The Central Finance Think Tank is committed to becoming the chief staff officer of sustainable development of enterprises, jointly building and sharing the cross-border integration of "government, business, media and consumption" with enterprises, win-win economic development with Chinese characteristics in the new era, and empowering enterprises with high-quality and sustainable development. Please discuss with us the gains and losses on the road of enterprise development, and exchange ideas with us.
Central Finance Think Tank
1. The first year of intelligence is opened,
Electrification of the second half of the competition
The policy sets the tone and the electrification of automobiles is a key part of the realization of carbon neutral carbon peaking. The "14th Five-Year Plan" and the outline of the 2035 long-term goals put forward that China will strive to achieve "carbon peak" in 2030 and "carbon neutrality" in 2060, and the energy revolution marked by "carbon peaking and carbon neutrality" has become a global consensus, and new energy vehicles as an important path for energy conservation and emission reduction are becoming the main battlefield of energy transformation.
Policy-driven towards market-driven, new energy vehicles to accelerate penetration. The trend of electrification has been determined, and in 2021, with the gradual retreat of the impact of the new crown epidemic and the increase in high-quality supply, new energy vehicles will accelerate penetration. According to the data of China Automobile Association, in 2021H1, China's new energy vehicles achieved sales of 1.206 million units, an increase of 201.5% year-on-year, driving the penetration rate of new energy vehicles to climb steadily, reaching 9.3% in 2021H1, an increase of 4.1pct over 5.2% in 2020.
Benchmarking the smart phone industry, the new energy automobile industry is moving from the introduction period to the growth period. Benchmarking the smart phone industry, high-quality supply drives new energy vehicles from the introduction period to the growth period. Assuming that the penetration rate of global and Chinese new energy vehicles in 2025 is 20% and 25% respectively, it is expected that global and Chinese new energy vehicle sales will be 13.815 million and 6.389 million units, respectively, with a compound growth rate of 36% from 2021 to 2025.
The trend of electrification has been determined, and intelligence has opened the second half of competition. In the future, the relationship between electric and intelligent will be more complementary and complementary in both directions:
1) The best carrier for electrification system intelligence: First, the vehicle EE architecture has moved from distributed to centralized (domain integration central computer), and the software OTA has given the vehicle the possibility of continuous upgrading and evolution; second, the energy consumption can meet the multiple growth of controller power consumption brought about by the increase in chip computing power; third, the security has the advantages of functional safety and information security to ensure that the system is robust and highly reliable;
2) The important promotion of intelligent electrification: In the early stage of the development of new energy vehicles, the problem of mileage anxiety greatly affects consumers' car purchase decisions, but with the breakthrough of battery technology and the improvement of vehicle mileage, the degree of impact of mileage on consumer car purchases (especially in some first- and second-tier cities with relatively complete infrastructure) is no longer obvious. In order to seek product differentiation selling points, new and old forces began to lock their eyes on intelligence, and the scheduling period accelerated its own electrification process through intelligence.
Automotive intelligence mainly reflects the two major categories of intelligent driving and intelligent cockpit:
1) Intelligent driving: data accumulation as the core. Including progressive and leapfrog two major implementation paths, of which the progressive to the traditional car companies and new car-making forces as the representative, to achieve from L0 to L5 automatic driving step by step, is currently in the ADAS accelerated penetration, L3 automatic driving initial introduction stage, Tesla is relatively leading, the next step of the expansion direction will be mainly reflected in the urban road application scenario expansion and functional continuity improvement; leapfrog to technology start-ups as representatives, typical such as Google Waymo, Baidu Apollo, etc., the scheduling step in place to achieve high-order self Moving driving, in general, "transporting goods faster than transporting people, low speed faster than high speed", specific scenarios such as ports, mines, etc. The relatively low difficulty of technical implementation is expected to meet the first landing, followed by high-speed freight, Robotaxi last. With the accelerated accumulation of data brought about by the efforts of car companies and technology Internet giants, breakthroughs in intelligent driving technology are expected to be faster than expected.
2) Intelligent cockpit: ecological construction as the core. As another important component of intelligent vehicles in addition to intelligent driving, the development of intelligent cockpits mainly benefits from the upgrade of the vehicle EE architecture, that is, from traditional distribution to the current stage of domain centralization, thereby realizing soft and hard decoupling and efficient interaction between multiple screens. Compared with intelligent driving, the implementation of smart cockpits is less difficult and more cost-effective, while benefiting from the growing demands of consumers for car comfort and safety and the gradual migration of consumer electronics application scenarios, which is expected to meet faster penetration in the short term. From the composition point of view, the smart cockpit is highly similar to the PC, the core lies in the operating system, and the technology Internet giant is expected to fully benefit from its strong ecological construction capabilities.
Under the opportunity of intelligent electric power, the century-old automobile industry order has been reshaped. Intelligent electric vehicles span the automotive, electronics, computers, IoT and other fields, spawning trillion-level market space, and software and services open up new profit channels, not only for the comprehensive transformation of traditional car companies, but also for the new forces of car manufacturing, actively enter the technology Internet enterprises to bring historic opportunities: the new force brand fully use the Internet thinking, select tracks to achieve a breakthrough, Huawei positioned intelligent electric vehicle incremental components provider, Baidu, Xiaomi has officially announced the next car, DJI, OPPO, Skyworth, etc. are also actively laid out. Multi-party cooperation to promote the vigorous development of the industry. At the same time, with the development of electric intelligence, the reshaping of the relationship between the traditional automobile industry chain, the emergence of Tier0.5 suppliers, the advancement of the industrial chain status, the vast space for domestic substitution of independent parts suppliers, and the improvement of technical capabilities, it is expected to rely on the Chinese market to rise as a global parts leader through globalization.
2. Electrification:
Intelligent best carrier,
Policy-driven accelerated development
The fundamental driving force of China's new energy vehicles has been driven by policy-driven transformation and supply-driven, the policy has shifted from subsidy policy to double integral policy, from "carrot" to "stick": the subsidy policy has declined sharply since 2019 and will end in 2023, and the double integration policy of taking over the baton will also be adjusted again in 2019, from encouraging the increase in mileage to encouraging energy-saving levels. From a global perspective, China's new energy transformation is at the forefront and has entered the stage of market drive. Foreign mainstream countries such as Europe, the United States and Japan, most countries' new energy penetration rate is at a low level, the policy has just entered the policy strong subsidy period, with the help of strong subsidies by policy, foreign new energy vehicle sales will enter a stage of rapid development in the future. Electrification as the base of intelligent development technology, the improvement of electrification and electrification capabilities can better support the future evolution of the vehicle's electronic and electrical architecture to the centralized, and promote the coordinated development of associated technologies such as wire control chassis, software algorithms, and vehicle Ethernet.
2.1. Overview:
Policy-driven to market-driven,
The industry is accelerating its penetration
The policy subsidies in the new energy automobile industry have achieved remarkable results, and the technical singularity has arrived. Through the early policy subsidies, domestic new energy vehicle technology has become competitive compared with fuel vehicles. Technically, the power battery as an important component of the new energy source vehicle, energy density continues to increase, the cost is decreasing year by year, from the world's leading battery supplier Ningde times power battery unit price, from 2.90 yuan / Wh in 2014 to 0.96 yuan / Wh in 2019, further decline to 0.65 yuan / Wh in 2020, is expected to decline to 0.63 yuan / Wh in 2023.
Hybrid technology can be used as a concentrated expression of the strength of Sandian technology. The technology by the engine, generator, electric machine, clutch, battery, variable speed transmission device, drive wheel composition, and through the hybrid strategy control, so that the engine always runs in the highest thermal efficiency range, of which the battery as a power reservoir, with the advantage of peak shaving valley filling, reduce fuel consumption, the current hybrid technology to BYD DMi, great wall DHT, etc. as representative, has been through highly integrated, customized fusion technology, to obtain high performance, low cost advantages, it is estimated that the cost of DMi is currently only higher than the general fuel vehicle 1 About 10,000 yuan, basically reached the level of fuel vehicle parity.
The decline of the subsidy policy accelerates the survival of the fittest in the industry, and the double integration policy accelerates the transformation of China's passenger car market to new energy. The 2017 version of the double credit policy failed to fully promote the high-quality development of new energy vehicles, and the new policy was formulated in 2019 to highlight energy conservation and efficiency.
Impact - Industry: Promote the transformation of the industry to low fuel consumption and new energy vehicles as a whole, and it is expected that the output of new energy source passenger cars will exceed 3.6 million units in 2023. In addition to meeting the NEV credit ratio requirements, passenger car companies also need to generate additional NEV positive points to deduct CAFC negative points, according to our conservative calculations, the policy requires that the corresponding new energy passenger car production in 2023 exceeds 3.6 million.
Impact - car companies: improve the energy saving of traditional fuel vehicles, while increasing investment and investment in high-performance new energy vehicles, expanding the production and sales of new energy need to pay attention to the decline in power consumption levels. Specifically:
Improve the energy saving of traditional fuel vehicles: The 2019 version of the double integral revision draft will include low fuel consumption passenger cars in the assessment, and its scope includes oil-electric hybrid with obvious fuel saving effect, and independent accounting for new energy and traditional energy source passenger cars, prompting car companies, especially independent brands, to increase research and development to improve the energy saving level of traditional fuel vehicles;
Increase investment and investment in high-performance new energy vehicles: The 2019 edition of the double integral revision draft has raised the threshold for pure electric integral calculation, and the impact on the joint venture brand is greater than the independent brand of the integral surplus, forcing the joint venture car companies to respond more actively and accelerate the launch of more competitive pure electric products. At the same time, the new version relaxes the power consumption requirements for high-end electric vehicles and encourages car companies to develop high-end electric vehicles;
In the process of expanding the production and marketing of new energy, it is necessary to pay attention to the decline in power consumption level: if the power consumption level can reach the standard, the highest can have a multiple effect of 1.5 times, which is extremely attractive to car companies, and the decline of the level of power consumption of 100 kilometers has higher requirements for the technical level, and the current power consumption indicators of A-class vehicles are relatively good, which may prompt car companies to increase the investment of A-class vehicles, which is also in line with the mainstream development direction.
China's new energy vehicle market has experienced a decline in sales in 2019-2020, and from 2021 onwards, technological progress has led to an increase in product strength and accelerated sales growth. Electrification brings a quiet and smooth driving experience, intelligent cockpits and self-driving cars bring intelligent experiences such as intelligent control and artificial intelligence, and technological improvements such as platform-based cars and batteries have brought about a decline in costs, prompting the new energy industry to change from policy-driven to supply-driven, and the C-end market continues to break out. From January to July 2021, China's new energy vehicle sales reached 1.478 million units, a cumulative increase of 197.1% year-on-year, and its new energy passenger car sales reached 1.398 million units, an increase of 212.3% year-on-year, while the cumulative increase in commercial vehicles was only 60.9%. From the perspective of the proportion of C-end sales, the proportion of C-end sales from 2019 to 2021 has increased significantly, and the average monthly proportion of C-end new energy in 2019 is about 60%, and by January 2021, sales will account for more than 80%. We believe that with the continuous release of models based on the latest electrification and intelligent technology, the penetration rate of the new energy automobile industry is experiencing a period of accelerated rise.
2.2. China vs Overseas:
China's new energy transformation is at the forefront
Global new energy vehicle sales maintained rapid growth, approaching 3 million units in 2020. According to Marklines data, before 2010, the annual sales of new energy vehicles in the world were less than 10,000 units, which increased sharply to 46,500 units in 2011, and then entered a high-speed growth channel, and the sales of new energy vehicles reached 2.99 million units in 2020, an increase of 44% year-on-year.
From a country perspective, China's new energy vehicle sales are at the forefront. By 2018, the proportion of global sales of new energy vehicles in China will increase to 55%, more than half of the global market share, the United States and Europe are 19% and 15%, and Japan's share is compressed to 2%. In 2019, due to the policy decline, China's new energy vehicle sales fell for the first time, accounting for 52%, Europe increased significantly to 22%, and the United States fell to 16%.
The penetration rate of new energy vehicles has increased year by year, and China's new energy market has shifted from policy-driven to market-driven, and the penetration rate has steadily increased. Globally, the penetration rate of new energy vehicles has accelerated from 3.9% in 2010 to 3.9% in 2020, but it is still at a low level and has a large room for growth. By country:
1) Nordic: Norway leads the world in the penetration of the new energy vehicle market. In 2020, the permeability rate of new energy vehicles in Norway will be 59.8%, far exceeding the global average of 3.9%. In addition, the penetration rate of new energy vehicles in Sweden and the Netherlands in 2020 was 27.2% and 23.2%, respectively, ranking the forefront of the world. Nordic countries have a strong sense of environmental protection, the government has a strong support for new energy vehicles, and has introduced a series of policies such as purchase tax reduction, toll road free, driving tax rebate, etc., which are attractive to consumers and make the penetration rate of New Energy Vehicles in Northern Europe lead the world.
2) Other major countries in Europe: The penetration rate of new energy vehicles has accelerated. Germany, the United Kingdom, France before 2020 new energy vehicle penetration rate increased steadily, but accelerated in 2020, 12.6%, 9.3%, 9.0%, respectively, the growth rate is much higher than the global average, mainly due to strong policy subsidies.
3) U.S.-China, South Korea, and Japan: Major countries are differentiated. China, the United States, Japan, South Korea in 2020 new energy penetration rate of 4.9%, 2.2%, 0.6%, 2.6%, affected by the subsidy policy China's penetration rate remains ahead, the United States, Japan and South Korea penetration rate is generally lower than the global average, of which The Japanese penetration rate is less than 1%, mainly because the Japanese car company HEV technology is mature, becoming the main energy-saving and emission-reduction technology route.
China's new energy vehicle industry is leading the world, accounting for 42% of the global new energy vehicle market in 2020. China's 2021H1 new energy vehicle sales of 1.2 million, is expected to grow at a rate of about 40% from 2022, but due to the expiration of the new energy vehicle subsidy policy in 2023, it is expected that the growth rate of China's new energy vehicles will stabilize at about 35% in 2023-2025, and the sales of new energy vehicles in 2025 is expected to reach 8.8 million. China's new energy vehicles have entered the market-driven stage, while foreign new energy vehicle subsidy policies are currently in a period of vigorous subsidies, and it is expected that global new energy vehicle sales are expected to reach 21 million by 2025.
2.3. Technical route:
Plug-in, extended range, pure electric trend judgment
At present, new energy vehicles have three mainstream technology paths of pure electric, extended range and plug-in, of which range extended range and plug-in type are hybrid technologies. Pure electric vehicle refers to the power battery as the power, with the motor to drive the wheels; hybrid technology is divided into series, parallel, mixed 3 modes, of which mixing is the current car companies mainly plug in the mixing technology path: 1) the tandem configuration of the engine and the wheel decoupling, the engine through the generator to generate electricity, and then by the motor to drive the wheel; 2) The parallel configuration engine and generator can drive the wheels at the same time, so as to achieve the complementarity and cooperation of the two power sources; 3) The mixed configuration can realize the function of series and parallel connection at the same time, the typical representative of which is the power shunt and the series parallel connection, respectively, using the planetary gear row and the string parallel structure, the control logic is the most complex. Based on the working principle of hybrid vehicles, hybrid vehicle power has a linear power output, and the driving experience is smooth, quiet, oilable and electric, which helps to solve consumer mileage anxiety, but it is usually more expensive than the same level of models.
In terms of convenience, economy, driving experience, and technical difficulty, the three schemes have their own advantages. Convenience aspect, extended range, plug-in and pure electricity compared to the convenience is better, extended range, plug-in can be oilable, no mileage anxiety, and do not rely on charging piles; economically, plug-in and mixing is greater than the increased range is greater than pure electric, mainly because the battery cost is still high, pure electric cost is greater than the range and plug-in; driving experience, pure electric is the best, the range increase is second, hybrid cars are relatively general; technical difficulty, plug-in difficulty, mechanical parts (engine, motor, gear mechanism), research and development difficulty, and the existing technology fortress / There are many patents.
