In 2015, the company pushed Tesla V7.0 version to unlock assisted driving functions such as Autosteer in Auto Pilot. It then registered and established Tesla China in Shanghai in 2018, and the Shanghai factory began trial production in 2019.

(Producer/Author: Kaiyuan Securities, Wu Liuyan)

1.1. Development history: Integration of industrial chain to enhance the competitiveness of the company

Tesla is the largest electric vehicle and solar energy company in the United States, and its main businesses are electric vehicle sales and leasing business, energy and energy storage business, services and other businesses. The company was established in 2003 and was listed on Nasdaq, USA on June 29, 2010. In 2012, the company Model S began to be delivered and entered the European and Chinese markets in 2013. In 2014, the company reached a cooperation agreement with Panasonic to build a Gigafactory in Nevada. In 2015, the company pushed Tesla V7.0 version to unlock assisted driving functions such as Autosteer (automatic assisted steering) in Auto Pilot. It then registered and established Tesla China in Shanghai in 2018, and the Shanghai factory began trial production in 2019. According to Clean Technica data disclosure, Tesla delivered a total of 935,950 new energy vehicles in 2021, with the delivery volume being the world's first new energy vehicle company, with a market share of about 14%. Since its establishment, Tesla has improved its production capacity and core technology accumulation by acquiring companies in different fields. On the production side, Tesla acquired Riviera Tool, Grohmann Engineering, Perbix Machine and Hibar Systems, respectively, with its business scope covering mold stamping, machine engineering, automation equipment manufacturing and battery pack assembly; on the technical side, the company acquired SolarCity, Maxwell and DeepScale, respectively, enhancing the company's strength in battery, solar energy and autonomous driving technologies.

1.2. Management: Rich in innovative technology thinking, continuous high research and development to ensure technological leadership

Tesla is a new energy company with innovative and technical genes. Company CEO Musk has entrepreneurial experience in the Internet software and new energy industry. Some executives graduated from hard-core majors from prestigious universities (such as Stanford University's engineering majors mainly from energy system engineering and power electronics technology ), and have excellent technical precipitation.

CEO Musk: He is innovative and has many entrepreneurial experiences. Musk once founded the network software company Zip2, the third-party payment platform PayPal, the space technology company SpaceX and the solar energy company SolarCity. Most of the companies were acquired at high prices. Zip2 was acquired by Compaq for US$370 million in 1999, and PayPal was acquired by EBay for US$1.5 billion in 2002. Many successful entrepreneurial experiences reflect Musk's innovative thinking to a certain extent, and innovation is also Tesla's more prominent gene.

Co-founder and Chief Technology Officer JB Straubel: Graduated from at Stanford University, holds a bachelor's degree in energy systems engineering from Stanford University and a master's degree in engineering mechanics from energy . He founded the aerospace company Volacom in 2002 and served as its CTO, and then joined Tesla in 2004. Andrew Baglino, Senior Vice President of Powertrain and Energy Engineering,

: Graduated from Stanford University, Bachelor of Electrical Engineering from Stanford University, . He joined Tesla in 2006 and served as the engineering director of Tesla Energy. He led the design of the Model S dual motor system and powertrain control algorithm .

The founder’s rich entrepreneurial experience and the technical background of the executives determine that the company attaches importance to innovation and adopts a strategy of high R&D investment. R&D investment continues to be ahead of the back-ranking manufacturers and helps maintain a leading position in technology. In terms of R&D investment, the company's R&D expenses fell slightly in 2019, and then the overall trend was on the rise. The R&D expenses in 2021 were 16.54 billion yuan, a year-on-year increase of 74% from 9.51 billion yuan in 2020. Compared with the new domestic car-making forces, Tesla's R&D expenses are far ahead. In 2021, Tesla's R&D expenses reached 16.54 billion yuan, while the R&D expenses of domestic NIO , Ideal, and Xiaopeng are 4.59 billion yuan, 3.29 billion yuan and 4.11 billion yuan respectively.

Tesla's equity and voting rights are relatively scattered, and Musk holds 23% of the shares.As of June 30, 2021, Musk, The Vanguard Group, Blackrock and Capital Ventures International held 23.1%, 5.9%, 5.1% and 5.0% respectively. Unlike domestic new companies, Tesla's voting rights are relatively scattered, with CEO Musk only 23.1% of the voting rights; while Ideal Li Xiang, Xiaopeng's He Xiaopeng's and NIO's Li Bin have 69.59%, 64.6%, and 39% of the voting rights respectively, and CEOs occupy a stronger dominant position in internal business decisions.

