride are respectively on the industrial and consumer markets (overall). The differences in quality requirements, market pricing, and market introduction cycles will not be discussed in detail here.But overall, the market elasticity of gallium nitride, which focuses on consumer-level applications, will be higher.In addition, because the development stage of silicon carbide power is relatively mature, there are basically a few relatively recognized high-quality companies in China from substrate, epitaxy, device manufacturing, or IDM, leaving the window of opportunity for entrepreneurs to be smaller.  
  2. Opportunities and advantages of domestic gallium nitride companies
  (1) “High growth + high concentration” brings opportunities to new playersAccording to data released by Yole, the overall market size in 2020 is only 46 million US dollars, and is expected to grow to 1.1 billion US dollars in 2026. The GaN power device market will maintain a growth rate of 70% from 2020 to 2026, 23 times in 6 years. Such a high industry growth rate directly brings about the problem of how to ensure the stability of the supply chain of gallium nitride devices. Currently, Navitas and Pi, the two major global shipping manufacturers, account for more than 70% of the global market share. Domestic brand manufacturers mainly use the gallium nitride charging solution of one of them.  In the short term,The upstream supply chain is highly concentrated, and there is a rigid demand for large brand power supply customers to diversify supply chain risks;In the long run,Compared with the current 65% market share of silicon-based MOS CR5, the current market concentration of gallium nitride is too high, and it is gradually approaching the silicon-based market structure. It is most likely a matter of time.    (2) The current overall domestic level of silicon-based gallium nitride in China is more advantageous than silicon-basedIn 2019, China Resources Micro and Silan Micro, the largest domestic MOS and IGBT companies, accounted for 3% and 2% of the global market. Innosec's current gallium nitride track shipments should have ranked third. Although there are several domestic gallium nitride start-ups that have shipped less, their current overall technology and industrial resource reserves also occupy a good position in the world. Traditional power leaders, such as Infineon and Ti’s gallium nitride on silicon products, are still struggling to make headway. In the domestic gallium nitride on silicon industry, although there is no overtaking in a corner, we look at the current global status of domestic companies and the core talent pool of Chinese people in the field of gallium nitride. The current overall progress position is indeed much better than that of silicon.  To sum up, in the face of a rapidly growing pie, domestic enterprises have the strength and talent to take a big piece of the pie.       02       “Concentrated + closed” supply chain, value and risk
          The existing supply chain is "concentrated + closed", causing the market to be driven by a small number of suppliers
  Concentration of existing supply chains.Unlike silicon-based products, the materials and manufacturing links are very mature, and the gallium nitride supply chain currently has very little production capacity that has passed batch verification. Foreign Foundries mainly include TSMC, Fujitsu, and X-Feb. Domestically, Sanan Integration is the main company, and some other Foundries are also making arrangements. Overall, domestic supply chain resources are very scarce.  The existing supply chain is closed.Currently, the global gallium nitride Foundry's main production capacity only serves one company, and design companies and the industry chain have a strong binding relationship. As the world's largest silicon-based gallium nitride device company in terms of shipments, Navitas currently occupies almost all of TSMC's gallium nitride production capacity, making it difficult for other customers to obtain production capacity. The "concentration + closure" of the existing supply chain has resulted in a very high degree of concentration in the market. Navitas, Pi, and Innosec are the three companies that control the supply of core production capacity and drive the market. It is difficult for brand power supply customers to have other choices.  
  Supply chains are expected to remain concentrated and closed   continuous concentrationGallium nitride is a new technology product, and the technological development of all aspects is still at a relatively early stage. Both end customers and design companies are relatively cautious about the upstream supply chain. Since proven technology is reliable, once a complete industry chain is verified and mature, design companies and power supply customers will accelerate their gathering in the supply chain. As an emerging technology, the entire ecosystem is relatively rudimentary.There is a lack of mature third-party epitaxial wafer providers, and talents for gallium nitride epitaxy and process are very rare, making it very difficult for companies to open up the production capacity of "epitaxial + manufacturing".In addition, the maturity of a new supply chain needs to be stabilized through shipments, as well as the yield rate to climb, which all take time.  Where do epitaxial wafers come from?At present, "epitaxy + manufacturing" is deeply integrated, and several Foundries that ship in batches have independent epitaxy production capabilities.Therefore, there is currently no independent third-party epitaxial wafer on the market that has shipped large quantities of epitaxial wafers.Therefore, when looking at whether it is the gallium nitride Foundry that is currently maturing or the IDM to be built, the first question that needs to be answered is the solution path and verification time of epitaxy.  Where are the talents for craftsmanship?The process of gallium nitride on silicon requires a full understanding of the shortcomings of the gallium nitride material itself. This understanding is based on a long period of trial and error, resulting in a shortage of core personnel with this ability in the world. Currently, there are some power manufacturers with a silicon-based background that are developing gallium nitride business lines. The key is where the process talents with a gallium nitride background are located.  