Pure electric takes the lead in the two-end market, plug-in hybrid acceleration fuel vehicle replacement. Based on the current level of technology, accessibility and cost problems hinder the replacement of pure electric fuel vehicle models, but we believe that pure electric vehicles have great potential in the scooter market, and customers in this market have almost no mileage anxiety, and pure electric cars have the advantages of low battery charging and low cost, and electric driving experience. In the high-end market, consumers are not sensitive to price, and pure electric vehicles are better than the same level of fuel vehicles in terms of intelligence and electrification, and pure electric vehicles can be realized to replace fuel vehicles in the high-end market. For plug-in and hybrid vehicles, the domestic OEMs represented by BYD have realized or will achieve the parity of fuel vehicles with plug-in and mixing technology, and then superimpose the electrification and intelligent experience of plug-in and mixed new energy vehicles, and plug-and-mix vehicles are rapidly replacing fuel vehicles.
Pure electric vehicles have obvious advantages based on dedicated pure electric platforms. For car companies, there are two modes for designing and producing new energy vehicles, one is to transform the traditional fuel vehicle platform, that is, AEP (Adapted Electric Platform), and the other is to use a dedicated platform developed with a new design, that is, NEP (New Electric Platform). The advantage of using AEP is that it can share the design, mold and parts with traditional fuel vehicles, etc., the early model launch speed is faster, and the upfront cost can be better controlled, in addition, the use of NEP has significant advantages:
Riding and storage space is more abundant: new energy vehicles in the powertrain and surrounding parts of the appearance of the size, layout and traditional fuel vehicles have obvious differences, the use of AEP, the power battery will often encroach on the rear seat space and trunk storage space;
Longer driving range: AEP does not reserve space for the power battery, the power battery layout cannot be optimal, and the matching of the thermal management system is more difficult, dragging down the driving range;
Better power performance: The unreasonable layout of the AEP powertrain leads to the inability to achieve optimal vehicle counterweight and battery thermal management, and it is impossible to give full play to the ultimate performance of the motor;
Lower production costs after new energy vehicles: When using AEP, power batteries are usually laid out in traditional fuel vehicle transmission systems, exhaust systems, fuel tanks, trunks, etc. The appearance profile is difficult to form a unified standard, compared with NEP, which usually lays the power battery flat on the chassis, which is easier to realize modularization and apply to different models, and the production cost of new energy vehicles is lower after discharge.
Domestic and foreign OEMs have released pure electric platforms,
The gap in platform-based car manufacturing is small:
1) Geely Haohan architecture to hardware layer, system layer and ecological layer to build a three-in-one three-dimensional layout, with the world's largest bandwidth, from A-class cars to E-class vehicles full-size coverage, can meet the manufacturing needs of cars, SUVs, MPVs, small city cars, sports cars, pickup trucks. At the same time, the vast architecture has reduced software development time by more than 50%.
2) The Great Wall ME platform integrates global resources, with large space (wheelbase length ratio can be greater than 70%), high safety (more than 60% of high-strength steel body, up to 4.0 roof pressure strength ratio, with super pressure resistance), lightweight (through new material application, cross-section optimization, integrated design), scalable (can cover A-D class models) and other characteristics.
3) GAC BJEV's second-generation pure electric special platform GEP (GAC Electric Platform), the biggest advantage of this platform is that the battery and electric drive system are located in the center of the vehicle, which is conducive to maximizing the space inside the vehicle.
4) BYD's e platform can integrate resources and improve product generalization, standardization and simplicity. The three electrics are highly integrated and standardized, and can be equipped with different power configurations and endurance requirements under the same platform, and can also be equipped with different suspension structures according to the needs. The design concept of simplifying complexity can also reduce the cost of the whole vehicle, the space for parts and components is squeezed out to be smaller, and the cost of supporting procurement can also be greatly reduced.
5) The Volkswagen MEB platform is highly malleable. The Volkswagen MEB has a modular design, which means that the size of the platform can be shortened, extended, or otherwise modified. The platform can be used in a variety of car types, divided by size, covering everything from SUVs to hatchback cars.
6) Toyota released the e-TNGA platform to differentiate the HEV market. Toyota's e-TNGA architecture is a hybrid platform that will be used to develop ten six-or-six models, including compact crossovers, mid-size crossovers, mid-size SUVs, mid-duty minivans, mid-size sedans, and large SUVs.
7) Daimler Benz EVA platform has core advantages: 1) suitable for all of Mercedes-Benz's current car series, including A-class, S-class, GL class, etc., while supporting two-wheel drive and four-wheel drive two-wheel drive working modes; 2) battery and electric motor architecture adjustability. The EVA platform battery is located in a shell under the floor between the axles, and the modular structure allows different battery capacities (60, 80, 95 and 110 kWh) to be accommodated in the same volume, and the power can be adjusted according to the specific vehicle type.
The policy guidance reflects the promising nature of hybrid technology. The "Energy-saving and New Energy Vehicle Technology Roadmap 2.0" led by the Society of Automotive Engineers of China has clearly defined the phased development goals: by 2025, 2030 and 2035, the sales volume of new energy vehicles (plug-in hybrid + pure electricity) will account for 20%, 40% and 50% respectively, and the sales volume of hybrid vehicles in traditional energy passenger cars will reach 50%, 75% and 100% respectively.
Hybrid fuel vehicles are affordable, and the replacement of fuel vehicles is accelerated. BYD DMi, Great Wall DHT, Chery Kunpeng and GAC all adopt dual-motor hybrid architectures; Ford, Toyota THS, and GM second-generation Voltec adopt power shunt architecture; Nissan e-power & ideally adopt electric extender architecture; Volkswagen DQ400E, Mercedes-Benz EQ Power, BMW 5-generation eDrive, Changan iDD, Geely Lynk & Co., P2.5, and SAIC SECOND-generation EDU all adopt single-motor P0-P4 architecture.
At present, the leading domestic hybrid enterprises have adopted the hybrid mode as the main technical path. Among them, BYD DMi took the lead in achieving the parity between the hybrid system and the fuel vehicle, and released the Qin, Song and Tang models based on the DMi platform, which has achieved explosive sales of fuel vehicles. Autonomous car companies such as Great Wall Lemon DHT, Geely GHS2.0, Chery Kunpeng and GAC Green Engine Technology will be installed in 2021H2 to 2022, or will accelerate into the era of fuel vehicle replacement.
2.4. Technical dimensions:
Electrification is the best carrier for intelligentization
Distributed electrical and electronic architecture can no longer meet the needs of the development of intelligent vehicles. With the development of automotive intelligent networking, the functions of automobiles have become more and more complex, and the ECU electronic controller units arranged on the whole vehicle are gradually increasing. A current passenger car can have up to 70-80 ECUs, and the total code volume of all ECUs is expected to reach about 100 million lines, which is far more complex than the Android mobile phone system. In the traditional automotive supply chain, different ECUs come from different suppliers, with different embedded software and underlying code. This distributed architecture creates considerable redundancy at the vehicle level, and software updates for traditional cars are almost synchronized with the car lifecycle, greatly affecting the user experience.
Towards a cross-domain centralized electrical and electronic architecture characterized by centralization and domain fusion. Distributed electronic and electrical architecture schemes no longer have advantages, and need to evolve to cross-domain centralized electrical and electronic architectures characterized by centralization and domain fusion, and the concept of "domain" is born. The whole vehicle needs to pay attention to the system solution and software integration control, and the single-function controller is no longer the mainstream. Only by mastering the concept of "domain" control can car companies continue to maintain the right to speak in the era of automotive intelligent networking that matches the status of the whole vehicle.
Traditional distributed electrical and electronic architecture systems usually divide functions into different module areas, such as powertrain, infotainment, chassis, body, etc. In each module area, the design of the controller is usually based on specific functions, such as: seat control unit SCU, tailgate controller PLG, etc. Information is passed between modules via the CAN bus, and their module division is generally based on the number of buses.
Advantages:
1) The functional division is clear,
Strict and clear boundaries between modules,
All "transgression" behavior is easy to control;
2) The independence between modules is strong,
Module developers don't have to think too much about interference from other modules.
limitations:
1) The independent development mode of the function controller leads to information blockage, limited bandwidth of the transmitted signal, resources cannot be shared, a large number of computing resources are wasted, and the performance is "partial";
2) There are bottlenecks in computing performance, communication bandwidth, deformation management and support for cross-domain functions;
3) With a highly embedded controller as the mainstay, hardware and software are highly integrated, and it is difficult to upgrade the software after the mass production of the vehicle, and it is difficult to support software innovation.
Cross-domain centralized electrical and electronic architecture better supports continuous software innovation and updates. The limitations of a modular, closed architecture of distributed electrical and electronic architectures can be tolerated in autonomous driving applications below L2, but are magnified under the requirements of L4 autonomous driving or ASIL-D functional safety, and become an obstacle to positive functional development. The cross-domain centralized electrical and electronic architecture provides the computing power and communication power required for the future car through domain controllers and Ethernet, concentrates vehicle-level software on domain controllers, standardizes highly embedded controllers, better supports deformation management and cross-domain functions, and solves the limitations of distributed architectures.
In the future, the centralized automotive EEL architecture will be divided into three layers:
The top level is the cloud computing service platform;
The middle layer is the on-board computing control platform (i.e. the domain controller);
The lower level is a standardized actuator and sensor controller for mechatronics.
Generally, the automotive electronic and electrical system is divided into five functional domains, namely the powertrain domain, the chassis domain, the body domain, the infotainment domain (intelligent cockpit domain), and the auxiliary/automatic driving domain. Thus, the calculation and control of the middle layer includes five domains of the master and Ethernet communication, wireless communication a total of seven elements.
The implementation of centralized solutions is constrained by the cost of realization. The concept of "domain" centralized scheme architecture is perfect, but in recent years, it has not been widely used in low-end models, and the cost of program implementation is the primary contradiction. With the increase in automotive electronics applications, the number of ECUs and the demand for computing power of the whole vehicle are growing, and the demand for computing bandwidth has also begun to explode, and the cost of automotive electronic systems has increased significantly. In order to cooperate with the concept of "functional domain" based on module division, wiring harnesses, arrangements, installations, and brackets have to be reshuffled in design, and the cost of mechanical structure transformation will also increase significantly.
Tesla's model 3 divides the domain according to the principle of physical space proximity, which has a cost advantage. In order to solve the high cost problem without losing the core concept of the "domain" software set, the Tesla Model 3 reclassifies the "domain". In the new concept, there is no longer a traditional body domain, power domain, etc., but replaced by an "area zone" based on the division of physical space, such as the middle domain, left domain and right domain. The new domains are divided according to the layout scheme, which is the core concept of the region.
The "Zone Zone" puts forward high requirements for software development and brings many challenges:
1) A single controller engineer needs to be responsible for more controllers and functions, for example, a body controller engineer may need to start studying radar drivers and algorithms;
2) The software development workload of the same function will be greatly improved, and the development of software at the ASIL-C and D levels of functional safety will gradually become standard;
3) Domain control development will no longer be limited to functions, software and hardware development will break the traditional functional division wall, more need to think from the perspective of the vehicle.
Domestic independent enterprises are entering the era of electronic and electrical architecture domain controllers, BYD, Nezha, Extreme Kr Automobile, Dongfeng Lantu have all released electric vehicle platforms with centralized domain control of electronic and electrical architectures, and Great Wall Motors will release a platform with centralized domain controllers in 2022. In the future, the platform architecture based on domain controllers will help the rapid development of intelligent and networked vehicles.
Centralized electrical and electronic architecture requires high software and data transmission, and requires SOA technical support, and the vehicle Ethernet is the best communication method to provide technical support for SOA. With the formation of the domain controller concept, the car has gradually become a platform equipped with new differentiating elements, including in-car entertainment systems, automatic driving and intelligent safety and other functions based on "high fault tolerance", which require software SOA cooperation, software through intelligent sensors and hardware integration, further into the digital stack, the stack will complete horizontal integration, and add new layers, thereby transforming the overall structure into SOA.
The advantages of introducing SOA in automobiles include: 1) high cohesiveness of services and easy reuse of software; 2) flexible deployment of services; 3) faster software update upgrades, SOA has the characteristics of "loose coupling", "interface standard accessibility" and "easy to expand", so that developers can respond to iterative and changing customer needs with minimal software changes, and signal-based communication methods (such as CAN bus) support data types that are too simple and scalable, and are not suitable for autonomous driving, software upgrades and other scenarios. The dynamic interaction of large amounts of data must be service-oriented communication methods to improve communication speed and efficiency. Vehicle communication on the basis of signal communication, must introduce SOA communication support, the current main focus on vehicle Ethernet technology.
Wire-by-wire technology is a necessary technology to achieve high-level autonomous driving, and electric vehicles are the best platform for the implementation of wire-controlled technology. In traditional chassis technology, when the driver makes the action of stepping on the brake pedal/accelerator pedal, turning the steering wheel or stepping on the clutch pedal and dialing the gear manipulator, the force is transmitted to the actuator through the mechanical connection device, and the relevant action is completed with the assistance of the hydraulic/pneumatic and other devices; the difference between the wire control chassis system is that when the driver makes the above related actions, each position sensor converts the force signal into an electrical signal, transmits it to the ECU to calculate the required force, and then the motor drives the actuator to complete the relevant action.
The core technology of the executive control mechanism mainly includes line control movement and steering:
Line control is an important part of the automatic driving execution system. ADAS is highly correlated with braking systems, because autonomous driving requires a shorter braking response speed (300ms120ms) at the executive layer, and new energy vehicles without engines to generate vacuum assist, improving energy recovery efficiency requires pedal decoupling. ESP-based braking systems can no longer meet the needs of new energy and autonomous vehicles, and line control can solve these two problems, of which the mainstream solution for driving brake centerline control will be the electrohydraulic braking system EHB (ElectroHydraulic Braking System).
Steer-by-wire is also accompanied by intelligent upgrades. With the maturity of autonomous driving technology, for L3 and above self-driving cars, part or the whole process will be out of the driver's control, so intelligent driving has higher control requirements for steering and other systems, and the previous steering system cannot meet these requirements. Therefore, steer-by-wire technology will become the development trend of the future. Compared with the current electronic power steering system EPS, the connection between the directional disc and the wheel is not in-line, but still mechanically connected.
3. Intelligent Driving:
The core element, data accumulation accelerates technological breakthroughs
The trend of electrification has been determined, and the next stage of competition in the automotive industry will come from intelligence, and intelligent driving as the core, which will subvert the traditional travel of human beings. According to the implementation path division: 1) progressive route: L3 function is initially introduced, but currently limited to high-speed and urban expressways (high-precision map coverage area), the next stage of the main expansion of urban areas while improving functional continuity; 2) leapfrog route: commercial scenarios take the lead in landing, transport goods faster than transporters, low speed faster than high-speed, head enterprises to open the technical solution output dimensionality reduction empowerment. According to the technical scheme: 1) bicycle intelligence: fusion cameras, millimeter wave radar, lidar and other sensors to achieve the perception of the surrounding environment, of which lidar is a new component of L3 and above, although slightly controversial but still considered by most car companies and Tier1 to be necessary for L3 and above; 2) V2X: based on the intranet, inter-vehicle network and on-board mobile Internet, high-end automatic driving is finally supplemented. We believe that with the accelerated accumulation of data brought about by car companies and technology Internet giants, the breakthrough of intelligent driving technology is expected to be faster than expected.
3.1. Overview:
The core elements of intelligent cars,
The policy is set at a high level
The U.S. SAE standard divides intelligent driving into six levels of L0-L5. At present, the global classification of autonomous vehicles is mainly based on the classification standards developed by the American Society of Automotive Engineers (SAE). According to SAE's classification criteria, autonomous driving technology is divided into six levels of L0-L5, from low to high: no automation, driving support, partial automation, conditional automation, high automation and full automation.
The Chinese version of the Intelligent Driving Classification was released by the Ministry of Industry and Information Technology on March 9, 2020. Compared with the Two Versions of the Standards, the difference is mainly reflected in the partial definition of L0-L2. In the Chinese version of the standard, the "target and event detection and response" of level 0 to level 2 autonomous driving is completed by the driver and the system, while in the AMERICAN SAE standard, the OEDR (target and event detection and decision-making task) of L0 to L2 autonomous vehicles is completed by the driver.
The refinement of legal responsibility promotes the process of automatic driving. Since 2017, countries, including China, have successively introduced laws and policies related to autonomous driving, clarifying the degree of automatic driving takeover and legal responsibility at all levels. The introduction of national standards and various support policies is conducive to the landing of automatic driving technology, and all kinds of enterprises can carry out targeted deployment to promote the accelerated mass production of intelligent driving vehicles at different levels.