1.3. Market positioning: Starting from the high-end, the high-end occupies the user's mind, the price drop drives significantly increase

Tesla product strategy starts at the high-end, seizes the user's mind in advance. In the early stages of the development of the new energy vehicle industry, due to the relatively immature industrial chain, the production cost is high, so the company has seized the minds of users in advance by launching relatively high-priced high-end products. Tesla's first three models are all high-end models priced at over $80,000. Tesla launched the high-end pure electric sports car Roadster in November 2008, priced at over $100,000; in June 2012, priced at about $80,000 to $100,000; in 2015, priced at about $80,000 to $120,000.

is based on advanced manufacturing processes to gradually decline production costs. The company launched parity products to achieve price sinking, driving a significant increase in delivery volume. Based on the company's CTP technology, integrated die-casting technology and scale effects brought by the increase in delivery volume, production costs have gradually declined, helping the company launch more affordable products. In 2017, Tesla launched the model 3, with the basic price of $35,000. The model used Tesla's latest CTP technology. Compared with Tesla's traditional solution, the number of modules in the company's large module solution decreased by about 70%, and the number of parts of the vehicle was reduced by 10%. The overall battery cost dropped from US$185/kWh to US$170/kWh. In order to offset the upstream cost pressure, Tesla gradually increased its price, and the latest price is about US$47,000-63,000; in 2020, Tesla launched the model Y, which was priced at 40,000-60,000. Model Y used Tesla's latest integrated die-cast rear floor assembly technology. The integrated die-casting technology will die-cast all parts at one time, with the number of parts being compared. The Model 3 has been reduced by 79 units, the solder joints have been reduced from 700-800 to 50 units, and the weight of the body assembly has been reduced by 30%, so the manufacturing cost has been reduced by 40%. The latest price is about 63,000 to 68,000 US dollars. To sum up, Tesla has achieved a significant drop in manufacturing costs through advanced production technology, helping to launch its affordable products.

The proportion of Model 3/Y delivery volume of relatively parity has increased year by year, from 2% in 2017 to 97% in 2021. At the same time, Tesla's total delivery number has achieved rapid growth, from 103,020 vehicles in 2017 to 935,950 vehicles in 2021, with a CAGR of 74% from 2017 to 2021. Tesla is based on advanced manufacturing processes to help product prices sink and achieve rapid growth in delivery number. (Report source: Future Think Tank)

. Product idea: Full stack self-development ensures technical and cost advantages, focusing on ADAS and three electric power, desalinating interior and smart cockpit

Tesla's product design idea: focus on overall cost control, implement minimalist in interior, focusing on intelligent driving and electrification. Tesla's three-electric technology and smart technology are both self-developed; on the cost side, the company implements minimalism in perceived hardware and interior to further control costs. On the technical side, since smart cars are highly coupled products of software and hardware, self-developed products can better combine software and hardware, which helps to iterate technology. The company's intelligent driving system and smart cockpit are both self-developed, which helps to technology iteration. According to Chejiahao's statistics, Tesla FSD has been updated in 2021 for 8 major versions, with nearly 130 minor versions + FSD Beta, and Ideal has been updated 6 times among the new domestic forces, and the software of Xiaopeng, NIO and Leapmotor have been updated 5 times respectively.

2.1, Three Electric: Full stack self-developed, continuously reducing costs and increasing efficiency

Tesla has excellent comprehensive performance in the fields of battery cells, battery packs, and battery management. Model 3 is equipped with a low battery capacity, but its unit endurance is ahead of comparable models at the same price; BMS has a technology leading, and the power consumption index of 100 kilometers per ton is outstanding among new energy models at the same price; the existing charging efficiency is temporarily behind, but it is expected to reach the industry-leading level after the supercharge network is updated. The battery life of

is mainly affected by battery cell, battery pack technology, and battery management technology. Benefiting from Tesla's technical accumulation in battery cell, battery pack and battery management, the company's unit battery life is ahead of domestic car companies. Tesla's unit life is 9.6 km/kwh, while the ZKR, Xiaopeng and BYD are 7.1, 8.3, and 7.9 km/kwh respectively.

(1) Battery Cell: Self-developed 4680 Battery

Tesla originally only used ternary lithium batteries, and then introduced lithium iron phosphate battery to achieve battery cost reduction. The company currently has three battery suppliers, including CATL , which supplies lithium iron phosphate batteries to Tesla, and Panasonic and LG Chemical supply ternary lithium batteries to Tesla. Compared with lithium iron phosphate batteries, ternary lithium batteries have higher energy density. The energy density of lithium iron phosphate batteries is about 180Wh/kg, while the energy density of ternary lithium batteries can reach 250Wh/kg or above. At the same time, the low-temperature performance of ternary lithium batteries is also better than lithium iron phosphate batteries. Tesla originally only used ternary lithium batteries, but as the subsidy for ternary lithium batteries declined, the cost advantage of lithium iron phosphate batteries was highlighted. Tesla began to use lithium iron phosphate batteries on some models. In the third quarter of 2021, Tesla announced that globally, its standard battery upgraded Model 3 and the standard battery life version of Model Y will use lithium iron phosphate batteries.

Tesla's self-developed 4680 battery has achieved a 5-fold increase in battery capacity, which is expected to increase the range and achieve a cost reduction. It was officially delivered in April 2022. On the "Battery Day" on September 22, 2020, Tesla launched a new 4680 battery. The new 4680 battery uses full-length ears and dry battery technology. Compared with the 21700 battery currently used by Tesla, the battery capacity of the new 4680 battery is 5 times higher and the output power is 6 times higher. It is expected to increase the car's range by 16% and reduce the cost by 14%. At present, the company's 4680 battery is relatively mature, and the first batch of Model Y equipped with 4680 battery was officially delivered in April 2022.