Iterate after on-site verification,Judging from TSMC's R&D experience, its overall R&D process of epitaxy and process is accompanied by problems discovered in product applications, which repeatedly promotes R&D iterations and brings stability to the entire supply chain. Therefore, without sufficient on-site verification data, whether it is epitaxy or process, it is still far from maturity. In addition, the yield problem caused by mass production needs to be solved. As the company with the strongest process engineering capabilities, TSMC took five or six years from gallium nitride research to stable mass production. Even though the current research and development foundation of the overall industry has made better progress than in the previous period, it is still very difficult to get through the upstream materials and manufacturing links.When we look at "mature" epitaxy and processes, we need to be in awe and have a verification cycle.   continued closureSince silicon-based gallium nitride is an emerging field of semiconductor foundry, from TSMC to domestic Sanan Integration, its gallium nitride production capacity has just started, so once a design company comes out, it will occupy most of the company's production capacity, and the impact of production capacity on orders will have a strong resonance effect:The design company has a large shipment volume → a large demand for production capacity → great Foundry support → the supply chain is guaranteed → the supply chain is more stable → the demand for customer orders is greater → the design company has a large shipment volume → a large demand for production capacity······  
  investment thinking, For a silicon-based gallium nitride company, a long-term and stable upstream supply chain is the key.   If you invest in Fabless, which has the fastest shipments, resources will be concentrated at an accelerated pace, and the risk of being slow is very high.due to various reasons，At present, domestic Fabless mainly relies on domestic Foundry, and domestic Foundry's epitaxy and technology are still in the stage of maturation, and the overall production capacity is small. Currently, the shipment volume of domestic Fabless GaN-on-Si companies is still small, and no one has a strong influence on Foundry's production capacity. Therefore, once a Fabless company can come out and get large-scale orders, it can occupy a very advantageous competitive position by locking in production capacity. The slow Fabless will be at great risk because it cannot obtain production capacity.    The key to investing in an IDM with deep experience in "epitaxy + manufacturing" technology is whether it can be made. If it is made, it will be a scarce resource in the world.It is currently very difficult for IDM to be able to go through the epitaxy and manufacturing of gallium nitride and meet the relevant requirements for yield and reliability, which requires awe. But once it runs through, the first-mover advantage it can bring is also very significant.        03       The technical route depends on technology, but also on ecology.
Gallium nitride transistors conduct electricity through a two-dimensional electron gas formed by the piezoelectric effect at the interface of two materials with different bandgaps (usually AlGaN and GaN). Since the two-dimensional electron gas only conducts electricity with a high concentration of electrons, there is no reverse recovery problem of the silicon MOSFET body diode. This means that when no voltage is applied between the gate and source, the drain and the component of the gallium nitride transistor are conductive, that is, it is a normally-on device. In order to deal with this problem, the industry usually has two solutions: One is to use a cascade (Cascode) structure to achieve turn-off control by connecting a MOS in series; the other is to use a single-tube enhancement mode (P-GaN) structure, that is, adding a P-type gallium nitride epitaxial layer to the gate to achieve turn-off control. The two technical routes have their own advantages and disadvantages when evaluated from a technical perspective, and it is difficult to compare them one by one. The main disadvantage of the single-tube enhancement structure is that the gate reliability is relatively low and the withstand voltage is low. The main disadvantage of the cascade structure is the Emi problem caused by parasitic inductance.  reliability.Gates with different technical routes have different impact resistance margins. The gate drive voltage levels of P-GaN and Cascode are generally about 7v and 18V, and the gate turn-on voltages are generally 6v and 8v. The direct problem is that the gate impact resistance margin of the P-GaN structure is only about 1-2V, which poses a greater challenge to reliability. From a structural perspective, the reliability of Cascode is much higher than that of P-Gan, and it is more suitable for industrial-level applications.  Pressure resistant.Device withstand voltage. Cascode has a relatively higher device voltage rating due to the voltage division of silicon tubes. Currently, P-GaN device voltage withstands are still below 650V. Cascode can reach 1200v, which can meet a large number of industrial-grade voltage requirements.  Electromagnetic interference.Cascode needs to seal the silicon tube and the gallium nitride device together, which brings a large parasitic inductance. If a higher frequency is turned on, the Emi will be larger. In addition, from a process perspective, Cascode and P-GaN require one and two epitaxy growths respectively. The overall process difficulty of Cascode is lower than that of P-GaN. From a driving perspective, Cascode driving is exactly the same as traditional silicon MOSFET driving, and there is no need to consider gate protection too much, and the requirements are relatively lower. From a cost perspective, in terms of wafers, Cascode uses only one epitaxy, so the wafer cost is lower, but the overall packaging cost is higher. Taken together, due to the current large overall shipment volume of P-GaN, the overall cost level is better.  Table 1 Comparison of advantages and disadvantages of GaN device structure  