3.2. Directions:
Progressive vs Span
From the perspective of realizing the path, intelligent driving includes two major paths: progressive and leapfrog. Among them, the progressive is represented by traditional car companies and new car-making forces, to achieve step-by-step advancement from L0 to L5 automatic driving, and is currently in the initial introduction stage of L3 automatic driving, Tesla is relatively leading, and the next stage mainly expands urban areas while improving functional continuity; leaps and bounds are represented by technology giants, such as Google Waymo, Baidu Apollo, etc., and the core lies in the scene landing and effective business model.
3.2.1. Progressive:
Tesla led the initial introduction of L3
Tesla is leading the way, and L3 autopilot has been initially introduced. At this stage, most of the intelligent driving of passenger cars stays in the L2, that is, the ADAS assisted driving stage, and the typical functions such as AEB and ACC are typical. Tesla, with its own software and hardware integration capabilities, took the lead in launching high-speed autonomous navigation driving, and the latest FSD Beta software added automatic auxiliary driving on urban streets, traffic light recognition, etc., and first explored L3 to lead the progressive route.
The binding effect between car companies and chip manufacturers has begun to appear, which can be roughly divided into four main lines. Perception algorithms (other planning and control algorithms, etc.) are the core difficulties of short-term intelligent driving software development, and their coupling with hardware is high, so it is easy for car companies to form a certain binding effect after selecting a computing platform. At present, in addition to Tesla's self-developed chips and mass production applications, other car companies have chosen chip manufacturer technology solutions, which can be roughly divided into four main lines: Nvidia, Huawei, Qualcomm and others (Mobileye, Horizon, Black Sesame, etc.).
1) Tesla
Master the integration of software and hardware, and fully integrate vertical integration. Tesla adheres to a pure visual perception route, relying on 1.35 million+ VEHICLEs equipped with AP hardware and 4+billion miles of data accumulation (by the end of 2020) to lead the progressive route. Its autonomous driving platform has undergone three generations of changes, from Mobileye to NVIDIA to self-research, gradually mastering the integration of software and hardware, vertical integration, and comprehensive access. Its new generation of FSD chips, in conjunction with Samsung, adopts a 5nm process process, which will empower the HW4.0 autonomous driving platform, and the hardware computing power will be about three times that of HW3.0, which is expected to meet at least the demand for L4 automatic driving.
2) NVIDIA
A I chip leader, software development ecology is complete. It has a complete software development ecosystem, covering the operation system Drive OS, middleware Drive Works, software stack Drive AV, and the toolchain is stable and open. The latest Orin chip hash rate of 200 TOPS, mass production in 2021, through different combinations can meet L2-L5 autonomous driving, up to 2,000 TOPS, April 13, 2021 released the latest generation of supercomputer chip Atlan, single chip hash rate of 1,000 TOPS, can meet L5 demand, is expected to provide samples in 2023, 2025 mass loading. From the perspective of the selection of the next-generation automatic driving platform program of major car companies, NVIDIA has significant advantages and is expected to reproduce Qualcomm's intelligent cockpit supremacy.
Participants in the Nvidia camp at this stage
Mainly including Xiaopeng, Weilai, Ideal, SAIC Zhiji, etc.:
Xiaopeng: The software is fully developed by itself, and it is the only one outside Tesla. With the ability to iteratively upgrade full-stack software development from perception, positioning, planning, control to data-based functions, the level of intelligence is in the forefront of autonomous car companies. Xpilot2.5 selected Moibleye EyeQ4, Xpilot3.0 switched to NVIDIA Xavier and based on the underlying operating system of NVIDIA Drive OS for upper-layer software development, Xpilot3.5 continued to use NVIDIA Xavier.
NIO: Restart the L4 self-developed route, and the ET7 configuration has been fully upgraded. In November 2020, WEILAI restarted the L4 self-developed route, made up for the shortcomings of visual perception, and laid out the full-stack development of automatic driving. Nio's first-generation platform also uses Moibleye EyeQ4, and the second-generation platform NT2.0 turns to NVIDIA Orin chips, with a combined computing power of up to 1,016 TOPS, which will be first launched in its first sedan, the ET7, which is equipped with 11 8MP HD cameras and 1 equivalent 300-line high-resolution lidar.
Ideal: Self-developed application layer algorithm, Orin chip to help accelerate catch-up. Ideal in the field of decision-making and control algorithms and Yihang Intelligent joint development, self-research capabilities are smaller Peng, Weilai slightly weaker. In February 2020, Ideal established the Shanghai R&D Center, focusing on the development of autonomous driving application layer algorithms and real-time operating system LiOS. The ideal ONE on the chip is equipped with Mobileye EyeQ4 chip, and the new model X01 in 2023 will be equipped with NVIDIA Orin chip, standard with L4 automatic driving hardware, accelerating the catch-up of Xiaopeng and Weilai.
SAIC Zhiji: Bringing together the group's resources to transform intelligent travel. SAIC Zhiji has established an intelligent travel research institute internally, and the group has an intelligent driving center (mainly attacking the algorithm layer) and a software subsidiary Zero Beam (mainly attacking architecture, jointly developing operating systems and AI chips), which has certain self-research capabilities. Zhiji selected NVIDIA Xavier (visual perception scheme) and Orin (compatible with lidar software and hardware redundancy scheme) chips, using a deep hardware embedding scheme, 15 high-definition visual cameras for the current mass production of the most, and compatible with lidar solutions.
3) Huawei
Provide automatic driving soft and hardware integrated full-stack solution,
From top to bottom, they include:
1) Cloud services, namely Octopus, provide three major services: data, training and simulation;
2) Algorithm, design and optimization for urban roads, highways, urban parking, etc.;
3) Development toolchain, that is, application development end-to-end toolset service;
4) Operating system AOS;
5) Ascend AI chip, 310 (car end) / 610 (car end) / 910 (cloud);
6) MDC Computing Platform, 210 (L2+, 48TOPS)/300F (Commercial Vehicle/Work Vehicle, 64TOPS) / 610(L3-L4, 200+TOPS)/810(L4-L5, 400+TOPS);
7) Sensor: LiDAR (96-line start) + millimeter wave radar (traditional +4D imaging 12T24R large array) + camera (HD 5.4MP start).
BAIC ARCFOX and Changan CHN are Huawei's main partners.
1) BAIC ARCFOX and Huawei in-depth cooperation, Jihu αS is the first landing model of "Huawei Inside", which is expected to be launched at the end of 2021. It is equipped with Huawei's autonomous driving ADS solution, equipped with 3 x 96-line lidar, 6 x millimeter waves, 12 cameras, 13 ultrasonic and 400 TOPS Huawei's self-developed chips;
2) Changan CHN's first mass-produced model, the E11, is expected to be launched in 2022, making it Huawei's second front-loading mass-production project after BAIC ARCFOX.
4) Qualcomm
New participants in autonomous driving, released a new computing platform to enter autonomous driving.
Relying on the technical reserves of consumer electronics chips, we have successfully entered the field of automotive intelligent cockpits, and launched the automotive system chip Snapdragon 820A at cess 2016. Judging from the selection of cockpit domain controller chips of mainstream car companies, Qualcomm has a high market share, and its latest release of the third-generation intelligent cockpit chip 8155 will be installed in NIO ET7, Xiaopeng P5, SAIC Zhiji L7, Great Wall Moka/VV7 and other models. In 2020, snapdragon Ride expandable computing platform into autonomous driving, its technical solutions cover heterogeneous system-level SoC chips, deep learning accelerators and algorithm stacks, with modular, scalable, passive/air-cooled heat dissipation and other features, computing power extension 30-700 TOPS, support L1-L5 automatic driving, the first model is expected to be put into production in 2022.
Great Wall Motors is a major player in the Qualcomm camp and a leader in autonomous driving in the intelligent era. It has successively established Xiandou Intelligent (intelligent cockpit), Millima Zhixing (intelligent driving), and Salon Zhixing (high-end brand), released the "331" strategy of intelligent driving, and accelerated the intelligent transformation. The computing platform based on Qualcomm's Snapdragon Ride platform and 8540+9000 processor (hash rate 360-720 TOPS) will first be applied to high-end models launched in 2022, and some models will also use Huawei MDC610. In addition to joining hands with Qualcomm and Huawei, the company will strategically invest in horizons to enter the chip industry in early 2021, and will develop rapidly in the chip industry through strategic investment, strategic cooperation and independent research and development in the future.
Horizontally, in addition to Tesla's relative lead in the gradual route, the gap between independent car companies is not significant. In addition, in the realization of high-end automatic driving technology, most of the leading domestic car companies choose the "software self-research + hardware embedding" strategy, and the scheduling period expands the design and operation area of automatic driving through the subsequent OTA upgrade iterative software algorithm, and the sensor configuration in addition to the traditional ADAS sensor (camera, millimeter wave radar, ultrasonic radar), laser radar begins to appear in the front-loading mass production plan of the car companies.
3.2.2. Leapfrog:
Specific scenarios take the lead in landing Head technical solution output
The leapfrog route is mainly aimed at two major application scenarios: Robotaxi and truck. High-end autonomous driving is of subversive significance for future travel, attracting many technology giants and start-ups to enter first, mainly for The two major application scenarios of Robotaxi and trucks, short-term bicycle cost, fleet size, and regulatory restrictions are still the core factors restricting the landing of high-level automatic driving.
Compared to Robotaxi,
Specific scenarios such as high-speed freight are expected to land earlier.
1) Robotaxi: Waymo was launched at the end of 2018, followed by domestic technology startups, followed by Xiaoma Zhixing, Baidu, etc., the key lies in data accumulation;
2) High-speed freight: As a trunk logistics scene, its traffic situation is relatively simple and controllable;
3) Specific scenarios: such as ports, mines and other structured scenarios, automatic driving technology is the least difficult to achieve, and it is expected to land as soon as possible.
Globally and in China,
Leapfrog route Google Waymo and Baidu Apollo are relatively ahead:
1) Google Waymo: the global leader in autonomous driving, the layout of autonomous driving rental + freight distribution + technical solution output, in March 2020 for the first time to open external financing. Waymo's self-driving vehicles have accumulated more than 20 million miles of road testing and more than 15 billion miles of simulation tests in addition to actual vehicle tests, and are currently working with Volvo, FCA, Jaguar Land Rover, Renault-Nissan-Mitsubishi Alliance, and Volvo.
2) Baidu Apollo: In 2020, Baidu Apollo opened Robotaxi operations in Changsha, Cangzhou, Beijing and other places, and is currently the holder with the largest number of cities and total licenses, the highest technical level and standards, and the most difficult test scenario among domestic T4 license enterprises. Its test mileage exceeds 200,000 miles in the United States and more than 7 million kilometers in China. Autonomous driving program Apollo Lite as the only L4 pure visual automatic driving technology in China to open the dimensionality reduction output, parking products AVP has taken the lead in landing on WM W6, pilot assisted driving ANP has also entered the mass production stage, and GAC, WM, Great Wall and other car companies to cooperate, within three years the first mass production target of 1 million units.
High and low levels open two-way interaction, and the breakthrough of intelligent driving technology is expected to be faster than expected. Different from the previous automatic driving financing fever, we believe that 2021 is expected to become a turning point in the true sense of autonomous driving, mainly reflected in the fact that the high and low orders are no longer independent, but begin to show a two-way interactive relationship: progressive ADAS accelerated penetration, L3 deep hardware embedding initial import, accelerated data accumulation; leapfrog specific scenarios landed first, and the head technology solution dimensionality reduction empowerment. Boosted by the two-way interaction of high and low orders, the breakthrough of intelligent driving technology is expected to be faster than expected, and 2022 may become a key year for the landing of L3.
3.3. Scheme:
Bike Intelligence vs V2X
Bicycle intelligence first, V2X long-term supplement. Intelligent driving mainly has two implementation methods of bicycle intelligence and V2X, of which bicycle intelligence refers to the automatic driving of bicycles through high-performance sensors and large computing power computing platforms, and continuously empowers intelligent driving "central brain" through data accumulation; V2X refers to the realization of vehicle access through a higher level of communication network, and the realization of terminal data collection by cloud processing platform, thereby evolving the automatic driving ability of the whole vehicle. In the short and medium term, we believe that bicycle intelligence will be the main implementation of intelligent driving, and V2X is more as a long-term supplement.
3.3.1. Bicycle Intelligence:
Perception, judgment, execution first
Multi-sensor fusion is the mainstream trend, and L3 and above lidar are of great importance. The advantages and disadvantages of various types of sensors for autonomous driving are different, and the current production car ADAS is mainly configured as a camera + millimeter wave radar + ultrasonic radar. In the L3 and above level, the importance of lidar is prominent, according to ElecFans, equipped with laser radar automatic driving system safety of up to 99.99%, while cameras, millimeter wave radar and other sensors can only guarantee 99%, although Tesla always adhere to the pure visual perception route, but lidar is still considered by most OEMs and Tier1 to be necessary for L3 and above.
1) LiDAR of the perceptual layer:
An important support for high-level automatic driving
Lidar is an active measurement device with a laser beam as the information carrier. Lidar is an active measuring device that uses the laser beam as a information carrier, uses phase, amplitude, frequency, etc. to carry information and increases the radiation source frequency to the optical frequency band, and measures the precise distance between the object and the sensor by emitting a laser to form a 3D point cloud data for the surrounding environment. According to the ranging method, it can be divided into time-of-flight ToF ranging method, FMCW ranging method based on coherence detection and triangulation method, etc., of which The ToF scheme has been widely used, and most of the FMCW lidar is still in the concept aircraft stage.
The technical solutions vary, and the short-term semi-solid state is the most suitable for application. Divided by scanning mode, it mainly includes three categories: mechanical (overall 360° rotation), semi-solid state (transceiver module is stationary, only the scanner undergoes mechanical movement, including rotor, MEMS) and solid state (no mechanical moving parts, including Flash, OPA). Mechanical laser radar has been widely used before, but mainly used in the field of L4-L5 unmanned driving, limited vehicle regulations, cost, shape and other issues are difficult to apply to L2-L3, semi-solid state (MEMS, rotating mirror, prism) at this stage is the most equipped with front-loading mass production application conditions, but in the medium and long term, the full solid state potential is huge.
The eve of front-loading mass production, large-scale application is imminent. Compared with cameras and millimeter wave radar, lidar combines the characteristics of long range, excellent angular resolution, and little influence by ambient light, which can significantly improve the reliability of the automatic driving system, and is a key sensor to solve the continuous experience of automatic driving, which is considered by most automakers and Tier1 to be a necessity for L3 and above automatic driving. Starting from the end of 2020, multi-car companies (such as BAIC, Xiaopeng, Changcheng, Volvo, etc.) announced that they will be equipped with lidar, and the model is expected to be on the market from the end of 2021 to the beginning of 2022, and the large-scale application will help further reduce the cost of lidar, and the industry as a whole is on the eve of front-loading mass production.
Prices are expected to gradually decline, with the global automotive lidar market reaching $8.1 billion by 2025. The application of new technologies such as large-scale mass production and chipping of lidar has boosted the continuous decline in the cost of lidar. According to ICVTank's forecast, the price of lidar will drop to $500 per unit in 2025, a 97.5% drop compared to 2018. With the introduction of L3 autonomous driving and the formation of a commercial model for autonomous driving, the lidar market is expected to meet rapid growth. According to Sullivan's forecast, the on-board lidar market size will reach $8.11 billion in 2025, and the CAGR will be as high as 61.2% in 2021-2025, leading the growth rate of autonomous driving sensors.
There are many market participants, and the traditional Tier1 competes with startups. The lidar industry generally presents the characteristics of traditional Tier1 and startups competing together, and the participants mainly include Velodyne, Luminar, Aeva, Ouster in the United States, Innoviz in Israel, Ibeo in Germany, and sagitar Juchuang and Radium Intelligence in China. In terms of technical capabilities, the gap between domestic manufacturers and overseas is small, basically in the same running line, the track layer opportunities are excellent, but the fault tolerance rate for the choice of technical routes is low, and more attention in the short term should focus on the acquisition of pre-installation mass production points.
2) Camera of the perception layer:
Advanced driver assistance core sensors
Smart car "double eyes", the core sensor of the vision scheme. The on-board camera is the core sensor of ADAS advanced driver assistance, and the vision processing chip processes the collected image information based on the neural network algorithm to identify the target and semantic information, and realize ADAS functions such as lane departure warning (LDW) and vehicle collision warning (FCW). According to the installation position, it can be divided into front view, rear view, surround view, and interior view car camera.
