As seen from the market, components of power semiconductors are mostly used in industrial fields, including motor control, rail transit, wireless power supply, energy control, and smart grid, which occupy more than 30% in the long run that is expected to arrive at 35% in 2021, followed by automotive applications at 29% that will gradually expand the automotive and high voltage MOSFET markets alongside the development of NEVs and EVs. In addition, consumer electronics also account for 18% of the elevated demand for notebooks, smartphones, wearable, and quick chargers. Communication and computing each take up 10% and 7%.

Demand for Consumer Electronics Applications: Quick Charging

The faster transmission of 5G smartphones compared to that of 4G requires additional radio frequency components, thus an increase in power consumption and the speed of battery drainage becomes inevitable. Brands have been releasing USB-PD quick charging products that are currently most adopted with Type-C under the rising requirements of consumers in charging efficiency, and a support of transmission voltage from higher specifications will require an integration with synchronous rectification MOSFET that is essential in adjusting optimization, as well as an increase in the quantity of MOSFET. In terms of materials, small chargers of high power density are steadily becoming market mainstream amidst the development and popularization of GaN technology, as well as the ever-changing market of USB Type-C PD chargers, and GaN that has a low calorific value and small dimension has become the optimal material for MOSFET pertaining to quick charging.

Demand for Communication Applications: 5G Base Station

5G base stations, adopted with the Massive MIMO technology, and require multi-channel architectures such as 32 channels (32T32R) and 64 channels (64T64R), are the core equipment of 5G network, and an increase in channel density will also lead to an increment in power consumption and cost for 5G base stations at the same time. 5G base stations consume double the power than 4G, while the demand for lowering power consumption has risen the demand for low depletion and high thermostability. In addition, the establishment of 5G network has elevated the scale of data centers and cloud services, while the installation of servers has also stimulated the demand for power management modules such as AC/DC converters and DC/DC converters. Hence, communication MOSFET is now one of the major demands.

Demand for Automotive Applications: NEV

The transition of the existing automotive industry that marches from traditional fossil fuel vehicles to NEVs requires extra power semiconductor components such as MOSFET to operate synergistically. For traditional fossil fuel vehicles, various power supply components are powered by the battery, which usually comes in either 24V or 12V. The voltage for the power battery of NEVs is usually 336V or 384V and can go up to 580-600V for large electric passenger cars, which is why power semiconductor components are necessary between high and low voltage systems of NEVs to implement voltage adjustments and achieve flowing of current between two systems that allow each electronic component to function.

The electronic components that complement the battery system of a NEV operate at different voltage levels, so voltage conversion is required as electric power moves through different components. As a MOSFET continuously switches, it gradually raises or lowers the voltage. Hence, the important considerations in MOSFET designs include current strength and withstand voltage. Additionally, different applications have their own design needs with respect to switching frequency, switching noise, oscillation damping, and DPM.

As the number of electronic components in a car increases, the number of automotive applications grows for power semiconductor components such as MOSFETs. Originally, a car with a traditional ICE has at least 90 MOSFET chips. As for NEVs, the number of MOSFET chips per vehicle is usually around 200 but can reach up to around 400 for high-end models. With more functions and features being added into a NEV, the chip number is expected to rise further in the future.

Demand for Industrial Applications: Automation and Charging Piles for NEVs

The industrial segment of the market for power semiconductor components has the widest range of applications. Most kinds of industrial equipment contain MOSFETs. At the same time, many manufacturing and processing sectors are adopting automation technologies in order to shorten process cycle time, reduce production costs, and minimize equipment idling. Besides the ongoing trend toward Industry 4.0, the shock of the COVID-19 pandemic has accelerated the progress in industrial automation during recent years because manufacturers have been compelled to find ways to keep their production lines running with limited manpower.