(2) Battery Pack: Using CTC technology

2022, it is expected to use CTC technology with higher integration to improve weight grouping efficiency and space utilization, and reduce the number of parts to achieve cost reduction and efficiency improvement. In 2017, Tesla used a large module solution for the first time on the Model 3. Compared with the 16 modules in the original Model S traditional solution, the total number of Model 3 modules using a large module solution has dropped to 4. In addition, the number of battery pack parts is 10% less than that of the traditional solution. Therefore, the weight of the Model 3 battery pack is 15% lower than that of the original Model S battery pack. The large module solution can reduce costs by reducing the overall number of parts and the weight of the battery pack and improve battery life.

Tesla launched CTC technology for 4680 batteries in 2022. At the same time, using integrated body die-casting and CTC technology is expected to reduce the weight of the new car by 10%, the battery life is expected to increase by 14%, and the body parts can be reduced by 370. Compared with CTP technology, the weight grouping efficiency of CTC technology is above 90%, higher than 80%-85% of CTP; at the same time, the space utilization rate of CTC technology is above 70%, higher than 60% of CTP technology; in terms of the number of parts, there are about 400 parts, lower than 470 of CTP technology. The Model Y, equipped with an integrated chassis of 4680 battery cells and CTC, has been officially launched and was officially delivered in the United States in April 2022.

(3) BMS: Leading technology, the power consumption index of 100 kilometers per ton is outstanding among new energy vehicles of the same price

BMS has three main functions, including battery protection, energy balance and SOC calculation functions.The battery protection function is mainly to prevent overcharge, overdischarge, overtemperature and overcurrent. When the battery temperature and current are too high, the system will automatically cut off the current path to ensure the safety of the battery and the system; the energy balance function is mainly to make up for the differences between individual battery cells, to do some active or passive charging and discharge management, ensure the consistency of the battery, and extend the life of the battery; the SOC calculation function is to calculate the power more accurately, so that the system can sense the power changes more accurately.

Tesla's self-developed battery management system adopts the master-slave architecture design. The main controller (BMU) is responsible for high voltage, insulation detection, high voltage interlocking, contactor control, external communication and other functions; the slave controller (BMB) is responsible for single-unit voltage and temperature detection, and reports to the BMU. Compared with other vehicles of the same price, Tesla's power consumption per 100 kilometers is 5.9kwh/ton, which is similar to Zekrch, and is lower than the Xiaopeng P7 and BYD's 6.2kwh/ton and 6.3kwh/ton, which highlights the superior performance of Tesla's battery management system on the side.

(4) Fast Charging: Using high current technology, the charging efficiency is expected to increase after the hypercharge network is updated

Tesla's existing charging speed is slightly insufficient compared with the fast charging technology currently launched, but it is expected to reach the industry-leading level after the hypercharge network is updated. Tesla launched a high-current V3 super charging pile in 2019, which can achieve a maximum charging power of 250KW, achieving a maximum rechargeable range of 250 kilometers in 15 minutes of charging. At the same time, the charging device will not divert, and every car on the network can charge with maximum power. Compared with the Porsche fast charging technology launched in the same year, Tesla's charging efficiency is slightly lower than Porsche's 5mins/100KM; compared with the Hyundai, Huawei and Geely , which launched high-voltage supercharging technology this year, it is still insufficient.

Musk mentioned in July 2021 that Tesla is currently updating the Supercharger V3 network, and its maximum output power will be increased from the current 250KW to 300KW. At the maximum charging rate of 300KW, Tesla's charging efficiency is expected to reach 5mins/161km, reaching the industry-leading level. At present, in theory, the large voltage solution has faster charging technology, but due to the long completion cycle, Tesla still adopts the high-current fast charging route in the medium and short term. Compared with the mainstream domestic large voltage solutions, the research and development of large current solutions is more difficult, but the completion cycle is short, and only the battery carrier needs to be changed. The research and development of large voltage solutions is less difficult, but the completion cycle is longer. It is necessary to link up the upstream and downstream supply chains and transform the three-electric system into large voltage.

high-current fast charging route has a great test for battery thermal management , and Tesla will face greater challenges in thermal management. Tesla currently uses high-current fast charging technology, that is, by increasing the current while keeping the voltage unchanged, the purpose of increasing the charging power is achieved. At a voltage of 400V, the charging power of 250A current is 100KW, and the charging power of 500A current is 200KW. According to the heat formula Q=I2R=UI, the heat generated by increasing the current when the output power is the same is higher than the voltage when the voltage is increased, so high current technology will put higher requirements on the thermal management of the battery.

2.2. Interior: Minimalism reflects Tesla's cost control strategy

Tesla adopts minimalism in interior to better control costs. It is not equipped with a full LCD instrument panel, ambient light, and car air purifier . The seat material is made of imitation leather instead of genuine leather. In terms of instrument panels, BYD Han, Xiaopeng P7 and Zekr 001, which are similar to Tesla's prices, all use full LCD instrument panels; in terms of seat materials, except for Tesla's imitation leather, other comparable models use higher-priced genuine leather; in terms of interior atmosphere, BYD Han, Xiaopeng P7 and Zekr 001 are all equipped with multi-color interior ambient lights, and are also equipped with on-board air purifiers.

