(Report Producer/Author: Industrial Securities, Yu Xiaoli)
1. Build a closed-loop new energy ecosystem and reshape the travel business model
Founded in 2003, Tesla is the world's largest company for electric vehicles and photovoltaic energy storage products. At this stage, Tesra has four major businesses: energy manufacturing and storage business, electric vehicle production and sales business, automotive service business and autonomous driving business. Solar Roof is responsible for energy production, energy storage devices Powerwall and Megapack are responsible for storing energy, electric vehicles Model S/3/X/Y, Cybertruck, Roadster (sports cars), Semi (trucks) are responsible for using energy, and based on the automotive business, they have developed services such as autonomous driving, super fast charging, and car insurance. With the landing of Tesla's autopilot, the next 10 years will be the 10 years of Tesla's autopilot shine.
1.1 Founding history
1. (2003-2004) Start-up period
On July 1, 2003, Martin Eberhard and Mark Tapening founded Tesla Motors, which the founders named "Tesla Motors" in honor of physicist Nikolai Tesla. In 2004, Elon Musk joined the company as a Series A investor, investing $6.5 million and becoming Tesla's chairman of the board and largest shareholder.
2. (2005-2009) Roadster period
Musk oversaw Roadster manufacturing in the early days, but was not deeply involved in day-to-day business operations. It wasn't until 2007, when Tesla founders Martin Eberhard and Mark Tappening left the company in 2008, that in October 2008, Musk became Tesla's fourth CEO and set Tesla's long-term goal of building affordable electric vehicles for the mass market. The company began its entry into the market from high-end sports cars in its early days, and in 2006 launched the Tesla Roadster, a pure electric sports car based on the British Lotus Evora sports car. Instead of following the traditional closed supply chain model of developing its own parts, the company borrows from the development of personal computers, and its parts are purchased from mature operators around the world using an international open architecture, and are integrated with their own battery management and charging technologies. The first model, the Roadster, was a full-fledged model with body panels from French supplier Sotira, chassis from British Lotus, single-speed transmission from BorgWarner, Michigan, and brakes and airbags from Siemens. In the early days of its establishment, the company used New Taipei City, Taiwan Province, as one of its production bases, to provide Tesla Roadster with electronic components including motors, electronic controls and batteries, and the proportion of electronic parts produced by Taiwanese companies once exceeded 30%. In 2009, in response to Obama's subsidy policy, the Taiwan branch was officially closed and the production line was moved back to Fremont, California.
3. (2010-2015) Post-IPO period
In 2010, Tesla acquired Toyota's Fremont plant in California for $42 million and began production of the Model S after the renovation. It was IPO on The NASDAQ on June 29, 2010, becoming the first U.S. car company to go public since the 1956 Ford IPO. Tesla stopped production of the Roadster in January 2012 and launched its second model, the Model S, in June. Tesla released Tesla Autopilot, a driver assistance system in partnership with Mobileye, in 2014. In September of the same year, Tesla equipped all cars with self-driving hardware, but required users to purchase it and activate the autopilot function through OTA. This concept of "hardware embedding, software paid purchase" has a far-reaching impact on the development and construction of subsequent models. Independent brands Xiaopeng & WEIlai & Ideal & Extreme Kr all use the method of embedded hardware. In April 2015, Tesla entered the energy storage market with Tesla Powerwall (home) and Tesla Powerpack (commercial) battery packs. Tesla began delivering its third model, the Hauser SUV Tesla Model X, in September 2015.
4. (2016-2018) SolarCity and Model 3 period
In November 2016, Tesla acquired SolarCity on a $2.6 billion all-stock deal and entered the photovoltaic market, where the solar installation business merged with Tesla's existing battery storage products division to form Tesla Energy Sub-Company. Tesla began selling its fourth model, the Model 3 sedan, in July 2017. Compared to the Tesla Cars before it, the Model 3 is a cheaper car and is designed for the mass market. Due to the excellent sport performance, the number of new models booked exceeded 455,000 units by August 2017. However, at that time, the supply chain of electric vehicles in the United States was not perfect, and the technology used by Tesla was leading, so it was plagued by production delays. This increased pressure on companies, which at the time were among the most shorted companies in the market.
In August 2018, CEO Elon Musk briefly considered taking Tesla private. But the plan did not materialize, and it sparked many controversies and lawsuits, including the SEC's securities fraud charges. By the end of 2018, production challenges had been solved, and the Model 3 became the world's best-selling plug-in electric vehicle that year.
5. (2019-present) Global expansion period
Tesla opened its first "gigafactory" outside the U.S. in Shanghai, China, in 2019. The Shanghai Super Factory is China's first wholly-owned automobile plant owned by a foreign company, built in less than 6 months and put into operation within 12 months. The following year, Tesla also began construction of new gigafactories in Berlin, Germany, and Texas, USA. In March 2020, Tesla began delivering its fifth model, the Model Y, a compact SUV. On January 10, 2020, Tesla's market capitalization reached $86 billion, breaking the record for the highest valuation of a U.S. automaker. On June 10, 2020, Tesla's market capitalization exceeded that of BMW, Daimler and Volkswagen combined. The following month, Tesla's valuation reached $206 billion, surpassing Toyota's $202 billion and becoming the world's most valuable automaker. On August 31, 2020, Tesla conducted a 1-to-5 stock split as value rose.
1.2 Institutional investors have larger positions
Tesla is loved by institutional investors, with 3,123 investors holding 41.54% of Tesla as of December 31, 2021, with Vanguard holding the most institutional investors in Tesla.
1.3 Introduction of Management
1.3.1 Core Leadership
At present, Tesla has three main leaders, namely CEO: Elon Musk, CFO: Zachary Kirkhorn and Vice President of Powertrain and Energy Engineering: Andrew Baglino. In 2021, Musk filed a document with the SEC, changing his and CFO job titles to "Technocking" and "Master of Coin," respectively.
1.3.2 The members of the Board of Directors are complementary in their profession
Currently, the company's board of directors consists of 8 members, of whom 6 are men and 2 are women. The board of directors covers Internet companies, investment institutions, media companies, etc., which can help Tesla's development and provide basic guarantee for Tesla's operation.
1.3.3 Industry-university-research integration of core technical personnel
Relying on the first-mover advantage of the well-established software and semiconductor industries in North America, Tesla can quickly establish a connection with the academic community and build a new technical system or reconstruct the existing technical system according to Tesla's needs. For example, Tesla can complete the tape-out of the FSD chip in 14 months, completely rewrite the C compiler for efficiency, or reconstruct the neural network architecture. This is all a reflection of Tesla's world-leading technology advantages. Because of its location in the United States, Tesla can easily recruit and fully cooperate with the world's best engineers and scientists.
Director of AI: Andrej Karpathy was born on October 23, 1986, and is Tesla's director of artificial intelligence. He studied under Li Feifei, a Chinese scientist at Stanford, and is the most well-known teaching assistant in the field of artificial intelligence, ST CS231.n, specializing in deep learning and computer vision, with strong engineering landing ability. During his time at school, he developed deep learning frameworks such as ConvNetJS, RecurrentJS, REINFORCEjs, t-sneJS, etc. in JavaScript. He joined the ARTIFICIAL intelligence group OpenAI in September 2016 as a research scientist and became Tesla's Director of Artificial Intelligence in June 2017. Director of Autonomous Driving: Ashok Elluswamy was the first employee of Tesla's autopilot team, and before joining Tesla, he had experience working in both Wabco and Volkswagen, responsible for Tesla's autopilot business.
Battery Director: Jeff Dahn is a professor in the Department of Physical and Atmospheric Sciences and the Department of Chemistry at Dalhousie University in Canada. He has published more than 640 academic papers and has applied or is applying for 65 invention patents. In battery research, Professor Jeff Dahn is recognized as a pioneer in lithium-ion battery technology. He has been engaged in research in this field since the invention of lithium-ion batteries. Jeff Dahn is widely recognized as having made significant contributions to the longevity of batteries, and his research has led to the commercialization of lithium-ion batteries. Currently, Jeff Dahn's research efforts focus on the potential growth and durability of battery energy density, while also focusing on new ways to reduce battery costs.
Chip Director: Ganesh Venkataramanan is responsible for all of Tesla's current chip designs and has 14 years of experience at AMD, designing Ryzen™ and Athlon CPUs. He joined Tesla in March 2016 and worked with Jim Keller to design Tesla's first FSD autopilot chip. After FSD was put into production in 2018, he was responsible for the development of Dojo chips.
1.4 Tesla stock price reconverted
Tesla was listed on the NASDAQ in June 2010 for an issue price of $17 and raised $226 million, and since then, despite the high topic, the stock price performance has continued to be sluggish. Released in 2012 and delivered, the Model S, which was in high demand in Northern Europe with its superior performance and longer battery life (twice as dense as the Nissan Leaf) spurred the stock price to rise from $6.8 in early 2013 to $30 in late 2013, an increase of 340%. Since then, although the company's Model X and Model 3 order data have been excellent, the North American new energy industry chain is not mature, unable to meet Tesla's production needs, and the output has not been able to go up. Moreover, at that time, the market still had doubts about the development of pure trams, believing that the mileage and residual value of pure trams would affect terminal consumption. As a result, the stock price fluctuated in the range of $30 to $70 (post-reinstatement) from 2014 to October 2019.
The turnaround in Tesla's stock price came in late 2019. Previously, Tesla released the Cybertruck electric pickup truck in November 2019, and its orders quickly exceeded 1 million. After that, on January 7, 2020, the Model 3 produced at Tesla's Shanghai factory was officially put into production and delivered. China's new energy vehicle industry chain has helped Tesla out of the production capacity trap, and the production capacity of the Shanghai factory has rapidly exceeded the annualization of 300,000 units. At the same time, the net profit attributable to the parent company in the first quarter of 2020 was positive, which greatly enhanced investor confidence in the company. From the beginning of 2020 to the august 2020 stock split, the company's stock price rose by more than 400%. With the continuous increase in the Shanghai factory, the company's operating capacity has been greatly improved, and the stock price has remained at a high level. On November 4, 2021, the company's share price reached a record high of $1243.49.