Compared with traditional Si-based MOSFETs and IGBTs, SiC-based MOSFETs have significant advantages in the industrial segment of the power semiconductor market because they are able to withstand higher voltages, perform faster switching, and offer lower on-resistance. Furthermore, the adoption of SiC can contribute to a reduction in power consumption and a more compact system. Hence, SiC-based solutions have become the focus of product development for power component suppliers.

Regarding charging piles for NEVs, these are considered industrial equipment and therefore have to conform to the certifications of the major industry bodies. Playing a key role in the proliferation of NEVs, charging piles are being developed to provide a shorter charging period as well as higher-power charging in different scenarios. In this application field, power semiconductor components are essential to electric power conversion. Electromechanical devices including AC/DC converter and DC/DC converter are the key parts of a charging pile because they perform voltage and frequency conversions. Due to the demand for an ever shorter charging period, power components have to achieve a higher standard in terms of withstand voltage. Hence, SiC-based MOSFETs are trickling into the charging pile market as component suppliers develop solutions that minimize both thermal resistance and switching loss that occurs during high-frequency switching.

（Image credit: Unsplash）

Following the peak period of production in 3Q21, the supply of DRAM will likely begin to outpace demand in 4Q21, according to TrendForce’s latest investigations (the surplus of DRAM supply is henceforth referred to as “sufficiency ratio”, expressed as a percentage). In addition, while DRAM suppliers are generally carrying a healthy level of inventory, most of their clients in the end-product markets are carrying a higher level of DRAM inventory than what is considered healthy, meaning these clients will be less willing to procure additional DRAM going forward. TrendForce therefore forecasts a downward trajectory for DRAM ASP in 4Q21. More specifically, DRAM products that are currently in oversupply may experience price drops of more than 5% QoQ, and the overall DRAM ASP will likely decline by about 3-8% QoQ in 4Q21.

PC DRAM prices are expected to decline by 5-10% QoQ as market demand for notebook computers weakens

Although WFH and distance learning applications previously generated high demand for notebook computers, increasingly widespread vaccinations in Europe and North America have now weakened this demand, particularly for Chromebooks. As a result, global production of notebooks is expected to decline in 4Q21, in turn propelling the sufficiency ratio of PC DRAM to 1.38%, which indicates that PC DRAM will no longer be in short supply in 4Q21. However, PC DRAM accounts for a relatively low share of DRAM manufacturers’ DRAM supply bits, since these suppliers have allocated more production capacities to server DRAM, which is in relatively high demand. Hence, there will unlikely be a severe surplus of PC DRAM in 4Q21. It should also be pointed out that, on average, the current spot prices of PC DRAM modules are far lower than their contract prices for 3Q21. TrendForce therefore expects an imminent 5-10% QoQ decline in PC DRAM contract prices for 4Q21, with potential for declines that are even greater than 10% for certain transactions, as PC OEMs anticipate further price drops in PC DRAM prices in the future.

Server DRAM prices are expected to decline for the first time this year, by 0-5% QoQ due to high client-side inventory

CSPs in North America and China currently carry more than eight weeks’ worth of server DRAM inventory, with some carrying more than 10 weeks’ worth of inventory, as they procured massive amounts of server DRAM in the previous two quarters to avoid shipment issues with whole server units caused by component shortages. In view of this aggressive procurement effort, the overall demand for server DRAM has gradually slowed, although certain Tier 2 data centers are still procuring server DRAM to make up for previous gaps. As server DRAM buyers continue to gravitate towards destocking their server DRAM inventory in 4Q21, demand will likely fall short of the previous quarters. Furthermore, due to long lead times for certain key components, shipment of whole servers is also expected to undergo quarterly declines. On the supply side, the three major DRAM suppliers (Samsung, SK hynix, and Micron) reallocated some of their production capacity for mobile DRAM to server DRAM in early 2Q21, and this reallocated capacity is expected to gradually begin outputting server DRAM in 4Q21. Given the slowdown in server DRAM demand, contract negotiations for server DRAM procurement in 3Q21 lasted until early August. Although server DRAM contract prices underwent a 5-10% QoQ increase in 3Q21 due to suppliers’ best attempts during contract negotiations, further price hikes going forward are unlikely. TrendForce expects server DRAM prices to undergo a decline for the first time this year in 4Q21 with a QoQ drop of 0-5%.