2.3. Smart cockpit: Use the AMD solution instead, taking into account hardware upgrades and cost controls

Tesla's second-generation smart cockpit has a gap with the new domestic car manufacturing forces, and its smart cockpit chip GPU computing power is relatively large.Tesla uses Intel 's Atom A3950 as a smart cockpit chip in the second-generation cockpit domain controller. Compared with the Qualcomm 820A used by new domestic forces, the AtomA3950 has a large gap in GPU computing power, with a computing power of 187GFLOPS, which is lower than NIO, Ideal, and Xiaopeng's 588GFLOPS.

Tesla uses AMD solutions in its latest Model S and domestic Model Y Performance, 2022 model model 3/Y, and its performance is significantly improved compared with the second-generation cockpit domain controller. The hardware upgrade is expected to improve the smart cockpit performance of Tesla . Tesla's latest AMD solution uses AMD Ryzen chip. Compared with the Intel AtomeA3950 chip of the second-generation solution, its GPU computing power is as high as 10TFLOPS, which is about 53 times the second-generation computing power. At the same time, its display and memory are 8GB and 10GB respectively, which is higher than the 4GB of the second-generation Intel AtomeA3950. We believe that Tesla's hardware upgrade is expected to bring an improvement in the smart cockpit experience.

Currently, Tesla's smart cockpit has implemented the following functions:

(1) Car sleep: Tesla can automatically turn off the display when the owner leaves the car, achieving energy-saving effect.

(2) Voice control: The voice command module uses a natural language processor to analyze and identify the voice request of the car owner, which is not limited to specific statements, and then converts it into operation instructions that the car can execute. The car owner can use voice control to realize temperature control, vehicle control, telephone and other functions.

(3) Energy Application: The ability to display the predicted range directly on the energy chart, intuitively highlighting the relationship between the predicted range and energy consumption, and by default, it displays an "instant" predicted range, reflecting the consumption of the last 1/10th of a mile.

(4) Software upgrade: It can provide software upgrade functions, and specific upgrade notifications will be displayed on the screen.

(5) Remote Start: Car owners can start Model S with their smartphone only. It can be used when the owner loses the car key, it can be accessed through the Tesla mobile application, and requires a password to be entered every time it is used.

(6)Power Management Options: The new power management option will enable Model S to enter energy-saving mode at night to help maximize the use of available energy. There is also an option to allow Tesla mobile apps to always connect to the car immediately while still saving power.

The difference in the hardware level of smart cockpits may have caused Tesla's smart cockpit to perform unsatisfactorily before chip replacement. In 2021, ModelY ranked 33rd in the total score in the Pacific Automobile smart cockpit evaluation. Pacific Auto’s smart cockpit score is mainly evaluated around screen effects, vehicle-machine performance, voice interaction and advanced intelligence needs, while Tesla Model Y’s smart cockpit ranking is 33.

Tesla Model Y scored 8 points in the screen effect evaluation and ranked 6th among the participating smart cockpit models, especially the screen size and resolution. The evaluation of screen effect is mainly based on four aspects: screen size, screen resolution, visual angle/brightness, and anti-scratch and anti-fingerprint coating. Among them, the screen size of Model Y is 15 inches, with a screen resolution of 1280*720, and it received full scores in both screen size and screen resolution. In the visual angle/brightness and anti-scratch and anti-fingerprint coating test, due to some reflections and subtle scratches on the screen, neither scored the two scores, which were 2 points and 1 point respectively. Tesla scored 8 points in this score, ranking 6th among the participating smart cockpits.

Tesla ModelY scored 28 points in the performance score of the car and ranked 30th among the participating smart cockpits, and there is still a lot of room for improvement; it performed well in the evaluation of the dashboard function richness, UI interaction design and central control screen function, but it was poor in the expansion of mobile phone interconnection. The evaluation of car performance is mainly based on the ease of use/interaction design of the car, the central control screen function and experience effect, the functional richness and expansion of the instrument panel. Model Y performs well in the dashboard feature richness, central control screen function & experience effect, and ease of use/UI interaction design, and has won full marks in many evaluations such as music player convenience, energy consumption information display, online maps, etc.However, in terms of scalability, Model Y has poor usability, does not support the mapping of Android, , Apple systems, nor does it support wireless mapping, and only get 0 points for a full score of 7 points.