1.5 Enterprise R&D Ideas - First Principles
The first principle refers to returning to the most basic conditions of things and splitting them into elements, conducting structural analysis, and finding a way to achieve the optimal path to the goal. The principle derives from a philosophical view put forward by the ancient Greek philosopher Aristotle: "There is a most basic proposition in every system, which cannot be violated or deleted." Tesla's use of first principles to solve problems is very different from the comparative approach used by other companies to solve problems. Other companies use the method of comparison, referring to what other companies have already done, so that the results of development can only produce small iterative development. Tesla uses first principles, starting from the perspective of physics, tracing things back to the most fundamental truth, and then starting from the core to reason, and finally finding a solution to the problem. We can find that Tesla solves many business problems with first principles.
In terms of reducing the cost of batteries, the first step in Tesla's analysis of the problem is to study the constituent materials of the battery. Tesla found that the battery is composed of lithium, iron, cobalt, nickel and other metals, in addition to the cost of buying these metals is absolutely impossible to reduce, the rest of the cost is generated in the process of human collaboration. The second step is to find new approaches. Tesla chose the 4680 battery shape to solve the problem of power and safety, using high nickel batteries to balance cost and performance, while setting up a battery cell factory to produce its own cells, and also signed a stable supply agreement with upstream North American local nickel suppliers to reduce costs. In terms of autonomous driving, Tesla believes that the autonomous driving of cars and the autonomous driving of humans are similar. People can use binocular vision to drive a car safely, so the amount of information carried in the optical information must be enough for the car to realize automatic driving. The only task to be completed is how to correctly extract and use the information in the optical signal, considering that the impact on the safety of automatic driving is the recognition ability and recognition speed of the machine, Tesla abandoned the lidar and took the pure visual route. At the same time, it relies on the research and development of FSD chips to solve the problem of computing power, and collects and simulates rare "long-tail scenarios" to solve the safety problem of automatic driving.
1.6 Tesla's Master Plan
Musk wrote two articles in 2006 and 2016, "Great Plan 1.0" and "Great Plan 2.0" respectively, pointing out the direction for Tesla's subsequent development.
2. The global transition to a low-carbon society is irreversible
2.1 Countries around the world have successively introduced carbon peaking and carbon neutrality policies
Carbon peaking is a historical inflection point in the increase and decrease of carbon dioxide emissions, marking the decoupling of carbon emissions from economic development, and the peaking target includes peaking years and peaks. China has been taking action to achieve the "double carbon" goal. China has pledged not to increase its carbon dioxide emissions by 2030. Since 1997, China's carbon emissions have shown a higher trend due to the high growth characteristics of China's economy, peaking around 2013 and fluctuating at a high level since then. The "14th Five-Year Plan" proposes that by 2025, China's forest coverage rate will increase from the current 23.04% by more than 1 percentage point to 24.1%, the wetland protection rate will be increased to 55%, the heavily polluted weather and urban black and odorous water bodies will be basically eliminated, the green transformation of the development mode will be accelerated, and the energy consumption per unit of GDP and carbon dioxide emissions will be reduced by 13.5% and 18% respectively compared with the end of 2020.
2.2 The competition of car companies shifts from electrification to intelligence and finally to the service of transportation and travel
With the advancement of the new energy process, consumers' expectations for car companies will change. Eventually, the core competitive elements between car companies will also change, and will shift from the current electrification competition to intelligent competition, and eventually to the transportation and travel service competition.
The first stage of competition is electrification competition. During this period, the electrification capabilities of car-sharing companies were compared, with the mass production of the Model 3 in 2017 as a landmark event. The main points of competition in this era are sandy technology, vehicle design, supply chain management and brand communication capabilities. Electric vehicles continue to gain users from old fuel vehicle companies with low cost of vehicles, excellent driving experience and high-end brand recognition. This view is validated by sales figures, with the global share of electric vehicles growing from 6.4% in 2020 to 12% in 2021. The outstanding car companies born at this stage are Tesla, Weilai, Ideal, Xiaopeng and BYD
The second stage of competition is intelligent competition. When new energy vehicle companies cannot compete in the field of electrification, intelligence will become the core of differentiated competition of tram companies. We can see that the newly released models in 2022, although they do not have the automatic driving of the Door to Door, they are equipped with a lot of automatic driving and smart cockpit hardware. It is believed that with the development of automatic driving, in the next 1-2 years, there will be leading companies taking the lead in automatic driving, so at that time it will depend on which company's automatic driving and intelligent cockpit experience is better, whose on-board intelligent hardware penetration rate can be further improved, and whose software upgrades can become the norm, thus leading the industry in a position. We believe that Tesla has a clear leading edge in the field of intelligence, and is expected to take the lead in opening the full version of FSD in North America at the end of 2022 or 2023, which will be a landmark event of intelligent competition.
The third stage of competition is the competition for mobility services. At this stage, car companies will not only produce vehicles, but also become transportation service providers. The winning car companies in the second stage of intelligentization will be the first to join the competition in this stage. In the past three years, the commercialization process of autonomous driving has accelerated, and we can see that the pace of the layout of autonomous driving travel services by various car companies has not stopped, and they are striving to achieve their commercialization. The acceptance of driverless taxis is gradually increasing, with more than half of people accepting the use of mobility services based on driverless taxis. At present, most people's concerns are still whether Robotaxi's automotive technology is safe and perfect. However, we believe that the commercialization of autonomous driving is expected to gradually advance from low-speed scenarios and closed roads to high-speed scenarios and general roads. Car companies that are the first to realize commercial travel can reduce transportation costs and obtain massive excess revenue. After solving technical problems and legal and ethical issues, the commercialization of autonomous driving will have broad prospects.
2.3 The pace of the global energy transition is accelerating
The United States, the European Union, Japan, China and other countries and regions have all made it clear in official policy documents that they will achieve carbon neutrality by 2025. The European Union's 2030 Climate Target Plan raises the 2030 greenhouse gas emission reduction target from the original 40% to 55%, and Japan's Global Warming Countermeasure Promotion Law specifies the Japanese government's target of achieving carbon neutrality by 2050. On the path to zero carbon, renewables will become the dominant energy source. The International Renewable Energy Agency proposed in the World Energy Transition Outlook that the proportion of global renewable energy generation will increase to 90% in 2050, of which photovoltaic and wind power will account for 63%, and the global photovoltaic installed capacity will exceed 14,000GW.
2.3.1 China's photovoltaic industry has a greater advantage than the United States
China's new PV installed capacity is always higher than that of the United States. From the perspective of new photovoltaic power generation installed capacity in 2015-2020, China's new installed capacity has always been much higher than that of the United States, and the gap gradually expanded in 2015-2017, and then with the phased reduction of China's photovoltaic installed capacity, the gap between China and the United States narrowed, and China's new installed capacity in 2021 was 54.88GW, which was 2.3 times the new installed capacity of photovoltaics in the United States.
The United States extends import tariffs on protective photovoltaic products. Since 2018, the United States has increased high import tariffs on imported photovoltaic products to support domestic enterprises. Tariffs are 30% from 2018 to 2020 and 15% in 2021. In February 2022, the United States extended its tariff protection policy for another 4 years. Therefore, we believe that US photovoltaic companies cannot participate in market-oriented competition with domestic enterprises.
2.3.2 Energy storage systems are developing rapidly, and electrochemical energy storage has broad prospects
Electrochemical energy storage is a key component of energy storage systems. At present, the mainstream energy storage technology route in the market can be divided into three types: thermal energy storage, electric energy storage and hydrogen energy storage, and different energy storage technologies are used in different scenarios, of which electric energy storage is the most widely used and most mainstream. Energy storage batteries are widely used in power generation, transmission and distribution, power consumption and other power system links with its functions of peak shaving and valley filling, system frequency regulation, and smooth new energy power output.
In 2021, the installed capacity of energy storage will grow explosively, and the installed capacity of energy storage in the United States will exceed 100GWh by 2030. Although the epidemic is still raging around the world, the installed capacity of energy storage in 2021 has reached a record 12GWh, and the figure in 2020 is only 4.9GWh. Among them, the United States ushered in the era of ultra-large-scale energy storage in 2021, deployed and put into operation a number of GWh-level energy storage projects, and the new scale exceeded 3.5GWh in the third quarter alone, and Connecticut, Virginia, and Maine set energy storage procurement targets. Both the U.S. Department of Energy (DOE) and the U.S. Energy Information Administration (EIA) have set a battery storage installation target of 14.5GW from 2021 to 2024 in their published reports. In the longer term, the American Energy Association has set an installed capacity target of 100GWh for 2030.
2.4 The freight market has great potential, and trunk logistics is expected to take the lead in achieving autonomous driving
2.4.1 The US truck freight market is huge and has obvious advantages in autonomous driving
The U.S. freight market is huge, with the highest proportion of truck transportation and a high concentration of trunk lines. At present, the size of the US freight market is about 1 trillion US dollars, while the US freight market will continue to grow at a CAGR rate of 3% between 2022 and 2030. In terms of segmentation, truck transportation accounted for 80%, rail transportation accounted for 9%, and air transportation accounted for 3% in 2020. Trucking is the absolute main force in the freight market, which can reach 8000 billion US dollars, while 80% of the freight transport in the United States is transported through 10% of the trunk line, and the concentration is high, so it is also promising to carry automatic driving on trucks.