Mobile DRAM prices are expected to remain relative unchanged from 3Q21 levels despite a possible price drop ahead of time at the end of the year

In light of fluctuations in the COVID-19 pandemic, the global demand for smartphones and the supply of smartphone components are both still at the risk of experiencing declines. In addition, after smartphone brands revised down their production targets at the end of 2Q21, brands and distributors alike have been facing the pressure of high smartphone inventory levels. In response to factors such as pandemic-related uncertainties and declines in mobile DRAM prices for 2022, smartphone brands will slow down their mobile DRAM procurement and prioritize inventory reduction instead. Hence, bit demand for mobile DRAM will decline even further in 4Q21. On the whole, given the uncertain state of the pandemic in the coming winter, smartphone brands will adopt a more conservative attitude towards both smartphone production and component procurement in 4Q21. As a result, even if DRAM suppliers are willing to lower mobile DRAM prices, such an effort will only result in limited sales growths. In addition, mobile DRAM still lags behind other DRAM product categories in terms of profitability, meaning a drop in mobile DRAM prices is unlikely. Taking these factors into account, TrendForce expects prices of discrete DRAM, eMCP, and uMCP to mostly hold flat in 4Q21 compared with 3Q21.

It should be noted that, by the end of the year, DRAM suppliers may potentially start supplying mobile DRAM at 1Q22 prices ahead of time, primarily for two reasons: First, DRAM suppliers will be faced with revenue performance pressures at the end of the year; second, smartphones and DRAM suppliers will enter into new LTAs (long term agreements) for 2022. These factors are expected to impact mobile DRAM ASP for 4Q21 and bring about a price drop ahead of time.

Graphics DRAM contract prices are expected to decline by 0-5% QoQ due to excess supply

Market demand for discrete graphics cards and notebook graphics cards still remains due to the stable market for commercial notebooks and the resurging cryptocurrency mining market, which saw cryptocurrency prices rebounding from rock bottom levels within the past two months. However, severe issues with the availability of components in the graphics card supply chain currently present the most significant bottleneck in graphics card production. In particular, components such as driver IC, PMIC, and other peripheral components are all in shortage, while graphics DRAM is in relative oversupply compared to these other components. Graphics card manufacturers are therefore revising down their graphics DRAM procurement. Consequently, even though DRAM suppliers have not significantly increased their graphics DRAM production, demand from the purchasing end will remain sluggish until the shortage of other components is resolved. Demand for graphics DRAM will unlikely see a resurgence before the end of 2021. On the supply side, the three major DRAM suppliers are primarily focused on GDDR6 for their current graphics DRAM production. As well, graphics card demand from the cryptocurrency mining market is generally aimed at newer graphics cards that feature GDDR6 memory. Accordingly, both production and sales of GDDR5 memory are relatively weak, and this bearish trend is especially reflected in spot prices. As spot prices are the first to enter a downturn, and the aforementioned market conditions lead to sluggish procurement activities, graphics DRAM prices are in turn expected to plummet from previous levels in 4Q21, although this decline is projected at a minor 0-5% QoQ owing to DRAM suppliers’ efforts to keep prices constant.