type	cascade	Enhance
reliability	excellent	inferior
Parasitic inductance	sensitive	insensitive
Pressure resistant	excellent	inferior
power	excellent	inferior
Process difficulty	more difficult	Disaster
Driving difficulty	easy	Disaster

        From a technical perspective, the biggest difference between the two is that the cascade structure temporarily solves industrial-level application problems.Since the reliability and voltage resistance characteristics of the cascade structure are more advantageous, Transphorm, as a representative of the cascade architecture, also released relatively ideal high-voltage device failure rate data at the end of last year. At present, the cascade structure has more advantages in the industrial market. The single-tube enhancement-mode structure is currently improving the reliability of P-GaN structure devices through more precise drive and power solutions, and improving the withstand voltage level through substrate technologies such as QST, but it will take time for the technology to gradually mature.            However, in the development of industrial technology routes, in addition to the technology itself, the ecology of the industry often plays a very important role.Overall, since the industrial chains of the cascade structure Pi and Transphorm are relatively closed, the industrial ecology of the more open single-tube enhanced structure is relatively For a hundred flowers to bloom, currently several Foundries that provide OEM services at home and abroad are following the single-tube enhanced process route. Most of the downstream power management IC manufacturers are also focusing on single-tube enhanced solutions. The entire ecosystem is relatively unified on this technology route. At present, it is expected that the technology iteration and cost optimization of the single-tube enhanced industry chain will continue to be strengthened, and it is expected to become the mainstream technology route in the short to medium term.      04       Application solutions, R&D resources and windows of opportunity
         
Gallium nitride is an emerging technology. Currently, the design and application of overall solutions are still driven by gallium nitride device manufacturers. A number of power management IC manufacturers have gradually emerged to form solutions with driver and control chips as the core.  Table 2 GaN main player types  

category		Representative enterprise	structure
GaN Dominant	GaN	Innosec	GaN
		Gan Systems	GaN
	GaN + driver control	Pi	Controlled drive GaN packaging
		Navitas	Driving GaN single die
Power management leads	Drive+Control	NXP	LLC control
		ON Semiconductor	Direct drive and LLC control
	GaN + driver control	Dongke	Driving GaN to control packaging
		biyiwei	Driver+GaN