The U.S. government promotes transportation innovation, with more than 40 states allowing self-driving truck testing. The U.S. government is committed to promoting ground transportation innovation to ensure that the United States is a world leader in the development and integration of autonomous vehicle (AV) technology, while prioritizing safety, security, and privacy. As of March 2021, 43 states allow L4 class autonomous semi-truck testing, of which 24 states allow commercial deployment of L4-level automated semi-trucks. As the U.S. government encourages autonomous driving, self-driving trucks will be a clear path, and self-driving software providers are expected to partner to create a safer trucking industry. (Source: Future Think Tank)
3. Supply chain and new technologies to build Tesla cost advantages
3.1 Continuously pursue an efficient and stable supply chain
3.1.1 Pursue supply chain integration and continuously expand upstream
In order to reduce the cost pressure caused by the limited raw materials or soaring prices, and strengthen the control of resources in the industrial chain, Tesla has continued to actively layout the upstream, focusing on material procurement, and established long-term and stable cooperative relations with a number of enterprises, realizing the integration of the supply chain and greatly improving competitiveness. At the same time, supply chain integration is also a manifestation of Tesla's application of first principles thinking. Taking the layout of nickel-cobalt lithium mines as an example, Tesla has actively "encircled" the raw materials in the field of battery raw materials and signed raw material supply agreements with local enterprises.
Signed a large contract for nickel mines, and actively locked nickel in the global attack. Since announcing the development direction of high nickel and low cobalt as the future battery development direction in 2019, Tesla will ensure the adequate supply of nickel as its top priority, and has signed nickel supply orders with a number of mining companies around the world. In March 2021, Tesla signed an agreement with the Goro Nickel Mine in New Caledonia to establish a nickel trading partnership. In July 2021, Tesla signed a nickel supply agreement with Australian mining giant BHP Billiton, where its Western Nickel West in Western Australia will supply about 18,000 tonnes of nickel to Tesla each year. In October 2021, Tesla and Prony Resources of New Caledonia reached a multi-year nickel supply agreement, which will purchase more than 42,000 tons of nickel in total. This is Tesla's second big move in nickel procurement in the second half of last year, and Prony has become Tesla's main nickel supplier. In January 2022, Tesla signed a six-year contract with U.S. miner Talon Metals to supply 75,000 tons of nickel, and Tesla will purchase at least 75,000 tons of nickel concentrate from Talon Metals' Tamarack mine in Minnesota over the next 6 years. This is the first nickel supply agreement in the U.S., and we believe this is a manifestation of Tesla's pursuit of supply chain integrity.
The high price of cobalt continues, and talking about "de-cobalt" is hoarding cobalt. Although there are high-nickel materials or alternatives to today's cobalt batteries, within the next 10 years, the possibility of cobalt-containing batteries being replaced by other technologies is very small. In the short term, cobalt, as an important factor in power batteries, will still be an indispensable rare metal on new energy vehicles before cobalt-free technology matures, so Tesla's control over cobalt is strengthening. Scarcity of resources, excessive concentration of reserves, and compliance during mining have resulted in the price of cobalt remaining high. In 2019, the world's proven cobalt reserves were approximately 7 million tonnes. Compared with the reserves of nickel, manganese, aluminum and other minerals in the range of billions or even tens of billions of tons, cobalt is a very scarce metal. Based on the global production capacity of 140,000 tons in 2019, cobalt can be mined for about 50 years. The characteristics of the concentrated distribution of cobalt ore are also very obvious. Congo, Australia and Cuba are the world's leading producers of cobalt, with the three countries combining about 70% of global cobalt reserves. Cobalt mining is concentrated in the hands of several major mining groups such as Glencore, Luoyang Molybdenum, and Eurasian Resources, which makes it difficult to have more room for negotiation on the price of cobalt.
The global sweeping goods stabilize the supply of lithium raw materials and serve people with "lithium". In March 2018, Tesla signed a 3-year lithium supply agreement with Australian lithium mining company Kidman Resources. In September 2020, Tesla signed a five-year lithium concentrate supply agreement with Piedmont Lithium, a 5-year lithium miner in Australia, where Piedmont will supply Tesla with one-third of its production (estimated at 53,000 tons) from its North Carolina deposit. The parties expect to begin deliveries of lithium mineral products between July 2022 and July 2023. In November 2021, Tesla and Chinese lithium giant Ganfeng Lithium once again signed a 3-year battery-grade lithium hydroxide procurement contract, which shows that there is still room for substantial growth in demand for ternary batteries in the future. This is not the first cooperation between Ganfeng Lithium and Tesla. Earlier in 2018, the two sides signed a three-year strategic cooperation agreement, when Tesla's annual purchase quantity was about 20% of the total production capacity of Ganfeng Lithium's product that year. This year, Tesla has increased its size of Australian lithium mines again, and on February 16, it signed a five-year spodumene concentrate supply agreement with Australian lithium miner Liontown Resources. Tesla will purchase 100,000 dry metric tons of spodumene concentrate from it in its first year starting in 2024 and increase to 150,000 dry metric tons per year in the following years, for a total of about 700,000 tons. On March 2, Tesla also reached a supply agreement with Australian lithium miner Core Lithium to supply Tesla with a total of up to 110,000 tons of spodumene concentrate over four years, with the supply scheduled to begin in the second half of 2023.
3.1.2 Promote the localization of supply chains and usher in development opportunities for local enterprises
Tesla sold 308,600 units against the trend in the fourth quarter of 2021, when the automotive industry was generally missing cores, thanks to Tesla's ongoing pursuit of localization of the supply chain. Taking Tesla's Shanghai factory as an example, tesla's Model 3 model has been provided by Japanese and Korean companies LG and Panasonic in the early days of production in the Shanghai Gigafactory. Shortly after Tesla started localization, in order to reduce transportation costs, CATL quickly became the exclusive power battery supplier of Tesla's Shanghai factory. CATL and Tesla have signed two cooperation agreements, one in February 2020, the locked supply period is from July 1, 2020 to June 30, 2022, and the other is in June 2021, the lock is from January 2022 to December 2025. These two times are the time when the batteries of the Ningde era began to be supplied to the domestic Model 3 and Model Y in batches. Not only power batteries, Tesla has begun to localize 90% of auto parts.
3.1.3 Multi-track supply strategy
In order to avoid the supply failure of one supplier from affecting its own production, Tesla adopts the strategy of designating multiple suppliers at the same time. Taking batteries as an example, as of April 2022, Tesla has a total of four battery suppliers around the world, which are panasonic, LG, CATL and Tesla. This strategy not only avoids supply chain fluctuations, but also reduces their own costs by competing with each other.
3.2 Promote cost reduction with new technologies
3.2.1 Multi-dimensional performance dominant 4680 large battery
The power battery is the decisive factor and cost unit of the performance of the electric vehicle, and the economy and performance of the battery determine the performance and economy of the electric vehicle. On Battery Day in September 2020, Tesla introduced the 4680 battery for the first time. Compared to the previous generation of 2170 cells, the 4680 cells have a 5x increase in energy, a 16% increase in range, a 6x increase in power, and a 14% reduction in cost per kWh at the battery pack level. Electrodeless ear design of the battery cell: After the copper foil and aluminum foil are rolled up, the positive and negative collectors are directly connected to the cover plate or housing. Electrodeless ear technology simplifies the winding and coating process in battery production, while reducing the internal resistance of the battery cell and removing the main heating components.
Dry electrodes are used for cell production: Compared with the traditional wet electrode production method, the core technology of the dry electrode reduces the solvent when the electrode is squeezed. The dry cell process has two major advantages: one is a significant improvement in efficiency, the battery has better conductivity; the other is to reduce costs, dry electrodes in the wet process simplified 10 processes, assembly line efficiency increased by 7 times. Positive and negative electrode material ratio adjustment, battery performance optimization: in terms of cathode materials, the use of cobalt-free "high nickel cathode", the cost per kWh is reduced by 15%; in terms of negative electrode materials, metallurgical silicon is used as raw materials, and the performance of silicon anode is improved by coating and special adhesive mixing by ionic conductive polymers. Increased silicon content to 20%, reduced costs by 5% to $1.2/kWh at the battery pack level, and increased vehicle endurance by 20%.
Tesla has three 4680 battery production sites. To meet the stable supply, they are: (1) Pilot Factory in California, USA: It is the research and development center of Tesla 4680 battery, and it is also the first battery production line to try. The production capacity in February 2022 will be 1 million units per month, corresponding to the battery demand of 1,000 vehicles. (2) Texas plant in the United States: By February 2022, the equipment will begin to enter the production line, and with the production capacity of 500,000 yuan in the Texas automobile plant, it is expected to start trial production in Q2 2022. We believe that at least 50GWh of capacity is allocated. On the one hand, it is ready to supply model Y, and on the other hand, it also needs to be allocated to Cybertruck. (3) Berlin, Germany: Planning to cooperate with the production capacity of the German automobile plant, the production capacity is predicted to be 20-30GWh. It is expected that the gradual release of production capacity will begin after Q2-Q3 2022.
3.2.2 Future evolution of production processes: chassis integration and body integration
Introduced one-piece chassis battery pack to reduce volume and improve safety. In October 2021, during the Berlin factory opening day, Tesla first displayed the one-piece chassis battery pack equipped with the new Model Y that will be put into production at the Berlin factory, and the new battery pack no longer has a battery module, but is directly covered with 4680 cells, which greatly reduces the middleware in the battery pack and significantly improves the volume utilization rate of the battery pack. The integrated battery pack leaves more space on both sides, mainly to leave room for lateral collisions in extreme cases, thereby improving the safety of battery collisions. After the new one-piece battery pack becomes the chassis structure, it is connected to the front and rear body to form Tesra's new "three-in-one" chassis.
Integrated die-cast body with a 10% weight reduction and 370 body parts. The Model Y produced in Berlin features an integrated die-cast body front and rear, which reduces welding processes and requires no coating and heat treatment than stamping parts assembly, increasing productivity and reducing costs. At the same time, the total body weight was reduced by 10%, the battery life was increased by 14%, and the body parts were reduced by 370. The new structure has high structural strength and stiffness, and the concentrated arrangement of the battery cells also reduces the vehicle's moment of inertia, which is more conducive to handling and steering response.