QoQ decline of DDR4 Consumer DRAM prices is expected to be among the highest drops, at 5-10% as procurement activities decelerate

Gradual easing of lockdowns in Europe and North America has led to a decline in consumer spending on home entertainment applications. This, along with the severe shortage in electronic components, has adversely affected the demand for consumer electronics, such as TVs, STBs (set-top boxes), and networking devices, as well as industrial-use products, thereby also reducing the procurement demand for consumer DRAM. On the other hand, while DRAM suppliers were in the process of transitioning from DDR3 manufacturing to other products, the massive price hike of DDR3 products in 1H21 led DRAM suppliers to slow this transition. Even so, certain market conditions are now placing downward pressure on DDR3 prices, so next year the three major suppliers may potentially speed up the transition of mature DDR3 manufacturing to other products, such as CMOS image sensors or other logic ICs, instead. As server and PC manufacturers’ DRAM inventory level rises, contract prices of those DRAM products will likely decline in 4Q21. Thus, given that the movement of DDR4 consumer DRAM prices is highly correlated with PC DRAM and server DRAM and has been trending relatively high, DDR4 consumer DRAM prices are expected to decline by 5-10% QoQ in 4Q21. Likewise, although the supply of DDR3 consumer DRAM has been gradually decreasing, DDR3 consumer DRAM prices will also undergo an overall decline, particularly for 4Gb chips. DDR3 consumer DRAM prices are expected to decline by 3-8% QoQ in 4Q21.

For more information on reports and market data from TrendForce’s Department of Semiconductor Research, please click here, or email Ms. Latte Chung from the Sales Department at lattechung@trendforce.com

In this press release, TrendForce details 10 major trends that are expected to take place across various segments in the tech industry, as follows:

Micro/Mini LED display development will revolve around active matrix solutions

A substantial number of technical bottlenecks in Micro LED development will still persist in 2022. While Micro LED manufacturing costs are expected to remain sky-high due to these bottlenecks, companies have not shown decreased willingness to participate in all segments of the Micro LED supply chain. On the contrary, these companies are actively expanding their respective production lines. Regarding the development of self-emitting Micro LED display products, TVs represent one of the major directions of mainstream Micro LED development, primarily because TVs, compared to IT products, have a relatively low technological barrier of entry. In other words, Micro LED TVs are easier to develop than are other Micro LED display products. For instance, after releasing a 110-inch commercial passive matrix Micro LED display, Samsung will likely continue to develop 88-inch (and under) consumer-grade active matrix Micro LED TVs. This extension of Micro LED technology from the large-sized commercial display segment to the household-use segment by Samsung is in turn indicative of the overall expansion of the Micro LED market.

Regarding display products equipped with Mini LED backlights, brands have been raising the number of Mini LED chips used per panel in an attempt to boost the specs of their display products and pursue 1:1,000,000 high contrast ratios that are comparable to OLED displays. As a result, Mini LED backlight panels’ LED chip consumption is more than 10 times higher compared to traditional LED backlight panels, in addition to the fact that Mini LED backplane manufacturing requires SMT equipment with a higher degree of accuracy and production capacity. While Mini LED backlights are primarily based on passive matrix solutions, they will move towards active matrix solutions going forward, with a corresponding surge in Mini LED chip consumption. Hence, the performance and capacity of SMT equipment will also become one of the key criteria in brands’ selection of potential supply chain partners.

More advanced AMOLED technology and under-display cameras will usher in the next stage of smartphone revolution

As the supply of and production capacity for AMOLED panels continue to rise, AMOLED technology has also become increasingly mature. Leading suppliers are still attempting to tack on additional functions and improved specs to their AMOLED panels in order to not only raise said panels’ added values, but also maintain the competitive advantages of the suppliers themselves. The primary value added to AMOLED panels in 2022 will likely continue to be the ever-improving foldable designs, which will feature optimized weight reduction and power efficiency. Apart from mainstream foldable phones that can unfold to reach tablet-like sizes, clamshell-like designs such as flip-up and flip-down smartphone bodies will also emerge as a form factor that more closely resembles the smartphones currently in use. Furthermore, retail prices of foldable phones are expected to generally fall within the price bands of mainstream flagships, thereby generating sales growths for the upcoming foldable models. Other foldable designs, including form factors with even more folds or rollable form factors, are expected to enter production within the near future. TrendForce expects foldable phones to reach a penetration rate of more than 1% in 2022 and 4% in 2024. LTPO panels, on the other hand, are an effective solution to power consumption issues arising from the adoption of 5G technology and high refresh rate displays. Hence, LTPO panels will likely gradually become the mainstream display panel for flagship smartphones. After two years of development and adjustments, under-display camera modules will finally make their appearance in various brands’ flagship models and enable the creation of smartphones with true full-screen displays.