The core device of the camera is the image sensor CMOS. The main components of the car camera include the lens group, image sensor (CMOS) and digital processing chip (DSP), which have high technical barriers; in terms of cost composition, the image sensor CMOS accounts for half of the cost of the car camera, the module package, optical lens accounted for 25% and 14% respectively, and the first three accounted for nearly 90% of the total.
The number of bicycles has increased, and the "chip removal" has driven the price of single products down. With the increase in the level of intelligent driving and the increase in the number of bicycle in-vehicle cameras, the L5 is expected to require 12-15 (perception + surround view + DMS/OMS). After the application of the domain controller, the camera "de-chip" will drive the price of the single product down, although the pixel increases the unit price, but the overall downward trend (with a chip and algorithm front-view camera about 1,000 yuan, no chip 5 million pixels only 200 yuan).
The global in-vehicle camera market space is expected to be $27 billion by 2025. According to ICVTank, the global in-vehicle camera market is expected to grow from $11.2 billion in 2019 to $27 billion in 2025, with a CAGR of 15.8%; with the gradual deepening of ADAS and autonomous driving, it is expected that the future in-vehicle camera market size will continue to grow at a high speed.
The competitive landscape is dominated by overseas suppliers and independent upstream cuts. Auto camera head manufacturers mainly include Panasonic, Valeo, Fujitsu, Continental, Magna, etc., and the industry CR3 in 2018 was 41.0%. In terms of autonomy, manufacturers with high market share in the field of mobile phone camera packaging such as Sunny Optics and OFILM are entering the automotive market with the accumulation of their consumer electronics industry processes. Core device image sensor CMOS head manufacturers are mostly overseas enterprises, Sony is a big one, the global market share in 2019 reached 39.1%, domestic companies such as Weier shares through the acquisition of Howe Technology to join the competition, Geke Micro has occupied a place, other local CMOS manufacturers as a whole small scale, technology is relatively backward, short-term technological breakthrough is difficult.
3) Millimeter wave radar of the perceptual layer:
The main backbone of the autonomous driving perception layer
Millimeter-wave radar is the main backbone of the perception layer, and currently looks at 77GHz with more potential. Millimeter wave radar refers to the use of wavelength 1-10mm, frequency 30-300GHZ millimeter wave detection sensor, with all-weather operation, distance depth of field information rich, high obstacle recognition rate advantages, but the detection point is sparse, low resolution, poor perception of pedestrians, in ADAS mainly with vehicle cameras. At present, the frequency bands allocated by various countries to vehicle-mounted millimeter-wave radar are mainly concentrated in 24GHz and 77GHz, with large bandwidth and short wavelength, which have more advantages in detection distance, accuracy and volume, and have greater room for development in the long run.
4D point cloud imaging millimeter-wave radar is expected to form an effective complement to lidar. Based on virtual antenna technology, 4D point cloud imaging radar has greatly improved performance (detection distance + angular resolution) compared with traditional radar, but the cost is basically similar. Although there is still a big gap between the number of point clouds and lidar, such as the number of point clouds in the Auco Eagle forward radar is roughly equivalent to the Valeo 4-line lidar Scala1, it is still expected to be widely used in L3 and above automatic driving to form an effective complement to lidar.
The core device of millimeter-wave radar is the radar main IC. Millimeter wave radar system is mainly composed of array antenna, front-end transceiver radio frequency components (MMIC chips), digital signal processors (DSP/FPGA) and control circuits, of which the antenna and front-end transceiver components as the core hardware. In terms of cost composition, the cost of the main RADAR IC accounted for 43%, and the cost of auxiliary chips, PCBs, and other electronic materials accounted for 11%, 16%, and 11% respectively.
L 3 and above is likely to require 6-8, and the global millimeter wave radar market is expected to exceed US$46 billion by 2025. Millimeter wave radar has the advantages of small environmental impact and long detection distance, and with the maturity of 4D point cloud millimeter wave radar technology, the application scenario is expected to be further expanded. At present, the ADAS application has reached up to 5 (1 forward radar + 4-angle radar), L3 and above are likely to need 6-8 radars; INP, the increase in the proportion of 77GHz and high-resolution radar is expected to drive an increase in the average price. According to DIGITIMES Research, the global automotive millimeter wave radar market will total about $46.2 billion in 2025, and the CAGR will be 40.7% from 2021 to 2025, which is slower than lidar but higher than that of in-vehicle cameras.
The competitive landscape is dominated by the traditional Tier1, and the domestic substitution is accelerated independently. The global millimeter wave radar market is mainly occupied by traditional Tier1s such as Bosch, Continental, hella, etc., and domestic companies such as Desay SV, Huayu Automobile, and Baolong Technology also have layouts, but overseas manufacturers still occupy the dominant position in the domestic and foreign markets. 2014-2016, a large number of millimeter wave radar start-ups and related listed enterprises have emerged in China, in terms of product indicators, domestic millimeter wave radar is not much different from competitors, all meet the requirements of the vehicle regulation level, but there are still certain differences in the signal-to-noise ratio, detection accuracy, yield rate, etc., and there is still a large space for import substitution in the long run.
4) Ultrasonic radar of the perception layer:
Safety aids with ultrasonic visits
Mature technology, wide range of applications, excellent cost performance. Ultrasonic radar is an extremely common on-board sensor, which is a safety aid when the car is parked or reversed, and can inform the driver of obstacles around the driver with sound or a more intuitive display. Generally mounted on the front and rear bumpers of the car and on the side of the car, the parking assistance system usually uses 6-12 ultrasonic radars, the lowest cost of all auxiliary driver sensors, and the unit price is roughly between $15-20.
The main players in the ultrasonic radar industry are divided into international Tier1, domestic Tier1 and start-up companies. Due to the relatively mature ultrasonic radar technology, the gap between players at home and abroad mainly lies in the stability and reliability of sensor implementation, but the overall difference is small.
5) The large computing power chip of the judgment layer:
Smart Car "Digital Engine"
The demand for computing power is growing rapidly with the improvement of autonomous driving level. According to Huawei's estimates, the L3 automatic driving computing power demand is 30-60 TOPS, the L4 demand is more than 100 TOPS, and the L5 demand is even 1,000 TOPS. At present, the suppliers that can provide autonomous driving large computing chips or computing platforms are mainly Huawei, Nvidia, Qualcomm, etc., and domestic AI chip suppliers Horizon and Black Sesame are also gradually developing from low computing power to high computing power.
Suppliers use chips as the cornerstone to build an industrial ecology. Most of the major chip suppliers do not stay in the chip itself, but gradually extend to the software layer to build an industrial ecology. In terms of technical solutions, NVIDIA and Qualcomm provide automatic driving computing platforms and basic software, and do not provide application-layer algorithms; Huawei provides full-stack solutions from chip algorithm sensors; Mobileye is similar to Horizon, mainly chip + perception algorithms. From the perspective of technical trends, chips are mainly developing in the three directions of large computing power, low power consumption and high process, and in terms of chip types, ASIC as-APPLIED circuit chips are expected to become mainstream.
Qualcomm is strong in, and Huawei and other independent opportunities have ushered in. Nvidia and Mobileye rely on their first-mover advantage and strong technical capabilities to take the lead in the early stage of autonomous driving development, but Mobileye has a low degree of openness and the launch of high-computing chips is relatively late, and Nvidia has an advantage in the next generation of autonomous driving platforms. As a new entrant, Qualcomm is expected to catch up with the experience in the field of smart cockpits and the cost advantages brought about by the co-production of mobile phone chips, but it is late to intervene, and it is difficult to make short-term breakthroughs. Huawei, Horizon, Black Sesame, etc. are expected to rely on the rapid response capabilities brought by localization to seize a certain market, while the rapid iteration of chip technology also brings certain opportunities for independent brands, but in the short term, it may be limited by chip foundry capabilities.
6) High-precision map of the judgment layer:
Self-driving car "clairvoyance"
Autonomous driving is an important safety redundancy, and the barriers to entry are high. As an important safety redundancy for autonomous driving, high-precision maps provide over-the-horizon perception. The collection methods are mainly divided into centralized collection (based on professional collection vehicles) and crowdsourcing collection (the car end is compared with the high-precision map based on actual perception, and the upload platform is uploaded when the information does not match).
The government requires that high-precision mapping must have corresponding qualifications, at present there are only 28 in the country, the barriers to entry are high, and the main market share is Baidu, NavInfo, AutoNavi and other enterprises.
7) Line control of the execution layer:
Intelligent driving is an important safety guarantee
The development of high-end autonomous driving has driven an increase in EHB penetration. Automotive braking system can be divided into service brake (foot brake) and parking brake (handbrake), the overall experience of mechanical braking - hydraulic braking - electronic control braking - line control action four stages, of which the line control action as the latest generation of braking technology, is expected to benefit from the long-term development of high-end automatic driving (the executive layer requires faster braking response speed (300ms120ms), to maximize the reliability and safety of the system).
Global and Chinese EHBs are expected to exceed RMB29/10 billion in 2030, respectively. The global EHB passenger car (fuel-fired vehicle + new energy vehicle) market size is expected to grow from 3.3 billion yuan in 2019 to 18.7 billion yuan in 2025, corresponding to a CAGR of 33%, and is expected to exceed 29 billion yuan in 2030. China's EHB passenger car (fuel vehicle + new energy vehicle) market size is expected to grow from 900 million yuan in 2019 to 5.5 billion yuan in 2025, corresponding to a CAGR of 34% and is expected to exceed 10 billion yuan in 2030.
Compared with two-box, one-box has more cost performance advantages. Compared with two-Box, One-Box has more advantages in performance and cost, and can better match L3 and above automatic driving, but the technical requirements are relatively higher: the One-Box solution integrates ESP, which needs to be based on mature ESP technology, and the Two-Box solution coordinates ESP, which can reduce technical difficulties by mining ESP.
The competitive landscape is undecided and is still dominated by the international Tier1. Due to the early start of overseas and the relative maturity of technology, the One-Box and Two-Box fields are mainly dominated by the mainland, Bosch, ZF and other international Tier1 occupy a leading position, but with the acceleration of the domestic substitution process, domestic enterprises are also gradually catching up, typical such as Bethel released Bethel WCBS products in 2019 and achieved mass production of multiple models in 2021, Anderson Electronics released Nathan NBooster products in 2018 and achieved mass production, but they are all equipped with independent low-end models. There is still a long way to go from the low end to the high end.
8) Wire steering of the execution layer:
The only way to improve the intelligent improvement of the car steering system
From EPS electric power steering to SBW steering-by-wire, it is the only way for the intelligent improvement of automotive steering systems. By improving steering efficiency and sensitivity, steer-by-wire not only changes the inherent mechanism of traditional steering systems, but also creates a development basis for unmanned driving systems that are flexible and require more space for layout.
Replacing mechanical connections is the trend. At present, EPS technology is mature and inexpensive, which is the mainstream of the current steering system. However, from the mechanical structure point of view, EPS by the motor to provide auxiliary torque, complex structure, and SBW cancel the steering wheel and steering wheel mechanical connection, by the sensor signal and the motor to achieve steering, and SBW variable transmission ratio can enhance the vehicle maneuverability and comfort, the future replacement of mechanical connection is the general trend.
The industry concentration is high, and the short-term technology is still immature. The concentration of the switch-by-wire steering industry is particularly obvious, mainly dominated by overseas head manufacturers such as Bosch, JTEKT, NSK, etc., and the domestic start-up participants are mainly Nasen Electronics, Hubei Hang Lung, Zhejiang Shibao, etc., but the overall scale is small and the technology is relatively backward. In the short term, SBW technology is still under development, there are still many uncertainties such as immature technology, expensive price, etc., so the current penetration rate is also low, and the Infiniti Q50 was first applied (clutch connection steering tube string provides safety and redundant backup), but due to defect problems, a large number of recalls.
3.3.2. Intelligent Networking V2X:
Long-term addition to autonomous driving
The policy sets the tone, emphasizing the coordination of intelligence and networking. The Internet of Vehicles is based on the intra-vehicle network, the inter-car network and the in-vehicle mobile Internet, which is the key prerequisite for achieving fully autonomous driving. In recent years, various policies have emphasized the positive role of 5G technology in the development of the Internet of Vehicles, and the "Intelligent Vehicle Innovation Development Strategy" jointly issued by 11 ministries and commissions in 2020 also emphasizes the synergy of intelligence and networking.
V2X interconnects the car and everything, and there are currently two routes available, DSRC and C-V2X. Through communication between V2V, V2N, V2R, V2I, V2P, the vehicle has a richer source of information, predicts dangers and prepares for emergencies in advance. DSRC and C-V2X are the current mainstream car networking communication technology standards, of which the DSRC standard was developed by the IEEE based on WIFI, and C-V2X was developed by 3GPP (Mobile Communication Partner Alliance) by expanding the communication LTE standard. Among them, C-V2X includes LTE-V2X and 5G-V2X, from the perspective of technology evolution, LTEV2X supports smooth evolution to 5G-V2X, with clear evolution to 5G. As the world's top two automotive markets, China and the United States strongly support C-V2X, which will accelerate the unification of communication standards as soon as possible, and C-V2X is expected to become a unified standard for the underlying communication technology of intelligent and connected vehicles in the world.
China mainly promotes C-V2X, and the industry ecology has been initially established. In terms of chip communication modules, Mainland completed LTE-V2X related tests in 2019, entered the mass production stage in 2020, and started the standard development of 5G-V2X in 2021. In terms of on-board terminals, in 2019, the local demonstration area will open the application, and the CV2X will be installed in front of the new car in 2020, and the new C-V2X vehicle loading rate will reach 50% in 2025. In terms of road testing facilities, from 2019 to 2021, the road testing facilities in the Vehicle Networking Demonstration Zone have been basically deployed, and it is planned to expand the coverage in typical cities and highways in 2022. In terms of vehicle verification, FAW, Changan, SAIC, Geely and many other automakers have actively expressed their support. From upstream communication chips and modules to downstream test verification and operation services, the C-V2X industry ecology has been initially established.
Short-term vehicle networking has led to a significant increase in demand for on-board communication equipment, and medium- and long-term TSPs will become the core. In the short term, the networking of traditional cars will directly drive the demand for on-board communication equipment to increase significantly, and drive the expansion of the vehicle networking market. In the medium and long term, in the case of hardware equipment becoming the standard configuration of automobiles, the development of the Internet of Vehicles will develop in the direction of enriching software categories and creating a service ecology, when all kinds of on-board content and services will become the main growth points and further promote the expansion of the scale of the Internet of Vehicles; by the maturity period of the development of intelligent networked vehicles, the increment will shift from hardware to software, and TSP (Vehicle Information Service Provider) will become the core.
3.4. Outlook:
In the short term, it is still based on ADAS penetration, L3 is gradually introduced into L3
In the early stage, short-term intelligent driving of passenger cars will still be dominated by ADAS penetration. According to IHS forecast, in 2018-2020, the penetration rate of L2 and above autonomous driving in China's passenger car market has increased from 3.0% to 13.0%, and it is expected to reach nearly 34.2% by 2025, of which L2, L3 and L4 account for 30.4%, 2.4%, and 1.5% respectively, with significant growth space, but the penetration scale of L3 and above is still small, and the key to its penetration inflection point lies in data accumulation, hardware cost convergence and consumer experience upgrade:
1) Data acceleration accumulation: data accumulation is the core of automatic driving, but from the current stage of technical level, the long-tail scenario is still a key factor restricting the safety of automatic driving, and all forces are seeking technology to accelerate data accumulation, which in turn constitutes a logical closed loop between software and data;
2) Hardware cost convergence: High-order automatic driving is highly dependent on high-performance sensors, although the cost has converged in recent years, but the penetration of L3-level automatic driving in a wider range still needs to be further reduced by sensor hardware represented by lidar to be realized;
3) Consumer experience upgrade: Different from the driver-centric in the ADAS stage, L3 and above are more driven by vehicles, and its core lies in reliability and usability, which are reflected in driving safety and functional continuity, that is, under the premise of ensuring driving safety, truly improving the consumer intelligent driving experience, thereby gradually enhancing user stickiness, truly using and daring to use.