Tesla ModelY only won 16 points in the voice interaction evaluation, ranking 39th among the participating smart cockpits; the speed of identifying voice commands is slow, and there is still a lot of room for improvement in error correction capabilities. Voice interaction evaluation mainly focuses on several aspects such as conventional function implementation, recognition speed, error correction ability, semantic association ability, etc. In terms of conventional function implementation and recognition speed, Model Y can recognize voice commands such as air conditioning, music, navigation, etc., but the recognition speed is slow and does not support voice control of the window. At the same time, there is no feedback after the Model Y voice is recognized successfully, and the entire recognition execution process is completed silently. In terms of error correction capabilities, the system failed to recognize the instructions in Simplified Chinese, but barely passed the recognition of Cantonese. In terms of semantic association, Model Y has high requirements for speech speed. Among the 6 commands tested, two commands related to navigation can be recognized and executed, while for air conditioning and cross-functional commands, they are intelligently recognized and cannot be executed.

Tesla Model Y performed well in the intelligent advanced demand assessment, with a full score of 10 points and ranked 4th among the smart cockpits that participated. Intelligent advanced demand assessment mainly focuses on the update of the car's function and the remote control function of the mobile phone. Tesla performed well in this assessment. Model Y supports FOTA car machine updates and car fun functions. At the same time, car owners can remotely override the vehicle through their mobile phones to realize the functions of opening windows, remotely controlling the car and moving the car, and parking the car.

2.4. Intelligent driving: Full stack of software and hardware, intelligent driving capabilities are leading in the world

2.4.1. Decision-making level: The computing power of self-developed chips is obvious, helping software algorithms iterate faster

Tesla's self-developed chips have obvious computing power advantages, helping the company's software algorithms iterate faster. Smart cars are highly coupled products with software and hardware. Self-developed products help the coordinated development of products, software and hardware, reduce the dependence of technology suppliers, and accelerate the speed of technology iteration. According to Che Jiahao's statistics, Tesla FSD has pushed 8 major version updates in 2021, and nearly 130 minor versions + FSD Beta has been pushed. Among the new domestic forces, Ideal has been updated 6 times, and the software of Xiaopeng, NIO and Leapan has been updated 5 times respectively.

Tesla used its own FSD processing platform in 2019. The computing power of a single chip is 72TOPS, and the final computing power (dual chip) is 144TOPS. With the increase in the level of autonomous driving, the required chip computing power will increase dozens of times. The computing power required for L2 level autonomous driving is 2-2.5TOPS, the computing power required for L3 level autonomous driving is 20-30TOPS, the computing power required for L4 level is 100TOPS+, and the computing power required for L5 level exceeds 1000TOPS+. Currently, Tesla's computing power can meet L4 level autonomous driving. Compared with models of the same price, Tesla's computing power has obvious advantages, which is higher than the 30TOPS and 48TOPS of Xiaopeng P7 and Zekr 001. It is expected that Tesla will use the newly developed FSD chip in HW4.0 in 2022. The computing power of this chip is expected to reach 216TOPS, and its power consumption is 65% lower than that of the previous generation chip.

2.4.2. Perception layer: adopts a pure vision solution to help cost control

Tesla adopts a pure vision solution dominated by a camera and combined with low-cost sensors such as mm wave radar .

pure visual solution has advantages in data fusion and applicability. In terms of data fusion, solutions using different sensors are more likely to have data fusion problems than pure visual solutions. When different sensors are inconsistent in perception, the system will confuse the environment and data judgments, resulting in security problems. In terms of applicability, the pure visual solution is more applicable, and the high-precision map + lidar solution has relatively large geographical restrictions. For example, Google 's waymo can only perform unmanned driving in places covered by Google's high-precision map. The disadvantage of pure visual solutions is that they are security and difficult to develop. In terms of safety, the camera relies on environmental lighting and is easily affected by harsh environments. It cannot intuitively judge the depth of field, and is prone to safety problems in harsh weather.In terms of R&D difficulty, Tesla's chief AI scientist Karpathy said at the 2021 CVPR 2021 Autonomous Driving Seminar that autonomous driving solutions based on pure vision are more difficult to implement technically because they require neural network to run very well based on video input alone. Tesla is expected to continue to accelerate its algorithm optimization based on massive data accumulation and supercomputer neural network training and gradually overcome application difficulties.

In terms of perception layer hardware, since Tesla's intelligent driving adopts a pure vision solution, its models still do not need to be equipped with lidar. Compared with new energy vehicles of the same price, Tesla has a lower number of sensors. Model 3 has only 21 sensors (including 8 cameras, 12 ultrasonic radars and 1 millimeter wave radar), which is lower than the 28 of the Zekrch 001 at a similar price and 31 of the Xiaopeng P7. The conservative number of sensors further reflects Tesla's cost control concept.

2.4.3. ADAS algorithm: Based on massive data + self-developed supercomputer, accelerated algorithm iteration

Tesla is one of the vehicle manufacturers that develop self-developed autonomous driving technology. Its advantage lies in having massive data and advanced computers to strengthen neural network training.

(1) Massive data

Tesla car equipped with the Autopilot system has traveled more than 2.2 billion miles in 2020, while domestic Weilai exceeded 100 million kilometers at the end of 2020. In May 2021, Xiaopeng Motors announced that its NGP (navigation assisted driving) users had traveled more than 5 million kilometers, and the data accumulation of both was far inferior to Tesla. Data is extremely important for autonomous driving. A large amount of data helps the rapid iteration of autonomous driving technology, and the improvement of intelligent software applications will help increase sales, thus forming a positive cycle.