The foundry industry welcomes the arrival of 3nm process technology courtesy of TSMC’s FinFET and Samsung’s GAA technologies

As semiconductor manufacturing processes gradually approach physical limits, chip development must now turn to either “changes in transistor architecture” or “breakthroughs in back-end packaging technology or materials” in order to achieve faster performances, reduced power consumption, and smaller footprints. After incorporating EUV lithography at the 7nm node in 2018, the semiconductor industry will welcome yet another revolutionary process technology in 2022 – the 3nm node. More specifically, TSMC and Samsung are expected to announce their respective 3nm process technologies in 2H22. While the former will continue to adopt the FinFET architecture that it has been using since the 1Xnm node, Samsung will for the first time utilize its own implementation of GAAFET, called MBCFET (multi-bridge channel field-effect transistor) for its 3nm process technology.

In contrast with the FinFET architecture, in which the gate makes contact with the source/drain channel on three sides, the GAAFET architecture consists of a gate that surrounds the nanowire or nanosheet channel on four sides, thus increasing the surface area of contact. The GAAFET design significantly reduces leakage currents by giving the gate a greater degree of control over the channel. Regarding possible applications, the first batch of products mass produced at the 3nm node in 2H22 is expected to primarily be HPC and smartphone chips since these products place a relatively high demand on performance, power consumption, and chip compactness.

While DDR5 products gradually enter mass production, NAND Flash stacking technology will advance past 200 layers

The three dominant DRAM suppliers (Samsung, SK Hynix, and Micron) will not only gradually kick off mass production of next-gen DDR5 products, but also continue to increase the penetration rate of LPDDR5 in the smartphone market in response to demand for 5G smartphones. With memory speed in excess of 4800Mbps, DDR5 DRAM can massively improve computing performances via their fast speed and low power consumption. As Intel releases its new CPUs that support DDR5 memory, with Alder Lake for the PC segment, followed by Eagle Stream for the server segment, DDR5 is expected to account for about 10-15% of DRAM suppliers’ total bit output by the end of 2022. Regarding process technologies, Samsung and SK hynix will kick off mass production of 1 alpha nm products manufactured with EUV lithography. These products’ market shares will likely increase on a quarterly basis next year.

Turning to NAND Flash products, their stacking technologies have yet to reach a bottleneck. Hence, after 176L products entered mass production in 2021, suppliers will continue to migrate towards 200L and above in 2022, although these upcoming products’ chip densities will remain at 512Gb/1Tb. Regarding storage interfaces, the market share of PCIe Gen4 SSDs will likely skyrocket in the consumer PC segment next year. In the server segment, as Intel Eagle Stream CPUs, which support PCIe Gen 5, enter mass production, the enterprise SSD market will also see the release of products that support this interface. Compared to the previous generation, PCIe Gen 5 features double the data transfer rate at 32GT/s and an expanded storage capacity for mainstream products at 4/8TB in order to meet the HPC demand of servers and data centers. Additionally, the release of PCIe Gen 5 SSDs is expected to quickly raise the average data storage capacity per server unit.

Regarding the server market, flexible pricing schemes and diverse services offered by CSPs have directly propelled the cloud service demand of enterprises in the past two years. From the perspective of the server supply chain, the predominant business model has gradually transformed from traditional server brands to ODM Direct, meaning that traditional server brands will see fundamental, structural changes, such as providing colocation servers or full-service cloud migration support, in their business models. This shift also means that enterprise clients will come to rely on more flexible pricing schemes and diverse risk mitigation measures in response to an uncertain global environment. In particular, while the pandemic accelerated changes in work and everyday life in 2020, hyperscalers are expected to account for nearly 50% of total demand for servers in 2022. In addition, the growth in ODM Direct server shipment is expected to surpass 10% YoY as well.