4. Smart Cockpit:
The third space differentiation appeal promotes rapid penetration
As another important component of intelligent vehicles in addition to intelligent driving, the development of intelligent cockpits mainly benefits from the upgrade of the vehicle EE architecture, that is, from traditional distribution to the current stage of domain centralization, thereby realizing soft and hard decoupling and efficient interaction between multiple screens. From the composition point of view, the intelligent cockpit is similar to the PC, of which the cockpit domain controller corresponds to the host, runs the operating system and application layer software internally, and the middleware serves as a bridge beam for the system and application layer software; the vehicle-mounted infotainment system (IVI), liquid crystal instrument, head-up display system (HUD) and so on correspond to the screen, mouse and other peripherals. Compared with intelligent driving, the intelligent cockpit is less difficult and more cost-effective, and at the same time, benefiting consumers' growing demands for car comfort and safety and the gradual migration of consumer electronics application scenarios, it is expected to meet faster penetration in the short term, and through the integration and development of hardware, human-computer interaction systems and software integration, it will continue to evolve into a "smart mobile third space" around the driving experience.
4.1. Overview:
Difficulty is low and cost-effective is expected to penetrate earlier than smart driving
The intelligent cockpit is an HMI (Human Machine Interface) system built from the perspective of consumer application scenarios, and the future focus tends to meet the emotional and personalized needs of consumers. Specific components include: 1) cockpit electronics: all kinds of electronic systems used by users in the car, the core of the hardware lies in the domain controller and chip, and the software includes the operating system, middleware and application layer software; 2) cockpit interior: to the intelligent evolution, the user can control all its functions, including seats, lights, air conditioning, etc.
The development trend of intelligent cockpit shows functional diversification and high-end configuration. In a narrow sense, the intelligent cockpit mainly includes the in-vehicle information entertainment system (IVI), head-up display system (HUD), driver monitoring system (DMS), streaming media rearview mirrors, full LCD instruments, etc. From the perspective of the new models launched by mainstream car companies in recent years, the functions of large screens, voice human-computer interaction and other functions have gradually been upgraded, and the penetration rate of HUD, DMS, and atmosphere lights has accelerated.
Differentiated appeals boost the rapid penetration of smart cockpits. According to the latest research results of IHS, the level of intelligent technology configuration in the cockpit is the second largest key element after the safety configuration, and its importance has exceeded the traditional key elements of car purchase such as power, space and price. At present, the penetration rate of new cars at the cockpit intelligent configuration level in the Chinese market is about 48.8%, and it is expected to exceed 75% in 2025.
4.2. Software:
The overall architecture is highly similar to that on the PC side
4.2.1. Hypervisor Virtual Layer
Hypervisor (also known as a virtual machine monitor, an intermediate layer of software that runs between the underlying physical server and the operating system, allowing multiple operating systems and applications to share hardware) allows multiple different operating systems to run on a single computer. Vehicle EE architecture from distributed to centralized direction of development, and in the cockpit domain controller, due to different safety requirements, need to run different operating systems (such as Linux/QNX responsible for instrumentation, Android responsible for infotainment system), through Hypervisor technology can be different operating systems can be run in the same high-performance SoC chip, a core multi-screen technology is becoming the next generation of intelligent cockpit mainstream trend.
Virtual machines have become an indispensable software system for cockpit electronics. In a virtualized environment, hardware resources such as CPU, memory, and I/O of the physical server are virtualized and scheduled by hypervisor, and multiple operating systems can share these virtualized hardware resources under hypervisor coordination, while each operating system can save independence from each other. QNX Hypervisor 2.0 features QNX SDP 7.0, a 64-bit embedded operating system that allows developers to unify multiple operating systems into a single computing platform or SoC chip capable of running operating systems such as QNX Neutrino, Linux, and Android; and QNX Hypervisor virtualization technology for cockpit platforms such as Visteon, Denso, Magneti Marelli, and WM Motors.
4.2.2. Operating System
The in-vehicle operating system is of great significance for the evolution of the car from low-level intelligence to high-level intelligence. The operating system is a bridge between hardware resources and software applications, and for complex underlying hardware resources, software developers must call through the operating system's instructions and interfaces. Through the operating system, it can organize and manage various software and hardware resources of the computer as effectively and reasonably as possible, organize the computer work process, control the execution of programs, and enable the entire computer system to run efficiently.
Car OS is divided into three worlds. The underlying vehicle operating system (OS, Operating System) mainly includes QNX, Linux, Android three camps, QNX due to its security advantages to become the new favorite of smart cars, Linux based on its own open source features and a large number of engineers continue to grab market share, Google fully exerts the advantages of Android open source and opens Android Automotive OS to third-party developers. The traditional WinCE is currently facing elimination, but with the entry of technology giants such as AliOS and Huawei Hongmeng OS and the public's proposal of self-developed VW. OS, the underlying in-vehicle operating system competition has intensified.
Most car companies develop exclusive operating systems based on the underlying in-vehicle operating system. In addition to Volkswagen's in-depth self-developed operation system, Weilai, Xiaopeng, BYD, Honda and other car companies generally choose to customize development based on open Android systems, Ford, Mercedes-Benz, BMW, Volkswagen, Volvo and other QNX system development, Tesla, Toyota, etc. are based on Linux system development, does not involve system kernel changes, belongs to semi-independent research and development.
Hongmeng OS is a microkernel-based all-scenario distributed OS. For the first time, Hongmeng OS uses a distributed architecture, which can achieve a seamless collaborative experience across terminals. Based on the microkernel design, when other modules have problems, it will not affect the operation of the entire system, and the stability of the system is significantly improved. Not only that, but the microkernel is also superior to the macrokernel in terms of scalability, maintainability, and debuggability. These advantages are exactly what the automotive OS needs, in April this year, Beijing Jihu Alpha S Huawei HI Edition officially confirmed the equipped Hongmeng OS intelligent interconnection cockpit, and then BAIC also said that this year will launch a new fuel SUV model, which will also be equipped with Hongmeng OS.
4.2.3. Middleware
Middleware is a stand-alone system software or service program that distributes applications that share resources between different technologies. With the continuous improvement of automotive application requirements, the total amount of software also increased rapidly, resulting in a sharp increase in system complexity and cost, in order to improve the management, portability, tailoring and quality of software, it is necessary to define a set of architecture, methodology and application interfaces to achieve standard interfaces, high-quality seamless integration, efficient development and management of complex systems through new models.
Middleware products are distributed computing architectures that have evolved from a two-tier structure to a three-tier structure. With the rapid development of computer and network technology, many software needs to run in a heterogeneous environment of hardware platforms and network protocols of different manufacturers, and the scale of applications has developed from local area networks to wide area networks, and the two-layer structure of the traditional "client / server" has been unable to meet the needs, and more and more users have put forward higher requirements for computer application systems.
Thanks to the acceleration of industry informatization construction, the domestic middleware market has maintained steady growth. The overall size of the middleware market in 2019 was 7.24 billion yuan, an increase of 11.4% year-on-year. With the popularization of digital technologies such as cloud computing, big data, and the Internet of Things, as well as the promotion of digital hotspot projects in industries such as government big data, smart cities, and enterprise cloud, it is expected to spawn a large number of new market demands and promote the sustained and rapid growth of market scale. At present, the domestic enterprises that provide middleware services mainly include Dongfangtong, Kingdee International, Puyuan Information, Baoland, etc.
At present, foreign manufacturers are in a leading position in the field of middleware, occupying a large share of products. Foreign companies like IBM, Oracle and other relatively early into the middleware field, there is a strong first-mover advantage (they have a complete database system, can form a complete solution), according to 2018 statistics, from the sales amount share, IBM and Oracle two foreign manufacturers occupy more than 50% of the market share, while the domestic intermediate parts market share is still low.
According to the "Kunpeng Computing Industry Development White Paper", the future middleware market has good prospects. In 2023, the global middleware market space is 43.4 billion US dollars, with a 5-year compound growth rate of 10.3%, and the Chinese middleware market space is 1.36 billion US dollars, with a 5-year compound growth rate of 15.7%. The accelerated landing of the xinchuang industry and the growing demand for cloud-based distributed application services, message queues and other middleware tools will promote the rapid development of the middleware market. According to the current middleware market structure, we predict that the global market size can reach 7.81 billion yuan in 2023 specific to in-vehicle middleware.
The expansion of demand in many aspects is driving the expansion of the middleware market. The rise of Internet technology has brought about a rich variety of new network application models, and increased the procurement demand for middleware by users in traditional industries such as telecommunications, finance, and government. At the same time, with the gradual maturity of the informatization construction of various industries, the demand for related industries is growing day by day. In addition to large-scale application systems, more and more small and medium-sized application systems have also begun to use middleware to build. Especially in the fields of e-government and small and medium-sized enterprises, middleware has become an inevitable choice for the construction of these systems, providing a broader space for the promotion of middleware.
The release of standard specifications will urge middleware manufacturers to supervise and upgrade technical standards. On the one hand, the new technical specification standard of avaEE8 defines the newly emerging technology and is compatible with the new popular development framework; on the other hand, it also points out the direction for the next technical evolution of application server middleware. With the further development of cloud computing related technologies, especially the gradual application of PAAS technology and Docker container technology, the underlying architecture of customers' business systems has gradually changed. In order to adapt to these new technologies and changes in the new environment, each middleware manufacturer needs to carry out corresponding technical upgrades and transformations of their own middleware software products.
4.2.4. Application Layer
The application layer is at the top of the software hierarchy and is responsible for the implementation of system functionality and business bare metal. With the current high degree of intelligence of the car, more and more in-vehicle applications are online, enriching the customer experience from the functional level, and the current applications are developing in a more refined, intelligent and diversified direction.
Car voice applications have gradually developed on the basis of mobile phone voice. From the initial entry-based voice to the current natural voice, NLP semantic parsing plays a very important role. Compared with the traditional speech system, the natural speech recognition system has the biggest feature of the Chinese language is deeply optimized, and the system can understand the driver's instructions without the need for stereotyped command vocabulary.
Nuance occupies a dominant position in the international market, and iFLYTEK has strong technical advantages in China. At present, Nuance occupies a dominant position in the international market, and the main cooperative car manufacturer is Ford Motor, and its technical advantage lies in the sound recognition ASR technology. In terms of the domestic voice application market, mainly iFLYTEK, Voice of the Cloud and Spich, iFLYTEK has a strong technical advantage (speech recognition technology and speech synthesis technology) with its high recognition rate of localized speech engines, and many cooperative car companies, including Volkswagen, Nissan, Toyota, Mazda, Lexus, Changan, SAIC, FAW, BAIC, Great Wall, Geely, Chery, Jacques, GAC, etc.
At present, the development direction of in-vehicle voice systems shows that the market has great potential for development. Many users have accepted and are accustomed to the current on-board voice applications, and the number of human-machine voice interactions and user stickiness have also performed well. In-vehicle voice is currently developing in the direction of simplifying operation, and as far as possible so that the car voice application can cover almost all the functions of the car, in addition, in the noise treatment, it will also be customized according to the needs of the in-car model, and the external noise is more complex pre-processed, so that users can form the habit of using voice applications.
According to the "Kunpeng Computing Industry Development White Paper", the software application market has good prospects for development. By 2023, the global enterprise application software market space will be $402.02 billion, with a five-year compound growth rate of 8.2%, and the China enterprise application software market space will be $15.58 billion, with a five-year compound growth rate of 11.7%. At present, there are many and scattered enterprise software in the market, strong regional attributes and industry attributes, and a high degree of customization. With the advent of new computing platforms and the high degree of compatibility of future systems, a series of new industry standards and models will be introduced, which will prompt enterprise applications to move towards a certain degree of concentration.
OEMs are currently becoming service providers and leaders of software applications. The final boundaries of the current on-board software system user service are not clear, when the software application has a problem, the user will be the first time to blame the poor sense of experience on the car manufacturer, rather than the software provider, so the car manufacturers will have to carry out the final responsibility and after-sales service of these software in the future, and with the increasingly diversified application "on the car", this problem will be more obvious.
Data attribution and software charging have become two major contradictions between OEMs and software applications. When there is a problem with the quality of software applications and services, automotive manufacturers need to directly obtain user data information for maintenance, and when software vendors want to improve the product experience, they need to obtain product data from manufacturers, and it is difficult to reach an agreement on data attribution issues, such as map information data. Secondly, application maintenance will create a series of cost problems, and both the car manufacturer and the user want to benefit from this aspect, so the current car manufacturers generally negotiate this aspect when looking for a third-party supplier.
The model of delivery of in-vehicle infotainment services is the key to the competition. There are two ways to provide on-board information and entertainment services: one is to provide it to the vehicle terminal through the vehicle networking service platform, mainly led by the vehicle manufacturer, and the Internet and suppliers support the development; the other is to project the content of the smart terminal such as mobile phones on the screen through the projection mode, mainly dominated by Internet companies, and the vehicle manufacturers and suppliers are in a passive position. At present, car manufacturers and Internet giants are laying out the in-vehicle infotainment service market.
4.3. Hardware: Incremental component penetration
4.3.1. Smart Cockpit Chip + Smart Cockpit Domain Controller
Intelligent cockpit one-core multi-screen technology has gradually become popular and has become a mainstream trend. Compared with the traditional multi-chip solution, the total cost of a single processor under the condition of multi-screen drive will be lower than that of a multi-processor, and it can provide a full-range intelligent interconnection experience such as multi-screen interaction, and the complexity of the multi-screen operating system is reduced to ensure driving safety. "One core, multiple screens" has become a development trend, and the chip itself will also develop in the direction of miniaturization, integration and high performance. From the perspective of car companies, Tesla Model 3, Volkswagen ID 3, etc. all use a core multi-screen solution. From the perspective of manufacturers, Desay SV and NVIDIA are also constantly developing a core multi-screen chip technology.
High-computing power smart cockpit chips will be a key factor in determining the performance of smart cockpits. With the gradual homogenization of automotive products, the traditional chip computing power can not support user needs, the user's demand for car cockpit function dimension continues to expand, from the big data survey can be seen, in 2020, the user's demand for intelligent configuration of the cockpit is as high as 78.7%, which makes the future cockpit more taking into account the "active intelligence" and "content + service". Traditional cockpit chips are difficult to support from a single "security" requirement to multiple service requirements, so the high-computing power AI intelligent SoC chip came into being.
7nm process, the mainstream of high-performance intelligent cockpit chips. With the emergence of the "one core, multiple systems" scheme, higher requirements have also been put forward for the computing power, power consumption, interface type and number of the main chip SoC. Therefore, in the terminal application market, the 7nm process has begun to become the mainstream of high-performance intelligent cockpit chips, and the main automotive chips at this stage are advancing towards 7nm or even 5nm. In addition to Qualcomm, at present, NVIDIA, Samsung, and so on have a layout, and new domestic chip forces, including Horizon, Xinchi Technology, and Xinqing Technology, have also begun to rise.
Smart cockpit chip manufacturers are increasingly competitive. At present, the main players of smart cockpit chips include traditional automotive chip manufacturers such as NXP, Texas Instruments, and Renesas Electronics, mainly for the low-end market. Manufacturers such as Qualcomm and Samsung in the field of consumer electronics are mainly for the high-end market. As the first NVIDIA company selected by luxury brands and equipped with intelligent cockpit systems, due to more energy in recent years in the field of automatic driving, and missed the first wave of intelligent cockpit "computing power" market competition, Resulting in Qualcomm almost sweeping the high-end car cockpit market in the past few years. However, as smart cockpits become standard in new cars, a new market growth cycle is coming.
Qualcomm's smart cockpit chip is high performance, and many automakers are seeking cooperation. Through the comparison of mainstream intelligent cockpit domain chip parameters, Qualcomm chip algorithms and processes are significantly better than other manufacturers. Qualcomm has partnered with 18 of the world's 25 largest automakers in smart cockpits. In addition, Qualcomm will continue to work with a number of automakers and Tier 1 suppliers to provide next-generation technology for vehicles. Some models of the top brands Mercedes-Benz, Audi and Porsche in traditional car companies, as well as the vehicles of Xiaopeng and Xiaopeng, which are ideally equipped with Qualcomm Snapdragon's digital cockpit platform, have been mass-produced and listed.