3.2.3 The degree of automation of the production line is strong
Acquisition of a number of automation equipment companies to increase the degree of automation of production lines. In order to improve efficiency and shorten the ramp-up time of production capacity, Tesla has acquired a number of automation equipment companies since 2015. The acquisition of the Riviera Tool for the production of stamping die systems in 2015 enables the company to cost-effectively produce new stamping and plastic parts and supports the development of new metal forming technologies. In 2016, it acquired the German company Grohmann Engineering for its automated production systems such as assembly lines, lasers and pressure welding, increasing vehicle production and cutting costs, while adding around 700 employees to the automation division. The acquisition of Perbix at the end of 2017 enabled Tesla to produce more automation equipment for use in automotive production lines. In the same year, the company's president acquired automotive equipment manufacturer Compass Automation to improve automated assembly and inspection technology, and the company's president was responsible for developing new automated robots after joining Tesla and helping to design new factories such as the Berlin Gigafactory. Acquired Canadian battery manufacturing company Hibar Systems in 2019, with its high-precision metering pumps, automated battery manufacturing equipment, and lithium-ion battery installations bringing the production line of alkaline batteries to a speed of 1,000 PPM. The acquisition of German company ATW Automation in 2020 will accelerate the automation of Tesla's battery production with its automotive battery module and battery pack assembly technology.
Master Plan 1.0: A secret that only you know
4.1 Build a sports car - Roadster
The Roadster is the originator of the pure electric model. It was an all-electric sports car based on the Lotus Elise chassis, produced between 2008 and 2012, and Tesla began to single-handedly promote the development of lithium-ion battery-powered electric vehicles. The Roadster is the first fully electric vehicle legally mass-produced on the highway, which can travel up to 320 kilometers per charge. The Roadster's greatest success is that it uses a three-phase four-pole AC induction motor to maximize its acceleration capabilities, with an acceleration time of 0 to 60 mph in 3.7 seconds, far exceeding the fuel sports cars of the same period. And it's inexpensive, starting at $100,000, far below other models with the same acceleration capability. At the same time, Tesla and the British BBC ace car program Top Gear added communication and topicality to Tesla on the Roadster dispute, and through the Roadster, Tesla and electric vehicles were equated, completing the education of consumer cognition and establishing the user's mind. Due to Tesla's contract with Lotus Motors for 2,500 chassis that expired at the end of 2011, Tesla officially discontinued the first-generation Roadster in January 2012.
4.2 Use the revenue from sports cars to build cars with relatively low prices
After completing production of the Roadster, Tesla fulfilled its promise to produce the relatively low-priced Model S and Model X in 2012 and 2015, respectively. Model S: Tesla's first luxury car, positioned for the public. Released in 2012 and mass-produced. The exterior is elegant and atmospheric, the front face design is quite modern, the hood contour is full, the body is young lines, with large thick-walled tires, and the taillights are soft. The interior is a perfect combination of simplicity and technology, with a 17-inch large central control screen, covering the super multi-functional in the car. Powerly, the Model S matches a single-speed transmission, the high-performance P100D 100 km/h acceleration time is only 2.7s, and the dual motor drive can make the power more powerful, and even achieve supercar-level acceleration performance. In addition, the extra large battery capacity can achieve a range of up to 658 kilometers, so long-distance driving is not a problem.
Model X: A crossover between an SUV and a coupe. Released in September 2015, it adopts a "2+3+2" seven-seat layout design, with a maximum endurance of about 386 kilometers, a maximum speed of 250km/h, and an acceleration time of 4.4 seconds per 100 kilometers, making it the fastest SUV with 100 kilometers of acceleration. If this acceleration does not meet the needs of users, users can also add money to upgrade to the "Ludicrous" acceleration mode to get faster acceleration of 100 kilometers. The Model S and Model X ushered in a redesign in the first quarter of 2021. Replacing the original Model S and Model X are the Model S Plaid and Model X Plaid versions, which retain a similar look but have a new interior technology. Refactoring Tesla software: The Plaid version of the car-powered system uses AMD Ryzen family processors with a new UI. Boosted battery voltage: The Power Battery voltage of the Plaid version is increased from 400V to 450V, which will allow the Plaid version of the car to maintain 250kW of power on the V3 overcharge for a longer period of time, and it will also make it possible to be compatible with Tesla's V4 overcharge in the future.
4.3 Use the money earned to build a model that is affordable to the people
After the Model S and Model X went into production, Tesla kept its promise to mass-produce the affordable Model 3 and Model Y in 2017 and 2020, respectively. Model 3: A landmark model that challenges traditional luxury cars. If the previously released Model S and Model X are defined as toys for the rich, then the Model 3 is the core product that Tesla uses to challenge the traditional luxury brands Mercedes-Benz, BMW, and Audi. After solving the production bottleneck, the Model 3 surpassed the Mercedes-Benz C-Class full series, BMW 3 Series 4 Series 5 Series, and Audi A4/A5/A6 to become the national small and medium-sized luxury car sales champion in 2019.
Model Y: Breaking the H6 Pin Myth. Released in March 2019, Model Y shares 75% of the spare parts with the Model 3 and begins production in the first quarter of 2020. In September 2021, the Model Y sold 33,033 units, surpassing the Haval H6's 23,341 units and breaking the Haval H6's 99-month sales record. The Model Y is currently Tesla's most popular model, and all of Tesla's gigafactories will produce the Model Y. The Berlin Gigafactory and the Texas Gigafactory joined production in March and April 2022, respectively, and we believe that the Model Y is expected to sell more than 850,000 units in 2022.
4.4 While doing these things, give users a zero-carbon solution
Tesla's acquisition of SolarCity in 2016 transformed Tesla into the world's only vertically integrated energy company, offering customers end-to-end clean energy products, including solar panels, home energy storage devices, and electric vehicles.
Master Plan 2.0: Tesla's True Ambitions Under Secrets
5.1 Photovoltaic + energy storage to create a sustainable energy ecosystem
5.1.1 Solar Roof: Powering homes with fully integrated solar systems
In 2016, Tesla launched Solar Roof, which uses a fully integrated solar system to power homes and help effectively control monthly electricity bills. Compared to the combination of traditional roofs and solar panels (Typical Roof + Solar Panels), the total cost of solar roofs is slightly lower. Aesthetics: Solar Roof is almost indistinguishable from traditional roofing materials, and the roof tiles look very smooth and more aesthetically pleasing. Smart: With the Tesla App, users can monitor electricity usage in real time and observe the rise and fall of electricity bills, including historical usage. With instant alarms and remote access, users can also control the system anytime, anywhere, manage their own solar system and power consumption. Durable: Solar Roof is more than 3 times stronger than standard roof tiles, making it easy to cope with all weather conditions and last for decades. Tesla offers users a 25-year warranty. On March 22, 2022, Tesla announced that it would update the solar roof version.
5.1.2 Energy Wall (Powerwall): Household energy storage system to ensure the uninterrupted operation of home appliances
In April 2015, Tesla launched a home energy storage battery, the Energy Wall (Powerwall). It is connected with the household photovoltaic power generation system (solar roof or solar panels) to form a self-sufficient energy system - micro-grid. During the day, Powerwall uses photovoltaic power generation to provide energy to the family and store excess energy, powering the family or Tesla products during peak or power outages at night, enhancing the independence of the system's electricity consumption. Powerwall is a large "charging treasure", the length, width and height are: 1150mm * 755mm * 155mm, weighing about 114kg, up to 10 can be used together. It has built-in rechargeable lithium battery pack, liquid cooling module, inverter, etc., energy capacity of 13.5kWh, discharge degree of 100%, charge and discharge cycle efficiency of 90%, peak power 7kW, continuous power 5kW, backup power seamless conversion. At the same time, Powerwall's design is stylish and compact, which can well fit a variety of residential styles and achieve a variety of indoor or outdoor space installations. On March 22, 2022, Tesla announced that it will launch the Powerwall 3, which has a larger storage capacity and lower cost.
5.1.3 Megapack: Utility energy storage product designed to provide an alternative to natural gas power plants (Peaker).
On July 30, 2019, Tesla unveiled its new utility-scale energy storage product, megapack, which is Tesla's third and largest energy storage system and the latest move to restructure and grow its energy storage business. Modeled on its giant battery system deployed in southern Australia, Megapack provides a stable renewable energy supply to grid systems, extends transmission and distribution investment life, provides voltage and capacity support, participates in power market regulation, and builds microgrids. Utilities can use Megapack to store excess solar or wind energy to support peak loads on the grid, rather than using natural gas power plants. Megapack each system is fully assembled, including battery modules, two-way inverters, thermal management systems, AC main switches and control systems, can store up to 3MWh of energy, inverter capacity of 1.5MW, multiple sets of Megapack series can also form a super energy storage device with energy storage of super 1GWh. Megapack requires 40% less space, one-tenth the number of parts in its class, and installs up to 10 times faster than comparable energy storage systems on the market. In May 2021, Tesla has decided to change the battery used in megapack to lithium iron phosphate battery to reduce production costs and drive up sales.
5.1.4 Three large plants in North America provide sufficient capacity
Super battery factory Nevada Giga. In June 2014, Tesla broke ground on the gigafactory outside Sparks, Neva, to provide enough batteries to support Tesla's projected demand for electric vehicles. This is Tesla's first self-built gigafactory. Today, the Gigafactory not only meets the Model 3's needs for motors and battery packs, but also covers the production of energy storage products such as Powerwall and Powerpack. At the same time, in the future, the roof of the factory will be fully paved with solar panels. As early as February 2018, solar panels were already laid on the roof of the Gigafactory. So far, 3,200kW of solar panels have been installed, with plans to increase to 24,000kW by the end of next year. After the plant is fully completed, the electricity required for production and operation will be provided by solar panels on the roof and wind power generation equipment around the factory, and all will use renewable and clean energy.