(2) Self-developed supercomputer strengthens neural network training

In August 2021, Tesla announced its self-developed supercomputer Dojo. Dojo has built-in Tesla's self-developed AI chip D1. A single D1 has the peak computing power of 362TFLOPS, and its power consumption does not exceed 400W. The ExaPOD cabinet consisting of 3,000 D1 chips has a computing power of up to 1.1EFLOPS, making it the fastest AI training computer in the world.

2.4.4.4. FSD payment rate: The price of FSD is gradually rising, and there is still a lot of room for improvement in the payment rate

According to TroyTeslike statistics, Tesla's FSD payment rate showed a trend of rising first and then falling from 2016Q4 to 2021Q2. The FSD payment rate in 2021Q2 is 11%. Due to the high price of FSD, the payment rate is not high. In the future, with the increase in FSD functions and the increase in user acceptance of ADAS, the FSD payment rate is expected to increase again. Tesla's FSD penetration rose from 12.2% in the fourth quarter of 2016 to 46.0% in the second quarter of 2019, and then fell to 11% in the second quarter of 2021. The penetration rate of FSD kits ushered in a significant increase in 2019, mainly because Tesla migrated the enhanced Autopilot function to the FSD kit; the payment rate of FSD began to decline significantly in 2019Q3. We believe that the main reason is that Tesla increased the price of FSD. The price of FSD gradually rose from US$5,000 in March 2019 to US$12,000 in January 2022, an increase of 140%.

2.4.5. Autonomous driving evaluation: assisted driving capabilities are leading, and the automatic parking level is slightly weak

Currently, Tesla's autonomous driving mainly includes the following functions:

(1) Automatic lane change

When driving on a highway, automatic lane change will position the car in the best lane to prepare for the parallel lane and exit when overtaking. This allows the driver to have a clear understanding of the upcoming lane changes and customization of the automatic lane change function.

(2) NOA

Autonomous driving from the highway entrance ramp to the exit ramp. Includes automatic lane change, traffic-aware cruise control with complete stop and reengagement, automatic steering and vehicles that exceed slow driving.

(3) Automatic parking

can currently realize vertical parking and parallel parking. The car will continuously monitor the surrounding space while driving at speeds below 15 mph, reminding owners of available parking spaces.

(4) Summon

is activated through the Tesla application. The car will find the owner anywhere in the parking lot, and can even park itself or cancel the parking space in a small space. This feature can be used in complex parking situations while obeying lane markings and stop signs to avoid pedestrians and obstacles such as traffic cones, garbage cans and shopping carts.

(5) Traffic lights and stop signs control

When a traffic light or stop sign is detected and slows down or stops.

Tesla's leading autonomous driving technology is reflected in the evaluation. In the car evaluation of Garage No. 42, Tesla performed well in the three major scoring items: basic assisted driving ability, navigation assisted driving ability, and automatic parking. The three scorings were first, second and sixth.

Tesla has excellent basic driving capabilities and ranks first in the 42Mark evaluation, among which lane keeping capabilities, curve capabilities and automatic assisted lane change capabilities are more outstanding. The evaluation of basic assisted driving capabilities is mainly based on the car's curve ability, lane keeping ability, special scenario usage experience, congestion scenario usage experience, etc. Tesla performed excellently in the curve ability test and was able to pass through the three major corners of the devil bend, high-speed bend and S bend smoothly. All three scores were full marks. At the same time, Tesla also performed excellently in lane capacity maintenance, with its night lane maintenance capability, rainy lane maintenance capability and medium and high-speed lane maintenance capability all scored.

But Tesla's interactive experience is relatively weak. In driver status detection, Tesla mainly senses the driver's driving state through the torque sensor of the steering wheel. The driver allows the car to sense the driver's presence by twisting the steering wheel or pressing the button on the steering wheel, but the overall perception effect is weak. The driver only needs to sense after the wheel when both hands are disengaged from the direction for a long time.

Tesla performed outstandingly according to navigation assisted driving, ranking second in the 42Mark evaluation, among which the lane change strategy is more outstanding. The evaluation of navigation assisted driving ability is mainly based on the car's usable range, ramp passing ability, lane change overtaking strategy and interactive experience. Compared with assisted driving, the navigation assisted driving function can control the vehicle's acceleration, braking and lane change operations based on the navigation destination set by the owner. In terms of lane change strategy, Tesla's high-speed scenarios have the shortest lane change distance and whether to accelerate when changing lane change are both full scores, and there is still room for improvement in lane change reaction speed.

At this stage, Tesla's NOA can only be used on closed sections such as highways, urban loops or elevated roads. Since there is no high-definition map installed, Tesla NOA's application scenarios will be slightly less than Xiaopeng NGP (can be enabled on some urban loops and most highways and some urban expressways) and NIO NOA (can be enabled on most highways and urban elevated roads within the coverage of high-precision maps).