Domestic intelligent cockpit chip technology is still in the stage of continuous improvement. Compared with the research and development of smart cockpit chips of foreign enterprises, domestic chips started late, and domestic smart cockpit technology is still in the learning stage. However, whether in terms of the degree of intelligence or the rich application ecology, the average level is higher than that of the international, and the layout strength and technology are constantly improving. For example, Huawei and TSINGHUA Unigroup have launched 5G communication intelligent cockpit chips; Unigroup Zhanrui launched the first 6nm process T7520 chip this year. We believe that China's smart cockpit chip market can be expected, and once the technology matures to the level of mass production, it will be able to seize some markets.
From distributed to centralized, cockpit domain controllers play a major role in the smart cockpit hardware platform. Cockpit Domain Controllers are able to integrate multiple or groups of electronic control unit ECUs into a single controller, achieving seamless human-computer interaction through a converged integrated interconnection ecosystem in terms of safety, small volume, small energy consumption, light weight, and low cost. Among them, the intelligent cockpit domain is the core of the future of the automobile, and it is also the key core point of the future differentiated competition of car companies to achieve hardware profitability, and the cockpit domain control will become the largest increment of intelligent driving.
Cockpit domain controllers are growing rapidly, with more than 6 million cockpit domain controllers shipped worldwide in 2025. Due to the low difficulty and relatively controllable cost of smart cockpit mass production, the application of automotive 5G networks around the world will accelerate the launch of smart cockpits. According to Visteon data, global cockpit domain controller shipments were approximately 400,000 units in 2019 and 800,000 units in 2020. We expect average annual growth of 50.9% in 2021-2025 and 18% per annum in 2026-2030. Shipments will exceed 6 million units by 2025 and more than 14 million units by 2030.
In recent years, many manufacturers around the world have laid out cockpit domain controller solutions. From the perspective of the global market, Visteon, Continental, and Bosch occupy a dominant position in the cockpit domain controller market, and domestic enterprises Huawei, Desay SV, Huayang Group, Hangsheng Electronics, Neusoft, etc. have also launched cockpit domain controller solutions, making the competitive trend of intelligent cockpit domain controllers more and more obvious. According to the comparison of the revenue of automotive electronics related to various manufacturers, we conclude that Denso accounts for the largest proportion of operating income among global companies, followed by mainland Germany. Focusing on the development of smart cockpit domain controllers, Visteon released the world's first cockpit domain controller, Smart Core, in September 2018, with revenues of $2,548 million in 2020. Desay SV, which is committed to the intelligent cockpit domain controller research in China, reached 1,042 million US dollars in automotive electronics revenue in 2020, an increase of 27.4% year-on-year.
Visteon accelerates the commercialization of integrated cockpit projects. In 2021, Visteon, Together with Ikatong Technology and Qualcomm, we announced that we will join forces to provide the latest advanced intelligent cockpit solutions for the global market. The solution can be widely used in traditional fuel models and new energy electric models, and will take the lead in mass production on Geely Automobile Xingyue L.
Together with BlackBerry, Desay SV launches the Smart Cockpit Controller. In March 2021, BlackBerry announced the launch of a dual-screen virtual smart cockpit domain controller with Desay SV for improved driving safety. The intelligent cockpit product has been mass-produced on a large scale in the models of GAC Passenger Vehicle, Great Wall Motor, Changan Automobile, Chery Automobile, Ideal Automobile, and many other leading domestic car companies, and integrates cockpit products and cockpit domain controllers to enter a new stage of rapid increase in large-scale sales, and has been officially applied to the Tiggo 8 Plus and Jietu X90 models under the Chery brand.
Smart cockpit controllers are in a phase of rapid growth, with a market size of $20 billion by 2025. From the mechanical age to the intelligent era, with the advent of the 5G era and the continuous upgrading of intelligence, the intelligent cockpit will usher in rapid penetration, of which the cockpit domain controller will also grow rapidly. From 2019 to 2020, the overall market size of China's intelligent cockpit domain controllers is 3.5 billion using $4.7 billion, we assume a compound annual growth of 33.7% from 2021 to 2025, and a growth of 23.7% from 2026 to 2030, calculating that the market size of 2025E will reach 20 billion US dollars, and the 2030E will reach 57.9 billion US dollars.
4.3.2. In-vehicle infotainment system IVI+ LCD meter
In-vehicle infotainment systems (IVI) are versatile and diverse. IVI is an on-board integrated information processing system based on the interconnection of body bus connections. IVI is divided into two categories: in-vehicle information system and in-vehicle entertainment system, including navigation and positioning, real-time road condition PTV, vehicle service, multimedia, wireless communication, online-based entertainment functions and TSP services and a series of applications, the development of IVI has greatly improved the level of vehicle electronics, networking and intelligent technology.
The in-vehicle infotainment system framework is mainly divided into two parts: hardware and software. Among them, the IVI hardware cost accounts for 69%, and the hardware framework is mainly composed of the main CPU + MCU and peripheral device control, power module, audio and video code, and Bluetooth module. The CPU is mainly the architecture of the OS and the processing of various applications, the MCU is mainly used to control and the communication of the in-car network, the mainstream CPU chips are Freescale's I.MAX6/8 and TI's jacinto 5/7, etc., and the MCU chips mainly include Renesas RH850 and so on.
In terms of underlying technology, one core and multiple screens are the future trend. The intelligent cockpit screen is expanding from the single screen in the past to multiple screens including the central control screen, co-driver, rear seat, electronic rearview mirror, reversing mirror, HUD, etc., which are responsible for vehicle information display, entertainment, navigation, driver monitoring and other functions. Comprehensive cost and security and other measurement factors, a core multi-screen will gradually replace the current mature multi-core multi-screen solution, Huawei Hongmeng, Tesla Model 3, Volkswagen ID 3, etc. have proposed a one-core multi-screen solution, ideal One for its four-screen interactive IVI chose a dual-core dual-system solution.
Compared with multi-core multi-screen, the technical threshold of one-core multi-screen is high compared with the previous investment. According to the data of Gaogong Intelligent Automobile Research, the cost of the car-machine system includes: 1) chip licensing fees of millions of yuan for chip manufacturers; 2) technical support costs of millions of yuan for software system developers; 3) development of operating system SOP solutions: tens of millions of yuan. Since the one-core multi-screen solution requires a more powerful processor and a more complex software operation system, its R&D investment in the process of technical threshold and architecture design verification will increase significantly. At present, the development price of the whole set of electronic cockpits based on the high-end platform is high, the return on investment is not high, and only the head enterprises are involved.
In terms of interaction mode, the IVI system has developed into a "person looking for a car" type of intelligent interaction, and takes the "car looking for a car" type of active intelligent interaction as the future development direction. As the core of the intelligent cockpit, the intelligent essence of the IVI system lies in the evolution of the interaction mode. Taking the on-board AI voice function as an example, the future human-vehicle interaction mode will be guided by the car, and through the algorithm design and the cooperation of hardware such as sensors, multi-modal interaction forms including lip movement detection, line of sight tracking, and global gesture recognition will be realized.
With the emergence of market diversification and consumer demand, the number and resolution of IVI in-vehicle displays are increasing. The mainstream of the display in the market is the IVI touch display, which has shown an increase in size and quantity in the mass production of new cars and concept cars of various car companies in recent years, such as the Audi A4L. We expect 3 in-vehicle displays to be the future, and the resolution is increasing year by year. It is estimated that by 2030, the resolution of more than 1 million can reach 41%. Future upgrades are more reflected in the diversification of screen forms, while integrating more functions to provide immersive, personalized, and customized AI services.
At present, the IVI industry market competition is large, and the industry presents three major echelons. The first echelon companies have the ability to develop automotive LCD instruments, leading the innovation and development of the industry, occupying most of the market share. Such as Ambov, continental, etc. The products of the second echelon can be matched with joint venture automobile OEMs and self-owned brand auto OEMs, such as Desay SV. The third echelon of companies have the ability to produce automotive LCD instruments, such companies have a large number of companies, and their market is mainly the automotive aftermarket repair market.
In the 2020 domestic IVI competition pattern, Aptiv has a strong momentum of development. Aptiv's annual report shows that due to the impact of the epidemic in 2020, the annual revenue fell by 9%, but Aptiv has been committed to the development of smart cars and accelerating closer to a more intelligent car model. Due to the company's market strategy and technology-based products, in the 2021 quarterly report, the first quarter turnover was $4 billion, an increase of 25% year-on-year. Actual turnover increased by 20% due to factors such as international market exchange rates, commodity changes and divestiture adjustments.
The Development Momentum of the Sino-German SEV IVI System in China is strong. Among the total revenue proportion of domestic IVI-related enterprises, Desay SV accounted for the highest 67.58%, the company's main business in 2016-2020 IVI five-year average accounted for 63%, in 2019 IVI accounted for 75.6%. The company's long-term core products infotainment system and liquid crystal instrumentation have achieved good development, of which display modules and systems, liquid crystal instruments have achieved sales growth of more than 100% year-on-year, sufficient order reserves will promote its sustained and rapid development. In 2020, the company's infotainment system business has successively broken through white-spot customers such as Toyota (Indonesia) and Maruti Suzuki (India), and has also obtained new project orders from customers such as FAW-Volkswagen, Changan Ford, and GAC Toyota.
IVI penetration and market size have increased significantly. According to the future think tank 2019-2020 IVI national market space of 320/355 billion yuan, we predict that in 2025 the domestic market space will reach 52 billion yuan, the global market will reach 191.4 billion yuan. Market penetration will increase from 81% in 2019 to 95% in 2025. Drive the rapid development of the smart cockpit field to bring customers a higher performance experience.
The automotive LCD instrument market is dominated by foreign capital, and its total market share accounts for about 50%. Among them, the market shares of bosch, DENSO and Continental group foreign-funded enterprises were 15.6%, 11.9%, 11.2% and 10.8% respectively. From the competition situation, it can be seen that Chinese enterprises have a certain gap with foreign enterprises in terms of technical ability and R&D level, but Chinese enterprises such as Desay SV are also emerging in the global market.
China's LCD instrument enterprises have begun to take shape, with development and production supporting capabilities. From the overall analysis, at present, China's domestic-funded automotive LCD instrument manufacturers are still in the catch-up stage in technology. Some domestic-funded automotive liquid instrument enterprises have increased their investment efforts, adopted the acquisition of technology companies, strategic cooperation and other cutting-in programs, and have achieved supporting facilities with independent brand automobile OEMs in the international market.
Lcd dashboards have been driven by the expansion of the smart cockpit market, and their market penetration has risen rapidly. At present, the value of LCD instrument cluster bicycles is about 2,500 yuan, assuming a yearly decline of about 2% in the future, automobile production will increase by 2% annually, we expect that the overall market penetration rate of LCD instruments will increase to about 70% in 2025, and the global market space in 2025 will be about 155 billion yuan, and the domestic LCD instrument market space will be about 48 billion yuan.
4.3.3. Head-up display HUD
HUDs improve driving safety, and AR-HUDs are the mainstream trend of the future. HuD can project driving information such as speed, fuel consumption, navigation, speed limit and so on on on the front windshield of the car, so that the driver can obtain information without looking down at the dashboard or navigation, improving driving safety. At present, it mainly includes C-HUD (application model: Onxella /CX-4), W-HUD (application model: Audi A8L/NIO ES8), AR-HUD (application model: new generation Mercedes-Benz S-Class/Volkswagen ID.6). In recent years, the cost of W-HUD has been reduced, becoming the mainstream configuration of HUD, AR-HUD will become a future trend because of its excellent visual effects and the characteristics of integration with ADAS.
The technical principle of HUD is that the image source is reflected into the human eye after multiple optical reflections. At present, the mainstream TFT and DLP image source technologies require optical reflection of the light source, and then projected to the front windshield, and finally reflected into the human eye. The optical lens used for general reflection is a free-curved lens, in order to ensure the quality of reflection, the lens needs to have high precision, high surface flatness, no rainbow pattern, anti-fog, anti-static and other characteristics.
AR-HUD has become a new outlet in the industry, and various car companies and technology companies have actively laid out. In 2020, approximately 5 million vehicles sold will be equipped with HUD technology. As major automakers continue to adopt AR technology to complement other enhanced connectivity capabilities in their vehicles, the number of vehicles equipped with the technology is expected to double by 2025. At the level of car companies, the 2021 Mercedes-Benz S-Class sedan, SAIC with ID.4X, and FAW Hongqi EHS-9 are all equipped with AR-HUD. At the level of technology companies, in December 2020, Future Black Technology and Horizon signed a strategic cooperation agreement on "jointly developing AR HUD business and exploring future human-computer interaction technologies". In July 2021, Corning partnered with Hyundai Mobis to launch the Corning Curved Rearview Mirror Solution for an immersive, connected driving experience.
Cost reduction and automotive intelligence will accelerate the development of HUD, and many car companies have already assembled HUDs. 1) The core device technology breakthrough brings about cost reduction. With the gradual maturity of technology and large-scale mass production, the cost of HUD has shown a downward trend, of which the cost of front stop imaging glass has dropped by more than 1000 yuan; 2) HUD is in line with the trend of automotive glass safety and intelligence. The UPcoming AR-HUD, which will be launched with a number of new models, can integrate ADAS to mark vehicle, pedestrian, and road sign information in real time, providing drivers with an intelligent driving experience. 3) The rapid penetration of ADAS and the gradual maturity of AR technology will promote the development of AR-HUD. ADAS penetration rate has increased rapidly, and AR car navigation has also been installed on Roewe Marvel X, Chery Starway and other models.
HuD has a considerable market penetration rate, and is currently mainly dominated by high-end models. We analyzed the models with more than 10,000 vehicles in 2020 and concluded that 11% of the models are equipped with huDs or optional HUDs, assuming that models with less than 10,000 insured vehicles are not equipped with HUDs, the market penetration rate of HUDs is about 9%-10%. Among them, the penetration rate of mid-to-high-end models is 25%, and that of low-end models is 1.8%. We expect HUD penetration to accelerate in 2021 and reach around 29% by 2025.
The main model of the HUD is still dominated by the German high-end model. The ratio of HIGH-END German models to HUDs reached 35%, followed by Japanese, autonomous and American models, with high-end model configuration ratios of 21%, 17.5% and 17%, respectively. In contrast, the Korean model performed below the market average, and the model was almost unassembled with a HUD system. We expect that in 2025, the penetration rate of German, Japanese, and autonomous HUDs will reach 42%, 30%, and 25%, respectively.
The global HUD market is expected to reach 6.2 million units in 2025. The main assumptions for 2020-2025 are that the growth rate of global automobile production will be 5.0%, 2.0%, 2.0% and 2.0% respectively, and the production structure will remain stable. HuD penetration rates for all types of commercial vehicles and passenger cars are increasing.
2021 is an important time period for AR-HUD to achieve mass production. Automotive AR-HUDs are becoming a new industry technology trend. In addition to the Mercedes-Benz S-Class and Volkswagen ID series, which are already equipped with AR-HUDs and launched in 2020, the new Audi Q4 e-tron in 2021 and the Mocha SUV of the Wey series will also have ARHUDs. Companies are re-launching HUD solutions, and several are already pushing huD technology forward. The intensive layout of AR-HUD mass production by domestic and foreign giants and start-ups will promote its large-scale production upgrade.
The global HUD market is highly concentrated, and individual start-ups are expected to enter the first array of the domestic market. At present, the global HUD market concentration is high, and the three major suppliers of Japan Seiki, Continental, and Denso account for 83%. Among them, Japan Seiki accounted for the largest proportion, reaching 35.4%; at present, the domestic HUD suppliers mainly include Huayang Group, Crystal Optoelectronics, BOE and other large listed companies, as well as Zejing Electronics, Future Black Technology, Ruisi Huachuang, Dianshi Innovation, and other start-up enterprises, among which Zejing Electronics, Future Black Technology, Ruisi Huachuang are expected to actively integrate into the pace of China's automotive industry 4.0 in terms of AR-HUD industrial layout, and collaborate with upstream and downstream industry chain partners to achieve AR-HUD application ecology. Take advantage of the smart cabin dividend to enter the first array of the domestic market.
4.3.4. Canopy glass
The industrial chain of automotive glass is divided into upstream float glass, PVB and downstream vehicle manufacturers. Automotive glass area accounts for about 1/3 of the total surface area of the car, according to the processing method, can be divided into: laminated glass, tempered glass, regional tempered glass. According to the different parts, it can be divided into: front windshield, rear windshield, sunroof glass, front triangular window glass, rear triangular window glass, front door glass, rear door glass.