An energy plant under construction mega Lathrop. In September 2021, Tesla officially started construction at its new Megafactory in Lathrop, California, covering an area of more than 46,000 square meters, which will specialize in the production of its grid-scale battery energy storage system Megapack. Before the Lathrop plant was built, Powerpack and Powerwall products were manufactured at the Nevada Gigafactory.
5.1.5 Autobidder: Virtual power plant, automated energy trading platform
In 2017, Tesla launched Autobidder, a smart energy management platform. Efficient resource allocation: Using the "distributed photovoltaic + energy storage + charge and discharge control" strategy, Autobidder can automatically dispatch energy and electricity in the Tesla ecosystem such as vehicles, batteries, photovoltaic equipment, and even the power grid, comprehensively utilize energy, achieve the maximum consumption of green electricity, and become the core and hub of Tesla's distributed energy layout. Maximum Business Benefits: Autobidder is a real-time trading and control platform for energy assets that automatically monetizes battery assets by providing features including price forecasting, load forecasting, power generation forecasting, dispatch optimization, and intelligent bidding. Value-based asset management and portfolio optimization enables owners and operators to maximize revenue based on business objectives and risk appetite. And through seamless integration of hardware and software, Autobidder allows users to generate revenue immediately and 24/7 after a project is launched. At present, Autobidder has managed more than 1.2 gigawatt hours of energy storage, and is basically in the commercial verification stage, and has several landing scenarios, such as centralized wind and solar + centralized storage, single-user optical storage and charging, multi-user optical storage and charging, etc.
5.1.6 Solar Roof has no cost advantage, and the prospects of Tesla's energy business are unclear
Tesla uses the PPA model to sign an agreement with users, consumers rent solar products from the company, the ownership of solar panels still belongs to the company, and the excess electricity produced is sold to the grid, thereby reducing the bill of all home appliances. This business model relies heavily on the combined cost of energy production, which requires the combined cost of solar power generation to be lower than that of directly drawing electricity from the grid. From the past performance, the cost of solar power generation is not lower than the cost of purchasing electricity, after the installation of Solar Roof products, a considerable number of users have reflected that their electricity bills have not decreased but increased, and Wal-Mart has also reported a number of fires caused by Tesla solar products. From the perspective of installed capacity, the installed capacity of photovoltaics per quarter has not risen significantly, and it is necessary to observe the follow-up cost reduction. The main customers of Tesla's energy storage business are local governments, Texas is its main customer, and the Texas power grid is independent of other power grids in the United States, so there is a large demand for peak shaving. But the demand for energy storage in other regions remains to be seen. On the whole, Tesla's energy business gross profit fluctuates greatly and shows an overall downward trend, falling to the lowest -100 million US dollars in the first quarter of 2021, which shows that the scale effect cannot improve gross margin at this stage. We believe that Tesla's energy business is more uncertain.
5.2 Expand the product line of electric vehicles, and the high-nickel storm is coming
In September 2020, Tesla mentioned "high nickelization" at battery day, announcing that future models will shift from "cobalt" models to "nickel" models. From a metal point of view, nickel has a high energy density and low cost, while the role of cobalt is to stabilize the battery structure, but the cost is high. Tesla noted that a very high nickel in the battery could make it possible for the battery to be completely cobalt-free and reach a 15% cost reduction. Tesla put forward three exploration ideas: one is to apply iron-based cathode materials to passenger cars and commercial energy storage with long cycle times; the other is to use nickel-manganese cathode materials, such as 2/3 nickel and 1/3 manganese, applied to long-range passenger cars and home energy storage; the third is high nickel in the absolute sense, which is applied to high-load vehicles such as electric pickups Cybertruck and long-distance commercial trucks Semi. (Source: Future Think Tank)
5.2.1 Semi Trucks is expected to be the first to realize autonomous driving of commercial trunk logistics
In November 2017, Tesla released the Electric Semi Truck. The Semi's exterior design uses aerodynamic principles, with a wind resistance coefficient of only 0.36, which can achieve good power saving; a four-motor design, the maximum power exceeds 735kW (about 1000Ps); as a heavy-duty truck, zero hundred acceleration time is only 5 seconds in the no-load state, zero hundred acceleration time is 20 seconds in the full load state; at the beginning of 2017, Tesla said that the Semi can last 500 miles (about 804 kilometers) per charge, and has a special " Megacharger "electric truck charging power station, half an hour can charge 80%. We believe that only the high-nickel 4680 battery can meet the high energy density required by the Semi, with a full load of 40 tons, the endurance is expected to exceed 1000 kilometers. Therefore, we believe that tesla will mass-produce the Semi after the 4680 battery capacity is fully charged in 2023, and the production location is in the Texas factory. At the same time, Semi is equipped with an autonomous driving system, combined with the use scenarios of trunk logistics, we believe that the commercial use of FSD will be the first to be applied from Semi to trunk logistics scenarios.
5.2.2 Cybertruck: The largest increase in the U.S. market in 2023
In November 2019, Tesla released its electric pickup Cybertruck. The number of existing orders has exceeded 1 million, and we think that in 2024, the Cybertruck will replace the best-selling pickup truck F-150 in the United States as the top pickup truck in North America. Subversive appearance: Cybertruck subverts the traditional pickup image, using a mech-style design, black and white color, in a very aggressive punk metal shape, the sense of technology is explosive. The dimensions are exactly the same as the F150, the best-selling pickup in the U.S., and the body is 15 to 20 percent smaller than the current model in the recent past, or in a smaller version. At present, the Cybertruck mass production version of the spy photo has been announced in January this year, the door is designed without handles, compared with the concept car, mainly the windshield added wipers, and the black rearview mirrors are installed on both sides of the body.
Exoskeleton construction: Cybertruck's body is made of super-hard 30X cold-rolled stainless steel and ultra-strong armored glass for extreme durability and passenger protection. At the same time, the utilization of space in the car is greatly improved: the front row can sit 3 people horizontally, the whole vehicle can sit 6 people, and the cargo compartment volume reaches 100 cubic feet. Roll-over performance: The Cybertruck has a powerful pull and up to 14,000 pounds of towing capacity to easily handle most towing scenes; zero-hundred acceleration time of just 2.9 seconds, marking the pickup's entry into the "3-second club", and a number of supercars; excellent off-road capabilities: 35 degrees close angle, 28 degrees departure angle, 406mm ground clearance, and maximum cruising range of up to 610 miles (about 976 kilometers). Camping Artifacts: Take the camping experience to a whole new level. Trailers, roof camping kits (e.g. tents on the roof) are supported, kitchen modules are supported and all modules can be controlled within the on-board system. With the introduction of the extended battery life configuration, you can optionally install solar panels to charge the battery.
Cybertruck will be sold primarily in North America in the future. Competitors include the Ford Lighting, Rivian R1T, Ford Oil-Fired Raptor, and GMC Electric Hummer. At present, the overseas pickup truck market has officially released and will soon be put on the market of five electric models, namely Ford F-150 Lightning, GMC Hummer EV, Rivian R1T, Tesla Cybertruck and Chevrolet Silverado EV.
5.2.3 Second Generation Roadster: The Pearl of Supercars
In 2017, Tesla released a new generation of Roadster. The new sports car accelerates from standstill to 96 km/h in just 1.9 seconds, has a built-in 200 kWh battery pack, lasts 620 miles (about 997 km) per charge, is powered by three motors and four-wheel drive, and has a torque of up to 10,000 Nm.
5.2.4 Cheap small car priced at $25,000: Locally designed, unreleased new car
In September 2020, Tesla said on Battery Day that it would launch a fully autonomous vehicle priced at $25,000 within three years. The media believes that the car will be named Model 2 or Model Q. We believe that this car will be a local model developed by Tesla's R&D centers in Europe and China in light of local needs. Tesla's China R&D center was put into operation in November 2021, and we think Tesla will release this local model in 2024 and will use it as a shared taxi in the future. Tesla's potential new car also has a Tesla Van, which we expect to be available in January 2024.
5.2.5 Four gigafactories provide 3 million production capacities in 2023
As of April 2022, Tesla has a total of four gigafactories around the world. With a reported capacity of 1.05 million units in 2021, we expect the Texas and Berlin plants to contribute a total of 1 million units per annum when they are commissioned this year, but due to the time required for plant commissioning, the two new plants are expected to provide a total of 600,000 units of actual capacity this year. At the Shanghai plant, we noted that 60,000 units were produced in January 2022 and 720,000 units per annum has been achieved. At the same time, Tesla will expand the scale of employees by 50% after the Spring Festival, so we conservatively estimate that Tesla's actual production capacity this year can reach 2.1 million vehicles, and by the end of 2022, the annualized capacity of Tesla's four-seat factory will exceed 3 million vehicles, which can meet the production capacity demand in 2022 and 2023.
Tesla said at its fourth quarter 2021 results meeting that they will announce the location of Tesla's new super factory by the end of 2022. We believe that Tesla's site selection will comprehensively consider the supporting industry chain, construction cycle, labor costs, logistics convenience and so on. Potential sites include China, Japan, South Korea, southeast Asia, etc.
In the next four years, Tesla will sell 7 models (Roadster/Model S/Model 3/Model X/Model Y/Semi/Cybertruck) at the same time, with a complete product matrix covering sports cars, cars, SUVs, pickup trucks and trucks. In terms of total sales, the Model 3 and Model Y will be the best-selling models, and we predict that these two models will reach the highest sales value in Q3 2023, with total quarterly sales expected to exceed 530,000 units. The three products that have been announced but have not yet been officially launched, Semi, Cybertruck and Second Generation Roadster, we expect to enter the market in 2023 Q1, 2023 Q2 and 2023 Q3 respectively, of which Cybertruck's sales and revenue will be strongly ahead of Semi and Second Generation Supercars, with sales of 300,000 vehicles and generating more than $15 billion in revenue by 2025 Q4.