Tesla's automatic parking capacity still has room for improvement. Its performance is average in terms of parking time and the limit length of automatic parking, and its overall automatic parking capacity ranks fifth. The evaluation of automatic parking capacity is mainly based on the evaluation of vertical parking capacity, side parking capacity, oblique parking capacity and parking experience. The main indicators include parking time, number of rebraking times, degree of automation, etc. Among the several models evaluated, Tesla's automatic parking performance still has room for improvement.

Tesla's weak vertical parking ability is mainly reflected in the longer parking time; the weak side parking ability is mainly reflected in the weakest extreme parking ability and poor parking posture. Among them, the extreme parking ability is mainly measured by the automatic parking limit length. In the evaluation, the length is 1300mm longer than the car; the parking attitude is mainly measured by the parking time and with the roadside example. Tesla's side parking time is 38 seconds, which is 37 cm away from the roadside. Its performance is not ideal, and there is still room for improvement.(Report source: Future Think Tank)

. Tesla's auto sales business gross profit margin is significantly higher than that of new domestic car-making forces, mainly due to the scale advantage of its self-built super factory and its own advanced manufacturing capabilities (leading body manufacturing technology and power battery system integration technology). In the future, the battery recycling business is expected to drive the gross profit margin to continue to increase after maturity. From 2018 to 2021, the gross profit margins of Tesla's automobile sales business were 22.4%, 20.1%, 24.8% and 28.9% respectively, while the gross profit margins of this business of new domestic car manufacturing forces were significantly lower.

3.1. Self-built super factory: The production capacity has significant advantages in scale, which is conducive to sharing costs and ensuring upstream procurement

Tesla's existing production capacity is leading, and the planned capacity release is concentrated. It is expected to quickly seize the market when the chip shortage eases. Currently, Tesla's total designed production capacity in 2021 is about 1.1 million vehicles, far higher than the production capacity level of new car-making forces such as Wei Xiaoli in China.

Tesla has manufacturing factories in the United States, Germany and China. In 2022, Tesla's design capacity of Austin plant in Texas, USA was 500,000 vehicles, and it was officially put into production on April 8, 2022, Beijing time; the design capacity of the Berlin plant in Germany was about 500,000 vehicles, and it was officially put into production on March 22, 2022, Beijing time; it is expected that the design capacity of the Shanghai Super Factory will be expanded to 1.1 million vehicles, and the production capacity of the Phreomont plant in the United States will remain the same, and the total design capacity is expected to reach 2.7 million vehicles. Tesla's production capacity is released large and concentrated, and it is expected that the automobile production capacity will double in 2022. Based on Tesla's current production capacity, global layout and capacity release time, we believe that Tesla is expected to continue to improve its sales performance as chip shortages ease and demand releases.

Tesla has stronger bargaining power in the supply chain compared with other new car manufacturers, and its gross profit margin advantage is expected to continue to maintain. Compared with other new forces, Tesla has its own battery cell factory, and its latest Berlin factory has a battery production capacity of 50GWh. Having its own battery cell factory gives Tesla a certain advantage in bargaining with battery suppliers. At the same time, Tesla's delivery volume in 2021 was 936,000, far higher than NIO, Ideal, and Xiaopeng's 91,000, 90,000 and 98,000. The sales gap is difficult to make up in the short term. Therefore, it is believed that Tesla's demand for automotive parts is much higher than that of new domestic forces, and has more advantages in component bargaining.

3.2. Its own advanced manufacturing capabilities: body manufacturing technology and power battery system integration technology are leading

In car body manufacturing, Tesla adopts integrated die-casting technology, which can die-cast all parts at one time to reduce the number of parts and the number of welding points. At the same time, the manufacturing time is also reduced from 1-2 hours of traditional processes to 3-5 minutes. At present, the body of the mass-produced models of domestic new energy vehicle companies basically uses single-piece stamping and then welded into a body assembly. There is no production method of mass-produced models using integrated die casting.

has always been leading in power battery system integration. It has applied large-module solutions to Model 3 in 2017, while CATL only launched the first generation of CTP technology in 2019, while BYD and Honeycomb Energy only launched the CTP solution in 2020. In October 2021, Tesla officially displayed its CTC solution, which directly integrates battery cells or modules into the chassis to realize the integration of chassis systems, autonomous driving-related modules, major three electrical systems and minor three electrical systems. Musk said that the technology combined with integrated die-casting technology can save 370 parts, reduce weight by 10% on the body and reduce battery cost per kilowatt-hour by 7%.

3.3. After the battery recycling business matures, it is expected to continue to drive the gross profit margin to increase

battery recycling capacity, and is expected to increase gross profit margin. In the first quarter of 2022, Tesla mentioned that the company's battery recycling capacity is expected to increase from the current 50 tons/week to 150 tons/week. Referring to the gross profit margin of Greenmei's power battery comprehensive utilization business in the second quarter of 2021, we assume that the positive and negative electrode raw materials in Tesla's battery recycling can bring a 20% cost savings.According to the report of China Automobile Center's "2020 Power Battery Industry Development Report", the cost of positive electrode materials for ternary lithium batteries accounts for 18%-25%, and the proportion of negative electrode materials is 13%-17%. Therefore, the cost savings brought by our battery recycling business are 37% (positive electrode: 22%, negative electrode 15%) can be transmitted to battery costs. According to the data of electric vehicle observers, the cost of battery system in electric vehicles with large capacity batteries accounts for 40% of the vehicle cost. Therefore, we speculate that after Tesla's battery recycling business matures, it can bring about a 3% increase in gross profit of about 3% of the automobile sales business.