In 2020, the equipping rate of new models (excluding modifications) of sunroof/sky curtain on the mainland is as high as 82.6%, and the equipment rate of panoramic sunroof/skylight is as high as 61.5%. Key models such as Geely Xingrui, Haval Big Dog, Hongqi H9, Changan UNI-T, Xiaopeng P7 and so on are equipped with panoramic sunroof/sky canopy. The average area of the electric sunroof of ordinary cars is only about 0.3 square meters, the average area of the panoramic sunroof is about 1.2 square meters, and the average area of the panoramic skylight is about 2.0 square meters, of which the panoramic canopy glass area of many models of SAIC Roewe is as high as 3.0 square meters.
The penetration rate of canopy glass is considerable, and it is expected that the penetration rate will reach 10.4% in 2025. The permeability of 2020/2021H1 curtain glass was 2.3%/2.6%, respectively. Among them, the penetration rate of high-end automobiles (more than 150,000) is 3.3%, the penetration rate of low-end (100,000-150,000) is 2.8%, and the penetration rate of low-end (less than 100,000) is 0.9%. We predict that by 2025, it will increase to about 10.4%, and the penetration rate of the canopy glass industry will increase rapidly after 2021.
Canopy glass has penetrated into low-end models. At present, the average selling price of vehicles equipped with panoramic canopy is about 305,000 yuan, 43.6% of the models are high-end models with a price of more than 150,000 yuan, and 44% of the models are low-end models with a price of less than 150,000 yuan, and there are many models with a price of less than 100,000 yuan, and the penetration of low-end models is considerable. We expect the penetration rate of low- and mid-end models to reach 12.8% in 2025, higher than the penetration rate of mid-to-high-end models of 11.3%.
The penetration rate of independent brands is significantly higher than that of other car series, and it is expected that the global sky glass market demand will reach 2.32 million sets in 2025. According to the penetration rate of the sub-vehicle series on sale in 2020, the penetration rate of the autonomous car series is 4.15% and the Penetration Rate of the Japanese series is 3.27%, and we expect the penetration rate of the autonomous and Japanese car series to reach 13.5%/11.1% in 2025. Assuming that the global automobile production growth rate from 2021 to 2025 is 5%, 2%, 2%, 2%, and 2%, respectively, and the production structure remains stable. We expect the market demand for canopy glass to reach 2.32 million units in 2025.
The global competition pattern is concentrated, and the position of the four major oligarchs of Fuyao Auto Glass is stable. In 2020, Fuyao's global auto glass market share reached 22.0%, achieving a revenue of 19.907 billion yuan, a year-on-year increase of -5.36%. In 2020, the revenue of AGC, Saint-Gobain, Plate Glass and Xinyi Auto Glass was equivalent to about RMB20.583 billion, RMB12.277 billion, RMB17.886 billion and RMB3.879 billion, respectively. According to estimates, the global vapor and glass market size in 2020 is about 80.077 billion yuan, so the market share of Fuyao, AGC, Saint-Bangor, Plate Glass and Xinyi is 22%, 26%, 15%, 22% and 5% respectively, and the market share of Fuyao is in second place. At the same time, the auto glass industry is a heavy asset industry, the entry barriers are high, and Fuyao, as the oligopoly enterprise with the second largest market share in the industry, is less affected by external risks.
Fuyao Glass is in better operating condition than competitors and continues to promote R&D and innovation. Fuyao gross profit margin and attributable net profit margin far exceed other manufacturers. From 2015 to 2020, Fuyao achieved gross profit margins of 42.4%, 43.1%, 42.8%, 42.6%, 37.5% and 40.0%. The gross profit margins of AGC, Saint-Gobain and Plate Glass in 2015-2020 were stable between 22.0%-28.0%, far below the level of Fuyao. 2015-2020 Fuyao achieved attributable net profit margins of 19.2%, 18.9%, 16.8%, 20.4%, 13.7%, 13.5%. The net profit margin attributable to the mother of AGC, Saint-Gobain and Plate Glass in 2015-2020 is less than 10%, and Saint-Gobain and Plate Glass even have negative attributable net profits, and the profitability is far worse than that of Fuyao.
Fuyao ROE continues to outperform its competitors. From 2015 to 2020, Fuyao achieved ROE of 20.7%, 18.3%, 17.0%, 21.0%, 13.9%, and 12.06%. In addition to the high fluctuations in the ROE of other manufacturers and many times negative, AGC and Saint-Gobain 2015-2020ROE are all below 9%, which is continuously lower than the level of Fuyao. Fuyao's high R&D expense rate ensures that future products keep up with the trend of the automotive industry: from 2015 to 2020, Fuyao's R&D expense ratio is 4.4%, 4.4%, 4.3%, 4.4%, 3.9%, 4.1%. The AGC R&D expense ratio is stable at about 3.1%, and Saint-Gobain is stable at about 1.1%, both lower than the Fuyao level.
In 2020, Fuyao's auto glass business was the least affected by the epidemic, and the siphon phenomenon may be strengthened after the epidemic. Due to the impact of the epidemic, the revenue of Fuyao Automobile Glass in 2020 was -5.36% year-on-year. In 2020, the revenue of AGC, Saint-Gobain and plate glass auto glass business was -28.74%, -28.0%, -26.8% year-on-year, and the decline rate was higher than that of Fuyao, and the degree of impact of the epidemic was seriously affected by Fuyao. Fuyao focuses on the auto glass business and provides customers with high-quality services. After the epidemic, Fuyao quickly resumed production, and its high reputation, fast response, and strong ability may further increase the head effect, enhance the siphon phenomenon, and increase the market share of Fuyao.
The rapid rise of the intelligent cockpit has caught up with foreign companies and become a world-class enterprise. Other auto glass manufacturers are laying out intelligent auto glass, and Fuyao may win with service capabilities and responsiveness. Fuyao has the highest R&D expense rate and the most complete layout of new products of automobile and glass. Fuyao focuses on the auto glass business, and its sensitivity, responsiveness and service capabilities to the auto glass market are stronger than other manufacturers, or help them further expand their market share. Fuyao resumed production the fastest after the epidemic, and the intelligent transformation of products was successful, or the market share of small manufacturers that failed to transform new products due to the collapse of the epidemic and the failure of new product transformation. Therefore, we believe that Fuyao Glass, as the leader of the domestic automotive glass industry, will continue to enhance the value of the enterprise, take advantage of the rapid rise of the intelligent cockpit, and catch up with foreign enterprises to become the world's first-class.
4.4. Outlook:
Ecological construction as the core,
The importance of software and hardware integration capabilities has increased
The basic composition of the intelligent ecosystem - the development of chip and algorithm enterprises. Cockpit chip suppliers and algorithm research and development Enterprises are the core participants of the intelligent ecosystem, which will provide strong technical support and computing power support for various scenario services in the "third space", and the continuous evolution of related product technical capabilities will also continuously improve the user experience and make the services of the intelligent ecosystem more abundant.
Ecosystem cooperation model. After the car cockpit enters the era of high intelligence, AI capabilities are greatly improved and strongly enter the cockpit system, and the future ecological cooperation mode will become inevitable. It is expected that the future ecosystem cooperation model will revolve around vehicle driving function services, mobility services, vehicle aftermarkets, and life and entertainment services involving driving and riding.
Ecosystem cooperation model - ecological synergy of the intelligent cockpit ecosystem. With the development of intelligent cockpit technology, its industrial chain will continue to extend, and more players will enter this big ecosystem to jointly provide consumers with more additional value-added services - the theme of the future development of the intelligent cockpit ecosystem will revolve around "ecological collaboration" and "cross-border extension", mainly manifested in the mainstream intelligent cockpit system to the low-relevance service expansion of the vehicle in the process of user service ecological construction.
The value of cockpit software will be greatly enhanced under the trend of intelligence. The development of intelligent cockpits has gradually led to the increasing complexity of automotive software applications, such as operating systems, infotainment and other important control domains, which require highly complex software to support, so the value of software and the value of additional experience will continue to increase.
The application of in-vehicle display collection 3D AR technology will be even more exciting. With the application of new technologies, in-vehicle display towards the direction of large-screen curved surface, multi-screen integration and interaction, 3D display, AR technology will also bring new breakthrough points, the future cabin on-board display system will also be more intelligent.
The future development trend and application trend of human-computer interaction. It is difficult to completely replace each other in a variety of ways of human-computer interaction, and the cockpit interaction experience also needs to be combined with a variety of input methods to optimize the driving operation: the future will present a multi-mode interaction and active interaction trend, bringing users an emotional, scenario-based, and personalized experience.
The overall loading trend of sensors will rise. Human-vehicle interaction will no longer be limited to keys, touch and voice, etc., voice assistants, gesture recognition, fingerprints, sound source positioning, face recognition, holograms and other human-vehicle interaction methods have appeared in the listed models. The design of different human-vehicle interaction scenarios and functions requires a large number of sensors as support, and the loading volume of sensors will be greatly increased in the future.
"Health + intelligence" has become an important direction. The new crown epidemic has promoted car companies to accelerate the research and development and innovation of healthy cars and healthy cabins. In the short term, car companies have launched intelligent air purification systems; In the long run, technologies such as real-time monitoring, active sensing, and biological monitoring will further promote the development of the "health + intelligence" cockpit.
Long-term cross-border extension of the industrial chain, in-depth cooperation, and differentiated competition. At present, various car companies, parts and components companies, Internet and technology companies have carried out in-depth cooperation across fields, forming a "network-like" industrial structure, breaking through the boundaries of the traditional industrial chain, or forming a new industry pattern.
The digital cockpit covers the whole scene of travel and becomes an "intelligent mobile space". With the development of digital technology, the future intelligent cockpit function will be intelligently changed from the five directions of interaction, environment, space, control and data, so as to realize the user's scenario-based and personalized experience.
5. Pattern Outlook:
Forces from all sides poured in, and the industrial order was reshaped
Intelligent electric vehicles span many fields such as automobiles, electronics, computers, and IoT, spawning trillion-level market space, and software and services open up new profit channels, not only for the comprehensive transformation of traditional car companies, but also for the new momentum of car manufacturing, and the active entry of science and technology Internet enterprises to bring historic opportunities: new power brands make full use of Internet thinking, select tracks to achieve breakthroughs, Huawei positioning intelligent electric vehicle incremental component providers, Baidu, Xiaomi first officially announced the next car manufacturing, DJI, OPPO, Skyworth, etc. also actively layout. Many parties work together to promote the vigorous development of the industry. At the same time, with the development of electric intelligence, the relationship between the traditional automobile industry chain is reshaped, Tier0.5 suppliers appear, the status of the industrial chain is moved forward, the domestic replacement space of independent parts suppliers is vast, and with the improvement of technical capabilities, it is expected to rely on the Chinese market to rise as a global parts leader through global supporting the trend.
5.1. Business model reshaping,
Software and services open new avenues for profitability
Intelligent focus to provide users with full life cycle services, the automotive industry ushered in a new profit growth point. The whole life cycle management of automobiles is divided into three stages, namely pre-management, mid-term operation and maintenance management, and late rotation and scrap management. Intelligence can effectively help car companies quickly respond to user service needs in the whole life cycle of the car, deeply participate in the management of the whole life cycle of the car, and further tap new profit growth points.
Reason one: Car sales are changing from "a hammer and a deal" to a "start to make a profit". The automobile sales business of traditional car companies often stops at delivering cars to users, and in the context of automobile intelligence, automobile sales are changing from "hammering the hammer" to "the beginning of profitability". Taking Tesla as an example, in the whole life cycle of the car, Tesla provides users with a one-time buyout/monthly subscription and OTA upgrade service of autopilot software, and the car can continuously introduce new features to improve vehicle performance through over-the-air software updates.
Reason two: Intelligent booster car companies from To B to To C transformation, closer to users. Traditional car companies mostly use the 4S shop distribution model, in this mode, the car company is only responsible for production and sales, while after-sales service, car maintenance, vehicle repair, etc. are usually completed in the 4S shop. Under the wave of intelligence, car companies innovate and adopt the direct sales model to connect front-end users and back-end research and development, so that they can directly understand the needs of users in the whole life cycle of the car and respond quickly.
Software is deeply involved in the entire life cycle of the car and reshapes the automotive business model. The overall architecture of intelligent vehicles includes: 1) hardware platform, heterogeneous distributed hardware architecture, 2) system software layer, 3) application middleware and development framework, and 4) application software layer. The role of software in automotive products has become increasingly important, and the emergence of "software-defined automobiles" has led to a change in the business model of the automotive industry, from a long-term dependence on the manufacture and sale of new cars for profits, to a larger ownership market fee.
Tesla pioneered software charging, opened a new channel for profitability, and led the business model reform of the automobile industry. Tesla software fee items can be divided into three categories, including: 1) Automatic Driving Optional Package (FSD), including advanced autonomous driving functions such as automatic navigation assistance driving and intelligent summoning; 2) OTA upgrade optional package, which continuously introduces new features and improves performance through over-the-air software updates; and 3) advanced internet of vehicle services, including real-time road conditions, karaoke, streaming and other functions. The FSD option currently contributes major software revenue to Tesla.
The FSD option rate continues to increase, and the shift in the billing model is expected to boost software revenue growth. In April 2019, Tesla Autopilot function standard, FSD began to be selected, with the gradual opening of the function, the price of FSD continued to climb, the latest price has reached 10,000 US dollars (64,000 yuan in China). In addition, Tesla launched a monthly subscription model in July this year, which costs $99 or $199/month, which is expected to attract more user experiences and use the FSD optional package, and software fee revenue is expected to continue to climb.
Under the one-time buyout fee model, FSD revenue is expected to reach $12.9 billion and gross profit of $9.7 billion in 2025. According to Tesla CFO Zachary J. Kirkhorn introduced at the 2020Q1 performance call, the recognition rules for Tesla software revenue under the one-time charging model are: users will pay cash in one lump sum after choosing to load FSD, and the company will recognize half of the revenue as current revenue, and the other half as deferred revenue, which will be recognized when the new features are launched later. Based on the above rules and the prediction of the three core variables of car sales, option rate and price, we expect FSD revenue to reach $12.9 billion and gross profit to reach $9.7 billion in 2025:
1) Car sales: Tesla is expected to sell more than 3.2 million units in 2025. On the one hand, global new energy vehicle sales continue to rise, 2020 sales growth rate of 41%, penetration rate of 4%; on the other hand, Tesla product matrix continues to enrich, the existing products include ModelS, ModelX, Model3, ModelY, from 2022 will be mass production delivery Cybertruck, Roadster2, Semi, Tesla's global market share is expected to further increase. So we expect Tesla to sell more than 3.2 million units worldwide in 2025;
2) Option rate: The FSD option rate is expected to reach 45% in 2025. According to foreign media, the current FSD selection rate in North America has reached 25%-30%, and in China it is 1%-2%. Tesla's rising sales will accelerate the accumulation of data, which in turn will increase the level of FSD localization, in addition to the advancement of autonomous driving technology will also increase the penetration rate of FSD, and it is expected that the FSD option rate will reach 45% in 2025;
3) Price: There is a continuous upward trend, and the FSD is expected to sell for $12,000 in 2025. Since the start of the option in April 2019, FSD has gone through four price increases, from $5,000 to $10,000. With the continuous enrichment and improvement of autonomous driving functions and the increase in demand, the price of FSD has continued to rise. With a conservative estimate of 5% price growth, FSD will sell at $12,000 in 2025.
Under the monthly subscription model, FSD incremental revenue is expected to reach $1.5 billion in 2025 and $7.9 billion in FSD incremental revenue in 2030. The monthly subscription model will effectively lower the purchase threshold and attract more stock users to experience and purchase, based on car ownership, option rates and price forecasts, it is expected that the unfladD incremental revenue will reach $7.9 billion in 2030.
NIO provides users with the NIO Pilot software for the autonomous driving system in the form of paid services. In October 2020, WEIO's autopilot assisted driving NOP function was officially launched, becoming the second car company to launch this function after Tesla's NOA function. NIO Pilot has two types for users to choose from, including the select package priced at 15,000 yuan, covering some of the common functions of the NIO Pilot automatic driving assistance system, and the full package priced at 39,000 yuan, equipped with all the functions of the NIO Pilot automatic driving assistance system.