5.3 Self-developed ecology has become the core advantage of automatic driving
Tesla is the only car company in the world that realizes independent car manufacturing, hardware self-research, software self-research, and continuous data iteration.
5.3.1 Evolution of autonomous driving hardware: from external procurement to self-research
Tesla's Autopilot platform has undergone three upgrade iterations: 1) the 2014 HW1.0 version, which uses a Mobile EyeQ3 processor, 1 camera, 1 millimeter-wave radar, and 12 ultrasonic radars; 2) the 2016 HW2.0 version, using the NVIDIA Drive PX2 computing platform, with 8 cameras to achieve 360-degree surround view; in 2017, the HW2.5 version was introduced to add computing power and chip redundancy on top of HW2.0; 3) it was launched in 2019 The HW3.0 version uses two self-developed FSD chips, while the radar was removed in early 2022, retaining only the camera. In 2022, Tesla is expected to launch HW4.0 to achieve hash rate upgrades.
There are two Tesla self-developed SoC chips on the Autopilolot HW3.0, providing redundant power supplies, interface types and sizes to ensure backward compatibility, supporting subsequent iterations and upgrades. Multiple cameras surrounding Tesla cars ensure that the field of view is redundantly covered, thus guaranteeing safety.
The FSD chip was designed by Tesla and launched for its own car in early 2019. Fabricated on a 14nm process at Samsung's plant in Austin, Texas, the chip encapsulates approximately 6 billion transistors on a 260mm2 silicon wafer with an FCBGA design and AEC-Q100 certification. The FSD chip is divided into five zones. The CPU adopts Cortex-A72 architecture, three sets of quad cores, a total of 12 cores, the highest operating frequency is 2.2GHz, this part of the processor core is used for general-purpose computing and tasks; the GPU frequency is the highest 1GHz, the highest computing power is about 600GFLPOS; the neural network accelerator (NPU), Tesla uses a self-developed architecture, designed 2 NNA cores, each core can perform 8-bit integer calculations, running frequency of 2GHz, A single NNA has a peak hash rate of 36.86TOPS and two NCAs have a peak hash rate of 73.7 TOPS; an image signal processor (ISP) with an internal 24-bit pipeline is designed to process eight HDR sensors on Tesla cars that can process a billion pixels of image information per second. Features such as tone mapping are added to the processing, and details such as shadows, highlights, and dark spots are allowed to be handled autonomously, as well as noise reduction designs; the security module contains a set of dual-core synchronized CPUs for performing arbitration of car information. It determines whether the execution plans emitted by the two FSD chips on the entire module of the FSD match and whether the process of driving the actuators is safe.
5.3.2 Software Self-Development: A Pure Visual Solution that Takes a Different Approach
In order to maintain competitiveness in autonomous driving, many automakers generally adopt multi-sensor fusion perception schemes based on cameras, millimeter-wave radar, lidar, etc., so that vehicles can detect in their environment. But From the perspective of first principles, Tesra believes that if people can drive, then the machine can also do it through the camera, so it takes a different route. Tesla used 8 cameras, 1 millimeter radar and 12 ultrasonic radars in the Model 3, without lidar. This design greatly reduces the cost of autonomous driving hardware, so that autonomous driving hardware can be embedded in all vehicle models. Take the Tesla Model 3, for example, a model released in 2017, but through software OTA and hardware replacement (from HW 2.5 to 3.0), Tesla can still run the latest self-driving features in the earliest Model 3.
Process image information: Build a Vector Space based on 8-camera images. Tesla uses 8 cameras as inputs, first calibrates the video stream, and then uses resNet and BiFPN neural networks to identify the image characteristics of objects, lane lines, and traffic lights in the environment. The image features are then converted to Key and Value to the Transformer of Multi-Head Attention, and the trained model retrieves the required features in the form of a lookup table for prediction, and finally successfully maps the features of the image space to the multidimensional Vector Space.
Processing time information: Outputs more stable results. The "vector space" obtained by processing image information is a single-frame information, without considering the timing information. To this end, Tesla added a feature queue module and a video module to the neural network to cache features on the timing and fusion timing information, and finally successfully integrated the information collected by 8 cameras into vector space. This method uses a model for integration, blending information on multi-camera timing and space, without the need for geometric transformation of the original video stream or the result annotation at the image layer. Labor costs are greatly reduced and end-to-end processes are implemented.
Auto-annotation system: Speed up training. Unlike other AI companies that rely on manual labeling, Tesla has developed an automatic labeling system for autonomous driving. During the training phase, Tesla re-projected the observer point of an object on Vector Space back into the camera space, compared it with the result of direct semantic segmentation in the original camera image, and then made joint optimizations across time and space dimensions on each camera to obtain high-quality reconstruction results. Since on Vertor Space, the space is continuous, multiple different vehicles pass through the same location in different ways at the same time, you can continuously optimize this Vector Space, and you can also crowdsource the Vector Space of other locations in this area to get a complete and accurate Vector Space of this location. Shadow mode collects data + self-learning mode simulates Cornor cases. In addition to coping with everyday driving scenarios, AI drivers also need to deal with some of the less common Corner cases. To this end, Tesla has developed data automatic labeling (Data Auto Labeling) and automatic training framework "Data Engine" in-house.
Decision Making: Use Monte Carlo Tree Search to balance the three major factors of decision making. In the process of autonomous driving, people do not want to spend too long (Less Traversal Time), do not want to produce Low Collision Risk, do not want to produce less Lateral Jerk, and do not want to accelerate or decelerate vertically (Less Lateral Accel). Tesla uses Monte Carlo tree search to plan paths, which significantly reduces the number of attempts compared to traditional A* calculations.
For the use of lidar: Used to provide Ground Truth. One of the big problems with cameras reaching the level of lidar is to be able to measure distance. In July 2021, Tesla used lidar as the Ground Truth, and has developed a "pseudo-lidar" algorithm based on pure visual ranging technology, which uses multiple cameras to achieve distance measurement of the target, and has achieved excellent results.
5.3.3 Self-developed neural network training chip Dojo solves the computing power problem
The next generation of supercomputing chip Dojo: AI's "training ground", unsupervised using massive data for efficient training. On August 20, 2021, at Tesla AI Day, Tesla released its self-developed AI chip D1 for training the supercomputer Dojo. It is the building block of the computing plane based on a large multi-chip module (MCM), which is composed of 120 MCM tiles and integrates to achieve ultra-high computing power of 1.1 EFLOPS, which is the fastest AI training computer in the world. In fact, Dojo is derived from the Japanese word for "dojo," and Tesla's name implies that it makes this supercomputer a training ground for autonomous driving.
Dojo is a supercomputer that can use massive amounts of video data to do unsupervised annotation and training. There are two key points: the first is the collection of massive amounts of data. As we all know, autonomous driving requires collecting a lot of data to train neural networks to make vehicles smarter; the second focus is on unsupervised labeling and training. Created from an array of 354 training nodes, the D1 chip uses a 7nm manufacturing process to implement 362TFLOPS machine learning calculations that automatically learn and identify pedestrians, animals, potholes, and other data on marked roads. After collecting massive amounts of data in Dojo, the algorithm evolution is continuously enhanced by automating deep neural network training without the need for a large number of researchers, greatly improving the training efficiency, and finally achieving fully automatic driving (FSD). Such a Dojo chip has built a "practice room" for AI. After walking out of this "practice room", dojo chips are likely to be difficult to find opponents in the short term.
5.3.4 FSD price increases in response to high server costs
In January 2022, Tesla announced that the price of the fully automated driving system FSD in the United States will rise to $12,000 on January 17. Tesla has been maintaining a large cloud computing cluster to process the collected driving information, and we judged that the FSD price increase was to cover the cost of the computing cluster. Tesla hopes to achieve a full-featured FSD by the end of 2022, so we believe that Tesla will increase its investment in computing clusters throughout 2022, and the price of FSD may rise further. After the official launch of the supercomputer Dojo in 2023 and 2024, the price of FSD will drop as Tesla sells more than 2 million vehicles and the scale effect appears.
The growth space of China's autonomous driving enterprises. China has introduced stricter data security management measures, resulting in the inability of Tesla's current domestic collection of data to be transmitted to North America, so domestic FSDs lag several versions behind North America. In the short term, Tesla may not be able to solve the problem of data security, and we believe that Tesla has to abandon the domestic autonomous driving market. In the long run, this will lead to the decoupling of domestic and foreign autonomous driving markets, which leaves sufficient market space for domestic enterprises.
5.4 Car sharing: Let the vehicle make money for the owner by sharing when it is idle
At the Autonomousy Day launch on April 22, 2019, Tesla unveiled robotaxi, a taxi that uses self-driving technology instead of a human driver for driving behavior. We believe that tesla's Robotaxi in the future has two business models, the first is to become an operator, that is, directly produce autonomous vehicles, into trunk logistics and manned taxis. The second provides an operating platform that allows car buyers to provide idle own vehicles to Tesla to operate and share revenue. By 2025, in the Robotaxi market, we expect 110,000 Teslas to be in operation, each with a quarterly mileage of 27,000 miles, assuming that Tesla's net profit per mile will reach $0.9, tesla will earn about $10 billion in profit for the whole year.
5.4.1 Robotaxi is highly disruptive and promises to reshape the competitive landscape in mobility.
Fuel and drivers are the biggest costs for commercial vehicles. According to the U.S. Energy Information Administration (EIA), the cost per mile traveled by u.S. fuel merchant trucks is $2.167, with driver costs and fuel costs accounting for the highest proportions, reaching 36% and 24%, respectively. Pure electric technology replaces fuel, and autonomous driving technology replaces drivers, which will greatly reduce the cost.