Mobile network operators will undertake more trial projects for 5G SA network slicing and low-latency applications

Mobile network operators have been actively release 5G SA (standalone) networks as the core network powering various services around the world, in turn accelerating the build-out of base stations in major cities, diversifying their network services (via network slicing and edge computing), and delivering end-to-end networks with a high degree of quality assurance. Moving to 2022, applications that are at the intersection of 5G, massive IoT, and critical IoT will emerge in response to enterprise demand. These applications, including light switches, sensors, and thermostats used in smart factories, involve the combination of network endpoints and data transmission. In particular, critical IoT applications include smart grid automation, telemedicine, traffic safety, and industrial automation, whereas critical IoT use cases within the context of Industry 4.0 include asset tracking, predictive maintenance, FSM (field service management), and logistics optimization.

Now that the pandemic has forced enterprises to engage in digital transformation and brought changes to the general public’s lifestyles, the importance of 5G deployment has become increasingly apparent. Private 5G networks, openRAN, unlicensed spectrums, and mmWave developments have also generated a diverse ecosystem that ranges from traditional mobile network operators to other emerging service providers, including OTT media service providers, CSPs, social media, and online businesses. In the future, mobile network operators will likely actively expand their enterprise 5G applications. For instance, O2’s 5G-ENCODE project explores new business models in industrial 5G networks, while Vodafone is collaborating with the MFM (Midlands Future Mobility) consortium to test networks for autonomous vehicles.

Satellite operators will compete over the low-earth orbit satellite market, with 3GPP now supporting non-terrestrial networks

3GPP recently announced that Release 17 Protocol Coding Freeze will take place in 2022. Release 17 represents the first time 3GPP has incorporated NTN (non-terrestrial network) communications into its releases and therefore marks a significant milestone for both the mobile communications industry and the satellite communications industry. Prior to this, mobile communications and satellite communications had been two separate, independently developing industries. That is why companies working across the two industries in the upstream, midstream, and downstream supply were generally different as well. After 3GPP includes NTN in its upcoming release, the two industries are likely to generate more opportunities for collaboration and co-create brand new innovations. Regarding the deployment of LEO (low earth orbit) satellites, US-based SpaceX has applied to launch the highest number of satellites among all satellite operators. Other major operators include Amazon, UK-based OneWeb, Canada-based Telesat, etc. Region-wise, US operators account for more than 50% of all satellites launched. Not only do LEO satellites have the advantage of signal coverage that is unaffected by geographical features, such as mountainous regions, oceans, and deserts, but they are also able to synergize with the 5G network. The ability of LEO satellites, as part of the NTN, to enhance 5G communications makes them a crucial component in the 3GPP Release 17. TrendForce therefore forecasts an increase in global satellite revenue in 2022.

While smart factories are among the first to leverage digital twins, IoT technologies are expected to become the backbone of the metaverse

The new normal that emerged in the wake of the COVID-19 pandemic continues to propel demand for contactless devices and digital transformations. As part of this evolution, IoT development in 2022 will likely focus on CPS (cyber-physical systems), which combines 5G, edge computing, and AI technologies to extract and analyze valuable information from vast data streams for the purpose of smart automation and prediction. A current example of CPS applications is the digital twin, used for such verticals as smart manufacturing and smart cities; while CPS integration for the former facilitates design, testing, and manufacturing simulations, the latter make use of CPS to monitor significant assets and assist in policymaking. Now that industrial realities have become more complex, and the interplay between usage cases and equipment have increasingly demanded attention, digital twins will subsequently be deployed to a wider range of applications. Paired with 3D sensing, VR, and AR capabilities, IoT-based metaverse will likely emerge as a smart, complete, real-time, and safe mirror to the physical world, and the first application of IoT-based metaverse is expected to be smart factories. Ultimately, technological innovations in data collection (including visual, auditory, and environmental data via sensors), data analysis (via AI platform integration), and data integrity (via blockchains) will also emerge as a result of IoT development.