4.1. Sales side: The direct sales model helps sales efficiency, store expansion accelerates sales growth

On the sales side, Tesla insists on the direct sales model. That is, the operating status of each retail point is directly operated, invested and managed by the company headquarters. Compared with the traditional distribution model, the direct sales model has higher single-store sales efficiency, which is conducive to improving long-term sales efficiency and also helps automobile companies establish brand image awareness among consumers. However, the direct sales model also faces problems such as slow market expansion in the early stage and high requirements for car companies' capital chain, asset turnover capabilities, and systematic management capabilities.

Compared with 4S stores, direct-operated stores do not require customers to make an appointment to view cars. Customers can experience brand new products when entering the door, which improves sales efficiency. The decrease in customer time costs and the improvement of sales efficiency will help enhance the brand image. In terms of location selection, Tesla's direct-operated stores are generally arranged in business districts, while traditional distributors such as 4S stores are generally arranged in suburbs, so direct-operated stores can effectively reduce the time cost of customers' arrival.

On the sales pricing side, car companies that adopt the direct sales model can achieve unified sales prices, prevent different pricing of different cars in different stores, avoid market prices chaos, and to a certain extent guarantee the premium ability of corporate products. However, the direct sales model has certain disadvantages in the market expansion end and cost control end. The expansion speed of the direct sales model in the short term will be slightly lower than that of the distribution model. At the same time, the direct sales model requires car companies to build their own stores, and channel inventory also needs to be borne by car companies, which has a great test of the car companies' capital chain and asset turnover capabilities.

Tesla insists on adopting the direct sales model. With the accumulation of sales experience, Tesla's sales efficiency has been improving year by year. Compared with dealers, vehicle manufacturers have a slight lack of experience in automobile sales, so they are less sales efficiency in the early stages of development. However, with the accumulation of experience, the sales efficiency of vehicle manufacturers that adopt the direct sales model will gradually improve. Tesla's single-store sales have maintained an upward trend in recent years, from 648 vehicles in 2018 to 1,429 vehicles in 2021, reflecting that Tesla's current sales system is relatively mature.

Tesla's sales store count maintains a rapid expansion. Tesla's car delivery volume increased from 244,920 in 2018 to 935,950 in 2021, of which the year-on-year growth rate of delivery volume in 2021 reached 88%, while Tesla's stores also rose from 378 in 2018 to 655 in 2021. At the same time, the expansion of stores is also accelerating, from 15% in 2018 to 25% in 2021. Considering that the chip shortage is expected to ease in the next 1-2 years, and Tesla will launch models such as Cybertruck and Semi, we expect Tesla's sales to maintain a high growth rate.

According to Troyteslike data, the delivery volume of the US, Europe and China market in 2021 accounted for 39%, 18% and 35% of the company's total delivery volume respectively. In terms of delivery volume, Tesla delivered the highest in China in 2021, at 321,000 vehicles, followed by the United States and Europe, at 299,000 and 170,000 respectively. In terms of market share, the company's sales share of new energy vehicles in the United States, China, and Europe is 56%, 10% and 7% respectively. Tesla has the highest market share in the United States, followed by China and Europe.

Considering that Tesla's new energy vehicle market share in China and Europe has room for continued growth, combined with Tesla's Shanghai Super Factory and its factory in Berlin, Europe, Germany, we expect Tesla may continue to expand its sales network in China and Europe.

4.2. After-sales end: The service system is well laid out, and the quality industry is leading

is on the after-sales end. Tesla has layout online and offline and outstanding service quality.Its after-sales system mainly includes four major modules: physical service center, self-operated/authorized body spraying center, virtual service center, and 400 customer service service. Among them, the physical service center mainly provides sales, delivery, after-sales service for car owners; the self-operated/sheet spraying center mainly provides car owners with a series of maintenance services such as repairing vehicle appearance sheet metal parts, vehicle frame main structure, etc.; the virtual service center provides car owners with online answers, reservation repairs and mobile services through intelligent back-end and remote diagnosis technology; the 400 customer service service is a bridge connecting all links, online 7*24 hours a day, providing car owners with road rescue and other business support.

Tesla's complete after-sales service system ensures Tesla's service quality to a certain extent. According to the complaint volume ranking of Cheyu Network, Tesla received a total of 151 complaints in 2021, NIO received a total of 90 complaints, and Xiaopeng received a total of 55 complaints. Although Tesla's total complaints are higher than that of NIO and Xiaopeng, Tesla's 10,000 complaints are only 4.7, that is, Tesla received 4.7 complaints for every 10,000 cars in 2021, lower than that of NIO and Xiaopeng, highlighting the excellent service quality of Tesla.

selected report source: [Future Think Tank]. Future Think Tank - Official website