Liberate productivity and turn the car from an expense into an asset: the car is currently only used for 5% of its entire life cycle, i.e. 95% of the time the car is just left in a garage or parking lot. Increasing the efficiency of vehicle use through Robotaxi will greatly stimulate the economy. We believe that Robotaxi can maximize this part of the productivity: Tesla owners can join their cars to the Tesla Network to share their Teslas, through the Robotaxi APP to match the car with the riders, so that the car can remain in use when it is idle. And, the final ride-hailing revenue is shared between Tesla and car owners, turning the car, which is currently an expense, into an asset. Tesla estimates that each Robotaxi generates about $30,000 a year in revenue for owners and can carry passengers for 11 consecutive years.
The "card position war" has begun, and various car companies are scrambling to seize the commercial first-mover advantage. Robotaxi can be said to be an innovative exploration of Tesla's business model at a strategic level. As we all know, with the rapid development of automobile intelligence, automatic driving has become a place where soldiers must compete. Tesla has achieved stronger autonomous driving than other companies through self-developed hardware and software and continuous data iteration. However, the road to autonomous driving is long and obstructed, and the road is coming. There will always be some companies that break through technical barriers with human hope. In this context, we believe that Robotaxi in the field of trunk material flow will most likely become the first landing scenario to achieve commercialization, and can also collect massive amounts of big data while conducting commercial exploration, so as to achieve continuous improvement at the technical level, and finally usher in a fundamental change in the business model of the automotive industry in the future.
We believe 2024 is the most likely time point in time for Robotaxi to enter commercialization. Since March 2019, Tesla has deployed FSD on the car and iteratively optimized it, and by March 2022, the V10.11 version has achieved excellent results. Judging from the actual feedback of Users in North America, Tesla has been able to handle various scenarios such as high-speed lane change, overtaking, contraverse meeting, and courtesy to pedestrians, and does not require a driver to take over in more than 3 hours of driving mileage. Although the reliability requirements of the system are completely different from the completely unmanned autonomous driving and the FSD that can be taken over by the driver at any time, we believe that Tesla will start from simple low-speed scenarios in 2024 and gradually expand to slightly more difficult open roads, such as airport logistics, factory logistics, park sanitation and other scenarios to gradually achieve Robotaxi. On the one hand, because these scenarios are relatively simple, there are not so many special scenes that need to be solved, on the other hand, if there is an accident, the division of rights and responsibilities is relatively clear and the losses are controllable. Waymo, another leading self-driving company, will also launch a customized vehicle specifically for North American Robotaxi models in 2024, which cross-validates the conclusion that 2024 will be the first year of commercial operation of autonomous driving from another dimension.
6. Service business: Low-cost strategy to maintain the normal operation of vehicle life cycle services
Over the past five years, Tesla's service business revenue has increased year by year, with the largest increase from 2020 to 2021, an increase of 65% year-on-year. Although the gross profit of the service business has always been negative, the loss has decreased year by year since 2019. Bicycle service revenue continued to decline in the first four years, rebounding in 2021, and bicycle gross profit has decreased significantly from 2020 onwards. Services revenue from 2019 to 2021 was $2.226/23.06/3.802 billion, gross profit was -$5.44/-3.65/- $104 million, bicycle revenue was $1078.73/906.13/1225.01, and bicycle gross profit was $2755.58/-1467.25/-1176.62, respectively.
6.1 Business model change: bypassing middlemen to reach users
Tesla was the first company to maintain a deep connection with its users, reshaping its business model with users. Unlike the traditional car companies and users in the past who generated new car sales and after-sales relations, Tesla has developed a new business model through multi-level exchanges with users through direct stores, charging networks, self-operated insurance institutions and OTA centers.
During the production phase, Tesla has a lot of hardware embedded in it and supports paying for OTA upgrades to vehicles. Tesla, for example, has embedded all the hardware needed for FSD, including chips and sensors. Users can open at the time of purchase, or they can pay for it at any time after purchasing the car. This kind of "hardware embedding, follow-up payment upgrade" approach can reduce production costs through the scale effect, with the continuous strengthening of the vehicle function, users can experience the continuous evolution of the vehicle, while the payment action can also cultivate the user's payment habits.
Tesla operates its own insurance business in North America, which charges users different premiums according to their usage habits. In the traditional model, car insurance is provided by insurance companies, and car companies are only the sales agencies of insurance companies. In terms of energy replenishment, Tesla has built a large number of charging piles to form a huge charging network. Unlike the third-party charging pile in the society, which is unmanaged and has a high failure rate, Tesla's self-operated charging pile has high reliability, fast charging speed and each user has a free charging quota, which can improve user stickiness through this way Tesla. Services are Tesla's next source of revenue growth. Tesla services are divided into insurance services, charging services, connected vehicle services and boutique services.
6.2 UBI Insurance: Promote a virtuous cycle of driving safety
Tesla launched its self-developed insurance business in North America. However, because the laws vary from state to state, the scope of insurance contracts is also different. In addition to California, Tesla launched a UBI insurance, Tesla will count the driving behavior of each driver's driving behavior score, according to the driver's driving style of radical degree, charging the driver different premiums. But in areas where regulations limit it, Tesla is introducing standardized insurance services where each person's premiums don't vary because of different driving patterns. According to the plan, Tesla will complete the national coverage of self-developed insurance business in the United States by the end of 2022, and will carry out Nordic self-developed insurance business in 2023.
Linking the user's driving habits with the premium can achieve a positive cycle of more cautious driving, the less likely it is to take risks, and the lower the future premium. Therefore, even if some users have aggressive driving styles and higher premiums, most users can enjoy the price advantage of the UBI model. According to statistics, Tesla's average premium is significantly lower than that of other insurance institutions. We predict that Tesla can acquire the vast majority of users through cheap insurance, but the profit margin of the insurance business is not high, and we predict that the future single-car premium income will be $1800 / year, and the gross margin of the insurance business will be 3%.
6.3 Charging network services: build a convenient energy replenishment system
Tesla offers three types of charging services, namely home charging, overcharge and destination charging station. Among them supercharge and purposeful charging stations are offered to join.
6.3.1 Home Charging
Home charging is the most common charging method used by car owners. Tesla offers a simultaneous installation of wall-mounted home charging piles when users buy electric vehicles, which cost 8,000 yuan in China. Last year, Tesla has officially released the third generation of home piles, the connector weighs only 5.5 kg, can provide 32A/7kW for all models under single-phase 220V; the three-phase 380V Model 3/Model Y models provide 16A/11Kw, and the Model S/Model X models provide 24A/16kW charging services, which also support Wi-Fi networking and OTA remote firmware upgrades. (Source: Future Think Tank)
6.3.2 Overcharge
Tesla currently operates more than 3,000 charging stations and more than 30,000 charging piles around the world. Charging stations are often located near restaurants, shopping malls, WiFi hotspots, etc. Tesla will come with a certain free charging quota with the car owner when buying the car, and the user's charging on the supercharge station will be recorded to the charging account where the vehicle is located. After 2016, for vehicles that have completed charging but have not driven away in time when the number of vacant parking spaces at the charging station is less than 50%, Tesla will charge them 2.6 yuan / minute (in China) and the "super charging station resource occupation fee" is not capped.
Tesla announced the launch of a higher-capacity "Megacharger" in November 2017, and also launched a prototype of the Semi semi-trailer truck, saying that in the future, Semi will use Tesla Megacharger for charging. The solar-assisted Megacharger will charge the truck 80 percent in 30 minutes, the Total Capacity of the Semi battery pack is 500 miles (about 805 kilometers), or 30 minutes, and the Semi can last 400 miles (about 644 kilometers). To achieve this, megacharger may have an output level of more than 1000kW.
6.3.3 Destination Charging Stations
Tesla also specially for restaurants, shopping malls, hotels, resorts and other venues that can stay for several hours or days equipped with charging piles to meet the charging needs of users. Most of these charging piles are free to charge Tesla users, and in return Tesla will mark these charging locations on the map to increase traffic in the area.
6.4 Internet of Vehicle Subscription Services: Paving the way for "soft revenue"
In order to enhance the in-car entertainment experience, Tesla offers Car Networking packages in most regions, and Car Networking subscriptions include navigation data, satellite maps, streaming video playback, in-car karaoke and other services. $9.99 per month. At the end of February this year, Tesla upgraded the subscription duration of the Internet of Vehicles service from a monthly subscription to an annual subscription with an annual price of $99. Tesla is developing an in-car app store where users buy apps on Tesla and pay app developers and Tesla. In addition to this, Tesla also introduced gamepads, and some of the games hidden after the introduction of the gamepads are only open to gamepad players.
6.5 Service Revenue Projections
We believe that Tesla's service business revenue sources include insurance business, car networking business, and user assembly or upgrade of FSD software. The charging business is mainly used to improve user stickiness, and the profit is not obvious. In the insurance business, Tesla implements a low-price strategy, so we predict that the gross profit margin of the insurance business will remain low. Connected vehicle business companies need to pay mobile operators traffic fees, which continue to increase as vehicle functions increase, and it is expected that revenue and cost will remain balanced in the future. In terms of FSD software sales, as the number of existing cars increases, the data used for training will also increase, and the number of FSD training clusters that Tesla needs to build in the background will also increase. As of August 2021, Tesla has more than 12,000 neural networks for training, and due to excessive compute cluster expenses, Tesla has to make multiple price adjustments to FSD to ensure breakeven. Therefore, we believe that Tesla service revenue will increase as Tesla model ownership increases, but gross profit remains low.