AR/VR equipment manufacturers aim to deliver fully immersive experiences via integration of additional sensors and AI processing

The COVID-19 pandemic has fundamentally changed the way people live and work. For enterprises, the pandemic not only accelerated their pace of digital transformation, but also increased their willingness to integrate emerging technologies into their existing operations. For instance, AR/VR adoption for applications such as virtual meetings, AR remote support, and virtual design has been on the rise recently. On the other hand, companies will likely focus on various remote interaction functionalities in virtual communities and online games as an important AR/VR market segment. TrendForce therefore believes that the AR/VR market will expand by a considerable margin in 2022 due to the falling retail prices of AR/VR hardware as well as the growing adoption of such hardware for various use cases. Furthermore, the market will also continue to pursue more realistic AR/VR effects, such as applications that feature more realistic images constructed by software tools or the creation of virtual responses from real-world data assisted by either AI processing or the integration of various sensors. For instance, eye-tracking functionalities will become an optional feature of consumer products released by Oculus and Sony. Apart from these examples, AR/VR solutions may even evolve to the point where they are able to provide partial haptic feedback to the user through controllers or other wearable devices in order to deepen user immersion.

A natural extension of autonomous driving technology, automated valet parking is set to resolve drivers’ pain points

As part of autonomous driving technology’s implementation aimed at improving everyday life, AVP (automated valet parking), an SAE level 4 driverless parking service, is expected to become an important optional function of high-end vehicles beginning in 2022. Relevant international standards are currently being drafted and are expected to facilitate the adoption of AVP going forward. However, since AVP systems differ according to vehicle specifications, they are subject to various restrictions related to driving conditions, including fixed/unfixed routes and private/public parking spaces, while parking lot conditions such as wireless network connectivity and the comprehensiveness of traffic markings can also affect the viability of AVP. The distance between people and the vehicle during AVP use, on the other hand, is governed by domestic laws. In view of automakers’ diverse technological roadmaps, AVP parking routes are generated by either local computing on the vehicle end or cloud computing, the latter of which requires sufficient network connectivity in order to function. The former is therefore expected to see usage in a wider variety of use cases. Alternatively, some vehicles may be equipped with both computing solutions. With other such factors as V2X and high-definition maps affecting the range of AVP applications, TrendForce expects an increasing number of different AVP solutions to be under development at the moment.

The third-generation semiconductor industry will move towards 8-inch wafers and new packaging technologies while expanding in production capacity

The gradual phasing out of ICE vehicles by various governments across the 2025-2050 period is set to both accelerate the pace of EV sales and increase the penetration rate of SiC and GaN devices/modules. Energy transition activities worldwide as well as the rapid growth of telecom applications such as 5G technology have also led to a persistent bull market for third-generation semiconductors, resulting in strong sales of SiC and Si substrates. However, as current efforts in substrate production and development are relatively limited, suppliers are able to ensure a steady yield of SiC and GaN substrates only by manufacturing them with 6-inch wafers. Such a limitation has, in turn, led to a long-term shortage in foundries’ and IDMs’ production capacities.

In response to this quandary, substrate suppliers, including Cree, II-VI, and Qromis, are now planning to not only expand their production capacities in 2022, but also migrate their SiC and GaN production to 8-inch wafers, in hopes that these plans will gradually alleviate the prevailing shortage in the third-generation semiconductor market. On the other hand, foundries such as TSMC and VIS are attempting to shift to 8-inch wafer fabrication for GaN on Si technology, while major IDM Infineon is releasing products based on the latest CoolSiC MOSFET, delivering trench designs that enable significant power efficiency for semiconductor devices. Finally, telecommunication component provider Qorvo has also released a new GaN MMIC copper flip chip packaging architecture for military applications.