Gallium nitride-led and power management-led companies have different advantages. The main advantages of gallium nitride device companies: Master the scarce supply chain resources of gallium nitride devices; Have a better understanding of gallium nitride devices; The time to launch relevant solutions has a first-mover advantage. The main advantages of power management chip companies: Design capabilities of drive and control ICs; Design experience of power supply solutions; ACDC sealing technology capabilities and packaging supply chain resources; Original power supply customer channels.
To make an ideal gallium nitride solution, silicon-based gallium nitride companies must not only make good power devices, but also drive and control IC technology and power supply solutions are also very important.Gallium Nitride on SiliconIt is a power company, and it was also a power supply company in the early stage.Compared with silicon-based solutions, the gallium nitride solution needs to solve problems, such as gate withstand voltage of gallium nitride devices, reliability caused by high frequency, and EMI problems; second, it needs to fully exploit the advantages and fully realize the characteristics of high efficiency and small size of gallium nitride in traditional topology structures. Therefore, the overall solution requires: 1. Fully understand the characteristics of gallium nitride devices; 2. Have strong design optimization capabilities at the solution level. The two need to fully understand each other (they need to understand each other very well), and they need to cooperate and iterate with each other at the R&D level (they need to be able to integrate R&D), in order to come up with a good power solution.Through full collaboration in the research and development of gallium nitride devices, drive circuits and control circuits, we can deliver a good solution.We see that both Pi and Navitas are focusing on solving the above problems internally, and domestic silicon-based gallium nitride companies are also building their own R&D ecosystem and that of power management IC companies. How domestic GaN-on-Si companies can achieve R&D collaboration on power devices and power ICs, and whether they have strong power solution team reserves, technical experience, and in-depth R&D partners, requires focus. In addition, the current mainstream solutions include Pi, Navitas, and Innosec. The integration level of the chips of the three companies is gradually decreasing. Among them, Pi uses a controller, driver and GaN device package, Navitas uses a single die integration of driver and gallium nitride device, and Innosec uses a gallium nitride single tube, which directly brings three different solutions. At present, the solutions of mainstream power supply manufacturers focus on the above three types. Therefore, the resistance to embracing mainstream solutions is relatively small. Differentiated solutions require self-built systems.      05       Digressions and Conclusions

There are currently some views like this: First, the gallium nitride track is not profitable; second, the technology threshold for gallium nitride is low and there is a lot of domestic production capacity under construction. Regarding economics, when the overall output value increases, the yield rate and capacity utilization rate of each company can reach a higher level. As an imperfectly competitive market, this track is likely to be more profitable than silicon-based power in the short to medium term. Regarding production capacity, there are currently many silicon-based gallium nitride production capacity projects announced by the media. If you are interested, you can investigate them one by one. How many actual effective production capacities are there? You can also evaluate how many production capacities are expected to be effective in the next two to three years. I won’t go into details here.  GaN-on-silicon is an emerging field. It is currently very difficult to develop the technology from devices to solutions and establish a stable and economical upstream supply chain. Such companies and teams are also very scarce.  

As long as it can solve the above problems well, it will be a rare silicon-based gallium nitride company that is not limited to business models, corporate backgrounds, wafer sizes, and technical routes for epitaxy and devices.

Disclaimer: This article is reprinted from "Three Meters Deep". This article only represents the author's personal views and does not represent the views of Sacco Micro and the industry. It is only reprinted and shared to support the protection of intellectual property rights. Please indicate the original source and author for reprinting. If there is any infringement, please contact us to delete it.

Company phone number: +86-0755-83044319
Fax/FAX: +86-0755-83975897
Email: 1615456225@qq.com
QQ: 3518641314 Manager Li

QQ: 332496225 Manager Qiu

Address: Room 809, Building C, Zhantao Technology Building, No. 1079 Minzhi Avenue, Longhua New District, Shenzhen