Master Plan 3.0: The Future of Science Fiction
In March 2022, Musk announced that he was working on his great plan, 3.0. Great Plan 1.0 was proposed in 2006 and realized in 2016, after 10 years. Great Plan 2.0 was introduced in 2016, and we expect great Plan 2.0 to be fully completed as Robotaxi is implemented at scale in 2025. So the essence of Tesla's great planning is Tesla's own 10-year plan for development. And the great plan 3.0 is actually Tesla's vision of a better life after 2030. We may wish to combine Tesla's past achievements to make some guesses about the future direction of the company.
7.1 Humanoid Robots: Solving productivity shortages
We think Tesla's next-generation grand plan will include the mass production of Optimus, a humanoid robot that Tesla will unveil on AI Day in August 2021.
The overall design is consistent with humans, sharing the Tesla Vehicle AI system O
Made of smooth, lightweight material, the ptimus is 5 feet 8 inches tall, about 172cm, weighs 125 pounds, about 56.7kg, has a load capacity of 45 pounds, about 20kg, and has a top walking speed of 5 mph, about 8km/h. Its head has a variety of sensors for obtaining information, and its face has a display for displaying user information. In terms of appearance, like humans, the robot has hands and legs, limbs using 40 electromechanical pushers, and power feedback sensors in both legs for balance and agility. What is more noteworthy is that the robot is a humanoid robot that shares tesla's car AI system, and Some of Tesla's technologies in artificial intelligence and automatic driving will be added to this humanoid robot. For example, the robot chest cavity is a Tesla FSD chip; the automatic driving camera will basically act as the robot's vision system; the fully autonomous driving computer will serve as the robot's internal organ to help it; in addition, the deep neural network architecture of the multi-camera vision architecture, including planning, automatic annotation, simulation, and Dojo training, will also be used to develop robots to help robots perform tasks effectively and efficiently.
7.2 General Artificial Intelligence (AGI): The kernel of robots and FSDs
General Artificial Intelligence (AGI) is an artificial intelligence that has the same intelligence as humans, or surpasses humans, and can perform all the intelligent behaviors that normal humans have. General AI is different from weak AI. Weak AI only deals with specific problems, it does not need to have complete human cognitive ability, or even no sensory cognitive ability that humans have, as long as it is designed to look intelligent. In other words, weak AI accomplishes specific tasks, while strong human intelligence can pass the Turing test to accomplish tasks that the vast majority of humans can accomplish. Tesla has the strongest AI team, and the development of general AI and the development of Optimus are two things that complement each other. We predict that developing turing-tested artificial intelligence will be one of the tasks Tesla will solve in Grand Plan 3.0. On the timeline, we believe that with the deployment of Dojo two years later, in 2025, Tesra General Artificial Intelligence will be successful, and will be the first to be introduced to users in Tesla vehicles, and then installed on the robot Optimus.
7.3 Promote underground tunnels: solve road traffic problems
Musk founded the tunnel construction company "the boring company", the company's mission is to solve the problem of traffic congestion within the city by digging tunnels under the city.
7.4 Starlink Program: Autonomous Driving from the Sky
SpaceX StarLink is a low-Earth orbit satellite cluster launched by Musk's space services company SpaceX to provide bandwidth Internet connectivity to underserved areas of Earth and affordable services for urban areas. The project is divided into two phases, the first phase is to launch about 4500 satellites in 1500km and 500km orbits, and the second phase is to launch about 7500 satellites in orbits around 350km. According to the schedule announced by the Starlink program, SpaceX will complete the launch of half of its satellites by March 2024 and all satellites by 2027. Therefore, we judge that in the next 10 years, Tesla's Robotaxi business will use starlink's Starlink Internet access capabilities to further expand the range of Robotaxi available.
8. Profit forecast
Tesla's current business segments are divided into four categories: automotive business, energy business, autonomous driving business and service industry.
Automotive: Maintain a long-term gross margin target of 30% to hit the 5.5 million unit sales target by 2025. Tesla will maintain the product matrix of Model S, 3, X, and Y in 2022. At the same time, in the first quarter of 2022, the new plants in Berlin and Texas and the expansion of the Shanghai plant will be completed, and the expansion of the California plant will be completed in the middle of the year. We expect that in 2022, Tesla will produce and sell 1.52 million new cars, of which Model 3 and Model Y are the main sales models, accounting for more than 40% and 56% of sales respectively, and Model S and Model X are positioned at the high end, and the total sales will account for about 6%.
In 2023, Tesla's main sales models will add Cybertruck, Semi and second-generation Roadster. The Roadster's position is a high-end supercar, and we predict that less than 700 units will be sold in 2023. Tesla has already piloted cybertrucks at its Texas facility in the first quarter of 2022. As the 4680 battery yield gradually improves, we expect Tesla to mass-produce the Cybertruck in 2023, producing more than 58,000 vehicles in the whole year and nearly 2.18 million new vehicles in 2023. The Semi is also a model to focus on in 2023 because it is equipped with the mission of realizing autonomous driving in trunk logistics.
In 2024, Tesla's product matrix will add the $25,000 model Model 2/Model Q and the van Tesla Van, which will carry Tesla's mission to serve a wider range of people, and we expect cumulative sales of these new models to exceed 2.7 million units between 2024 and 2025. Tesla's advanced manufacturing capabilities are the core competitiveness of the current stage, Tesla has a total of 4 super factories in the world, actively develop local suppliers in the super factory, according to the specific conditions of parts and components using multi-track supply or exclusive self-production management mode, in the chip, battery and other parts supply shortage still ensures the stability of the supply chain, sales continue to rise. We expect Tesla production and sales to reach 152/218/381/5.39 million units from 2022 to 2025, respectively.
In terms of gross margin changes, Tesla's automotive business had a gross margin of 30% in Q1 2022 and has two new plants in 2022: the Texas Gigafactory and the Berlin Gigafactory. From Tesla's past experience in putting into operation a new factory, the gross profit margin will decline briefly for each super factory, so we believe that after the two factories are put into operation in the first half of 2022, the gross profit margin of the automobile business may be slightly reduced. However, Tesla implements an effective cost control strategy 1) supply chain integration, 2) supply chain localization, 3) multi-track supply strategy, and with the use of 4680 batteries and integrated die casting, the gross profit margin will rebound to Tesla's long-term gross margin target of 30%. Tesla's self-produced 4680 battery can increase the battery life by 16% compared to the previous 2170 battery, and increase the power by 6 times. Tesla's long-term goal for 2030 is to reach 3TWh of installed capacity. We believe that this goal is risky, we believe that Tesla's self-produced battery core installed capacity will reach 20Gwh in 2022, and 300GWh of self-produced battery cells in 2025. The profitability of this part of the business has not yet been reflected.
Energy business: cost-free, business processes to be connected. We forecast that the capacity of our Nevada, New York and Lathrop plants will reach 40GWh in 2023 and 100GWh in 2025, respectively. However, Tesla's two current gigafactories responsible for energy business are in North America, and from the perspective of the industrial chain, if Tesla insists on only producing photovoltaic and energy storage equipment in North America, the business will maintain a low single-digit gross margin. There are three reasons: 1) There is no advantage in the cost of photovoltaic products in the United States. According to the cost of electricity, the average cost in China is lower than the average cost in the United States, and it is necessary to achieve photovoltaic parity to look forward to the explosive development of the industry. 2) The energy storage business relies on government investment, and most of the customers are governments. 3) The two-party system affects the stable development of the new energy industry in the United States. In 2017, the Republican Party took office to accelerate the development of fossil fuels, abolished Obama-era incentives for new energy sources, repealed the Climate Action Plan and the Clean Water Act, and formally withdrew from the Paris Climate Agreement to control greenhouse gas emissions in November 2020. Democratic Biden returned to the Paris Agreement in February 2021, and on the morning of November 19, 2021, the House passed Biden's $1.75 trillion Build Back Better Act. Although the bill passed smoothly, all Republican lawmakers voted against the voting process, and the voting results were extremely sticky. We believe that the two-party energy transition in the United States will not be resolute enough, and there is a possibility of iteration, which will also put a question mark on the profitability prospects of Tesra's photovoltaic and energy storage businesses.
Autonomous driving business: Taking the lead in iterating from data and models has the advantage of being a first mover. Tesla's fully automated driver assistance (FSD) with a pure vision solution opened for testing in 2019, with a first-mover advantage in data accumulation and data usage, with a total of 7 major version upgrades by April 2022, and driving ability, safety and comfort have made great progress. In time, we expect the full FSD feature to be rolled out in North America by the end of 2022 and the beginning of 2023. We believe that FSD software penetration will gradually increase as FSD software continues to improve, reaching a weighted penetration rate of 33% by 2025. Considering that the FSD has grown from $8,000 in 2019 to $12,000, the possibility of unit prices continuing to rise is not ruled out. We believe FSD autonomous driving revenue will reach $41/108/195/248 billion from 2022 to 2025. Tesla in china's autopilot market will gradually withdraw due to data security problems, which will leave a broad market space for Chinese autopilot companies.
In the follow-up 2024, Tesla will start the commercial operation of autonomous driving, and the field of trunk logistics will be the first breakthrough. We believe that the Semi, a pure electric truck equipped with FSD, will be the first model to be used in the trunk logistics field, and the $25,000 model that the company is planning will be the first model to realize autonomous operation in the passenger car field. The commercial operation of autonomous driving will greatly reduce logistics costs, increase the rate of material circulation in the whole society, detonate the revolution in the transportation industry, and promote Tesla's transformation from manufacturing to service.
Service business: low profit strategy, maintain stable operation. Tesla's revenue will come mainly from the insurance business. We expect the insurance business to generate $1,800 in premiums per year, and premium margins will remain low due to Tesla's low-cost strategy of maintaining premiums. While other service businesses, such as maintenance, internet of vehicles, etc. are not Tesla's main source of profit, we believe that strategically Tesla will maintain the service business gross margin at a low single digit.
(This article is for informational purposes only and does not represent any of our investment advice.) For usage information, see the original report. ）
Featured report source: [Future Think Tank].