Xiaopeng provides users with the software of the automatic driving system XPILOT 3.0 in the form of paid services. In January 2021, Xiaopeng NGP function was opened for the first time, which is the third after Tesla and Weilai. Xiaopeng XPILOT 3.0 lifetime subscription before payment 20,000 yuan, delivery lifetime subscription 36,000 yuan. As of the end of February 2021, about 20% of P7 owners purchased and activated the feature, with a 25% payment rate for P7 software in March, and 2021Q1 the company recognized revenue of $80 million for autonomous driving software for the first time ($30 million from P750 million from 21Q1 sales from previous year deferred confirmations).
In the short term, electric vehicles will become "new infrastructure", and software charging dominated by automatic driving systems is expected to become the "new kinetic energy" of the automobile industry. The essential difference between electric vehicles and traditional fuel vehicles is in the form of driving, but the change of driving form will not bring long-term high valuation to enterprises, and software services are the "boosters" behind the high valuation of enterprises. As electrification accelerates penetration, electric vehicles will become infrastructure, and software services are expected to be the driving force behind gross profit growth.
In the medium and long term, under the trend of automatic driving standardization, capacity charges and channel charges will become the "source of living water" for car companies' profits. At present, the basis for the commercialization of autonomous driving software is that it can bring users a forward-looking driving experience, and different levels of autonomous driving technology further differentiate the user experience, so in the short term, software charges can help car companies make profits. With the major car companies to accelerate the layout of intelligent driving, automatic driving is expected to become the standard, in the automatic driving standardization trend, once a car company chooses the free form, other car companies will fall into a passive situation, the profit space of software charges will be compressed. Therefore, in the medium and long term, with reference to the transformation of Apple's business model, capacity charges based on shared travel and channel charges based on software ecology will become the source of corporate profits.
The game of hardware sales and software services will reshape the automotive business model. We believe that under the wave of software-defined cars, the core competitiveness performance carrier of the whole vehicle is gradually shifting from hardware + machinery to hardware + software + services, and under this dual profit structure, the profit from bicycle sales will remain reasonably low in the future, and software and services will bring major profits to enterprises. We are optimistic about the potential of software and service business in the whole life cycle of the automobile, and the business model of the traditional automobile industry will be reshaped under the background of intelligence.
5.2. Technology giants enter the game,
Nugget intelligence
Forces from all sides have entered the game one after another, and the century-old automobile industry order has been reshaped. Intelligent electric vehicles span automotive, electronics, computers, IoT and other fields, spawning trillion-level market space, and software and services open up new profit channels, not only for the comprehensive transformation of traditional car companies, but also for the new forces of car manufacturing, actively enter the technology Internet enterprises to bring historic opportunities: the new force brand fully uses Internet thinking, selected tracks to achieve breakthroughs, Huawei positioning intelligent electric vehicle incremental component providers, Baidu, Xiaomi has officially announced the next car manufacturing, DJI, OPPO, Skyworth, etc. are also actively laid out. Multi-party cooperation to promote the vigorous development of the industry.
5.2.1. Huawei:
Huawei does not build cars The vision is to bring the digital world to every car
More than 30 years of deep cultivation in the field of ICT,
Cross-border entry into the automotive industry:
In 2013, the Vehicle Networking Business Department was established and the on-board module ME909T was launched;
In 2014, the Internet of Vehicles Laboratory was established and officially entered the Internet of Vehicles;
In 2015, designed a communication architecture for autonomous vehicles and high-speed Internet connectivity;
In 2018, the OceanConnect Internet of Vehicles was released;
In April 2019, it participated in the exhibition for the first time as a supplier of incremental components for smart cars at the Shanghai Auto Show;
In May 2019, Ren Zhengfei issued Huawei's 2019 Document No. 223 and officially established the Intelligent Vehicle Solutions Division (BU);
In February 2020, MDC Intelligent Driving Calculation Platform and Intelligent Electric Platform obtained ISO Functional Safety Management System Certification;
In October 2020, the smart car solution brand HI was released;
In November 2020, it was announced that it will jointly build a smart car brand with CATL and Changan Automobile;
In November 2020, the jurisdiction of the BU business of smart car solutions was adjusted from ICT to consumer BG;
In April 2021, HI's first landing model, BAIC ARCFOX αS Huawei HI Edition, was released;
In April 2021, xilix Huawei Smart Select SF5 officially entered the Huawei flagship store and sold through Huawei's national retail channel network.
Automotive BU ownership adjustment, To B turned to To C directly to the customer. On May 27, 2019, Huawei Intelligent Automotive Solutions BU was established, under the ICT Management Committee, and on November 25, 2020, the business jurisdiction was adjusted from ICT to consumer BG, from ToB to ToC, and in 2021, it plans to invest US$1 billion in research and development, with more than 5,000 R&D personnel.
What Huawei is doing:
"1+5+N" full-stack intelligent car solution,
Incremental Parts Providers:
1) Architecture layer:
A new computing and communication architecture CCA
2) System layer:
Five major systems
Intelligent driving: MDC intelligent driving platform (based on Shengteng chip, running AOS intelligent driving operation system) + lidar and other sensors;
Intelligent cockpit: CDC intelligent cockpit platform (based on Kirin chip, running HOS-A intelligent cockpit operation system) + AR-HUD + AI visual recognition + natural voice interaction;
Intelligent electric: VDC vehicle control platform (running VOS intelligent vehicle control operating system) + mPower (electric drive, charging and battery management system) + TMS;
Intelligent network connection: vehicle mobile communication module + Tbox + Ethernet off;
Smart Car Cloud: Octopus Octopus Autonomous Driving Cloud Service + Ocean connect Car Connectivity Service + Entertainment Service.
3) Component layer:
Lidar, AR-HUD and other 30+ intelligent components, with each system application.
Five major systems specifically,
Intelligent driving and intelligent cockpit are the short-term core:
Smart Driving:
Standardized output, full-stack solution.
Huawei's Smart Driving Full Stack Solution (ADS) includes:
1) Cloud Services:
Octopus,
Provide data, training and simulation services;
2) Algorithm:
Design and optimization for urban roads, highways, urban parking, etc.;
3) Development toolchain:
Provide an end-to-end toolset for application development;
4) Ascend AI chip:
310 (vehicle side)/610(vehicle side)/910 (cloud); 5) MDC computing platform: 210 (L2+, 48TOPS)/300F (commercial vehicle/work vehicle, 64TOPS)/610 (L3-L4, 200+TOPS)/810 (L4-L5,400+TOPS); 6) Sensor: Laser radar + millimeter wave radar (conventional +4D imaging) + camera.
Smart Cockpit:
Three major platform layouts,
The core of the difference personalized appeal is ecological construction.
1) Intelligent hardware platform: CDC computing platform (Kirin high-performance car machine module pluggable design, easy to upgrade) + display platform (AR-HUD, vehicle smart screen);
2) Software platform: Hongmeng car machine operating system HOS-A + 7 HMS-A service framework and core Kit (HMS for Auto, including vision, voice, sound, interconnection, AR-HUD, AI, OTA, enabling car companies, Tier1, developer differentiated development) + incremental development of 12 HOS-A subsystem middleware (overlay HOS common components & subsystems) for vehicle scenarios;
3) Vehicle ecosystem platform: application ecology (open API, including 1,300+ HOS APIs, 500+ Car APIs, 200+ HMS-A APIs, short-term Huawei Hicar transition) + hardware ecology (jointly build hardware interface standards with third parties to reduce the difficulty and cost of docking).
5.2.2. Baidu:
Triple positioning of supplier + service provider (Robotaxi) + manufacturer
ACE traffic engine + LEGO-style car intelligent solutions to empower smart transportation.
In 2019, Baidu formally released the ACE (Autonomous Driving, Connected Road, Efficient Mobility) intelligent transportation solution, which adopts the "1+2+N" system architecture, covering a digital base, two intelligent pilots, and N application ecosystems to empower urban transportation; in December 2020, Baidu officially released the Apollo LEGO-style car intelligent solution, including smart driving, smart cabin, and smart map According to the company, in the second half of 2021, Apollo will meet the peak of mass production, a new car will be listed every month, and it is expected to be mass-produced in the next 3-5 years.
Established an automobile company and joined hands with Geely to officially build a car. In January 2021, Baidu and Geely Automobile cooperated to formally establish an automobile company, Jidu Automobile, which is oriented to the passenger car market and maintains independent operation independently of the parent company system. Jidu Automobile plans to invest 50 billion yuan in five years, and the first model is based on the Geely Haohan architecture platform, which will be mass-produced and launched in 2024 at the latest, and one model will be launched every 1 to 1.5 years.
5.2.3. Xiaomi:
Announced the next car,
Multi-dimensional advantage Nugget smart car
Xiaomi has officially entered the game, and the forward-looking layout of car manufacturing has been for a long time. Xiaomi announced on March 30, 2021 that it will build a car, planning to invest 10 billion yuan in the first phase and a total investment of 10 billion US dollars in the next 10 years. Xiaomi car-making plan has been brewing for a long time, which can be seen from its existing layout: 1) strategic investment: more than 10 auto industry-related enterprises such as Ganfeng Lithium Battery, Kailide, Weilai, Xiaopeng, PATEO, Lingming Photon, etc.; 2) Technical reserves: the establishment of the "Xiaomi Car Union" layout vehicle networking project, and apply for more than ten patents such as vehicle cruise and vehicle navigation; 3) chip planning: Close cooperation with Qualcomm, is expected to receive corresponding support at the chip level, and has successively invested in 8 chip companies such as Yunyinggu Technology and BYD Semiconductor.
Since the entry of consumer electronics, multi-dimensional advantages nugget smart cars. At the capability level, Xiaomi's mobile phone business ranks third in the world, and its core capabilities (such as hardware and the Internet) have a strong correlation with the smart electric vehicle industry; at the brand level, the monthly active users of Xiaomi MIUI system reach 368 million (as of 2020Q3), and the brand foundation is solid; at the capital level, as of the end of 2020, Xiaomi's cash balance is 108 billion yuan, with abundant capital reserves; at the channel level, Xiaomi Home stores exceed 2,000, covering 270 counties in 30 provinces across the country, laying the foundation for follow-up marketing Fixed channel base. According to Lei Jun, the first model is expected to be a high-end model of 100,000-300,000 yuan, and Xiaomi is expected to occupy a place with its experience in the field of consumer electronics.
Other participants include DJI, OPPO, Skyworth, etc. DJI through the launch of the intelligent driving business brand "DJI car" and layout of the lidar business to start, together with SAIC-GM-Wuling, Volkswagen, etc. to open the output of intelligent driving solutions; smart phone giant OPPO began to prepare for car manufacturing in April 2021, has applied for more than 60 car patents, and reserves more than dozens of autonomous driving patents, and is currently working with SAIC to explore intelligent car ecology; home appliance giant Skyworth also officially announced the car in April 2021, and officially released Skyworth Auto EV6;360 in July In May 2021, Nezha Automobile officially announced the cooperation in car manufacturing; in addition, start-ups such as Xiaoma Zhixing and Tucson Future also crossed the border to prepare for car manufacturing.
Tech giants are friends and rivals, and the short-term impact is limited and the long-term impact cannot be underestimated. We believe that the entry of technology giants will be a powerful empowerment, which will bring the world's top technology and talents on the one hand, and on the other hand, it will also force traditional car companies to accelerate their transformation. However, limited by its own manufacturing capabilities, it is very likely that the initial cooperation with the main engine factory will be adopted to build cars, giving birth to a new joint venture model (traditional car companies + technology giants, Jidu automobile is a typical representative). From the perspective of the time node of the launch of the model, it is expected that it will still take 3 years for the technology giants to truly form a climate, which will provide a valuable time window for traditional car companies, and compared with the technology giants to supplement the manufacturing capacity, it is more difficult for traditional car companies to make up for intelligence, and the determination and final results of transformation may become the key to future survival.
5.3. Vehicle endgame discussion,
Hardware super chassis + software upper layer ecology
In the era of the software-defined automobile, the car base may no longer be a "car". Product definition, supply relationship, research and development process, talent organization, business model, etc. will undergo fundamental changes, the core capabilities will also be transformed from production and manufacturing to software development and ecological construction, at this stage of the vehicle end in addition to Tesla's electric intelligence level is relatively prominent, other domestic and foreign car companies including new car-making forces can be roughly understood as being in the same running line, who can win is still difficult to determine, take the initiative to change is the core.
In terms of endgame, we think the endgame of the car will be Maas, or mobility as a service. The car itself will be composed of two parts: "super chassis (hardware) + upper layer ecology (software)", which will accelerate the iterative development of models and shorten the research and development and verification cycle by installing different bodyworks. thereinto:
1) The super chassis is similar to the chassis we understand at this stage, including tires, brakes, steering, suspension, etc., but they are all in-line control, which are called by the upper layer software in the form of interfaces; the power battery will be integrated as a standard component in the form of CTC, and the model will optionally have battery capacity on demand;
2) The upper ecological refers to software, covering intelligent driving and intelligent cockpit, of which intelligent driving will become a standardized configuration after maturity, and flying cars will greatly reduce the difficulty of intelligent driving development; intelligent cockpit is more similar to the current stage of PC machines, the core lies in the operating system and ecological construction.
The transformation of the automotive industry led by smart electric can be analogous to the smartphone industry of that year, but it will give birth to one Apple and two Androids. An Apple refers to the "super chassis (hardware) + upper layer ecology (software)" are independently developed, the current time to see Tesla has the most opportunities; two Androids: 1) software side: Microsoft in the PC stage, Android/IOS in the smart mobile phone, the operating system of the Internet of Everything in the era of intelligent electric vehicles, the current node looks at Huawei Hongmeng OS and Android have the most potential, but it is expected that Apple will also become a favorable competitor in the future; 2) Hardware side: The super chassis is expected to spawn giants, and the scale will be done to the extreme. Its influence and discourse power will exceed Foxconn's simple FOUNDRY, but the profit margin level is lower than that of the software side.
5.4. Reshaping the order of the industrial chain,
Chinese parts from 1 to 100
With the development of electric intelligence, the reshaping of the traditional automobile industry chain, the emergence of Tier0.5 suppliers, the status of the industrial chain has moved forward, and the independent parts suppliers with core technical strength are expected to rely on the Chinese market to rise to become the global parts leader through globalization, and the potential entry impact outside the original automobile supply system cannot be underestimated. The serious imbalance between the status of the mainland vehicle industry and the status of the parts industry is expected to be alleviated.
Tesla broke the game and reshaped the entire zero relationship. Compared with traditional car companies, Tesla's supply chain has fewer levels and a high degree of flattening, and parts companies have more opportunities to achieve assembly support, upgrading from the original Tier 2 to Tier 1 or even Tier 0.5, and at the same time using Tesla to enter the global supporting system and accelerate the globalization process.
Traditional parts from 1 to N, optimistic about the rapid response ability and cost advantages of the company. A. Interior parts, chassis parts, tires and other technical barriers are low, with cost-effective and rapid response capabilities of independent suppliers will achieve From China to the world, from single products to assemblies, the growth curve will become more steep; b. Chassis electronics, lamps and other products are iteratively upgraded, and independent suppliers with innovative capabilities will accelerate domestic substitution; c. Small and beautiful invisible champion: focus on small piece manufacturing, profitability and operating efficiency are both high, share increase and globalization are parallel.
Incremental components from 0 to 1, the new machine breeds a new pattern and high growth. Compared with traditional fuel vehicles, the main incremental components of intelligent electric vehicles include: 1) electric vehicles as the base: three electric systems, new energy thermal management systems, etc., high bicycle value + high technical barriers, as standard products, the most benefit from the increase in the penetration rate of electric vehicles; electric vehicles have a more urgent demand for lightweight, integrated die casting will further accelerate the lightweight of automobiles; 2) intelligent acceleration penetration: intelligent cockpit to create a third space, canopy glass, HUD, car entertainment, intelligent interior lights and other product penetration rate increase Automatic driving as the core element, cameras, millimeter wave radar, lidar and other sensors and chips, domain controllers increased demand, the implementation of layer control movement, steering by wire, air suspension and other gradual penetration. With the accelerated development of electric intelligence, incremental components will show a high growth trend, the overall competitive pattern is undecided, in addition to suppliers within the system, there will be many new cross-industry entrants, which is expected to give birth to new leaders.
(This article is for informational purposes only and does not represent any of our investment advice.) For usage information, see the original report. ）